diff options
Diffstat (limited to 'generic/tclUtil.c')
| -rw-r--r-- | generic/tclUtil.c | 3817 |
1 files changed, 2697 insertions, 1120 deletions
diff --git a/generic/tclUtil.c b/generic/tclUtil.c index b327b99..866b6ae 100644 --- a/generic/tclUtil.c +++ b/generic/tclUtil.c @@ -1,118 +1,466 @@ -/* +/* * tclUtil.c -- * - * This file contains utility procedures that are used by many Tcl + * This file contains utility functions that are used by many Tcl * commands. * * Copyright (c) 1987-1993 The Regents of the University of California. * Copyright (c) 1994-1998 Sun Microsystems, Inc. - * Copyright (c) 2001 by Kevin B. Kenny. All rights reserved. + * Copyright (c) 2001 by Kevin B. Kenny. All rights reserved. * - * See the file "license.terms" for information on usage and redistribution - * of this file, and for a DISCLAIMER OF ALL WARRANTIES. + * See the file "license.terms" for information on usage and redistribution of + * this file, and for a DISCLAIMER OF ALL WARRANTIES. */ #include "tclInt.h" -#include "tclPort.h" +#include <float.h> +#include <math.h> /* - * The following variable holds the full path name of the binary - * from which this application was executed, or NULL if it isn't - * know. The value of the variable is set by the procedure - * Tcl_FindExecutable. The storage space is dynamically allocated. + * The absolute pathname of the executable in which this Tcl library is + * running. */ -char *tclExecutableName = NULL; -char *tclNativeExecutableName = NULL; +static ProcessGlobalValue executableName = { + 0, 0, NULL, NULL, NULL, NULL, NULL +}; /* - * The following values are used in the flags returned by Tcl_ScanElement - * and used by Tcl_ConvertElement. The value TCL_DONT_USE_BRACES is also - * defined in tcl.h; make sure its value doesn't overlap with any of the - * values below. - * - * TCL_DONT_USE_BRACES - 1 means the string mustn't be enclosed in - * braces (e.g. it contains unmatched braces, - * or ends in a backslash character, or user - * just doesn't want braces); handle all - * special characters by adding backslashes. - * USE_BRACES - 1 means the string contains a special - * character that can be handled simply by - * enclosing the entire argument in braces. - * BRACES_UNMATCHED - 1 means that braces aren't properly matched - * in the argument. + * The following values are used in the flags arguments of Tcl*Scan*Element and + * Tcl*Convert*Element. The values TCL_DONT_USE_BRACES and TCL_DONT_QUOTE_HASH + * are defined in tcl.h, like so: + * +#define TCL_DONT_USE_BRACES 1 +#define TCL_DONT_QUOTE_HASH 8 + * + * Those are public flag bits which callers of the public routines + * Tcl_Convert*Element() can use to indicate: + * + * TCL_DONT_USE_BRACES - 1 means the caller is insisting that brace + * quoting not be used when converting the list + * element. + * TCL_DONT_QUOTE_HASH - 1 means the caller insists that a leading hash + * character ('#') should *not* be quoted. This + * is appropriate when the caller can guarantee + * the element is not the first element of a + * list, so [eval] cannot mis-parse the element + * as a comment. + * + * The remaining values which can be carried by the flags of these routines + * are for internal use only. Make sure they do not overlap with the public + * values above. + * + * The Tcl*Scan*Element() routines make a determination which of 4 modes of + * conversion is most appropriate for Tcl*Convert*Element() to perform, and + * sets two bits of the flags value to indicate the mode selected. + * + * CONVERT_NONE The element needs no quoting. Its literal string + * is suitable as is. + * CONVERT_BRACE The conversion should be enclosing the literal string + * in braces. + * CONVERT_ESCAPE The conversion should be using backslashes to escape + * any characters in the string that require it. + * CONVERT_MASK A mask value used to extract the conversion mode from + * the flags argument. + * Also indicates a strange conversion mode where all + * special characters are escaped with backslashes + * *except for braces*. This is a strange and unnecessary + * case, but it's part of the historical way in which + * lists have been formatted in Tcl. To experiment with + * removing this case, set the value of COMPAT to 0. + * + * One last flag value is used only by callers of TclScanElement(). The flag + * value produced by a call to Tcl*Scan*Element() will never leave this bit + * set. + * + * CONVERT_ANY The caller of TclScanElement() declares it can make + * no promise about what public flags will be passed to + * the matching call of TclConvertElement(). As such, + * TclScanElement() has to determine the worst case + * destination buffer length over all possibilities, and + * in other cases this means an overestimate of the + * required size. + * + * For more details, see the comments on the Tcl*Scan*Element and + * Tcl*Convert*Element routines. */ -#define USE_BRACES 2 -#define BRACES_UNMATCHED 4 +#define COMPAT 1 +#define CONVERT_NONE 0 +#define CONVERT_BRACE 2 +#define CONVERT_ESCAPE 4 +#define CONVERT_MASK (CONVERT_BRACE | CONVERT_ESCAPE) +#define CONVERT_ANY 16 /* - * The following values determine the precision used when converting - * floating-point values to strings. This information is linked to all - * of the tcl_precision variables in all interpreters via the procedure - * TclPrecTraceProc. + * The following key is used by Tcl_PrintDouble and TclPrecTraceProc to + * access the precision to be used for double formatting. */ -static char precisionString[10] = "12"; - /* The string value of all the tcl_precision - * variables. */ -static char precisionFormat[10] = "%.12g"; - /* The format string actually used in calls - * to sprintf. */ -TCL_DECLARE_MUTEX(precisionMutex) +static Tcl_ThreadDataKey precisionKey; /* - * Prototypes for procedures defined later in this file. + * Prototypes for functions defined later in this file. */ -static void UpdateStringOfEndOffset _ANSI_ARGS_((Tcl_Obj* objPtr)); -static int SetEndOffsetFromAny _ANSI_ARGS_((Tcl_Interp* interp, - Tcl_Obj* objPtr)); +static void ClearHash(Tcl_HashTable *tablePtr); +static void FreeProcessGlobalValue(ClientData clientData); +static void FreeThreadHash(ClientData clientData); +static Tcl_HashTable * GetThreadHash(Tcl_ThreadDataKey *keyPtr); +static int SetEndOffsetFromAny(Tcl_Interp* interp, + Tcl_Obj* objPtr); +static void UpdateStringOfEndOffset(Tcl_Obj* objPtr); /* - * The following is the Tcl object type definition for an object - * that represents a list index in the form, "end-offset". It is - * used as a performance optimization in TclGetIntForIndex. The - * internal rep is an integer, so no memory management is required - * for it. + * The following is the Tcl object type definition for an object that + * represents a list index in the form, "end-offset". It is used as a + * performance optimization in TclGetIntForIndex. The internal rep is an + * integer, so no memory management is required for it. */ Tcl_ObjType tclEndOffsetType = { "end-offset", /* name */ - (Tcl_FreeInternalRepProc*) NULL, /* freeIntRepProc */ - (Tcl_DupInternalRepProc*) NULL, /* dupIntRepProc */ + NULL, /* freeIntRepProc */ + NULL, /* dupIntRepProc */ UpdateStringOfEndOffset, /* updateStringProc */ - SetEndOffsetFromAny + SetEndOffsetFromAny }; + +/* + * * STRING REPRESENTATION OF LISTS * * * + * + * The next several routines implement the conversions of strings to and + * from Tcl lists. To understand their operation, the rules of parsing + * and generating the string representation of lists must be known. Here + * we describe them in one place. + * + * A list is made up of zero or more elements. Any string is a list if + * it is made up of alternating substrings of element-separating ASCII + * whitespace and properly formatted elements. + * + * The ASCII characters which can make up the whitespace between list + * elements are: + * + * \u0009 \t TAB + * \u000A \n NEWLINE + * \u000B \v VERTICAL TAB + * \u000C \f FORM FEED + * \u000D \r CARRIAGE RETURN + * \u0020 SPACE + * + * NOTE: differences between this and other places where Tcl defines a role + * for "whitespace". + * + * * Unlike command parsing, here NEWLINE is just another whitespace + * character; its role as a command terminator in a script has no + * importance here. + * + * * Unlike command parsing, the BACKSLASH NEWLINE sequence is not + * considered to be a whitespace character. + * + * * Other Unicode whitespace characters (recognized by + * [string is space] or Tcl_UniCharIsSpace()) do not play any role + * as element separators in Tcl lists. + * + * * The NUL byte ought not appear, as it is not in strings properly + * encoded for Tcl, but if it is present, it is not treated as + * separating whitespace, or a string terminator. It is just + * another character in a list element. + * + * The interpretaton of a formatted substring as a list element follows + * rules similar to the parsing of the words of a command in a Tcl script. + * Backslash substitution plays a key role, and is defined exactly as it is + * in command parsing. The same routine, TclParseBackslash() is used in both + * command parsing and list parsing. + * + * NOTE: This means that if and when backslash substitution rules ever + * change for command parsing, the interpretation of strings as lists also + * changes. + * + * Backslash substitution replaces an "escape sequence" of one or more + * characters starting with + * \u005c \ BACKSLASH + * with a single character. The one character escape sequent case happens + * only when BACKSLASH is the last character in the string. In all other + * cases, the escape sequence is at least two characters long. + * + * The formatted substrings are interpreted as element values according to + * the following cases: + * + * * If the first character of a formatted substring is + * \u007b { OPEN BRACE + * then the end of the substring is the matching + * \u007d } CLOSE BRACE + * character, where matching is determined by counting nesting levels, + * and not including any brace characters that are contained within a + * backslash escape sequence in the nesting count. Having found the + * matching brace, all characters between the braces are the string + * value of the element. If no matching close brace is found before the + * end of the string, the string is not a Tcl list. If the character + * following the close brace is not an element separating whitespace + * character, or the end of the string, then the string is not a Tcl list. + * + * NOTE: this differs from a brace-quoted word in the parsing of a + * Tcl command only in its treatment of the backslash-newline sequence. + * In a list element, the literal characters in the backslash-newline + * sequence become part of the element value. In a script word, + * conversion to a single SPACE character is done. + * + * NOTE: Most list element values can be represented by a formatted + * substring using brace quoting. The exceptions are any element value + * that includes an unbalanced brace not in a backslash escape sequence, + * and any value that ends with a backslash not itself in a backslash + * escape sequence. + * + * * If the first character of a formatted substring is + * \u0022 " QUOTE + * then the end of the substring is the next QUOTE character, not counting + * any QUOTE characters that are contained within a backslash escape + * sequence. If no next QUOTE is found before the end of the string, the + * string is not a Tcl list. If the character following the closing QUOTE + * is not an element separating whitespace character, or the end of the + * string, then the string is not a Tcl list. Having found the limits + * of the substring, the element value is produced by performing backslash + * substitution on the character sequence between the open and close QUOTEs. + * + * NOTE: Any element value can be represented by this style of formatting, + * given suitable choice of backslash escape sequences. + * + * * All other formatted substrings are terminated by the next element + * separating whitespace character in the string. Having found the limits + * of the substring, the element value is produced by performing backslash + * substitution on it. + * + * NOTE: Any element value can be represented by this style of formatting, + * given suitable choice of backslash escape sequences, with one exception. + * The empty string cannot be represented as a list element without the use + * of either braces or quotes to delimit it. + * + * This collection of parsing rules is implemented in the routine + * TclFindElement(). + * + * In order to produce lists that can be parsed by these rules, we need + * the ability to distinguish between characters that are part of a list + * element value from characters providing syntax that define the structure + * of the list. This means that our code that generates lists must at a + * minimum be able to produce escape sequences for the 10 characters + * identified above that have significance to a list parser. + * + * * * CANONICAL LISTS * * * * * + * + * In addition to the basic rules for parsing strings into Tcl lists, there + * are additional properties to be met by the set of list values that are + * generated by Tcl. Such list values are often said to be in "canonical + * form": + * + * * When any canonical list is evaluated as a Tcl script, it is a script + * of either zero commands (an empty list) or exactly one command. The + * command word is exactly the first element of the list, and each argument + * word is exactly one of the following elements of the list. This means + * that any characters that have special meaning during script evaluation + * need special treatment when canonical lists are produced: + * + * * Whitespace between elements may not include NEWLINE. + * * The command terminating character, + * \u003b ; SEMICOLON + * must be BRACEd, QUOTEd, or escaped so that it does not terminate + * the command prematurely. + * * Any of the characters that begin substitutions in scripts, + * \u0024 $ DOLLAR + * \u005b [ OPEN BRACKET + * \u005c \ BACKSLASH + * need to be BRACEd or escaped. + * * In any list where the first character of the first element is + * \u0023 # HASH + * that HASH character must be BRACEd, QUOTEd, or escaped so that it + * does not convert the command into a comment. + * * Any list element that contains the character sequence + * BACKSLASH NEWLINE cannot be formatted with BRACEs. The + * BACKSLASH character must be represented by an escape + * sequence, and unless QUOTEs are used, the NEWLINE must + * be as well. + * + * * It is also guaranteed that one can use a canonical list as a building + * block of a larger script within command substitution, as in this example: + * set script "puts \[[list $cmd $arg]]"; eval $script + * To support this usage, any appearance of the character + * \u005d ] CLOSE BRACKET + * in a list element must be BRACEd, QUOTEd, or escaped. + * + * * Finally it is guaranteed that enclosing a canonical list in braces + * produces a new value that is also a canonical list. This new list has + * length 1, and its only element is the original canonical list. This + * same guarantee also makes it possible to construct scripts where an + * argument word is given a list value by enclosing the canonical form + * of that list in braces: + * set script "puts {[list $one $two $three]}"; eval $script + * This sort of coding was once fairly common, though it's become more + * idiomatic to see the following instead: + * set script [list puts [list $one $two $three]]; eval $script + * In order to support this guarantee, every canonical list must have + * balance when counting those braces that are not in escape sequences. + * + * Within these constraints, the canonical list generation routines + * TclScanElement() and TclConvertElement() attempt to generate the string + * for any list that is easiest to read. When an element value is itself + * acceptable as the formatted substring, it is usually used (CONVERT_NONE). + * When some quoting or escaping is required, use of BRACEs (CONVERT_BRACE) + * is usually preferred over the use of escape sequences (CONVERT_ESCAPE). + * There are some exceptions to both of these preferences for reasons of + * code simplicity, efficiency, and continuation of historical habits. + * Canonical lists never use the QUOTE formatting to delimit their elements + * because that form of quoting does not nest, which makes construction of + * nested lists far too much trouble. Canonical lists always use only a + * single SPACE character for element-separating whitespace. + * + * * * FUTURE CONSIDERATIONS * * * + * + * When a list element requires quoting or escaping due to a CLOSE BRACKET + * character or an internal QUOTE character, a strange formatting mode is + * recommended. For example, if the value "a{b]c}d" is converted by the + * usual modes: + * + * CONVERT_BRACE: a{b]c}d => {a{b]c}d} + * CONVERT_ESCAPE: a{b]c}d => a\{b\]c\}d + * + * we get perfectly usable formatted list elements. However, this is not + * what Tcl releases have been producing. Instead, we have: + * + * CONVERT_MASK: a{b]c}d => a{b\]c}d + * + * where the CLOSE BRACKET is escaped, but the BRACEs are not. The same + * effect can be seen replacing ] with " in this example. There does not + * appear to be any functional or aesthetic purpose for this strange + * additional mode. The sole purpose I can see for preserving it is to + * keep generating the same formatted lists programmers have become accustomed + * to, and perhaps written tests to expect. That is, compatibility only. + * The additional code complexity required to support this mode is significant. + * The lines of code supporting it are delimited in the routines below with + * #if COMPAT directives. This makes it easy to experiment with eliminating + * this formatting mode simply with "#define COMPAT 0" above. I believe + * this is worth considering. + * + * Another consideration is the treatment of QUOTE characters in list elements. + * TclConvertElement() must have the ability to produce the escape sequence + * \" so that when a list element begins with a QUOTE we do not confuse + * that first character with a QUOTE used as list syntax to define list + * structure. However, that is the only place where QUOTE characters need + * quoting. In this way, handling QUOTE could really be much more like + * the way we handle HASH which also needs quoting and escaping only in + * particular situations. Following up this could increase the set of + * list elements that can use the CONVERT_NONE formatting mode. + * + * More speculative is that the demands of canonical list form require brace + * balance for the list as a whole, while the current implementation achieves + * this by establishing brace balance for every element. + * + * Finally, a reminder that the rules for parsing and formatting lists are + * closely tied together with the rules for parsing and evaluating scripts, + * and will need to evolve in sync. + */ + +/* + *---------------------------------------------------------------------- + * + * TclMaxListLength -- + * + * Given 'bytes' pointing to 'numBytes' bytes, scan through them and + * count the number of whitespace runs that could be list element + * separators. If 'numBytes' is -1, scan to the terminating '\0'. + * Not a full list parser. Typically used to get a quick and dirty + * overestimate of length size in order to allocate space for an + * actual list parser to operate with. + * + * Results: + * Returns the largest number of list elements that could possibly + * be in this string, interpreted as a Tcl list. If 'endPtr' is not + * NULL, writes a pointer to the end of the string scanned there. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +int +TclMaxListLength( + CONST char *bytes, + int numBytes, + CONST char **endPtr) +{ + int count = 0; + + if ((numBytes == 0) || ((numBytes == -1) && (*bytes == '\0'))) { + /* Empty string case - quick exit */ + goto done; + } + /* No list element before leading white space */ + count += 1 - TclIsSpaceProc(*bytes); + + /* Count white space runs as potential element separators */ + while (numBytes) { + if ((numBytes == -1) && (*bytes == '\0')) { + break; + } + if (TclIsSpaceProc(*bytes)) { + /* Space run started; bump count */ + count++; + do { + bytes++; + numBytes -= (numBytes != -1); + } while (numBytes && TclIsSpaceProc(*bytes)); + if ((numBytes == 0) || ((numBytes == -1) && (*bytes == '\0'))) { + break; + } + /* (*bytes) is non-space; return to counting state */ + } + bytes++; + numBytes -= (numBytes != -1); + } + + /* No list element following trailing white space */ + count -= TclIsSpaceProc(bytes[-1]); + + done: + if (endPtr) { + *endPtr = bytes; + } + return count; +} /* *---------------------------------------------------------------------- * * TclFindElement -- * - * Given a pointer into a Tcl list, locate the first (or next) - * element in the list. + * Given a pointer into a Tcl list, locate the first (or next) element in + * the list. * * Results: - * The return value is normally TCL_OK, which means that the - * element was successfully located. If TCL_ERROR is returned - * it means that list didn't have proper list structure; - * the interp's result contains a more detailed error message. + * The return value is normally TCL_OK, which means that the element was + * successfully located. If TCL_ERROR is returned it means that list + * didn't have proper list structure; the interp's result contains a more + * detailed error message. * * If TCL_OK is returned, then *elementPtr will be set to point to the * first element of list, and *nextPtr will be set to point to the * character just after any white space following the last character - * that's part of the element. If this is the last argument in the - * list, then *nextPtr will point just after the last character in the - * list (i.e., at the character at list+listLength). If sizePtr is - * non-NULL, *sizePtr is filled in with the number of characters in the - * element. If the element is in braces, then *elementPtr will point - * to the character after the opening brace and *sizePtr will not - * include either of the braces. If there isn't an element in the list, - * *sizePtr will be zero, and both *elementPtr and *termPtr will point - * just after the last character in the list. Note: this procedure does - * NOT collapse backslash sequences. + * that's part of the element. If this is the last argument in the list, + * then *nextPtr will point just after the last character in the list + * (i.e., at the character at list+listLength). If sizePtr is non-NULL, + * *sizePtr is filled in with the number of bytes in the element. If + * the element is in braces, then *elementPtr will point to the character + * after the opening brace and *sizePtr will not include either of the + * braces. If there isn't an element in the list, *sizePtr will be zero, + * and both *elementPtr and *nextPtr will point just after the last + * character in the list. If literalPtr is non-NULL, *literalPtr is set + * to a boolean value indicating whether the substring returned as + * the values of **elementPtr and *sizePtr is the literal value of + * a list element. If not, a call to TclCopyAndCollapse() is needed + * to produce the actual value of the list element. Note: this function + * does NOT collapse backslash sequences, but uses *literalPtr to tell + * callers when it is required for them to do so. * * Side effects: * None. @@ -121,25 +469,27 @@ Tcl_ObjType tclEndOffsetType = { */ int -TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, - bracePtr) - Tcl_Interp *interp; /* Interpreter to use for error reporting. - * If NULL, then no error message is left - * after errors. */ - CONST char *list; /* Points to the first byte of a string +TclFindElement( + Tcl_Interp *interp, /* Interpreter to use for error reporting. If + * NULL, then no error message is left after + * errors. */ + CONST char *list, /* Points to the first byte of a string * containing a Tcl list with zero or more * elements (possibly in braces). */ - int listLength; /* Number of bytes in the list's string. */ - CONST char **elementPtr; /* Where to put address of first significant + int listLength, /* Number of bytes in the list's string. */ + CONST char **elementPtr, /* Where to put address of first significant * character in first element of list. */ - CONST char **nextPtr; /* Fill in with location of character just + CONST char **nextPtr, /* Fill in with location of character just * after all white space following end of * argument (next arg or end of list). */ - int *sizePtr; /* If non-zero, fill in with size of + int *sizePtr, /* If non-zero, fill in with size of * element. */ - int *bracePtr; /* If non-zero, fill in with non-zero/zero - * to indicate that arg was/wasn't - * in braces. */ + int *literalPtr) /* If non-zero, fill in with non-zero/zero to + * indicate that the substring of *sizePtr + * bytes starting at **elementPtr is/is not + * the literal list element and therefore + * does not/does require a call to + * TclCopyAndCollapse() by the caller. */ { CONST char *p = list; CONST char *elemStart; /* Points to first byte of first element. */ @@ -148,16 +498,17 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, int inQuotes = 0; int size = 0; /* lint. */ int numChars; + int literal = 1; CONST char *p2; - + /* - * Skim off leading white space and check for an opening brace or - * quote. We treat embedded NULLs in the list as bytes belonging to - * a list element. + * Skim off leading white space and check for an opening brace or quote. + * We treat embedded NULLs in the list as bytes belonging to a list + * element. */ limit = (list + listLength); - while ((p < limit) && (isspace(UCHAR(*p)))) { /* INTL: ISO space. */ + while ((p < limit) && (TclIsSpaceProc(*p))) { p++; } if (p == limit) { /* no element found */ @@ -173,9 +524,6 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, p++; } elemStart = p; - if (bracePtr != 0) { - *bracePtr = openBraces; - } /* * Find element's end (a space, close brace, or the end of the string). @@ -183,123 +531,119 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, while (p < limit) { switch (*p) { - /* * Open brace: don't treat specially unless the element is in * braces. In this case, keep a nesting count. */ - case '{': - if (openBraces != 0) { - openBraces++; - } - break; + case '{': + if (openBraces != 0) { + openBraces++; + } + break; /* * Close brace: if element is in braces, keep nesting count and * quit when the last close brace is seen. */ - case '}': - if (openBraces > 1) { - openBraces--; - } else if (openBraces == 1) { - size = (p - elemStart); - p++; - if ((p >= limit) - || isspace(UCHAR(*p))) { /* INTL: ISO space. */ - goto done; - } + case '}': + if (openBraces > 1) { + openBraces--; + } else if (openBraces == 1) { + size = (p - elemStart); + p++; + if ((p >= limit) || TclIsSpaceProc(*p)) { + goto done; + } - /* - * Garbage after the closing brace; return an error. - */ - - if (interp != NULL) { - char buf[100]; - - p2 = p; - while ((p2 < limit) - && (!isspace(UCHAR(*p2))) /* INTL: ISO space. */ - && (p2 < p+20)) { - p2++; - } - sprintf(buf, - "list element in braces followed by \"%.*s\" instead of space", - (int) (p2-p), p); - Tcl_SetResult(interp, buf, TCL_VOLATILE); + /* + * Garbage after the closing brace; return an error. + */ + + if (interp != NULL) { + p2 = p; + while ((p2 < limit) && (!TclIsSpaceProc(*p2)) + && (p2 < p+20)) { + p2++; } - return TCL_ERROR; + Tcl_SetObjResult(interp, Tcl_ObjPrintf( + "list element in braces followed by \"%.*s\" " + "instead of space", (int) (p2-p), p)); } - break; + return TCL_ERROR; + } + break; /* - * Backslash: skip over everything up to the end of the - * backslash sequence. + * Backslash: skip over everything up to the end of the backslash + * sequence. */ - case '\\': { - TclParseBackslash(p, limit - p, &numChars, NULL); - p += (numChars - 1); - break; + case '\\': + if (openBraces == 0) { + /* + * A backslash sequence not within a brace quoted element + * means the value of the element is different from the + * substring we are parsing. A call to TclCopyAndCollapse() + * is needed to produce the element value. Inform the caller. + */ + literal = 0; } + TclParseBackslash(p, limit - p, &numChars, NULL); + p += (numChars - 1); + break; /* * Space: ignore if element is in braces or quotes; otherwise * terminate element. */ - case ' ': - case '\f': - case '\n': - case '\r': - case '\t': - case '\v': - if ((openBraces == 0) && !inQuotes) { - size = (p - elemStart); - goto done; - } - break; + case ' ': + case '\f': + case '\n': + case '\r': + case '\t': + case '\v': + if ((openBraces == 0) && !inQuotes) { + size = (p - elemStart); + goto done; + } + break; /* * Double-quote: if element is in quotes then terminate it. */ - case '"': - if (inQuotes) { - size = (p - elemStart); - p++; - if ((p >= limit) - || isspace(UCHAR(*p))) { /* INTL: ISO space */ - goto done; - } + case '"': + if (inQuotes) { + size = (p - elemStart); + p++; + if ((p >= limit) || TclIsSpaceProc(*p)) { + goto done; + } - /* - * Garbage after the closing quote; return an error. - */ - - if (interp != NULL) { - char buf[100]; - - p2 = p; - while ((p2 < limit) - && (!isspace(UCHAR(*p2))) /* INTL: ISO space */ - && (p2 < p+20)) { - p2++; - } - sprintf(buf, - "list element in quotes followed by \"%.*s\" %s", - (int) (p2-p), p, "instead of space"); - Tcl_SetResult(interp, buf, TCL_VOLATILE); + /* + * Garbage after the closing quote; return an error. + */ + + if (interp != NULL) { + p2 = p; + while ((p2 < limit) && (!TclIsSpaceProc(*p2)) + && (p2 < p+20)) { + p2++; } - return TCL_ERROR; + Tcl_SetObjResult(interp, Tcl_ObjPrintf( + "list element in quotes followed by \"%.*s\" " + "instead of space", (int) (p2-p), p)); } - break; + return TCL_ERROR; + } + break; } p++; } - /* * End of list: terminate element. */ @@ -321,8 +665,8 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, size = (p - elemStart); } - done: - while ((p < limit) && (isspace(UCHAR(*p)))) { /* INTL: ISO space. */ + done: + while ((p < limit) && (TclIsSpaceProc(*p))) { p++; } *elementPtr = elemStart; @@ -330,6 +674,9 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, if (sizePtr != 0) { *sizePtr = size; } + if (literalPtr != 0) { + *literalPtr = literal; + } return TCL_OK; } @@ -338,7 +685,7 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, * * TclCopyAndCollapse -- * - * Copy a string and eliminate any backslashes that aren't in braces. + * Copy a string and substitute all backslash escape sequences * * Results: * Count bytes get copied from src to dst. Along the way, backslash @@ -353,10 +700,10 @@ TclFindElement(interp, list, listLength, elementPtr, nextPtr, sizePtr, */ int -TclCopyAndCollapse(count, src, dst) - int count; /* Number of bytes to copy from src. */ - CONST char *src; /* Copy from here... */ - char *dst; /* ... to here. */ +TclCopyAndCollapse( + int count, /* Number of byte to copy from src. */ + CONST char *src, /* Copy from here... */ + char *dst) /* ... to here. */ { int newCount = 0; @@ -390,21 +737,19 @@ TclCopyAndCollapse(count, src, dst) * Splits a list up into its constituent fields. * * Results - * The return value is normally TCL_OK, which means that - * the list was successfully split up. If TCL_ERROR is - * returned, it means that "list" didn't have proper list - * structure; the interp's result will contain a more detailed - * error message. - * - * *argvPtr will be filled in with the address of an array - * whose elements point to the elements of list, in order. - * *argcPtr will get filled in with the number of valid elements - * in the array. A single block of memory is dynamically allocated - * to hold both the argv array and a copy of the list (with - * backslashes and braces removed in the standard way). - * The caller must eventually free this memory by calling free() - * on *argvPtr. Note: *argvPtr and *argcPtr are only modified - * if the procedure returns normally. + * The return value is normally TCL_OK, which means that the list was + * successfully split up. If TCL_ERROR is returned, it means that "list" + * didn't have proper list structure; the interp's result will contain a + * more detailed error message. + * + * *argvPtr will be filled in with the address of an array whose elements + * point to the elements of list, in order. *argcPtr will get filled in + * with the number of valid elements in the array. A single block of + * memory is dynamically allocated to hold both the argv array and a copy + * of the list (with backslashes and braces removed in the standard way). + * The caller must eventually free this memory by calling free() on + * *argvPtr. Note: *argvPtr and *argcPtr are only modified if the + * function returns normally. * * Side effects: * Memory is allocated. @@ -413,54 +758,40 @@ TclCopyAndCollapse(count, src, dst) */ int -Tcl_SplitList(interp, list, argcPtr, argvPtr) - Tcl_Interp *interp; /* Interpreter to use for error reporting. - * If NULL, no error message is left. */ - CONST char *list; /* Pointer to string with list structure. */ - int *argcPtr; /* Pointer to location to fill in with - * the number of elements in the list. */ - CONST char ***argvPtr; /* Pointer to place to store pointer to - * array of pointers to list elements. */ +Tcl_SplitList( + Tcl_Interp *interp, /* Interpreter to use for error reporting. If + * NULL, no error message is left. */ + CONST char *list, /* Pointer to string with list structure. */ + int *argcPtr, /* Pointer to location to fill in with the + * number of elements in the list. */ + CONST char ***argvPtr) /* Pointer to place to store pointer to array + * of pointers to list elements. */ { - CONST char **argv; - CONST char *l; + CONST char **argv, *end, *element; char *p; - int length, size, i, result, elSize, brace; - CONST char *element; + int length, size, i, result, elSize; /* - * Figure out how much space to allocate. There must be enough - * space for both the array of pointers and also for a copy of - * the list. To estimate the number of pointers needed, count - * the number of space characters in the list. + * Allocate enough space to work in. A (CONST char *) for each + * (possible) list element plus one more for terminating NULL, + * plus as many bytes as in the original string value, plus one + * more for a terminating '\0'. Space used to hold element separating + * white space in the original string gets re-purposed to hold '\0' + * characters in the argv array. */ - for (size = 2, l = list; *l != 0; l++) { - if (isspace(UCHAR(*l))) { /* INTL: ISO space. */ - size++; - /* Consecutive space can only count as a single list delimiter */ - while (1) { - char next = *(l + 1); - if (next == '\0') { - break; - } - ++l; - if (isspace(UCHAR(next))) { - continue; - } - break; - } - } - } - length = l - list; + size = TclMaxListLength(list, -1, &end) + 1; + length = end - list; argv = (CONST char **) ckalloc((unsigned) ((size * sizeof(char *)) + length + 1)); + for (i = 0, p = ((char *) argv) + size*sizeof(char *); *list != 0; i++) { CONST char *prevList = list; - - result = TclFindElement(interp, list, length, &element, - &list, &elSize, &brace); + int literal; + + result = TclFindElement(interp, list, length, &element, &list, + &elSize, &literal); length -= (list - prevList); if (result != TCL_OK) { ckfree((char *) argv); @@ -478,14 +809,13 @@ Tcl_SplitList(interp, list, argcPtr, argvPtr) return TCL_ERROR; } argv[i] = p; - if (brace) { - memcpy((VOID *) p, (VOID *) element, (size_t) elSize); + if (literal) { + memcpy(p, element, (size_t) elSize); p += elSize; *p = 0; p++; } else { - TclCopyAndCollapse(elSize, element, p); - p += elSize+1; + p += 1 + TclCopyAndCollapse(elSize, element, p); } } @@ -500,17 +830,15 @@ Tcl_SplitList(interp, list, argcPtr, argvPtr) * * Tcl_ScanElement -- * - * This procedure is a companion procedure to Tcl_ConvertElement. - * It scans a string to see what needs to be done to it (e.g. add - * backslashes or enclosing braces) to make the string into a - * valid Tcl list element. + * This function is a companion function to Tcl_ConvertElement. It scans + * a string to see what needs to be done to it (e.g. add backslashes or + * enclosing braces) to make the string into a valid Tcl list element. * * Results: - * The return value is an overestimate of the number of characters - * that will be needed by Tcl_ConvertElement to produce a valid - * list element from string. The word at *flagPtr is filled in - * with a value needed by Tcl_ConvertElement when doing the actual - * conversion. + * The return value is an overestimate of the number of bytes that + * will be needed by Tcl_ConvertElement to produce a valid list element + * from src. The word at *flagPtr is filled in with a value needed by + * Tcl_ConvertElement when doing the actual conversion. * * Side effects: * None. @@ -519,12 +847,12 @@ Tcl_SplitList(interp, list, argcPtr, argvPtr) */ int -Tcl_ScanElement(string, flagPtr) - register CONST char *string; /* String to convert to list element. */ - register int *flagPtr; /* Where to store information to guide - * Tcl_ConvertCountedElement. */ +Tcl_ScanElement( + register CONST char *src, /* String to convert to list element. */ + register int *flagPtr) /* Where to store information to guide + * Tcl_ConvertCountedElement. */ { - return Tcl_ScanCountedElement(string, -1, flagPtr); + return Tcl_ScanCountedElement(src, -1, flagPtr); } /* @@ -532,19 +860,17 @@ Tcl_ScanElement(string, flagPtr) * * Tcl_ScanCountedElement -- * - * This procedure is a companion procedure to - * Tcl_ConvertCountedElement. It scans a string to see what - * needs to be done to it (e.g. add backslashes or enclosing - * braces) to make the string into a valid Tcl list element. - * If length is -1, then the string is scanned up to the first - * null byte. + * This function is a companion function to Tcl_ConvertCountedElement. It + * scans a string to see what needs to be done to it (e.g. add + * backslashes or enclosing braces) to make the string into a valid Tcl + * list element. If length is -1, then the string is scanned from src up + * to the first null byte. * * Results: - * The return value is an overestimate of the number of characters - * that will be needed by Tcl_ConvertCountedElement to produce a - * valid list element from string. The word at *flagPtr is - * filled in with a value needed by Tcl_ConvertCountedElement - * when doing the actual conversion. + * The return value is an overestimate of the number of bytes that + * will be needed by Tcl_ConvertCountedElement to produce a valid list + * element from src. The word at *flagPtr is filled in with a value + * needed by Tcl_ConvertCountedElement when doing the actual conversion. * * Side effects: * None. @@ -553,115 +879,266 @@ Tcl_ScanElement(string, flagPtr) */ int -Tcl_ScanCountedElement(string, length, flagPtr) - CONST char *string; /* String to convert to Tcl list element. */ - int length; /* Number of bytes in string, or -1. */ - int *flagPtr; /* Where to store information to guide +Tcl_ScanCountedElement( + CONST char *src, /* String to convert to Tcl list element. */ + int length, /* Number of bytes in src, or -1. */ + int *flagPtr) /* Where to store information to guide * Tcl_ConvertElement. */ { - int flags, nestingLevel; - register CONST char *p, *lastChar; + int flags = CONVERT_ANY; + int numBytes = TclScanElement(src, length, &flags); - /* - * This procedure and Tcl_ConvertElement together do two things: - * - * 1. They produce a proper list, one that will yield back the - * argument strings when evaluated or when disassembled with - * Tcl_SplitList. This is the most important thing. - * - * 2. They try to produce legible output, which means minimizing the - * use of backslashes (using braces instead). However, there are - * some situations where backslashes must be used (e.g. an element - * like "{abc": the leading brace will have to be backslashed. - * For each element, one of three things must be done: - * - * (a) Use the element as-is (it doesn't contain any special - * characters). This is the most desirable option. - * - * (b) Enclose the element in braces, but leave the contents alone. - * This happens if the element contains embedded space, or if it - * contains characters with special interpretation ($, [, ;, or \), - * or if it starts with a brace or double-quote, or if there are - * no characters in the element. - * - * (c) Don't enclose the element in braces, but add backslashes to - * prevent special interpretation of special characters. This is a - * last resort used when the argument would normally fall under case - * (b) but contains unmatched braces. It also occurs if the last - * character of the argument is a backslash or if the element contains - * a backslash followed by newline. - * - * The procedure figures out how many bytes will be needed to store - * the result (actually, it overestimates). It also collects information - * about the element in the form of a flags word. - * - * Note: list elements produced by this procedure and - * Tcl_ConvertCountedElement must have the property that they can be - * enclosing in curly braces to make sub-lists. This means, for - * example, that we must not leave unmatched curly braces in the - * resulting list element. This property is necessary in order for - * procedures like Tcl_DStringStartSublist to work. - */ + *flagPtr = flags; + return numBytes; +} + +/* + *---------------------------------------------------------------------- + * + * TclScanElement -- + * + * This function is a companion function to TclConvertElement. It + * scans a string to see what needs to be done to it (e.g. add + * backslashes or enclosing braces) to make the string into a valid Tcl + * list element. If length is -1, then the string is scanned from src up + * to the first null byte. A NULL value for src is treated as an + * empty string. The incoming value of *flagPtr is a report from the + * caller what additional flags it will pass to TclConvertElement(). + * + * Results: + * The recommended formatting mode for the element is determined and + * a value is written to *flagPtr indicating that recommendation. This + * recommendation is combined with the incoming flag values in *flagPtr + * set by the caller to determine how many bytes will be needed by + * TclConvertElement() in which to write the formatted element following + * the recommendation modified by the flag values. This number of bytes + * is the return value of the routine. In some situations it may be + * an overestimate, but so long as the caller passes the same flags + * to TclConvertElement(), it will be large enough. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ - nestingLevel = 0; - flags = 0; - if (string == NULL) { - string = ""; - } - if (length == -1) { - length = strlen(string); +int +TclScanElement( + CONST char *src, /* String to convert to Tcl list element. */ + int length, /* Number of bytes in src, or -1. */ + int *flagPtr) /* Where to store information to guide + * Tcl_ConvertElement. */ +{ + CONST char *p = src; + int nestingLevel = 0; /* Brace nesting count */ + int forbidNone = 0; /* Do not permit CONVERT_NONE mode. Something + needs protection or escape. */ + int requireEscape = 0; /* Force use of CONVERT_ESCAPE mode. For some + * reason bare or brace-quoted form fails. */ + int extra = 0; /* Count of number of extra bytes needed for + * formatted element, assuming we use escape + * sequences in formatting. */ + int bytesNeeded; /* Buffer length computed to complete the + * element formatting in the selected mode. */ +#if COMPAT + int preferEscape = 0; /* Use preferences to track whether to use */ + int preferBrace = 0; /* CONVERT_MASK mode. */ + int braceCount = 0; /* Count of all braces '{' '}' seen. */ +#endif + + if ((p == NULL) || (length == 0) || ((*p == '\0') && (length == -1))) { + /* Empty string element must be brace quoted. */ + *flagPtr = CONVERT_BRACE; + return 2; } - lastChar = string + length; - p = string; - if ((p == lastChar) || (*p == '{') || (*p == '"')) { - flags |= USE_BRACES; + + if ((*p == '{') || (*p == '"')) { + /* + * Must escape or protect so leading character of value is not + * misinterpreted as list element delimiting syntax. + */ + forbidNone = 1; +#if COMPAT + preferBrace = 1; +#endif } - for ( ; p < lastChar; p++) { + + while (length) { switch (*p) { - case '{': - nestingLevel++; - break; - case '}': - nestingLevel--; - if (nestingLevel < 0) { - flags |= TCL_DONT_USE_BRACES|BRACES_UNMATCHED; - } - break; - case '[': - case '$': - case ';': - case ' ': - case '\f': - case '\n': - case '\r': - case '\t': - case '\v': - flags |= USE_BRACES; + case '{': +#if COMPAT + braceCount++; +#endif + extra++; /* Escape '{' => '\{' */ + nestingLevel++; + break; + case '}': +#if COMPAT + braceCount++; +#endif + extra++; /* Escape '}' => '\}' */ + nestingLevel--; + if (nestingLevel < 0) { + /* Unbalanced braces! Cannot format with brace quoting. */ + requireEscape = 1; + } + break; + case ']': + case '"': +#if COMPAT + forbidNone = 1; + extra++; /* Escapes all just prepend a backslash */ + preferEscape = 1; + break; +#else + /* FLOW THROUGH */ +#endif + case '[': + case '$': + case ';': + case ' ': + case '\f': + case '\n': + case '\r': + case '\t': + case '\v': + forbidNone = 1; + extra++; /* Escape sequences all one byte longer. */ +#if COMPAT + preferBrace = 1; +#endif + break; + case '\\': + extra++; /* Escape '\' => '\\' */ + if ((length == 1) || ((length == -1) && (p[1] == '\0'))) { + /* Final backslash. Cannot format with brace quoting. */ + requireEscape = 1; break; - case '\\': - if ((p+1 == lastChar) || (p[1] == '\n')) { - flags = TCL_DONT_USE_BRACES | BRACES_UNMATCHED; - } else { - int size; - - TclParseBackslash(p, lastChar - p, &size, NULL); - p += size-1; - flags |= USE_BRACES; - } + } + if (p[1] == '\n') { + extra++; /* Escape newline => '\n', one byte longer */ + /* Backslash newline sequence. Brace quoting not permitted. */ + requireEscape = 1; + length -= (length > 0); + p++; break; + } + if ((p[1] == '{') || (p[1] == '}') || (p[1] == '\\')) { + extra++; /* Escape sequences all one byte longer. */ + length -= (length > 0); + p++; + } + forbidNone = 1; +#if COMPAT + preferBrace = 1; +#endif + break; + case '\0': + if (length == -1) { + goto endOfString; + } + /* TODO: Panic on improper encoding? */ + break; } + length -= (length > 0); + p++; } + + endOfString: if (nestingLevel != 0) { - flags = TCL_DONT_USE_BRACES | BRACES_UNMATCHED; + /* Unbalanced braces! Cannot format with brace quoting. */ + requireEscape = 1; } - *flagPtr = flags; - /* - * Allow enough space to backslash every character plus leave - * two spaces for braces. - */ + /* We need at least as many bytes as are in the element value... */ + bytesNeeded = p - src; - return 2*(p-string) + 2; + if (requireEscape) { + /* + * We must use escape sequences. Add all the extra bytes needed + * to have room to create them. + */ + bytesNeeded += extra; + /* Make room to escape leading #, if needed. */ + if ((*src == '#') && !(*flagPtr & TCL_DONT_QUOTE_HASH)) { + bytesNeeded++; + } + *flagPtr = CONVERT_ESCAPE; + goto overflowCheck; + } + if (*flagPtr & CONVERT_ANY) { + /* + * The caller has not let us know what flags it will pass to + * TclConvertElement() so compute the max size we might need for + * any possible choice. Normally the formatting using escape + * sequences is the longer one, and a minimum "extra" value of 2 + * makes sure we don't request too small a buffer in those edge + * cases where that's not true. + */ + if (extra < 2) { + extra = 2; + } + *flagPtr &= ~CONVERT_ANY; + *flagPtr |= TCL_DONT_USE_BRACES; + } + if (forbidNone) { + /* We must request some form of quoting of escaping... */ +#if COMPAT + if (preferEscape && !preferBrace) { + /* + * If we are quoting solely due to ] or internal " characters + * use the CONVERT_MASK mode where we escape all special + * characters except for braces. "extra" counted space needed + * to escape braces too, so substract "braceCount" to get our + * actual needs. + */ + bytesNeeded += (extra - braceCount); + /* Make room to escape leading #, if needed. */ + if ((*src == '#') && !(*flagPtr & TCL_DONT_QUOTE_HASH)) { + bytesNeeded++; + } + /* + * If the caller reports it will direct TclConvertElement() to + * use full escapes on the element, add back the bytes needed to + * escape the braces. + */ + if (*flagPtr & TCL_DONT_USE_BRACES) { + bytesNeeded += braceCount; + } + *flagPtr = CONVERT_MASK; + goto overflowCheck; + } +#endif + if (*flagPtr & TCL_DONT_USE_BRACES) { + /* + * If the caller reports it will direct TclConvertElement() to + * use escapes, add the extra bytes needed to have room for them. + */ + bytesNeeded += extra; + /* Make room to escape leading #, if needed. */ + if ((*src == '#') && !(*flagPtr & TCL_DONT_QUOTE_HASH)) { + bytesNeeded++; + } + } else { + /* Add 2 bytes for room for the enclosing braces. */ + bytesNeeded += 2; + } + *flagPtr = CONVERT_BRACE; + goto overflowCheck; + } + + /* So far, no need to quote or escape anything. */ + if ((*src == '#') && !(*flagPtr & TCL_DONT_QUOTE_HASH)) { + /* If we need to quote a leading #, make room to enclose in braces. */ + bytesNeeded += 2; + } + *flagPtr = CONVERT_NONE; + + overflowCheck: + if (bytesNeeded < 0) { + Tcl_Panic("TclScanElement: string length overflow"); + } + return bytesNeeded; } /* @@ -669,16 +1146,15 @@ Tcl_ScanCountedElement(string, length, flagPtr) * * Tcl_ConvertElement -- * - * This is a companion procedure to Tcl_ScanElement. Given - * the information produced by Tcl_ScanElement, this procedure - * converts a string to a list element equal to that string. + * This is a companion function to Tcl_ScanElement. Given the information + * produced by Tcl_ScanElement, this function converts a string to a list + * element equal to that string. * * Results: - * Information is copied to *dst in the form of a list element - * identical to src (i.e. if Tcl_SplitList is applied to dst it - * will produce a string identical to src). The return value is - * a count of the number of characters copied (not including the - * terminating NULL character). + * Information is copied to *dst in the form of a list element identical + * to src (i.e. if Tcl_SplitList is applied to dst it will produce a + * string identical to src). The return value is a count of the number of + * characters copied (not including the terminating NULL character). * * Side effects: * None. @@ -687,10 +1163,10 @@ Tcl_ScanCountedElement(string, length, flagPtr) */ int -Tcl_ConvertElement(src, dst, flags) - register CONST char *src; /* Source information for list element. */ - register char *dst; /* Place to put list-ified element. */ - register int flags; /* Flags produced by Tcl_ScanElement. */ +Tcl_ConvertElement( + register CONST char *src, /* Source information for list element. */ + register char *dst, /* Place to put list-ified element. */ + register int flags) /* Flags produced by Tcl_ScanElement. */ { return Tcl_ConvertCountedElement(src, -1, dst, flags); } @@ -700,17 +1176,15 @@ Tcl_ConvertElement(src, dst, flags) * * Tcl_ConvertCountedElement -- * - * This is a companion procedure to Tcl_ScanCountedElement. Given - * the information produced by Tcl_ScanCountedElement, this - * procedure converts a string to a list element equal to that - * string. + * This is a companion function to Tcl_ScanCountedElement. Given the + * information produced by Tcl_ScanCountedElement, this function converts + * a string to a list element equal to that string. * * Results: - * Information is copied to *dst in the form of a list element - * identical to src (i.e. if Tcl_SplitList is applied to dst it - * will produce a string identical to src). The return value is - * a count of the number of characters copied (not including the - * terminating NULL character). + * Information is copied to *dst in the form of a list element identical + * to src (i.e. if Tcl_SplitList is applied to dst it will produce a + * string identical to src). The return value is a count of the number of + * characters copied (not including the terminating NULL character). * * Side effects: * None. @@ -719,118 +1193,177 @@ Tcl_ConvertElement(src, dst, flags) */ int -Tcl_ConvertCountedElement(src, length, dst, flags) - register CONST char *src; /* Source information for list element. */ - int length; /* Number of bytes in src, or -1. */ - char *dst; /* Place to put list-ified element. */ - int flags; /* Flags produced by Tcl_ScanElement. */ +Tcl_ConvertCountedElement( + register CONST char *src, /* Source information for list element. */ + int length, /* Number of bytes in src, or -1. */ + char *dst, /* Place to put list-ified element. */ + int flags) /* Flags produced by Tcl_ScanElement. */ { - register char *p = dst; - register CONST char *lastChar; + int numBytes = TclConvertElement(src, length, dst, flags); + dst[numBytes] = '\0'; + return numBytes; +} + +/* + *---------------------------------------------------------------------- + * + * TclConvertElement -- + * + * This is a companion function to TclScanElement. Given the + * information produced by TclScanElement, this function converts + * a string to a list element equal to that string. + * + * Results: + * Information is copied to *dst in the form of a list element identical + * to src (i.e. if Tcl_SplitList is applied to dst it will produce a + * string identical to src). The return value is a count of the number of + * characters copied (not including the terminating NULL character). + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ - /* - * See the comment block at the beginning of the Tcl_ScanElement - * code for details of how this works. - */ +int TclConvertElement( + register CONST char *src, /* Source information for list element. */ + int length, /* Number of bytes in src, or -1. */ + char *dst, /* Place to put list-ified element. */ + int flags) /* Flags produced by Tcl_ScanElement. */ +{ + int conversion = flags & CONVERT_MASK; + char *p = dst; - if (src && length == -1) { - length = strlen(src); + /* Let the caller demand we use escape sequences rather than braces. */ + if ((flags & TCL_DONT_USE_BRACES) && (conversion & CONVERT_BRACE)) { + conversion = CONVERT_ESCAPE; } - if ((src == NULL) || (length == 0)) { - p[0] = '{'; - p[1] = '}'; - p[2] = 0; - return 2; + + /* No matter what the caller demands, empty string must be braced! */ + if ((src == NULL) || (length == 0) || ((*src == '\0') && (length == -1))) { + src = tclEmptyStringRep; + length = 0; + conversion = CONVERT_BRACE; + } + + /* Escape leading hash as needed and requested. */ + if ((*src == '#') && !(flags & TCL_DONT_QUOTE_HASH)) { + if (conversion == CONVERT_ESCAPE) { + p[0] = '\\'; + p[1] = '#'; + p += 2; + src++; + length -= (length > 0); + } else { + conversion = CONVERT_BRACE; + } } - lastChar = src + length; - if ((flags & USE_BRACES) && !(flags & TCL_DONT_USE_BRACES)) { + + /* No escape or quoting needed. Copy the literal string value. */ + if (conversion == CONVERT_NONE) { + if (length == -1) { + /* TODO: INT_MAX overflow? */ + while (*src) { + *p++ = *src++; + } + return p - dst; + } else { + memcpy(dst, src, length); + return length; + } + } + + /* Formatted string is original string enclosed in braces. */ + if (conversion == CONVERT_BRACE) { *p = '{'; p++; - for ( ; src != lastChar; src++, p++) { - *p = *src; + if (length == -1) { + /* TODO: INT_MAX overflow? */ + while (*src) { + *p++ = *src++; + } + } else { + memcpy(p, src, length); + p += length; } *p = '}'; p++; - } else { - if (*src == '{') { - /* - * Can't have a leading brace unless the whole element is - * enclosed in braces. Add a backslash before the brace. - * Furthermore, this may destroy the balance between open - * and close braces, so set BRACES_UNMATCHED. - */ + return p - dst; + } - p[0] = '\\'; - p[1] = '{'; - p += 2; - src++; - flags |= BRACES_UNMATCHED; - } - for (; src != lastChar; src++) { - switch (*src) { - case ']': - case '[': - case '$': - case ';': - case ' ': - case '\\': - case '"': - *p = '\\'; - p++; - break; - case '{': - case '}': - /* - * It may not seem necessary to backslash braces, but - * it is. The reason for this is that the resulting - * list element may actually be an element of a sub-list - * enclosed in braces (e.g. if Tcl_DStringStartSublist - * has been invoked), so there may be a brace mismatch - * if the braces aren't backslashed. - */ - - if (flags & BRACES_UNMATCHED) { - *p = '\\'; - p++; - } - break; - case '\f': - *p = '\\'; - p++; - *p = 'f'; - p++; - continue; - case '\n': - *p = '\\'; - p++; - *p = 'n'; - p++; - continue; - case '\r': - *p = '\\'; - p++; - *p = 'r'; - p++; - continue; - case '\t': - *p = '\\'; - p++; - *p = 't'; - p++; - continue; - case '\v': - *p = '\\'; - p++; - *p = 'v'; - p++; - continue; + /* conversion == CONVERT_ESCAPE or CONVERT_MASK */ + + /* Formatted string is original string converted to escape sequences. */ + for ( ; length; src++, length -= (length > 0)) { + switch (*src) { + case ']': + case '[': + case '$': + case ';': + case ' ': + case '\\': + case '"': + *p = '\\'; + p++; + break; + case '{': + case '}': +#if COMPAT + if (conversion == CONVERT_ESCAPE) { +#endif + *p = '\\'; + p++; +#if COMPAT } - *p = *src; +#endif + break; + case '\f': + *p = '\\'; + p++; + *p = 'f'; + p++; + continue; + case '\n': + *p = '\\'; + p++; + *p = 'n'; + p++; + continue; + case '\r': + *p = '\\'; p++; + *p = 'r'; + p++; + continue; + case '\t': + *p = '\\'; + p++; + *p = 't'; + p++; + continue; + case '\v': + *p = '\\'; + p++; + *p = 'v'; + p++; + continue; + case '\0': + if (length == -1) { + return p - dst; + } + /* + * If we reach this point, there's an embedded NULL in the + * string range being processed, which should not happen when + * the encoding rules for Tcl strings are properly followed. + * If the day ever comes when we stop tolerating such things, + * this is where to put the Tcl_Panic(). + */ + break; } + *p = *src; + p++; } - *p = '\0'; - return p-dst; + return p - dst; } /* @@ -838,15 +1371,14 @@ Tcl_ConvertCountedElement(src, length, dst, flags) * * Tcl_Merge -- * - * Given a collection of strings, merge them together into a - * single string that has proper Tcl list structured (i.e. - * Tcl_SplitList may be used to retrieve strings equal to the - * original elements, and Tcl_Eval will parse the string back - * into its original elements). + * Given a collection of strings, merge them together into a single + * string that has proper Tcl list structured (i.e. Tcl_SplitList may be + * used to retrieve strings equal to the original elements, and Tcl_Eval + * will parse the string back into its original elements). * * Results: - * The return value is the address of a dynamically-allocated - * string containing the merged list. + * The return value is the address of a dynamically-allocated string + * containing the merged list. * * Side effects: * None. @@ -855,16 +1387,25 @@ Tcl_ConvertCountedElement(src, length, dst, flags) */ char * -Tcl_Merge(argc, argv) - int argc; /* How many strings to merge. */ - CONST char * CONST *argv; /* Array of string values. */ +Tcl_Merge( + int argc, /* How many strings to merge. */ + CONST char * CONST *argv) /* Array of string values. */ { # define LOCAL_SIZE 20 - int localFlags[LOCAL_SIZE], *flagPtr; - int numChars; - char *result; - char *dst; - int i; + int localFlags[LOCAL_SIZE], *flagPtr = NULL; + int i, bytesNeeded = 0; + char *result, *dst; + const int maxFlags = UINT_MAX / sizeof(int); + + if (argc == 0) { + /* + * Handle empty list case first, so logic of the general case + * can be simpler. + */ + result = ckalloc(1); + result[0] = '\0'; + return result; + } /* * Pass 1: estimate space, gather flags. @@ -872,31 +1413,48 @@ Tcl_Merge(argc, argv) if (argc <= LOCAL_SIZE) { flagPtr = localFlags; + } else if (argc > maxFlags) { + /* + * We cannot allocate a large enough flag array to format this + * list in one pass. We could imagine converting this routine + * to a multi-pass implementation, but for sizeof(int) == 4, + * the limit is a max of 2^30 list elements and since each element + * is at least one byte formatted, and requires one byte space + * between it and the next one, that a minimum space requirement + * of 2^31 bytes, which is already INT_MAX. If we tried to format + * a list of > maxFlags elements, we're just going to overflow + * the size limits on the formatted string anyway, so just issue + * that same panic early. + */ + Tcl_Panic("max size for a Tcl value (%d bytes) exceeded", INT_MAX); } else { flagPtr = (int *) ckalloc((unsigned) argc*sizeof(int)); } - numChars = 1; for (i = 0; i < argc; i++) { - numChars += Tcl_ScanElement(argv[i], &flagPtr[i]) + 1; + flagPtr[i] = ( i ? TCL_DONT_QUOTE_HASH : 0 ); + bytesNeeded += TclScanElement(argv[i], -1, &flagPtr[i]); + if (bytesNeeded < 0) { + Tcl_Panic("max size for a Tcl value (%d bytes) exceeded", INT_MAX); + } + } + if (bytesNeeded > INT_MAX - argc + 1) { + Tcl_Panic("max size for a Tcl value (%d bytes) exceeded", INT_MAX); } + bytesNeeded += argc; /* * Pass two: copy into the result area. */ - result = (char *) ckalloc((unsigned) numChars); + result = ckalloc((unsigned) bytesNeeded); dst = result; for (i = 0; i < argc; i++) { - numChars = Tcl_ConvertElement(argv[i], dst, flagPtr[i]); - dst += numChars; + flagPtr[i] |= ( i ? TCL_DONT_QUOTE_HASH : 0 ); + dst += TclConvertElement(argv[i], -1, dst, flagPtr[i]); *dst = ' '; dst++; } - if (dst == result) { - *dst = 0; - } else { - dst[-1] = 0; - } + dst[-1] = 0; if (flagPtr != localFlags) { ckfree((char *) flagPtr); @@ -912,10 +1470,10 @@ Tcl_Merge(argc, argv) * Figure out how to handle a backslash sequence. * * Results: - * The return value is the character that should be substituted - * in place of the backslash sequence that starts at src. If - * readPtr isn't NULL then it is filled in with a count of the - * number of characters in the backslash sequence. + * The return value is the character that should be substituted in place + * of the backslash sequence that starts at src. If readPtr isn't NULL + * then it is filled in with a count of the number of characters in the + * backslash sequence. * * Side effects: * None. @@ -924,11 +1482,11 @@ Tcl_Merge(argc, argv) */ char -Tcl_Backslash(src, readPtr) - CONST char *src; /* Points to the backslash character of - * a backslash sequence. */ - int *readPtr; /* Fill in with number of characters read - * from src, unless NULL. */ +Tcl_Backslash( + CONST char *src, /* Points to the backslash character of a + * backslash sequence. */ + int *readPtr) /* Fill in with number of characters read from + * src, unless NULL. */ { char buf[TCL_UTF_MAX]; Tcl_UniChar ch; @@ -941,73 +1499,228 @@ Tcl_Backslash(src, readPtr) /* *---------------------------------------------------------------------- * + * TclTrimRight -- + * Takes two counted strings in the Tcl encoding which must both be + * null terminated. Conceptually trims from the right side of the + * first string all characters found in the second string. + * + * Results: + * The number of bytes to be removed from the end of the string. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +int +TclTrimRight( + const char *bytes, /* String to be trimmed... */ + int numBytes, /* ...and its length in bytes */ + const char *trim, /* String of trim characters... */ + int numTrim) /* ...and its length in bytes */ +{ + const char *p = bytes + numBytes; + int pInc; + + if ((bytes[numBytes] != '\0') || (trim[numTrim] != '\0')) { + Tcl_Panic("TclTrimRight works only on null-terminated strings"); + } + + /* Empty strings -> nothing to do */ + if ((numBytes == 0) || (numTrim == 0)) { + return 0; + } + + /* Outer loop: iterate over string to be trimmed */ + do { + Tcl_UniChar ch1; + const char *q = trim; + int bytesLeft = numTrim; + + p = Tcl_UtfPrev(p, bytes); + pInc = TclUtfToUniChar(p, &ch1); + + /* Inner loop: scan trim string for match to current character */ + do { + Tcl_UniChar ch2; + int qInc = TclUtfToUniChar(q, &ch2); + + if (ch1 == ch2) { + break; + } + + q += qInc; + bytesLeft -= qInc; + } while (bytesLeft); + + if (bytesLeft == 0) { + /* No match; trim task done; *p is last non-trimmed char */ + p += pInc; + break; + } + } while (p > bytes); + + return numBytes - (p - bytes); +} + +/* + *---------------------------------------------------------------------- + * + * TclTrimLeft -- + * Takes two counted strings in the Tcl encoding which must both be + * null terminated. Conceptually trims from the left side of the + * first string all characters found in the second string. + * + * Results: + * The number of bytes to be removed from the start of the string. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +int +TclTrimLeft( + const char *bytes, /* String to be trimmed... */ + int numBytes, /* ...and its length in bytes */ + const char *trim, /* String of trim characters... */ + int numTrim) /* ...and its length in bytes */ +{ + const char *p = bytes; + + if ((bytes[numBytes] != '\0') || (trim[numTrim] != '\0')) { + Tcl_Panic("TclTrimLeft works only on null-terminated strings"); + } + + /* Empty strings -> nothing to do */ + if ((numBytes == 0) || (numTrim == 0)) { + return 0; + } + + /* Outer loop: iterate over string to be trimmed */ + do { + Tcl_UniChar ch1; + int pInc = TclUtfToUniChar(p, &ch1); + const char *q = trim; + int bytesLeft = numTrim; + + /* Inner loop: scan trim string for match to current character */ + do { + Tcl_UniChar ch2; + int qInc = TclUtfToUniChar(q, &ch2); + + if (ch1 == ch2) { + break; + } + + q += qInc; + bytesLeft -= qInc; + } while (bytesLeft); + + if (bytesLeft == 0) { + /* No match; trim task done; *p is first non-trimmed char */ + break; + } + + p += pInc; + numBytes -= pInc; + } while (numBytes); + + return p - bytes; +} + +/* + *---------------------------------------------------------------------- + * * Tcl_Concat -- * * Concatenate a set of strings into a single large string. * * Results: - * The return value is dynamically-allocated string containing - * a concatenation of all the strings in argv, with spaces between - * the original argv elements. + * The return value is dynamically-allocated string containing a + * concatenation of all the strings in argv, with spaces between the + * original argv elements. * * Side effects: - * Memory is allocated for the result; the caller is responsible - * for freeing the memory. + * Memory is allocated for the result; the caller is responsible for + * freeing the memory. * *---------------------------------------------------------------------- */ +/* The whitespace characters trimmed during [concat] operations */ +#define CONCAT_WS " \f\v\r\t\n" +#define CONCAT_WS_SIZE (int) (sizeof(CONCAT_WS "") - 1) + char * -Tcl_Concat(argc, argv) - int argc; /* Number of strings to concatenate. */ - CONST char * CONST *argv; /* Array of strings to concatenate. */ +Tcl_Concat( + int argc, /* Number of strings to concatenate. */ + CONST char * CONST *argv) /* Array of strings to concatenate. */ { - int totalSize, i; - char *p; - char *result; + int i, needSpace = 0, bytesNeeded = 0; + char *result, *p; - for (totalSize = 1, i = 0; i < argc; i++) { - totalSize += strlen(argv[i]) + 1; - } - result = (char *) ckalloc((unsigned) totalSize); + /* Dispose of the empty result corner case first to simplify later code */ if (argc == 0) { - *result = '\0'; + result = (char *) ckalloc(1); + result[0] = '\0'; return result; } - for (p = result, i = 0; i < argc; i++) { - CONST char *element; - int length; + /* First allocate the result buffer at the size required */ + for (i = 0; i < argc; i++) { + bytesNeeded += strlen(argv[i]); + if (bytesNeeded < 0) { + Tcl_Panic("Tcl_Concat: max size of Tcl value exceeded"); + } + } + if (bytesNeeded + argc - 1 < 0) { /* - * Clip white space off the front and back of the string - * to generate a neater result, and ignore any empty - * elements. + * Panic test could be tighter, but not going to bother for + * this legacy routine. */ + Tcl_Panic("Tcl_Concat: max size of Tcl value exceeded"); + } + /* All element bytes + (argc - 1) spaces + 1 terminating NULL */ + result = (char *) ckalloc((unsigned) (bytesNeeded + argc)); + for (p = result, i = 0; i < argc; i++) { + int trim, elemLength; + const char *element; + element = argv[i]; - while (isspace(UCHAR(*element))) { /* INTL: ISO space. */ - element++; - } - for (length = strlen(element); - (length > 0) - && (isspace(UCHAR(element[length-1]))) /* INTL: ISO space. */ - && ((length < 2) || (element[length-2] != '\\')); - length--) { - /* Null loop body. */ - } - if (length == 0) { + elemLength = strlen(argv[i]); + + /* Trim away the leading whitespace */ + trim = TclTrimLeft(element, elemLength, CONCAT_WS, CONCAT_WS_SIZE); + element += trim; + elemLength -= trim; + + /* + * Trim away the trailing whitespace. Do not permit trimming + * to expose a final backslash character. + */ + + trim = TclTrimRight(element, elemLength, CONCAT_WS, CONCAT_WS_SIZE); + trim -= trim && (element[elemLength - trim - 1] == '\\'); + elemLength -= trim; + + /* If we're left with empty element after trimming, do nothing */ + if (elemLength == 0) { continue; } - memcpy((VOID *) p, (VOID *) element, (size_t) length); - p += length; - *p = ' '; - p++; - } - if (p != result) { - p[-1] = 0; - } else { - *p = 0; + + /* Append to the result with space if needed */ + if (needSpace) { + *p++ = ' '; + } + memcpy(p, element, (size_t) elemLength); + p += elemLength; + needSpace = 1; } + *p = '\0'; return result; } @@ -1020,8 +1733,8 @@ Tcl_Concat(argc, argv) * object with spaces between the original strings. * * Results: - * The return value is a new string object containing a concatenation - * of the strings in objv. Its ref count is zero. + * The return value is a new string object containing a concatenation of + * the strings in objv. Its ref count is zero. * * Side effects: * A new object is created. @@ -1030,27 +1743,29 @@ Tcl_Concat(argc, argv) */ Tcl_Obj * -Tcl_ConcatObj(objc, objv) - int objc; /* Number of objects to concatenate. */ - Tcl_Obj *CONST objv[]; /* Array of objects to concatenate. */ +Tcl_ConcatObj( + int objc, /* Number of objects to concatenate. */ + Tcl_Obj *CONST objv[]) /* Array of objects to concatenate. */ { - int allocSize, finalSize, length, elemLength, i; - char *p; - char *element; - char *concatStr; - Tcl_Obj *objPtr; + int i, elemLength, needSpace = 0, bytesNeeded = 0; + const char *element; + Tcl_Obj *objPtr, *resPtr; /* - * Check first to see if all the items are of list type. If so, - * we will concat them together as lists, and return a list object. - * This is only valid when the lists have no current string - * representation, since we don't know what the original type was. - * An original string rep may have lost some whitespace info when - * converted which could be important. + * Check first to see if all the items are of list type or empty. If so, + * we will concat them together as lists, and return a list object. This + * is only valid when the lists are in canonical form. */ + for (i = 0; i < objc; i++) { + int length; + objPtr = objv[i]; - if ((objPtr->typePtr != &tclListType) || (objPtr->bytes != NULL)) { + if (TclListObjIsCanonical(objPtr)) { + continue; + } + Tcl_GetStringFromObj(objPtr, &length); + if (length > 0) { break; } } @@ -1058,92 +1773,90 @@ Tcl_ConcatObj(objc, objv) Tcl_Obj **listv; int listc; - objPtr = Tcl_NewListObj(0, NULL); + resPtr = NULL; for (i = 0; i < objc; i++) { /* - * Tcl_ListObjAppendList could be used here, but this saves - * us a bit of type checking (since we've already done it) - * Use of INT_MAX tells us to always put the new stuff on - * the end. It will be set right in Tcl_ListObjReplace. + * Tcl_ListObjAppendList could be used here, but this saves us a + * bit of type checking (since we've already done it). Use of + * INT_MAX tells us to always put the new stuff on the end. It + * will be set right in Tcl_ListObjReplace. + * Note that all objs at this point are either lists or have an + * empty string rep. */ - Tcl_ListObjGetElements(NULL, objv[i], &listc, &listv); - Tcl_ListObjReplace(NULL, objPtr, INT_MAX, 0, listc, listv); - } - return objPtr; - } - allocSize = 0; - for (i = 0; i < objc; i++) { - objPtr = objv[i]; - element = Tcl_GetStringFromObj(objPtr, &length); - if ((element != NULL) && (length > 0)) { - allocSize += (length + 1); + objPtr = objv[i]; + if (objPtr->bytes && objPtr->length == 0) { + continue; + } + TclListObjGetElements(NULL, objPtr, &listc, &listv); + if (listc) { + if (resPtr) { + Tcl_ListObjReplace(NULL, resPtr, INT_MAX, 0, listc, listv); + } else { + resPtr = TclListObjCopy(NULL, objPtr); + } + } } - } - if (allocSize == 0) { - allocSize = 1; /* enough for the NULL byte at end */ + if (!resPtr) { + resPtr = Tcl_NewObj(); + } + return resPtr; } /* - * Allocate storage for the concatenated result. Note that allocSize - * is one more than the total number of characters, and so includes - * room for the terminating NULL byte. + * Something cannot be determined to be safe, so build the concatenation + * the slow way, using the string representations. */ - - concatStr = (char *) ckalloc((unsigned) allocSize); + /* First try to pre-allocate the size required */ + for (i = 0; i < objc; i++) { + element = TclGetStringFromObj(objv[i], &elemLength); + bytesNeeded += elemLength; + if (bytesNeeded < 0) { + break; + } + } /* - * Now concatenate the elements. Clip white space off the front and back - * to generate a neater result, and ignore any empty elements. Also put - * a null byte at the end. + * Does not matter if this fails, will simply try later to build up + * the string with each Append reallocating as needed with the usual + * string append algorithm. When that fails it will report the error. */ + TclNewObj(resPtr); + Tcl_AttemptSetObjLength(resPtr, bytesNeeded + objc - 1); + Tcl_SetObjLength(resPtr, 0); - finalSize = 0; - if (objc == 0) { - *concatStr = '\0'; - } else { - p = concatStr; - for (i = 0; i < objc; i++) { - objPtr = objv[i]; - element = Tcl_GetStringFromObj(objPtr, &elemLength); - while ((elemLength > 0) && (UCHAR(*element) < 127) - && isspace(UCHAR(*element))) { /* INTL: ISO C space. */ - element++; - elemLength--; - } + for (i = 0; i < objc; i++) { + int trim; + + element = TclGetStringFromObj(objv[i], &elemLength); - /* - * Trim trailing white space. But, be careful not to trim - * a space character if it is preceded by a backslash: in - * this case it could be significant. - */ + /* Trim away the leading whitespace */ + trim = TclTrimLeft(element, elemLength, CONCAT_WS, CONCAT_WS_SIZE); + element += trim; + elemLength -= trim; - while ((elemLength > 0) && (UCHAR(element[elemLength-1]) < 127) - && isspace(UCHAR(element[elemLength-1])) /* INTL: ISO C space. */ - && ((elemLength < 2) || (element[elemLength-2] != '\\'))) { - elemLength--; - } - if (elemLength == 0) { - continue; /* nothing left of this element */ - } - memcpy((VOID *) p, (VOID *) element, (size_t) elemLength); - p += elemLength; - *p = ' '; - p++; - finalSize += (elemLength + 1); - } - if (p != concatStr) { - p[-1] = 0; - finalSize -= 1; /* we overwrote the final ' ' */ - } else { - *p = 0; - } + /* + * Trim away the trailing whitespace. Do not permit trimming + * to expose a final backslash character. + */ + + trim = TclTrimRight(element, elemLength, CONCAT_WS, CONCAT_WS_SIZE); + trim -= trim && (element[elemLength - trim - 1] == '\\'); + elemLength -= trim; + + /* If we're left with empty element after trimming, do nothing */ + if (elemLength == 0) { + continue; + } + + /* Append to the result with space if needed */ + if (needSpace) { + Tcl_AppendToObj(resPtr, " ", 1); + } + Tcl_AppendToObj(resPtr, element, elemLength); + needSpace = 1; } - - TclNewObj(objPtr); - objPtr->bytes = concatStr; - objPtr->length = finalSize; - return objPtr; + return resPtr; } /* @@ -1154,10 +1867,9 @@ Tcl_ConcatObj(objc, objv) * See if a particular string matches a particular pattern. * * Results: - * The return value is 1 if string matches pattern, and - * 0 otherwise. The matching operation permits the following - * special characters in the pattern: *?\[] (see the manual - * entry for details on what these mean). + * The return value is 1 if string matches pattern, and 0 otherwise. The + * matching operation permits the following special characters in the + * pattern: *?\[] (see the manual entry for details on what these mean). * * Side effects: * None. @@ -1166,12 +1878,12 @@ Tcl_ConcatObj(objc, objv) */ int -Tcl_StringMatch(string, pattern) - CONST char *string; /* String. */ - CONST char *pattern; /* Pattern, which may contain special +Tcl_StringMatch( + CONST char *str, /* String. */ + CONST char *pattern) /* Pattern, which may contain special * characters. */ { - return Tcl_StringCaseMatch(string, pattern, 0); + return Tcl_StringCaseMatch(str, pattern, 0); } /* @@ -1179,14 +1891,13 @@ Tcl_StringMatch(string, pattern) * * Tcl_StringCaseMatch -- * - * See if a particular string matches a particular pattern. - * Allows case insensitivity. + * See if a particular string matches a particular pattern. Allows case + * insensitivity. * * Results: - * The return value is 1 if string matches pattern, and - * 0 otherwise. The matching operation permits the following - * special characters in the pattern: *?\[] (see the manual - * entry for details on what these mean). + * The return value is 1 if string matches pattern, and 0 otherwise. The + * matching operation permits the following special characters in the + * pattern: *?\[] (see the manual entry for details on what these mean). * * Side effects: * None. @@ -1195,127 +1906,133 @@ Tcl_StringMatch(string, pattern) */ int -Tcl_StringCaseMatch(string, pattern, nocase) - CONST char *string; /* String. */ - CONST char *pattern; /* Pattern, which may contain special +Tcl_StringCaseMatch( + CONST char *str, /* String. */ + CONST char *pattern, /* Pattern, which may contain special * characters. */ - int nocase; /* 0 for case sensitive, 1 for insensitive */ + int nocase) /* 0 for case sensitive, 1 for insensitive */ { int p, charLen; CONST char *pstart = pattern; Tcl_UniChar ch1, ch2; - + while (1) { p = *pattern; - + /* - * See if we're at the end of both the pattern and the string. If - * so, we succeeded. If we're at the end of the pattern but not at - * the end of the string, we failed. + * See if we're at the end of both the pattern and the string. If so, + * we succeeded. If we're at the end of the pattern but not at the end + * of the string, we failed. */ - + if (p == '\0') { - return (*string == '\0'); + return (*str == '\0'); } - if ((*string == '\0') && (p != '*')) { + if ((*str == '\0') && (p != '*')) { return 0; } /* - * Check for a "*" as the next pattern character. It matches - * any substring. We handle this by calling ourselves - * recursively for each postfix of string, until either we - * match or we reach the end of the string. + * Check for a "*" as the next pattern character. It matches any + * substring. We handle this by calling ourselves recursively for each + * postfix of string, until either we match or we reach the end of the + * string. */ - + if (p == '*') { /* * Skip all successive *'s in the pattern */ + while (*(++pattern) == '*') {} p = *pattern; if (p == '\0') { return 1; } + /* * This is a special case optimization for single-byte utf. */ + if (UCHAR(*pattern) < 0x80) { ch2 = (Tcl_UniChar) - (nocase ? tolower(UCHAR(*pattern)) : UCHAR(*pattern)); + (nocase ? tolower(UCHAR(*pattern)) : UCHAR(*pattern)); } else { Tcl_UtfToUniChar(pattern, &ch2); if (nocase) { ch2 = Tcl_UniCharToLower(ch2); } } + while (1) { /* * Optimization for matching - cruise through the string * quickly if the next char in the pattern isn't a special * character */ + if ((p != '[') && (p != '?') && (p != '\\')) { if (nocase) { - while (*string) { - charLen = TclUtfToUniChar(string, &ch1); + while (*str) { + charLen = TclUtfToUniChar(str, &ch1); if (ch2==ch1 || ch2==Tcl_UniCharToLower(ch1)) { break; } - string += charLen; + str += charLen; } } else { /* * There's no point in trying to make this code - * shorter, as the number of bytes you want to - * compare each time is non-constant. + * shorter, as the number of bytes you want to compare + * each time is non-constant. */ - while (*string) { - charLen = TclUtfToUniChar(string, &ch1); + + while (*str) { + charLen = TclUtfToUniChar(str, &ch1); if (ch2 == ch1) { break; } - string += charLen; + str += charLen; } } } - if (Tcl_StringCaseMatch(string, pattern, nocase)) { + if (Tcl_StringCaseMatch(str, pattern, nocase)) { return 1; } - if (*string == '\0') { + if (*str == '\0') { return 0; } - string += TclUtfToUniChar(string, &ch1); + str += TclUtfToUniChar(str, &ch1); } } /* - * Check for a "?" as the next pattern character. It matches - * any single character. + * Check for a "?" as the next pattern character. It matches any + * single character. */ if (p == '?') { pattern++; - string += TclUtfToUniChar(string, &ch1); + str += TclUtfToUniChar(str, &ch1); continue; } /* - * Check for a "[" as the next pattern character. It is followed - * by a list of characters that are acceptable, or by a range - * (two characters separated by "-"). + * Check for a "[" as the next pattern character. It is followed by a + * list of characters that are acceptable, or by a range (two + * characters separated by "-"). */ if (p == '[') { Tcl_UniChar startChar, endChar; pattern++; - if (UCHAR(*string) < 0x80) { + if (UCHAR(*str) < 0x80) { ch1 = (Tcl_UniChar) - (nocase ? tolower(UCHAR(*string)) : UCHAR(*string)); - string++; + (nocase ? tolower(UCHAR(*str)) : UCHAR(*str)); + str++; } else { - string += Tcl_UtfToUniChar(string, &ch1); + str += Tcl_UtfToUniChar(str, &ch1); if (nocase) { ch1 = Tcl_UniCharToLower(ch1); } @@ -1325,8 +2042,8 @@ Tcl_StringCaseMatch(string, pattern, nocase) return 0; } if (UCHAR(*pattern) < 0x80) { - startChar = (Tcl_UniChar) - (nocase ? tolower(UCHAR(*pattern)) : UCHAR(*pattern)); + startChar = (Tcl_UniChar) (nocase + ? tolower(UCHAR(*pattern)) : UCHAR(*pattern)); pattern++; } else { pattern += Tcl_UtfToUniChar(pattern, &startChar); @@ -1340,9 +2057,8 @@ Tcl_StringCaseMatch(string, pattern, nocase) return 0; } if (UCHAR(*pattern) < 0x80) { - endChar = (Tcl_UniChar) - (nocase ? tolower(UCHAR(*pattern)) - : UCHAR(*pattern)); + endChar = (Tcl_UniChar) (nocase + ? tolower(UCHAR(*pattern)) : UCHAR(*pattern)); pattern++; } else { pattern += Tcl_UtfToUniChar(pattern, &endChar); @@ -1374,8 +2090,8 @@ Tcl_StringCaseMatch(string, pattern, nocase) } /* - * If the next pattern character is '\', just strip off the '\' - * so we do exact matching on the character that follows. + * If the next pattern character is '\', just strip off the '\' so we + * do exact matching on the character that follows. */ if (p == '\\') { @@ -1386,11 +2102,11 @@ Tcl_StringCaseMatch(string, pattern, nocase) } /* - * There's no special character. Just make sure that the next - * bytes of each string match. + * There's no special character. Just make sure that the next bytes of + * each string match. */ - string += TclUtfToUniChar(string, &ch1); + str += TclUtfToUniChar(str, &ch1); pattern += TclUtfToUniChar(pattern, &ch2); if (nocase) { if (Tcl_UniCharToLower(ch1) != Tcl_UniCharToLower(ch2)) { @@ -1405,14 +2121,16 @@ Tcl_StringCaseMatch(string, pattern, nocase) /* *---------------------------------------------------------------------- * - * TclMatchIsTrivial -- + * TclByteArrayMatch -- * - * Test whether a particular glob pattern is a trivial pattern. - * (i.e. where matching is the same as equality testing). + * See if a particular string matches a particular pattern. Does not + * allow for case insensitivity. + * Parallels tclUtf.c:TclUniCharMatch, adjusted for char* and sans nocase. * * Results: - * A boolean indicating whether the pattern is free of all of the - * glob special chars. + * The return value is 1 if string matches pattern, and 0 otherwise. The + * matching operation permits the following special characters in the + * pattern: *?\[] (see the manual entry for details on what these mean). * * Side effects: * None. @@ -1421,32 +2139,222 @@ Tcl_StringCaseMatch(string, pattern, nocase) */ int -TclMatchIsTrivial(pattern) - CONST char *pattern; +TclByteArrayMatch( + const unsigned char *string, /* String. */ + int strLen, /* Length of String */ + const unsigned char *pattern, /* Pattern, which may contain special + * characters. */ + int ptnLen, /* Length of Pattern */ + int flags) { - CONST char *p = pattern; + const unsigned char *stringEnd, *patternEnd; + unsigned char p; + + stringEnd = string + strLen; + patternEnd = pattern + ptnLen; while (1) { - switch (*p++) { - case '\0': - return 1; - case '*': - case '?': - case '[': - case '\\': + /* + * See if we're at the end of both the pattern and the string. If so, + * we succeeded. If we're at the end of the pattern but not at the end + * of the string, we failed. + */ + + if (pattern == patternEnd) { + return (string == stringEnd); + } + p = *pattern; + if ((string == stringEnd) && (p != '*')) { + return 0; + } + + /* + * Check for a "*" as the next pattern character. It matches any + * substring. We handle this by skipping all the characters up to the + * next matching one in the pattern, and then calling ourselves + * recursively for each postfix of string, until either we match or we + * reach the end of the string. + */ + + if (p == '*') { + /* + * Skip all successive *'s in the pattern. + */ + + while ((++pattern < patternEnd) && (*pattern == '*')) { + /* empty body */ + } + if (pattern == patternEnd) { + return 1; + } + p = *pattern; + while (1) { + /* + * Optimization for matching - cruise through the string + * quickly if the next char in the pattern isn't a special + * character. + */ + + if ((p != '[') && (p != '?') && (p != '\\')) { + while ((string < stringEnd) && (p != *string)) { + string++; + } + } + if (TclByteArrayMatch(string, stringEnd - string, + pattern, patternEnd - pattern, 0)) { + return 1; + } + if (string == stringEnd) { + return 0; + } + string++; + } + } + + /* + * Check for a "?" as the next pattern character. It matches any + * single character. + */ + + if (p == '?') { + pattern++; + string++; + continue; + } + + /* + * Check for a "[" as the next pattern character. It is followed by a + * list of characters that are acceptable, or by a range (two + * characters separated by "-"). + */ + + if (p == '[') { + unsigned char ch1, startChar, endChar; + + pattern++; + ch1 = *string; + string++; + while (1) { + if ((*pattern == ']') || (pattern == patternEnd)) { + return 0; + } + startChar = *pattern; + pattern++; + if (*pattern == '-') { + pattern++; + if (pattern == patternEnd) { + return 0; + } + endChar = *pattern; + pattern++; + if (((startChar <= ch1) && (ch1 <= endChar)) + || ((endChar <= ch1) && (ch1 <= startChar))) { + /* + * Matches ranges of form [a-z] or [z-a]. + */ + break; + } + } else if (startChar == ch1) { + break; + } + } + while (*pattern != ']') { + if (pattern == patternEnd) { + pattern--; + break; + } + pattern++; + } + pattern++; + continue; + } + + /* + * If the next pattern character is '\', just strip off the '\' so we + * do exact matching on the character that follows. + */ + + if (p == '\\') { + if (++pattern == patternEnd) { + return 0; + } + } + + /* + * There's no special character. Just make sure that the next bytes of + * each string match. + */ + + if (*string != *pattern) { return 0; } + string++; + pattern++; } } /* *---------------------------------------------------------------------- * + * TclStringMatchObj -- + * + * See if a particular string matches a particular pattern. + * Allows case insensitivity. This is the generic multi-type handler + * for the various matching algorithms. + * + * Results: + * The return value is 1 if string matches pattern, and 0 otherwise. The + * matching operation permits the following special characters in the + * pattern: *?\[] (see the manual entry for details on what these mean). + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +int +TclStringMatchObj( + Tcl_Obj *strObj, /* string object. */ + Tcl_Obj *ptnObj, /* pattern object. */ + int flags) /* Only TCL_MATCH_NOCASE should be passed or 0. */ +{ + int match, length, plen; + + /* + * Promote based on the type of incoming object. + * XXX: Currently doesn't take advantage of exact-ness that + * XXX: TclReToGlob tells us about + trivial = nocase ? 0 : TclMatchIsTrivial(TclGetString(ptnObj)); + */ + + if (strObj->typePtr == &tclStringType) { + Tcl_UniChar *udata, *uptn; + + udata = Tcl_GetUnicodeFromObj(strObj, &length); + uptn = Tcl_GetUnicodeFromObj(ptnObj, &plen); + match = TclUniCharMatch(udata, length, uptn, plen, flags); + } else if ((strObj->typePtr == &tclByteArrayType) && !flags) { + unsigned char *data, *ptn; + + data = Tcl_GetByteArrayFromObj(strObj, &length); + ptn = Tcl_GetByteArrayFromObj(ptnObj, &plen); + match = TclByteArrayMatch(data, length, ptn, plen, 0); + } else { + match = Tcl_StringCaseMatch(TclGetString(strObj), + TclGetString(ptnObj), flags); + } + return match; +} + +/* + *---------------------------------------------------------------------- + * * Tcl_DStringInit -- * - * Initializes a dynamic string, discarding any previous contents - * of the string (Tcl_DStringFree should have been called already - * if the dynamic string was previously in use). + * Initializes a dynamic string, discarding any previous contents of the + * string (Tcl_DStringFree should have been called already if the dynamic + * string was previously in use). * * Results: * None. @@ -1458,8 +2366,8 @@ TclMatchIsTrivial(pattern) */ void -Tcl_DStringInit(dsPtr) - Tcl_DString *dsPtr; /* Pointer to structure for dynamic string. */ +Tcl_DStringInit( + Tcl_DString *dsPtr) /* Pointer to structure for dynamic string. */ { dsPtr->string = dsPtr->staticSpace; dsPtr->length = 0; @@ -1472,66 +2380,63 @@ Tcl_DStringInit(dsPtr) * * Tcl_DStringAppend -- * - * Append more characters to the current value of a dynamic string. + * Append more bytes to the current value of a dynamic string. * * Results: * The return value is a pointer to the dynamic string's new value. * * Side effects: - * Length bytes from string (or all of string if length is less - * than zero) are added to the current value of the string. Memory - * gets reallocated if needed to accomodate the string's new size. + * Length bytes from "bytes" (or all of "bytes" if length is less than + * zero) are added to the current value of the string. Memory gets + * reallocated if needed to accomodate the string's new size. * *---------------------------------------------------------------------- */ char * -Tcl_DStringAppend(dsPtr, string, length) - Tcl_DString *dsPtr; /* Structure describing dynamic string. */ - CONST char *string; /* String to append. If length is -1 then - * this must be null-terminated. */ - int length; /* Number of characters from string to - * append. If < 0, then append all of string, - * up to null at end. */ +Tcl_DStringAppend( + Tcl_DString *dsPtr, /* Structure describing dynamic string. */ + CONST char *bytes, /* String to append. If length is -1 then this + * must be null-terminated. */ + int length) /* Number of bytes from "bytes" to append. If + * < 0, then append all of bytes, up to null + * at end. */ { int newSize; char *dst; CONST char *end; if (length < 0) { - length = strlen(string); + length = strlen(bytes); } newSize = length + dsPtr->length; /* - * Allocate a larger buffer for the string if the current one isn't - * large enough. Allocate extra space in the new buffer so that there - * will be room to grow before we have to allocate again. + * Allocate a larger buffer for the string if the current one isn't large + * enough. Allocate extra space in the new buffer so that there will be + * room to grow before we have to allocate again. */ if (newSize >= dsPtr->spaceAvl) { dsPtr->spaceAvl = newSize * 2; if (dsPtr->string == dsPtr->staticSpace) { - char *newString; + char *newString = ckalloc((unsigned) dsPtr->spaceAvl); - newString = (char *) ckalloc((unsigned) dsPtr->spaceAvl); - memcpy((VOID *) newString, (VOID *) dsPtr->string, - (size_t) dsPtr->length); + memcpy(newString, dsPtr->string, (size_t) dsPtr->length); dsPtr->string = newString; } else { - dsPtr->string = (char *) ckrealloc((VOID *) dsPtr->string, + dsPtr->string = ckrealloc((void *) dsPtr->string, (size_t) dsPtr->spaceAvl); } } /* - * Copy the new string into the buffer at the end of the old - * one. + * Copy the new string into the buffer at the end of the old one. */ - for (dst = dsPtr->string + dsPtr->length, end = string+length; - string < end; string++, dst++) { - *dst = *string; + for (dst = dsPtr->string + dsPtr->length, end = bytes+length; + bytes < end; bytes++, dst++) { + *dst = *bytes; } *dst = '\0'; dsPtr->length += length; @@ -1549,62 +2454,67 @@ Tcl_DStringAppend(dsPtr, string, length) * The return value is a pointer to the dynamic string's new value. * * Side effects: - * String is reformatted as a list element and added to the current - * value of the string. Memory gets reallocated if needed to - * accomodate the string's new size. + * String is reformatted as a list element and added to the current value + * of the string. Memory gets reallocated if needed to accomodate the + * string's new size. * *---------------------------------------------------------------------- */ char * -Tcl_DStringAppendElement(dsPtr, string) - Tcl_DString *dsPtr; /* Structure describing dynamic string. */ - CONST char *string; /* String to append. Must be +Tcl_DStringAppendElement( + Tcl_DString *dsPtr, /* Structure describing dynamic string. */ + CONST char *element) /* String to append. Must be * null-terminated. */ { - int newSize, flags, strSize; - char *dst; - - strSize = ((string == NULL) ? 0 : strlen(string)); - newSize = Tcl_ScanCountedElement(string, strSize, &flags) - + dsPtr->length + 1; + char *dst = dsPtr->string + dsPtr->length; + int needSpace = TclNeedSpace(dsPtr->string, dst); + int flags = needSpace ? TCL_DONT_QUOTE_HASH : 0; + int newSize = dsPtr->length + needSpace + + TclScanElement(element, -1, &flags); /* - * Allocate a larger buffer for the string if the current one isn't - * large enough. Allocate extra space in the new buffer so that there - * will be room to grow before we have to allocate again. - * SPECIAL NOTE: must use memcpy, not strcpy, to copy the string - * to a larger buffer, since there may be embedded NULLs in the - * string in some cases. + * Allocate a larger buffer for the string if the current one isn't large + * enough. Allocate extra space in the new buffer so that there will be + * room to grow before we have to allocate again. SPECIAL NOTE: must use + * memcpy, not strcpy, to copy the string to a larger buffer, since there + * may be embedded NULLs in the string in some cases. */ if (newSize >= dsPtr->spaceAvl) { dsPtr->spaceAvl = newSize * 2; if (dsPtr->string == dsPtr->staticSpace) { - char *newString; + char *newString = ckalloc((unsigned) dsPtr->spaceAvl); - newString = (char *) ckalloc((unsigned) dsPtr->spaceAvl); - memcpy((VOID *) newString, (VOID *) dsPtr->string, - (size_t) dsPtr->length); + memcpy(newString, dsPtr->string, (size_t) dsPtr->length); dsPtr->string = newString; } else { - dsPtr->string = (char *) ckrealloc((VOID *) dsPtr->string, + dsPtr->string = (char *) ckrealloc((void *) dsPtr->string, (size_t) dsPtr->spaceAvl); } + dst = dsPtr->string + dsPtr->length; } /* - * Convert the new string to a list element and copy it into the - * buffer at the end, with a space, if needed. + * Convert the new string to a list element and copy it into the buffer at + * the end, with a space, if needed. */ - dst = dsPtr->string + dsPtr->length; - if (TclNeedSpace(dsPtr->string, dst)) { + if (needSpace) { *dst = ' '; dst++; dsPtr->length++; + + /* + * If we need a space to separate this element from preceding stuff, + * then this element will not lead a list, and need not have it's + * leading '#' quoted. + */ + + flags |= TCL_DONT_QUOTE_HASH; } - dsPtr->length += Tcl_ConvertCountedElement(string, strSize, dst, flags); + dsPtr->length += TclConvertElement(element, -1, dst, flags); + dsPtr->string[dsPtr->length] = '\0'; return dsPtr->string; } @@ -1613,25 +2523,24 @@ Tcl_DStringAppendElement(dsPtr, string) * * Tcl_DStringSetLength -- * - * Change the length of a dynamic string. This can cause the - * string to either grow or shrink, depending on the value of - * length. + * Change the length of a dynamic string. This can cause the string to + * either grow or shrink, depending on the value of length. * * Results: * None. * * Side effects: - * The length of dsPtr is changed to length and a null byte is - * stored at that position in the string. If length is larger - * than the space allocated for dsPtr, then a panic occurs. + * The length of dsPtr is changed to length and a null byte is stored at + * that position in the string. If length is larger than the space + * allocated for dsPtr, then a panic occurs. * *---------------------------------------------------------------------- */ void -Tcl_DStringSetLength(dsPtr, length) - Tcl_DString *dsPtr; /* Structure describing dynamic string. */ - int length; /* New length for dynamic string. */ +Tcl_DStringSetLength( + Tcl_DString *dsPtr, /* Structure describing dynamic string. */ + int length) /* New length for dynamic string. */ { int newsize; @@ -1640,15 +2549,15 @@ Tcl_DStringSetLength(dsPtr, length) } if (length >= dsPtr->spaceAvl) { /* - * There are two interesting cases here. In the first case, the user - * may be trying to allocate a large buffer of a specific size. It + * There are two interesting cases here. In the first case, the user + * may be trying to allocate a large buffer of a specific size. It * would be wasteful to overallocate that buffer, so we just allocate - * enough for the requested size plus the trailing null byte. In the + * enough for the requested size plus the trailing null byte. In the * second case, we are growing the buffer incrementally, so we need - * behavior similar to Tcl_DStringAppend. The requested length will - * usually be a small delta above the current spaceAvl, so we'll end up - * doubling the old size. This won't grow the buffer quite as quickly, - * but it should be close enough. + * behavior similar to Tcl_DStringAppend. The requested length will + * usually be a small delta above the current spaceAvl, so we'll end + * up doubling the old size. This won't grow the buffer quite as + * quickly, but it should be close enough. */ newsize = dsPtr->spaceAvl * 2; @@ -1658,14 +2567,12 @@ Tcl_DStringSetLength(dsPtr, length) dsPtr->spaceAvl = length + 1; } if (dsPtr->string == dsPtr->staticSpace) { - char *newString; + char *newString = ckalloc((unsigned) dsPtr->spaceAvl); - newString = (char *) ckalloc((unsigned) dsPtr->spaceAvl); - memcpy((VOID *) newString, (VOID *) dsPtr->string, - (size_t) dsPtr->length); + memcpy(newString, dsPtr->string, (size_t) dsPtr->length); dsPtr->string = newString; } else { - dsPtr->string = (char *) ckrealloc((VOID *) dsPtr->string, + dsPtr->string = (char *) ckrealloc((void *) dsPtr->string, (size_t) dsPtr->spaceAvl); } } @@ -1678,21 +2585,22 @@ Tcl_DStringSetLength(dsPtr, length) * * Tcl_DStringFree -- * - * Frees up any memory allocated for the dynamic string and - * reinitializes the string to an empty state. + * Frees up any memory allocated for the dynamic string and reinitializes + * the string to an empty state. * * Results: * None. * * Side effects: - * The previous contents of the dynamic string are lost, and - * the new value is an empty string. + * The previous contents of the dynamic string are lost, and the new + * value is an empty string. * - *---------------------------------------------------------------------- */ + *---------------------------------------------------------------------- + */ void -Tcl_DStringFree(dsPtr) - Tcl_DString *dsPtr; /* Structure describing dynamic string. */ +Tcl_DStringFree( + Tcl_DString *dsPtr) /* Structure describing dynamic string. */ { if (dsPtr->string != dsPtr->staticSpace) { ckfree(dsPtr->string); @@ -1708,29 +2616,28 @@ Tcl_DStringFree(dsPtr) * * Tcl_DStringResult -- * - * This procedure moves the value of a dynamic string into an - * interpreter as its string result. Afterwards, the dynamic string - * is reset to an empty string. + * This function moves the value of a dynamic string into an interpreter + * as its string result. Afterwards, the dynamic string is reset to an + * empty string. * * Results: * None. * * Side effects: - * The string is "moved" to interp's result, and any existing - * string result for interp is freed. dsPtr is reinitialized to - * an empty string. + * The string is "moved" to interp's result, and any existing string + * result for interp is freed. dsPtr is reinitialized to an empty string. * *---------------------------------------------------------------------- */ void -Tcl_DStringResult(interp, dsPtr) - Tcl_Interp *interp; /* Interpreter whose result is to be reset. */ - Tcl_DString *dsPtr; /* Dynamic string that is to become the +Tcl_DStringResult( + Tcl_Interp *interp, /* Interpreter whose result is to be reset. */ + Tcl_DString *dsPtr) /* Dynamic string that is to become the * result of interp. */ { Tcl_ResetResult(interp); - + if (dsPtr->string != dsPtr->staticSpace) { interp->result = dsPtr->string; interp->freeProc = TCL_DYNAMIC; @@ -1740,7 +2647,7 @@ Tcl_DStringResult(interp, dsPtr) } else { Tcl_SetResult(interp, dsPtr->string, TCL_VOLATILE); } - + dsPtr->string = dsPtr->staticSpace; dsPtr->length = 0; dsPtr->spaceAvl = TCL_DSTRING_STATIC_SIZE; @@ -1752,14 +2659,14 @@ Tcl_DStringResult(interp, dsPtr) * * Tcl_DStringGetResult -- * - * This procedure moves an interpreter's result into a dynamic string. + * This function moves an interpreter's result into a dynamic string. * * Results: * None. * * Side effects: - * The interpreter's string result is cleared, and the previous - * contents of dsPtr are freed. + * The interpreter's string result is cleared, and the previous contents + * of dsPtr are freed. * * If the string result is empty, the object result is moved to the * string result, then the object result is reset. @@ -1768,26 +2675,23 @@ Tcl_DStringResult(interp, dsPtr) */ void -Tcl_DStringGetResult(interp, dsPtr) - Tcl_Interp *interp; /* Interpreter whose result is to be reset. */ - Tcl_DString *dsPtr; /* Dynamic string that is to become the - * result of interp. */ +Tcl_DStringGetResult( + Tcl_Interp *interp, /* Interpreter whose result is to be reset. */ + Tcl_DString *dsPtr) /* Dynamic string that is to become the result + * of interp. */ { Interp *iPtr = (Interp *) interp; - + if (dsPtr->string != dsPtr->staticSpace) { ckfree(dsPtr->string); } /* - * If the string result is empty, move the object result to the - * string result, then reset the object result. + * If the string result is empty, move the object result to the string + * result, then reset the object result. */ - if (*(iPtr->result) == 0) { - Tcl_SetResult(interp, TclGetString(Tcl_GetObjResult(interp)), - TCL_VOLATILE); - } + (void) Tcl_GetStringResult(interp); dsPtr->length = strlen(iPtr->result); if (iPtr->freeProc != NULL) { @@ -1796,7 +2700,7 @@ Tcl_DStringGetResult(interp, dsPtr) dsPtr->spaceAvl = dsPtr->length+1; } else { dsPtr->string = (char *) ckalloc((unsigned) (dsPtr->length+1)); - strcpy(dsPtr->string, iPtr->result); + memcpy(dsPtr->string, iPtr->result, (unsigned) dsPtr->length+1); (*iPtr->freeProc)(iPtr->result); } dsPtr->spaceAvl = dsPtr->length+1; @@ -1809,9 +2713,9 @@ Tcl_DStringGetResult(interp, dsPtr) dsPtr->string = (char *) ckalloc((unsigned) (dsPtr->length + 1)); dsPtr->spaceAvl = dsPtr->length + 1; } - strcpy(dsPtr->string, iPtr->result); + memcpy(dsPtr->string, iPtr->result, (unsigned) dsPtr->length+1); } - + iPtr->result = iPtr->resultSpace; iPtr->resultSpace[0] = 0; } @@ -1821,9 +2725,9 @@ Tcl_DStringGetResult(interp, dsPtr) * * Tcl_DStringStartSublist -- * - * This procedure adds the necessary information to a dynamic - * string (e.g. " {" to start a sublist. Future element - * appends will be in the sublist rather than the main list. + * This function adds the necessary information to a dynamic string + * (e.g. " {") to start a sublist. Future element appends will be in the + * sublist rather than the main list. * * Results: * None. @@ -1835,8 +2739,8 @@ Tcl_DStringGetResult(interp, dsPtr) */ void -Tcl_DStringStartSublist(dsPtr) - Tcl_DString *dsPtr; /* Dynamic string. */ +Tcl_DStringStartSublist( + Tcl_DString *dsPtr) /* Dynamic string. */ { if (TclNeedSpace(dsPtr->string, dsPtr->string + dsPtr->length)) { Tcl_DStringAppend(dsPtr, " {", -1); @@ -1850,10 +2754,9 @@ Tcl_DStringStartSublist(dsPtr) * * Tcl_DStringEndSublist -- * - * This procedure adds the necessary characters to a dynamic - * string to end a sublist (e.g. "}"). Future element appends - * will be in the enclosing (sub)list rather than the current - * sublist. + * This function adds the necessary characters to a dynamic string to end + * a sublist (e.g. "}"). Future element appends will be in the enclosing + * (sub)list rather than the current sublist. * * Results: * None. @@ -1865,8 +2768,8 @@ Tcl_DStringStartSublist(dsPtr) */ void -Tcl_DStringEndSublist(dsPtr) - Tcl_DString *dsPtr; /* Dynamic string. */ +Tcl_DStringEndSublist( + Tcl_DString *dsPtr) /* Dynamic string. */ { Tcl_DStringAppend(dsPtr, "}", -1); } @@ -1876,14 +2779,14 @@ Tcl_DStringEndSublist(dsPtr) * * Tcl_PrintDouble -- * - * Given a floating-point value, this procedure converts it to - * an ASCII string using. + * Given a floating-point value, this function converts it to an ASCII + * string using. * * Results: - * The ASCII equivalent of "value" is written at "dst". It is - * written using the current precision, and it is guaranteed to - * contain a decimal point or exponent, so that it looks like - * a floating-point value and not an integer. + * The ASCII equivalent of "value" is written at "dst". It is written + * using the current precision, and it is guaranteed to contain a decimal + * point or exponent, so that it looks like a floating-point value and + * not an integer. * * Side effects: * None. @@ -1892,43 +2795,158 @@ Tcl_DStringEndSublist(dsPtr) */ void -Tcl_PrintDouble(interp, value, dst) - Tcl_Interp *interp; /* Interpreter whose tcl_precision - * variable used to be used to control - * printing. It's ignored now. */ - double value; /* Value to print as string. */ - char *dst; /* Where to store converted value; - * must have at least TCL_DOUBLE_SPACE - * characters. */ +Tcl_PrintDouble( + Tcl_Interp *interp, /* Interpreter whose tcl_precision variable + * used to be used to control printing. It's + * ignored now. */ + double value, /* Value to print as string. */ + char *dst) /* Where to store converted value; must have + * at least TCL_DOUBLE_SPACE characters. */ { char *p, c; - Tcl_UniChar ch; + int exponent; + int signum; + char* digits; + char* end; - Tcl_MutexLock(&precisionMutex); - sprintf(dst, precisionFormat, value); - Tcl_MutexUnlock(&precisionMutex); + int *precisionPtr = Tcl_GetThreadData(&precisionKey, (int)sizeof(int)); /* - * If the ASCII result looks like an integer, add ".0" so that it - * doesn't look like an integer anymore. This prevents floating-point - * values from being converted to integers unintentionally. - * Check for ASCII specifically to speed up the function. - */ + * Handle NaN. + */ - for (p = dst; *p != 0; ) { - if (UCHAR(*p) < 0x80) { - c = *p++; - } else { - p += Tcl_UtfToUniChar(p, &ch); - c = UCHAR(ch); + if (TclIsNaN(value)) { + TclFormatNaN(value, dst); + return; } - if ((c == '.') || isalpha(UCHAR(c))) { /* INTL: ISO only. */ + + /* + * Handle infinities. + */ + + if (TclIsInfinite(value)) { + /* + * Remember to copy the terminating NUL too. + */ + + if (value < 0) { + memcpy(dst, "-Inf", 5); + } else { + memcpy(dst, "Inf", 4); + } return; } + + /* + * Ordinary (normal and denormal) values. + */ + + if (*precisionPtr == 0) { + digits = TclDoubleDigits(value, -1, TCL_DD_SHORTEST, + &exponent, &signum, &end); + } else { + /* + * There are at least two possible interpretations for tcl_precision. + * + * The first is, "choose the decimal representation having + * $tcl_precision digits of significance that is nearest to the + * given number, breaking ties by rounding to even, and then + * trimming trailing zeros." This gives the greatest possible + * precision in the decimal string, but offers the anomaly that + * [expr 0.1] will be "0.10000000000000001". + * + * The second is "choose the decimal representation having at + * most $tcl_precision digits of significance that is nearest + * to the given number. If no such representation converts + * exactly to the given number, choose the one that is closest, + * breaking ties by rounding to even. If more than one such + * representation converts exactly to the given number, choose + * the shortest, breaking ties in favour of the nearest, breaking + * remaining ties in favour of the one ending in an even digit." + * + * Tcl 8.4 implements the first of these, which gives rise to + * anomalies in formatting: + * + * % expr 0.1 + * 0.10000000000000001 + * % expr 0.01 + * 0.01 + * % expr 1e-7 + * 9.9999999999999995e-08 + * + * For human readability, it appears better to choose the second rule, + * and let [expr 0.1] return 0.1. But for 8.4 compatibility, we + * prefer the first (the recommended zero value for tcl_precision + * avoids the problem entirely). + * + * Uncomment TCL_DD_SHORTEN_FLAG in the next call to prefer the + * method that allows floating point values to be shortened if + * it can be done without loss of precision. + */ + + digits = TclDoubleDigits(value, *precisionPtr, + TCL_DD_E_FORMAT /* | TCL_DD_SHORTEN_FLAG */, + &exponent, &signum, &end); + } + if (signum) { + *dst++ = '-'; + } + p = digits; + if (exponent < -4 || exponent > 16) { + /* + * E format for numbers < 1e-3 or >= 1e17. + */ + + *dst++ = *p++; + c = *p; + if (c != '\0') { + *dst++ = '.'; + while (c != '\0') { + *dst++ = c; + c = *++p; + } + } + /* + * Tcl 8.4 appears to format with at least a two-digit exponent; \ + * preserve that behaviour when tcl_precision != 0 + */ + if (*precisionPtr == 0) { + sprintf(dst, "e%+d", exponent); + } else { + sprintf(dst, "e%+03d", exponent); + } + } else { + /* + * F format for others. + */ + + if (exponent < 0) { + *dst++ = '0'; + } + c = *p; + while (exponent-- >= 0) { + if (c != '\0') { + *dst++ = c; + c = *++p; + } else { + *dst++ = '0'; + } + } + *dst++ = '.'; + if (c == '\0') { + *dst++ = '0'; + } else { + while (++exponent < -1) { + *dst++ = '0'; + } + while (c != '\0') { + *dst++ = c; + c = *++p; + } + } + *dst++ = '\0'; } - p[0] = '.'; - p[1] = '0'; - p[2] = 0; + ckfree(digits); } /* @@ -1936,33 +2954,33 @@ Tcl_PrintDouble(interp, value, dst) * * TclPrecTraceProc -- * - * This procedure is invoked whenever the variable "tcl_precision" - * is written. + * This function is invoked whenever the variable "tcl_precision" is + * written. * * Results: - * Returns NULL if all went well, or an error message if the - * new value for the variable doesn't make sense. + * Returns NULL if all went well, or an error message if the new value + * for the variable doesn't make sense. * * Side effects: - * If the new value doesn't make sense then this procedure - * undoes the effect of the variable modification. Otherwise - * it modifies the format string that's used by Tcl_PrintDouble. + * If the new value doesn't make sense then this function undoes the + * effect of the variable modification. Otherwise it modifies the format + * string that's used by Tcl_PrintDouble. * *---------------------------------------------------------------------- */ /* ARGSUSED */ char * -TclPrecTraceProc(clientData, interp, name1, name2, flags) - ClientData clientData; /* Not used. */ - Tcl_Interp *interp; /* Interpreter containing variable. */ - CONST char *name1; /* Name of variable. */ - CONST char *name2; /* Second part of variable name. */ - int flags; /* Information about what happened. */ +TclPrecTraceProc( + ClientData clientData, /* Not used. */ + Tcl_Interp *interp, /* Interpreter containing variable. */ + CONST char *name1, /* Name of variable. */ + CONST char *name2, /* Second part of variable name. */ + int flags) /* Information about what happened. */ { - CONST char *value; - char *end; + Tcl_Obj* value; int prec; + int *precisionPtr = Tcl_GetThreadData(&precisionKey, (int) sizeof(int)); /* * If the variable is unset, then recreate the trace. @@ -1974,54 +2992,39 @@ TclPrecTraceProc(clientData, interp, name1, name2, flags) TCL_GLOBAL_ONLY|TCL_TRACE_READS|TCL_TRACE_WRITES |TCL_TRACE_UNSETS, TclPrecTraceProc, clientData); } - return (char *) NULL; + return NULL; } /* - * When the variable is read, reset its value from our shared - * value. This is needed in case the variable was modified in - * some other interpreter so that this interpreter's value is - * out of date. + * When the variable is read, reset its value from our shared value. This + * is needed in case the variable was modified in some other interpreter + * so that this interpreter's value is out of date. */ - Tcl_MutexLock(&precisionMutex); if (flags & TCL_TRACE_READS) { - Tcl_SetVar2(interp, name1, name2, precisionString, + Tcl_SetVar2Ex(interp, name1, name2, Tcl_NewIntObj(*precisionPtr), flags & TCL_GLOBAL_ONLY); - Tcl_MutexUnlock(&precisionMutex); - return (char *) NULL; + return NULL; } /* - * The variable is being written. Check the new value and disallow - * it if it isn't reasonable or if this is a safe interpreter (we - * don't want safe interpreters messing up the precision of other - * interpreters). + * The variable is being written. Check the new value and disallow it if + * it isn't reasonable or if this is a safe interpreter (we don't want + * safe interpreters messing up the precision of other interpreters). */ if (Tcl_IsSafe(interp)) { - Tcl_SetVar2(interp, name1, name2, precisionString, - flags & TCL_GLOBAL_ONLY); - Tcl_MutexUnlock(&precisionMutex); return "can't modify precision from a safe interpreter"; } - value = Tcl_GetVar2(interp, name1, name2, flags & TCL_GLOBAL_ONLY); - if (value == NULL) { - value = ""; - } - prec = strtoul(value, &end, 10); - if ((prec <= 0) || (prec > TCL_MAX_PREC) || (prec > 100) || - (end == value) || (*end != 0)) { - Tcl_SetVar2(interp, name1, name2, precisionString, - flags & TCL_GLOBAL_ONLY); - Tcl_MutexUnlock(&precisionMutex); + value = Tcl_GetVar2Ex(interp, name1, name2, flags & TCL_GLOBAL_ONLY); + if (value == NULL + || Tcl_GetIntFromObj((Tcl_Interp*) NULL, value, &prec) != TCL_OK + || prec < 0 || prec > TCL_MAX_PREC) { return "improper value for precision"; } - TclFormatInt(precisionString, prec); - sprintf(precisionFormat, "%%.%dg", prec); - Tcl_MutexUnlock(&precisionMutex); - return (char *) NULL; + *precisionPtr = prec; + return NULL; } /* @@ -2029,9 +3032,8 @@ TclPrecTraceProc(clientData, interp, name1, name2, flags) * * TclNeedSpace -- * - * This procedure checks to see whether it is appropriate to - * add a space before appending a new list element to an - * existing string. + * This function checks to see whether it is appropriate to add a space + * before appending a new list element to an existing string. * * Results: * The return value is 1 if a space is appropriate, 0 otherwise. @@ -2043,24 +3045,25 @@ TclPrecTraceProc(clientData, interp, name1, name2, flags) */ int -TclNeedSpace(start, end) - CONST char *start; /* First character in string. */ - CONST char *end; /* End of string (place where space will - * be added, if appropriate). */ +TclNeedSpace( + CONST char *start, /* First character in string. */ + CONST char *end) /* End of string (place where space will be + * added, if appropriate). */ { /* - * A space is needed unless either + * A space is needed unless either: * (a) we're at the start of the string, or */ + if (end == start) { return 0; } /* - * (b) we're at the start of a nested list-element, quoted with an - * open curly brace; we can be nested arbitrarily deep, so long - * as the first curly brace starts an element, so backtrack over - * open curly braces that are trailing characters of the string; and + * (b) we're at the start of a nested list-element, quoted with an open + * curly brace; we can be nested arbitrarily deep, so long as the + * first curly brace starts an element, so backtrack over open curly + * braces that are trailing characters of the string; and */ end = Tcl_UtfPrev(end, start); @@ -2073,39 +3076,39 @@ TclNeedSpace(start, end) /* * (c) the trailing character of the string is already a list-element - * separator (according to TclFindElement); that is, one of these - * characters: - * \u0009 \t TAB - * \u000A \n NEWLINE - * \u000B \v VERTICAL TAB - * \u000C \f FORM FEED - * \u000D \r CARRIAGE RETURN - * \u0020 SPACE - * with the condition that the penultimate character is not a - * backslash. + * separator (according to TclFindElement); that is, one of these + * characters: + * \u0009 \t TAB + * \u000A \n NEWLINE + * \u000B \v VERTICAL TAB + * \u000C \f FORM FEED + * \u000D \r CARRIAGE RETURN + * \u0020 SPACE + * with the condition that the penultimate character is not a + * backslash. */ if (*end > 0x20) { /* - * Performance tweak. All ASCII spaces are <= 0x20. So get - * a quick answer for most characters before comparing against - * all spaces in the switch below. + * Performance tweak. All ASCII spaces are <= 0x20. So get a quick + * answer for most characters before comparing against all spaces in + * the switch below. * - * NOTE: Remove this if other Unicode spaces ever get accepted - * as list-element separators. + * NOTE: Remove this if other Unicode spaces ever get accepted as + * list-element separators. */ return 1; } switch (*end) { - case ' ': - case '\t': - case '\n': - case '\r': - case '\v': - case '\f': - if ((end == start) || (end[-1] != '\\')) { - return 0; - } + case ' ': + case '\t': + case '\n': + case '\r': + case '\v': + case '\f': + if ((end == start) || (end[-1] != '\\')) { + return 0; + } } return 1; } @@ -2120,7 +3123,9 @@ TclNeedSpace(start, end) * inserted at the start of the buffer. A null character is inserted at * the end of the formatted characters. It is the caller's * responsibility to ensure that enough storage is available. This - * procedure has the effect of sprintf(buffer, "%d", n) but is faster. + * procedure has the effect of sprintf(buffer, "%ld", n) but is faster + * as proven in benchmarks. This is key to UpdateStringOfInt, which + * is a common path for a lot of code (e.g. int-indexed arrays). * * Results: * An integer representing the number of characters formatted, not @@ -2162,8 +3167,7 @@ TclFormatInt(buffer, n) intVal = -n; /* [Bug 3390638] Workaround for*/ if (n == -n || intVal == n) { /* broken compiler optimizers. */ - sprintf(buffer, "%ld", n); - return strlen(buffer); + return sprintf(buffer, "%ld", n); } /* @@ -2199,133 +3203,119 @@ TclFormatInt(buffer, n) /* *---------------------------------------------------------------------- * - * TclLooksLikeInt -- - * - * This procedure decides whether the leading characters of a - * string look like an integer or something else (such as a - * floating-point number or string). - * - * Results: - * The return value is 1 if the leading characters of p look - * like a valid Tcl integer. If they look like a floating-point - * number (e.g. "e01" or "2.4"), or if they don't look like a - * number at all, then 0 is returned. - * - * Side effects: - * None. - * - *---------------------------------------------------------------------- - */ - -int -TclLooksLikeInt(bytes, length) - register CONST char *bytes; /* Points to first byte of the string. */ - int length; /* Number of bytes in the string. If < 0 - * bytes up to the first null byte are - * considered (if they may appear in an - * integer). */ -{ - register CONST char *p; - - if ((bytes == NULL) && (length > 0)) { - Tcl_Panic("TclLooksLikeInt: cannot scan %d bytes from NULL", length); - } - - if (length < 0) { - length = (bytes? strlen(bytes) : 0); - } - - p = bytes; - while (length && isspace(UCHAR(*p))) { /* INTL: ISO space. */ - length--; p++; - } - if (length == 0) { - return 0; - } - if ((*p == '+') || (*p == '-')) { - p++; length--; - } - - return (0 != TclParseInteger(p, length)); -} - -/* - *---------------------------------------------------------------------- - * * TclGetIntForIndex -- * - * This procedure returns an integer corresponding to the list index - * held in a Tcl object. The Tcl object's value is expected to be - * either an integer or a string of the form "end([+-]integer)?". + * This function returns an integer corresponding to the list index held + * in a Tcl object. The Tcl object's value is expected to be in the + * format integer([+-]integer)? or the format end([+-]integer)?. * * Results: * The return value is normally TCL_OK, which means that the index was - * successfully stored into the location referenced by "indexPtr". If - * the Tcl object referenced by "objPtr" has the value "end", the - * value stored is "endValue". If "objPtr"s values is not of the form - * "end([+-]integer)?" and - * can not be converted to an integer, TCL_ERROR is returned and, if - * "interp" is non-NULL, an error message is left in the interpreter's - * result object. + * successfully stored into the location referenced by "indexPtr". If the + * Tcl object referenced by "objPtr" has the value "end", the value + * stored is "endValue". If "objPtr"s values is not of one of the + * expected formats, TCL_ERROR is returned and, if "interp" is non-NULL, + * an error message is left in the interpreter's result object. * * Side effects: - * The object referenced by "objPtr" might be converted to an - * integer, wide integer, or end-based-index object. + * The object referenced by "objPtr" might be converted to an integer, + * wide integer, or end-based-index object. * *---------------------------------------------------------------------- */ int -TclGetIntForIndex(interp, objPtr, endValue, indexPtr) - Tcl_Interp *interp; /* Interpreter to use for error reporting. - * If NULL, then no error message is left - * after errors. */ - Tcl_Obj *objPtr; /* Points to an object containing either - * "end" or an integer. */ - int endValue; /* The value to be stored at "indexPtr" if +TclGetIntForIndex( + Tcl_Interp *interp, /* Interpreter to use for error reporting. If + * NULL, then no error message is left after + * errors. */ + Tcl_Obj *objPtr, /* Points to an object containing either "end" + * or an integer. */ + int endValue, /* The value to be stored at "indexPtr" if * "objPtr" holds "end". */ - int *indexPtr; /* Location filled in with an integer + int *indexPtr) /* Location filled in with an integer * representing an index. */ { - if (Tcl_GetIntFromObj(NULL, objPtr, indexPtr) == TCL_OK) { + int length; + char *opPtr, *bytes; + + if (TclGetIntFromObj(NULL, objPtr, indexPtr) == TCL_OK) { return TCL_OK; } if (SetEndOffsetFromAny(NULL, objPtr) == TCL_OK) { /* - * If the object is already an offset from the end of the - * list, or can be converted to one, use it. + * If the object is already an offset from the end of the list, or can + * be converted to one, use it. */ *indexPtr = endValue + objPtr->internalRep.longValue; + return TCL_OK; + } + + bytes = TclGetStringFromObj(objPtr, &length); + + /* + * Leading whitespace is acceptable in an index. + */ + + while (length && TclIsSpaceProc(*bytes)) { + bytes++; + length--; + } + + if (TclParseNumber(NULL, NULL, NULL, bytes, length, (const char **)&opPtr, + TCL_PARSE_INTEGER_ONLY | TCL_PARSE_NO_WHITESPACE) == TCL_OK) { + int code, first, second; + char savedOp = *opPtr; + + if ((savedOp != '+') && (savedOp != '-')) { + goto parseError; + } + if (TclIsSpaceProc(opPtr[1])) { + goto parseError; + } + *opPtr = '\0'; + code = Tcl_GetInt(interp, bytes, &first); + *opPtr = savedOp; + if (code == TCL_ERROR) { + goto parseError; + } + if (TCL_ERROR == Tcl_GetInt(interp, opPtr+1, &second)) { + goto parseError; + } + if (savedOp == '+') { + *indexPtr = first + second; + } else { + *indexPtr = first - second; + } + return TCL_OK; + } + + /* + * Report a parse error. + */ + + parseError: + if (interp != NULL) { + char *bytes = Tcl_GetString(objPtr); - } else { /* - * Report a parse error. + * The result might not be empty; this resets it which should be both + * a cheap operation, and of little problem because this is an + * error-generation path anyway. */ - if (interp != NULL) { - char *bytes = Tcl_GetString(objPtr); - /* - * The result might not be empty; this resets it which - * should be both a cheap operation, and of little problem - * because this is an error-generation path anyway. - */ - Tcl_ResetResult(interp); - Tcl_AppendStringsToObj(Tcl_GetObjResult(interp), - "bad index \"", bytes, - "\": must be integer or end?-integer?", - (char *) NULL); - if (!strncmp(bytes, "end-", 3)) { - bytes += 3; - } - TclCheckBadOctal(interp, bytes); + Tcl_ResetResult(interp); + Tcl_AppendResult(interp, "bad index \"", bytes, + "\": must be integer?[+-]integer? or end?[+-]integer?", NULL); + if (!strncmp(bytes, "end-", 4)) { + bytes += 4; } - - return TCL_ERROR; + TclCheckBadOctal(interp, bytes); } - - return TCL_OK; + + return TCL_ERROR; } /* @@ -2342,16 +3332,15 @@ TclGetIntForIndex(interp, objPtr, endValue, indexPtr) * Side effects: * Stores a valid string in the object's string rep. * - * This procedure does NOT free any earlier string rep. If it is - * called on an object that already has a valid string rep, it will - * leak memory. + * This function does NOT free any earlier string rep. If it is called on an + * object that already has a valid string rep, it will leak memory. * *---------------------------------------------------------------------- */ static void -UpdateStringOfEndOffset(objPtr) - register Tcl_Obj* objPtr; +UpdateStringOfEndOffset( + register Tcl_Obj* objPtr) { char buffer[TCL_INTEGER_SPACE + sizeof("end") + 1]; register int len; @@ -2362,8 +3351,8 @@ UpdateStringOfEndOffset(objPtr) buffer[len++] = '-'; len += TclFormatInt(buffer+len, -(objPtr->internalRep.longValue)); } - objPtr->bytes = ckalloc((unsigned) (len+1)); - strcpy(objPtr->bytes, buffer); + objPtr->bytes = ckalloc((unsigned) len+1); + memcpy(objPtr->bytes, buffer, (unsigned) len+1); objPtr->length = len; } @@ -2372,100 +3361,104 @@ UpdateStringOfEndOffset(objPtr) * * SetEndOffsetFromAny -- * - * Look for a string of the form "end-offset" and convert it - * to an internal representation holding the offset. + * Look for a string of the form "end[+-]offset" and convert it to an + * internal representation holding the offset. * * Results: * Returns TCL_OK if ok, TCL_ERROR if the string was badly formed. * * Side effects: - * If interp is not NULL, stores an error message in the - * interpreter result. + * If interp is not NULL, stores an error message in the interpreter + * result. * *---------------------------------------------------------------------- */ static int -SetEndOffsetFromAny(interp, objPtr) - Tcl_Interp* interp; /* Tcl interpreter or NULL */ - Tcl_Obj* objPtr; /* Pointer to the object to parse */ +SetEndOffsetFromAny( + Tcl_Interp *interp, /* Tcl interpreter or NULL */ + Tcl_Obj *objPtr) /* Pointer to the object to parse */ { int offset; /* Offset in the "end-offset" expression */ - Tcl_ObjType* oldTypePtr = objPtr->typePtr; - /* Old internal rep type of the object */ register char* bytes; /* String rep of the object */ int length; /* Length of the object's string rep */ - /* If it's already the right type, we're fine. */ + /* + * If it's already the right type, we're fine. + */ if (objPtr->typePtr == &tclEndOffsetType) { return TCL_OK; } - /* Check for a string rep of the right form. */ + /* + * Check for a string rep of the right form. + */ - bytes = Tcl_GetStringFromObj(objPtr, &length); + bytes = TclGetStringFromObj(objPtr, &length); if ((*bytes != 'e') || (strncmp(bytes, "end", (size_t)((length > 3) ? 3 : length)) != 0)) { if (interp != NULL) { Tcl_ResetResult(interp); - Tcl_AppendStringsToObj(Tcl_GetObjResult(interp), - "bad index \"", bytes, - "\": must be end?-integer?", - (char*) NULL); + Tcl_AppendResult(interp, "bad index \"", bytes, + "\": must be end?[+-]integer?", NULL); } return TCL_ERROR; } - /* Convert the string rep */ + /* + * Convert the string rep. + */ if (length <= 3) { offset = 0; - } else if ((length > 4) && (bytes[3] == '-')) { + } else if ((length > 4) && ((bytes[3] == '-') || (bytes[3] == '+'))) { /* - * This is our limited string expression evaluator. Pass everything + * This is our limited string expression evaluator. Pass everything * after "end-" to Tcl_GetInt, then reverse for offset. */ + + if (TclIsSpaceProc(bytes[4])) { + return TCL_ERROR; + } if (Tcl_GetInt(interp, bytes+4, &offset) != TCL_OK) { return TCL_ERROR; } - offset = -offset; + if (bytes[3] == '-') { + offset = -offset; + } } else { /* - * Conversion failed. Report the error. + * Conversion failed. Report the error. */ + if (interp != NULL) { Tcl_ResetResult(interp); - Tcl_AppendStringsToObj(Tcl_GetObjResult(interp), - "bad index \"", bytes, - "\": must be integer or end?-integer?", - (char *) NULL); + Tcl_AppendResult(interp, "bad index \"", bytes, + "\": must be end?[+-]integer?", NULL); } return TCL_ERROR; } /* - * The conversion succeeded. Free the old internal rep and set - * the new one. + * The conversion succeeded. Free the old internal rep and set the new + * one. */ - if ((oldTypePtr != NULL) && (oldTypePtr->freeIntRepProc != NULL)) { - oldTypePtr->freeIntRepProc(objPtr); - } - + TclFreeIntRep(objPtr); objPtr->internalRep.longValue = offset; objPtr->typePtr = &tclEndOffsetType; return TCL_OK; -} +} /* *---------------------------------------------------------------------- * * TclCheckBadOctal -- * - * This procedure checks for a bad octal value and appends a - * meaningful error to the interp's result. + * This function checks for a bad octal value and appends a meaningful + * error to the interp's result. * * Results: * 1 if the argument was a bad octal, else 0. @@ -2477,41 +3470,48 @@ SetEndOffsetFromAny(interp, objPtr) */ int -TclCheckBadOctal(interp, value) - Tcl_Interp *interp; /* Interpreter to use for error reporting. - * If NULL, then no error message is left - * after errors. */ - CONST char *value; /* String to check. */ +TclCheckBadOctal( + Tcl_Interp *interp, /* Interpreter to use for error reporting. If + * NULL, then no error message is left after + * errors. */ + CONST char *value) /* String to check. */ { register CONST char *p = value; /* - * A frequent mistake is invalid octal values due to an unwanted - * leading zero. Try to generate a meaningful error message. + * A frequent mistake is invalid octal values due to an unwanted leading + * zero. Try to generate a meaningful error message. */ - while (isspace(UCHAR(*p))) { /* INTL: ISO space. */ + while (TclIsSpaceProc(*p)) { p++; } if (*p == '+' || *p == '-') { p++; } if (*p == '0') { + if ((p[1] == 'o') || p[1] == 'O') { + p+=2; + } while (isdigit(UCHAR(*p))) { /* INTL: digit. */ p++; } - while (isspace(UCHAR(*p))) { /* INTL: ISO space. */ + while (TclIsSpaceProc(*p)) { p++; } if (*p == '\0') { - /* Reached end of string */ + /* + * Reached end of string. + */ + if (interp != NULL) { /* - * Don't reset the result here because we want this result - * to be added to an existing error message as extra info. + * Don't reset the result here because we want this result to + * be added to an existing error message as extra info. */ + Tcl_AppendResult(interp, " (looks like invalid octal number)", - (char *) NULL); + NULL); } return 1; } @@ -2522,28 +3522,353 @@ TclCheckBadOctal(interp, value) /* *---------------------------------------------------------------------- * + * ClearHash -- + * + * Remove all the entries in the hash table *tablePtr. + * + *---------------------------------------------------------------------- + */ + +static void +ClearHash( + Tcl_HashTable *tablePtr) +{ + Tcl_HashSearch search; + Tcl_HashEntry *hPtr; + + for (hPtr = Tcl_FirstHashEntry(tablePtr, &search); hPtr != NULL; + hPtr = Tcl_NextHashEntry(&search)) { + Tcl_Obj *objPtr = (Tcl_Obj *) Tcl_GetHashValue(hPtr); + Tcl_DecrRefCount(objPtr); + Tcl_DeleteHashEntry(hPtr); + } +} + +/* + *---------------------------------------------------------------------- + * + * GetThreadHash -- + * + * Get a thread-specific (Tcl_HashTable *) associated with a thread data + * key. + * + * Results: + * The Tcl_HashTable * corresponding to *keyPtr. + * + * Side effects: + * The first call on a keyPtr in each thread creates a new Tcl_HashTable, + * and registers a thread exit handler to dispose of it. + * + *---------------------------------------------------------------------- + */ + +static Tcl_HashTable * +GetThreadHash( + Tcl_ThreadDataKey *keyPtr) +{ + Tcl_HashTable **tablePtrPtr = (Tcl_HashTable **) + Tcl_GetThreadData(keyPtr, (int) sizeof(Tcl_HashTable *)); + + if (NULL == *tablePtrPtr) { + *tablePtrPtr = (Tcl_HashTable *)ckalloc(sizeof(Tcl_HashTable)); + Tcl_CreateThreadExitHandler(FreeThreadHash, (ClientData)*tablePtrPtr); + Tcl_InitHashTable(*tablePtrPtr, TCL_ONE_WORD_KEYS); + } + return *tablePtrPtr; +} + +/* + *---------------------------------------------------------------------- + * + * FreeThreadHash -- + * + * Thread exit handler used by GetThreadHash to dispose of a thread hash + * table. + * + * Side effects: + * Frees a Tcl_HashTable. + * + *---------------------------------------------------------------------- + */ + +static void +FreeThreadHash( + ClientData clientData) +{ + Tcl_HashTable *tablePtr = (Tcl_HashTable *) clientData; + + ClearHash(tablePtr); + Tcl_DeleteHashTable(tablePtr); + ckfree((char *) tablePtr); +} + +/* + *---------------------------------------------------------------------- + * + * FreeProcessGlobalValue -- + * + * Exit handler used by Tcl(Set|Get)ProcessGlobalValue to cleanup a + * ProcessGlobalValue at exit. + * + *---------------------------------------------------------------------- + */ + +static void +FreeProcessGlobalValue( + ClientData clientData) +{ + ProcessGlobalValue *pgvPtr = (ProcessGlobalValue *) clientData; + + pgvPtr->epoch++; + pgvPtr->numBytes = 0; + ckfree(pgvPtr->value); + pgvPtr->value = NULL; + if (pgvPtr->encoding) { + Tcl_FreeEncoding(pgvPtr->encoding); + pgvPtr->encoding = NULL; + } + Tcl_MutexFinalize(&pgvPtr->mutex); +} + +/* + *---------------------------------------------------------------------- + * + * TclSetProcessGlobalValue -- + * + * Utility routine to set a global value shared by all threads in the + * process while keeping a thread-local copy as well. + * + *---------------------------------------------------------------------- + */ + +void +TclSetProcessGlobalValue( + ProcessGlobalValue *pgvPtr, + Tcl_Obj *newValue, + Tcl_Encoding encoding) +{ + CONST char *bytes; + Tcl_HashTable *cacheMap; + Tcl_HashEntry *hPtr; + int dummy; + + Tcl_MutexLock(&pgvPtr->mutex); + + /* + * Fill the global string value. + */ + + pgvPtr->epoch++; + if (NULL != pgvPtr->value) { + ckfree(pgvPtr->value); + } else { + Tcl_CreateExitHandler(FreeProcessGlobalValue, (ClientData) pgvPtr); + } + bytes = Tcl_GetStringFromObj(newValue, &pgvPtr->numBytes); + pgvPtr->value = ckalloc((unsigned) pgvPtr->numBytes + 1); + memcpy(pgvPtr->value, bytes, (unsigned) pgvPtr->numBytes + 1); + if (pgvPtr->encoding) { + Tcl_FreeEncoding(pgvPtr->encoding); + } + pgvPtr->encoding = encoding; + + /* + * Fill the local thread copy directly with the Tcl_Obj value to avoid + * loss of the intrep. Increment newValue refCount early to handle case + * where we set a PGV to itself. + */ + + Tcl_IncrRefCount(newValue); + cacheMap = GetThreadHash(&pgvPtr->key); + ClearHash(cacheMap); + hPtr = Tcl_CreateHashEntry(cacheMap, + (char *) INT2PTR(pgvPtr->epoch), &dummy); + Tcl_SetHashValue(hPtr, (ClientData) newValue); + Tcl_MutexUnlock(&pgvPtr->mutex); +} + +/* + *---------------------------------------------------------------------- + * + * TclGetProcessGlobalValue -- + * + * Retrieve a global value shared among all threads of the process, + * preferring a thread-local copy as long as it remains valid. + * + * Results: + * Returns a (Tcl_Obj *) that holds a copy of the global value. + * + *---------------------------------------------------------------------- + */ + +Tcl_Obj * +TclGetProcessGlobalValue( + ProcessGlobalValue *pgvPtr) +{ + Tcl_Obj *value = NULL; + Tcl_HashTable *cacheMap; + Tcl_HashEntry *hPtr; + int epoch = pgvPtr->epoch; + + if (pgvPtr->encoding) { + Tcl_Encoding current = Tcl_GetEncoding(NULL, NULL); + + if (pgvPtr->encoding != current) { + /* + * The system encoding has changed since the master string value + * was saved. Convert the master value to be based on the new + * system encoding. + */ + + Tcl_DString native, newValue; + + Tcl_MutexLock(&pgvPtr->mutex); + pgvPtr->epoch++; + epoch = pgvPtr->epoch; + Tcl_UtfToExternalDString(pgvPtr->encoding, pgvPtr->value, + pgvPtr->numBytes, &native); + Tcl_ExternalToUtfDString(current, Tcl_DStringValue(&native), + Tcl_DStringLength(&native), &newValue); + Tcl_DStringFree(&native); + ckfree(pgvPtr->value); + pgvPtr->value = ckalloc((unsigned int) + Tcl_DStringLength(&newValue) + 1); + memcpy(pgvPtr->value, Tcl_DStringValue(&newValue), + (size_t) Tcl_DStringLength(&newValue) + 1); + Tcl_DStringFree(&newValue); + Tcl_FreeEncoding(pgvPtr->encoding); + pgvPtr->encoding = current; + Tcl_MutexUnlock(&pgvPtr->mutex); + } else { + Tcl_FreeEncoding(current); + } + } + cacheMap = GetThreadHash(&pgvPtr->key); + hPtr = Tcl_FindHashEntry(cacheMap, (char *) INT2PTR(epoch)); + if (NULL == hPtr) { + int dummy; + + /* + * No cache for the current epoch - must be a new one. + * + * First, clear the cacheMap, as anything in it must refer to some + * expired epoch. + */ + + ClearHash(cacheMap); + + /* + * If no thread has set the shared value, call the initializer. + */ + + Tcl_MutexLock(&pgvPtr->mutex); + if ((NULL == pgvPtr->value) && (pgvPtr->proc)) { + pgvPtr->epoch++; + (*(pgvPtr->proc))(&pgvPtr->value, &pgvPtr->numBytes, + &pgvPtr->encoding); + if (pgvPtr->value == NULL) { + Tcl_Panic("PGV Initializer did not initialize"); + } + Tcl_CreateExitHandler(FreeProcessGlobalValue, (ClientData)pgvPtr); + } + + /* + * Store a copy of the shared value in our epoch-indexed cache. + */ + + value = Tcl_NewStringObj(pgvPtr->value, pgvPtr->numBytes); + hPtr = Tcl_CreateHashEntry(cacheMap, + (char *) INT2PTR(pgvPtr->epoch), &dummy); + Tcl_MutexUnlock(&pgvPtr->mutex); + Tcl_SetHashValue(hPtr, (ClientData) value); + Tcl_IncrRefCount(value); + } + return (Tcl_Obj *) Tcl_GetHashValue(hPtr); +} + +/* + *---------------------------------------------------------------------- + * + * TclSetObjNameOfExecutable -- + * + * This function stores the absolute pathname of the executable file + * (normally as computed by TclpFindExecutable). + * + * Results: + * None. + * + * Side effects: + * Stores the executable name. + * + *---------------------------------------------------------------------- + */ + +void +TclSetObjNameOfExecutable( + Tcl_Obj *name, + Tcl_Encoding encoding) +{ + TclSetProcessGlobalValue(&executableName, name, encoding); +} + +/* + *---------------------------------------------------------------------- + * + * TclGetObjNameOfExecutable -- + * + * This function retrieves the absolute pathname of the application in + * which the Tcl library is running, usually as previously stored by + * TclpFindExecutable(). This function call is the C API equivalent to + * the "info nameofexecutable" command. + * + * Results: + * A pointer to an "fsPath" Tcl_Obj, or to an empty Tcl_Obj if the + * pathname of the application is unknown. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +Tcl_Obj * +TclGetObjNameOfExecutable(void) +{ + return TclGetProcessGlobalValue(&executableName); +} + +/* + *---------------------------------------------------------------------- + * * Tcl_GetNameOfExecutable -- * - * This procedure simply returns a pointer to the internal full - * path name of the executable file as computed by - * Tcl_FindExecutable. This procedure call is the C API - * equivalent to the "info nameofexecutable" command. + * This function retrieves the absolute pathname of the application in + * which the Tcl library is running, and returns it in string form. + * + * The returned string belongs to Tcl and should be copied if the caller + * plans to keep it, to guard against it becoming invalid. * * Results: - * A pointer to the internal string or NULL if the internal full - * path name has not been computed or unknown. + * A pointer to the internal string or NULL if the internal full path + * name has not been computed or unknown. * * Side effects: - * The object referenced by "objPtr" might be converted to an - * integer object. + * None. * *---------------------------------------------------------------------- */ CONST char * -Tcl_GetNameOfExecutable() +Tcl_GetNameOfExecutable(void) { - return tclExecutableName; + int numBytes; + const char *bytes = + Tcl_GetStringFromObj(TclGetObjNameOfExecutable(), &numBytes); + + if (numBytes == 0) { + return NULL; + } + return bytes; } /* @@ -2551,7 +3876,9 @@ Tcl_GetNameOfExecutable() * * TclpGetTime -- * - * Deprecated synonym for Tcl_GetTime. + * Deprecated synonym for Tcl_GetTime. This function is provided for the + * benefit of extensions written before Tcl_GetTime was exported from the + * library. * * Results: * None. @@ -2559,15 +3886,265 @@ Tcl_GetNameOfExecutable() * Side effects: * Stores current time in the buffer designated by "timePtr" * - * This procedure is provided for the benefit of extensions written - * before Tcl_GetTime was exported from the library. - * *---------------------------------------------------------------------- */ void -TclpGetTime(timePtr) - Tcl_Time* timePtr; +TclpGetTime( + Tcl_Time *timePtr) { Tcl_GetTime(timePtr); } + +/* + *---------------------------------------------------------------------- + * + * TclGetPlatform -- + * + * This is a kludge that allows the test library to get access the + * internal tclPlatform variable. + * + * Results: + * Returns a pointer to the tclPlatform variable. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +TclPlatformType * +TclGetPlatform(void) +{ + return &tclPlatform; +} + +/* + *---------------------------------------------------------------------- + * + * TclReToGlob -- + * + * Attempt to convert a regular expression to an equivalent glob pattern. + * + * Results: + * Returns TCL_OK on success, TCL_ERROR on failure. If interp is not + * NULL, an error message is placed in the result. On success, the + * DString will contain an exact equivalent glob pattern. The caller is + * responsible for calling Tcl_DStringFree on success. If exactPtr is not + * NULL, it will be 1 if an exact match qualifies. + * + * Side effects: + * None. + * + *---------------------------------------------------------------------- + */ + +int +TclReToGlob( + Tcl_Interp *interp, + const char *reStr, + int reStrLen, + Tcl_DString *dsPtr, + int *exactPtr) +{ + int anchorLeft, anchorRight, lastIsStar, numStars; + char *dsStr, *dsStrStart, *msg; + const char *p, *strEnd; + + strEnd = reStr + reStrLen; + Tcl_DStringInit(dsPtr); + + /* + * "***=xxx" == "*xxx*", watch for glob-sensitive chars. + */ + + if ((reStrLen >= 4) && (memcmp("***=", reStr, 4) == 0)) { + /* + * At most, the glob pattern has length 2*reStrLen + 2 to + * backslash escape every character and have * at each end. + */ + Tcl_DStringSetLength(dsPtr, 2*reStrLen + 2); + dsStr = dsStrStart = Tcl_DStringValue(dsPtr); + *dsStr++ = '*'; + for (p = reStr + 4; p < strEnd; p++) { + switch (*p) { + case '\\': case '*': case '[': case ']': case '?': + /* Only add \ where necessary for glob */ + *dsStr++ = '\\'; + /* fall through */ + default: + *dsStr++ = *p; + break; + } + } + *dsStr++ = '*'; + Tcl_DStringSetLength(dsPtr, dsStr - dsStrStart); + if (exactPtr) { + *exactPtr = 0; + } + return TCL_OK; + } + + /* + * At most, the glob pattern has length reStrLen + 2 to account + * for possible * at each end. + */ + + Tcl_DStringSetLength(dsPtr, reStrLen + 2); + dsStr = dsStrStart = Tcl_DStringValue(dsPtr); + + /* + * Check for anchored REs (ie ^foo$), so we can use string equal if + * possible. Do not alter the start of str so we can free it correctly. + * + * Keep track of the last char being an unescaped star to prevent + * multiple instances. Simpler than checking that the last star + * may be escaped. + */ + + msg = NULL; + p = reStr; + anchorRight = 0; + lastIsStar = 0; + numStars = 0; + + if (*p == '^') { + anchorLeft = 1; + p++; + } else { + anchorLeft = 0; + *dsStr++ = '*'; + lastIsStar = 1; + } + + for ( ; p < strEnd; p++) { + switch (*p) { + case '\\': + p++; + switch (*p) { + case 'a': + *dsStr++ = '\a'; + break; + case 'b': + *dsStr++ = '\b'; + break; + case 'f': + *dsStr++ = '\f'; + break; + case 'n': + *dsStr++ = '\n'; + break; + case 'r': + *dsStr++ = '\r'; + break; + case 't': + *dsStr++ = '\t'; + break; + case 'v': + *dsStr++ = '\v'; + break; + case 'B': case '\\': + *dsStr++ = '\\'; + *dsStr++ = '\\'; + anchorLeft = 0; /* prevent exact match */ + break; + case '*': case '[': case ']': case '?': + /* Only add \ where necessary for glob */ + *dsStr++ = '\\'; + anchorLeft = 0; /* prevent exact match */ + /* fall through */ + case '{': case '}': case '(': case ')': case '+': + case '.': case '|': case '^': case '$': + *dsStr++ = *p; + break; + default: + msg = "invalid escape sequence"; + goto invalidGlob; + } + break; + case '.': + anchorLeft = 0; /* prevent exact match */ + if (p+1 < strEnd) { + if (p[1] == '*') { + p++; + if (!lastIsStar) { + *dsStr++ = '*'; + lastIsStar = 1; + numStars++; + } + continue; + } else if (p[1] == '+') { + p++; + *dsStr++ = '?'; + *dsStr++ = '*'; + lastIsStar = 1; + numStars++; + continue; + } + } + *dsStr++ = '?'; + break; + case '$': + if (p+1 != strEnd) { + msg = "$ not anchor"; + goto invalidGlob; + } + anchorRight = 1; + break; + case '*': case '+': case '?': case '|': case '^': + case '{': case '}': case '(': case ')': case '[': case ']': + msg = "unhandled RE special char"; + goto invalidGlob; + break; + default: + *dsStr++ = *p; + break; + } + lastIsStar = 0; + } + if (numStars > 1) { + /* + * Heuristic: if >1 non-anchoring *, the risk is large that glob + * matching is slower than the RE engine, so report invalid. + */ + msg = "excessive recursive glob backtrack potential"; + goto invalidGlob; + } + + if (!anchorRight && !lastIsStar) { + *dsStr++ = '*'; + } + Tcl_DStringSetLength(dsPtr, dsStr - dsStrStart); + + if (exactPtr) { + *exactPtr = (anchorLeft && anchorRight); + } + +#if 0 + fprintf(stderr, "INPUT RE '%.*s' OUTPUT GLOB '%s' anchor %d:%d \n", + reStrLen, reStr, + Tcl_DStringValue(dsPtr), anchorLeft, anchorRight); + fflush(stderr); +#endif + return TCL_OK; + + invalidGlob: +#if 0 + fprintf(stderr, "INPUT RE '%.*s' NO OUTPUT GLOB %s (%c)\n", + reStrLen, reStr, msg, *p); + fflush(stderr); +#endif + if (interp != NULL) { + Tcl_AppendResult(interp, msg, NULL); + } + Tcl_DStringFree(dsPtr); + return TCL_ERROR; +} + +/* + * Local Variables: + * mode: c + * c-basic-offset: 4 + * fill-column: 78 + * End: + */ |