add compilation of [string is]

7 files changed, 417 insertions, 4 deletions

diff --git a/generic/tclAssembly.c b/generic/tclAssembly.c
index 55a0c3f..d1866c8 100644
--- a/generic/tclAssembly.c
+++ b/generic/tclAssembly.c
@@ -438,6 +438,7 @@ static const TalInstDesc TalInstructionTable[] = {
     {"nop",		ASSEM_1BYTE,	INST_NOP,		0,	0},
     {"not",		ASSEM_1BYTE,	INST_LNOT,		1,	1},
     {"nsupvar",		ASSEM_LVT4,	INST_NSUPVAR,		2,	1},
+    {"numericType",	ASSEM_1BYTE,	INST_NUM_TYPE,		1,	1},
     {"originCmd",	ASSEM_1BYTE,	INST_ORIGIN_COMMAND,	1,	1},
     {"over",		ASSEM_OVER,	INST_OVER,		INT_MIN,-1-1},
     {"pop",		ASSEM_1BYTE,	INST_POP,		1,	0},
@@ -478,6 +479,7 @@ static const TalInstDesc TalInstructionTable[] = {
     {"tclooIsObject",	ASSEM_1BYTE,	INST_TCLOO_IS_OBJECT,	1,	1},
     {"tclooNamespace",	ASSEM_1BYTE,	INST_TCLOO_NS,		1,	1},
     {"tclooSelf",	ASSEM_1BYTE,	INST_TCLOO_SELF,	0,	1},
+    {"tryCvtToBoolean",	ASSEM_1BYTE,	INST_TRY_CVT_TO_BOOLEAN,1,	2},
     {"tryCvtToNumeric",	ASSEM_1BYTE,	INST_TRY_CVT_TO_NUMERIC,1,	1},
     {"uminus",		ASSEM_1BYTE,	INST_UMINUS,		1,	1},
     {"unset",		ASSEM_BOOL_LVT4,INST_UNSET_SCALAR,	0,	0},
@@ -517,7 +519,8 @@ static const unsigned char NonThrowingByteCodes[] = {
     INST_RESOLVE_COMMAND,					/* 154 */
     INST_STR_TRIM, INST_STR_TRIM_LEFT, INST_STR_TRIM_RIGHT,	/* 166-168 */
     INST_CONCAT_STK,						/* 169 */
-    INST_STR_UPPER, INST_STR_LOWER, INST_STR_TITLE		/* 170-172 */
+    INST_STR_UPPER, INST_STR_LOWER, INST_STR_TITLE,		/* 170-172 */
+    INST_NUM_TYPE						/* 180 */
 };
 
 /*
diff --git a/generic/tclCmdMZ.c b/generic/tclCmdMZ.c
index c02cd1a..00c9f2f 100644
--- a/generic/tclCmdMZ.c
+++ b/generic/tclCmdMZ.c
@@ -3332,7 +3332,7 @@ TclInitStringCmd(
 	{"equal",	StringEqualCmd,	TclCompileStringEqualCmd, NULL, NULL, 0},
 	{"first",	StringFirstCmd,	TclCompileStringFirstCmd, NULL, NULL, 0},
 	{"index",	StringIndexCmd,	TclCompileStringIndexCmd, NULL, NULL, 0},
-	{"is",		StringIsCmd,	NULL, NULL, NULL, 0},
+	{"is",		StringIsCmd,	TclCompileStringIsCmd, NULL, NULL, 0},
 	{"last",	StringLastCmd,	TclCompileStringLastCmd, NULL, NULL, 0},
 	{"length",	StringLenCmd,	TclCompileStringLenCmd, NULL, NULL, 0},
 	{"map",		StringMapCmd,	TclCompileStringMapCmd, NULL, NULL, 0},
diff --git a/generic/tclCompCmdsSZ.c b/generic/tclCompCmdsSZ.c
index 4b14f24..e6ec0a6 100644
--- a/generic/tclCompCmdsSZ.c
+++ b/generic/tclCompCmdsSZ.c
@@ -425,6 +425,284 @@ TclCompileStringIndexCmd(
 }
 
 int
+TclCompileStringIsCmd(
+    Tcl_Interp *interp,		/* Used for error reporting. */
+    Tcl_Parse *parsePtr,	/* Points to a parse structure for the command
+				 * created by Tcl_ParseCommand. */
+    Command *cmdPtr,		/* Points to defintion of command being
+				 * compiled. */
+    CompileEnv *envPtr)		/* Holds resulting instructions. */
+{
+    DefineLineInformation;	/* TIP #280 */
+    Tcl_Token *tokenPtr = TokenAfter(parsePtr->tokenPtr);
+    static const char *const isClasses[] = {
+	"alnum",	"alpha",	"ascii",	"control",
+	"boolean",	"digit",	"double",	"entier",
+	"false",	"graph",	"integer",	"list",
+	"lower",	"print",	"punct",	"space",
+	"true",		"upper",	"wideinteger",	"wordchar",
+	"xdigit",	NULL
+    };
+    enum isClasses {
+	STR_IS_ALNUM,	STR_IS_ALPHA,	STR_IS_ASCII,	STR_IS_CONTROL,
+	STR_IS_BOOL,	STR_IS_DIGIT,	STR_IS_DOUBLE,	STR_IS_ENTIER,
+	STR_IS_FALSE,	STR_IS_GRAPH,	STR_IS_INT,	STR_IS_LIST,
+	STR_IS_LOWER,	STR_IS_PRINT,	STR_IS_PUNCT,	STR_IS_SPACE,
+	STR_IS_TRUE,	STR_IS_UPPER,	STR_IS_WIDE,	STR_IS_WORD,
+	STR_IS_XDIGIT
+    };
+    int t, range, allowEmpty = 0, end;
+    InstStringClassType strClassType;
+    Tcl_Obj *isClass;
+
+    if (parsePtr->numWords < 3 || parsePtr->numWords > 6) {
+	return TCL_ERROR;
+    }
+    isClass = Tcl_NewObj();
+    if (!TclWordKnownAtCompileTime(tokenPtr, isClass)) {
+	Tcl_DecrRefCount(isClass);
+	return TCL_ERROR;
+    } else if (Tcl_GetIndexFromObj(interp, isClass, isClasses, "class", 0,
+	    &t) != TCL_OK) {
+	Tcl_DecrRefCount(isClass);
+	TclCompileSyntaxError(interp, envPtr);
+	return TCL_OK;
+    }
+    Tcl_DecrRefCount(isClass);
+
+#define GotLiteral(tokenPtr, word) \
+    ((tokenPtr)->type == TCL_TOKEN_SIMPLE_WORD &&			\
+     (tokenPtr)[1].size > 1 &&						\
+     (tokenPtr)[1].start[0] == word[0] &&				\
+     strncmp((tokenPtr)[1].start, (word), (tokenPtr)[1].size) == 0)
+
+    /*
+     * Cannot handle the -failindex option at all, and that's the only legal
+     * way to have more than 4 arguments.
+     */
+
+    if (parsePtr->numWords != 3 && parsePtr->numWords != 4) {
+	return TCL_ERROR;
+    }
+
+    tokenPtr = TokenAfter(tokenPtr);
+    if (parsePtr->numWords == 3) {
+	allowEmpty = 1;
+    } else {
+	if (!GotLiteral(tokenPtr, "-strict")) {
+	    return TCL_ERROR;
+	}
+	tokenPtr = TokenAfter(tokenPtr);
+    }
+#undef GotLiteral
+
+    /*
+     * Compile the code. There are several main classes of check here.
+     *	1. Character classes
+     *	2. Booleans
+     *	3. Integers
+     *	4. Floats
+     *	5. Lists
+     */
+
+    CompileWord(envPtr, tokenPtr, interp, parsePtr->numWords-1);
+
+    switch ((enum isClasses) t) {
+    case STR_IS_ALNUM:
+	strClassType = STR_CLASS_ALNUM;
+	goto compileStrClass;
+    case STR_IS_ALPHA:
+	strClassType = STR_CLASS_ALPHA;
+	goto compileStrClass;
+    case STR_IS_ASCII:
+	strClassType = STR_CLASS_ASCII;
+	goto compileStrClass;
+    case STR_IS_CONTROL:
+	strClassType = STR_CLASS_CONTROL;
+	goto compileStrClass;
+    case STR_IS_DIGIT:
+	strClassType = STR_CLASS_DIGIT;
+	goto compileStrClass;
+    case STR_IS_GRAPH:
+	strClassType = STR_CLASS_GRAPH;
+	goto compileStrClass;
+    case STR_IS_LOWER:
+	strClassType = STR_CLASS_LOWER;
+	goto compileStrClass;
+    case STR_IS_PRINT:
+	strClassType = STR_CLASS_PRINT;
+	goto compileStrClass;
+    case STR_IS_PUNCT:
+	strClassType = STR_CLASS_PUNCT;
+	goto compileStrClass;
+    case STR_IS_SPACE:
+	strClassType = STR_CLASS_SPACE;
+	goto compileStrClass;
+    case STR_IS_UPPER:
+	strClassType = STR_CLASS_UPPER;
+	goto compileStrClass;
+    case STR_IS_WORD:
+	strClassType = STR_CLASS_WORD;
+	goto compileStrClass;
+    case STR_IS_XDIGIT:
+	strClassType = STR_CLASS_XDIGIT;
+    compileStrClass:
+	if (allowEmpty) {
+	    OP1(	STR_CLASS, strClassType);
+	} else {
+	    int over, over2;
+
+	    OP(		DUP);
+	    OP1(	STR_CLASS, strClassType);
+	    JUMP1(	JUMP_TRUE, over);
+	    OP(		POP);
+	    PUSH(	"0");
+	    JUMP1(	JUMP, over2);
+	    FIXJUMP1(over);
+	    PUSH(	"");
+	    OP(		STR_NEQ);
+	    FIXJUMP1(over2);
+	}
+	return TCL_OK;
+
+    case STR_IS_BOOL:
+    case STR_IS_FALSE:
+    case STR_IS_TRUE:
+	OP(		TRY_CVT_TO_BOOLEAN);
+	switch (t) {
+	    int over, over2;
+
+	case STR_IS_BOOL:
+	    if (allowEmpty) {
+		JUMP1(	JUMP_TRUE, over);
+		PUSH(	"");
+		OP(	STR_EQ);
+		JUMP1(	JUMP, over2);
+		FIXJUMP1(over);
+		OP(	POP);
+		PUSH(	"1");
+		FIXJUMP1(over2);
+	    } else {
+		OP4(	REVERSE, 2);
+		OP(	POP);
+	    }
+	    return TCL_OK;
+	case STR_IS_TRUE:
+	    JUMP1(	JUMP_TRUE, over);
+	    if (allowEmpty) {
+		PUSH(	"");
+		OP(	STR_EQ);
+	    } else {
+		OP(	POP);
+		PUSH(	"0");
+	    }
+	    FIXJUMP1(	over);
+	    OP(		LNOT);
+	    OP(		LNOT);
+	    return TCL_OK;
+	case STR_IS_FALSE:
+	    JUMP1(	JUMP_TRUE, over);
+	    if (allowEmpty) {
+		PUSH(	"");
+		OP(	STR_NEQ);
+	    } else {
+		OP(	POP);
+		PUSH(	"1");
+	    }
+	    FIXJUMP1(	over);
+	    OP(		LNOT);
+	    return TCL_OK;
+	}
+
+    case STR_IS_DOUBLE: {
+	int satisfied, isEmpty;
+
+	if (allowEmpty) {
+	    OP(		DUP);
+	    PUSH(	"");
+	    OP(		STR_EQ);
+	    JUMP1(	JUMP_TRUE, isEmpty);
+	    OP(		NUM_TYPE);
+	    JUMP1(	JUMP_TRUE, satisfied);
+	    PUSH(	"0");
+	    JUMP1(	JUMP, end);
+	    FIXJUMP1(	isEmpty);
+	    OP(		POP);
+	    FIXJUMP1(	satisfied);
+	} else {
+	    OP(		NUM_TYPE);
+	    JUMP1(	JUMP_TRUE, satisfied);
+	    PUSH(	"0");
+	    JUMP1(	JUMP, end);
+	    TclAdjustStackDepth(-1, envPtr);
+	    FIXJUMP1(	satisfied);
+	}
+	PUSH(		"1");
+	FIXJUMP1(	end);
+	return TCL_OK;
+    }
+
+    case STR_IS_INT:
+    case STR_IS_WIDE:
+    case STR_IS_ENTIER:
+	if (allowEmpty) {
+	    int testNumType;
+
+	    OP(		DUP);
+	    OP(		NUM_TYPE);
+	    OP(		DUP);
+	    JUMP1(	JUMP_TRUE, testNumType);
+	    OP(		POP);
+	    PUSH(	"");
+	    OP(		STR_EQ);
+	    JUMP1(	JUMP, end);
+	    TclAdjustStackDepth(1, envPtr);
+	    FIXJUMP1(	testNumType);
+	    OP4(	REVERSE, 2);
+	    OP(		POP);
+	} else {
+	    OP(		NUM_TYPE);
+	    OP(		DUP);
+	    JUMP1(	JUMP_FALSE, end);
+	}
+
+	switch (t) {
+	case STR_IS_INT:
+	    PUSH(	"1");
+	    OP(		EQ);
+	    break;
+	case STR_IS_WIDE:
+	    PUSH(	"2");
+	    OP(		LE);
+	    break;
+	case STR_IS_ENTIER:
+	    PUSH(	"3");
+	    OP(		LE);
+	    break;
+	}
+	FIXJUMP1(	end);
+	return TCL_OK;
+
+    case STR_IS_LIST:
+	range = TclCreateExceptRange(CATCH_EXCEPTION_RANGE, envPtr);
+	OP4(		BEGIN_CATCH4, range);
+	ExceptionRangeStarts(envPtr, range);
+	OP(		DUP);
+	OP(		LIST_LENGTH);
+	OP(		POP);
+	ExceptionRangeEnds(envPtr, range);
+	ExceptionRangeTarget(envPtr, range, catchOffset);
+	OP(		POP);
+	OP(		PUSH_RETURN_CODE);
+	OP(		END_CATCH);
+	OP(		LNOT);
+	return TCL_OK;
+    }
+
+    return TclCompileBasicMin0ArgCmd(interp, parsePtr, cmdPtr, envPtr);
+}
+
+int
 TclCompileStringMatchCmd(
     Tcl_Interp *interp,		/* Used for error reporting. */
     Tcl_Parse *parsePtr,	/* Points to a parse structure for the command
@@ -939,6 +1217,41 @@ TclCompileStringToTitleCmd(
 }
 
 /*
+ * Support definitions for the [string is] compilation.
+ */
+
+static int
+UniCharIsAscii(
+    int character)
+{
+    return (character >= 0) && (character < 0x80);
+}
+
+static int
+UniCharIsHexDigit(
+    int character)
+{
+    return (character >= 0) && (character < 0x80) && isxdigit(character);
+}
+
+StringClassDesc const tclStringClassTable[] = {
+    {"alnum",	Tcl_UniCharIsAlnum},
+    {"alpha",	Tcl_UniCharIsAlpha},
+    {"ascii",	UniCharIsAscii},
+    {"control", Tcl_UniCharIsControl},
+    {"digit",	Tcl_UniCharIsDigit},
+    {"graph",	Tcl_UniCharIsGraph},
+    {"lower",	Tcl_UniCharIsLower},
+    {"print",	Tcl_UniCharIsPrint},
+    {"punct",	Tcl_UniCharIsPunct},
+    {"space",	Tcl_UniCharIsSpace},
+    {"upper",	Tcl_UniCharIsUpper},
+    {"word",	Tcl_UniCharIsWordChar},
+    {"xdigit",	UniCharIsHexDigit},
+    {NULL,	NULL}
+};
+
+/*
  *----------------------------------------------------------------------
  *
  * TclCompileSubstCmd --
diff --git a/generic/tclCompile.c b/generic/tclCompile.c
index f75ac83..c5d0107 100644
--- a/generic/tclCompile.c
+++ b/generic/tclCompile.c
@@ -638,6 +638,18 @@ InstructionDesc const tclInstructionTable[] = {
 	 * that are the response back on top of the stack when it resumes.
 	 * Stack:  ... [list ns cmd arg1 ... argN] => ... resumeList */
 
+    {"numericType",	 1,	0,	  0,	{OPERAND_NONE}},
+	/* Pushes the numeric type code of the word at the top of the stack.
+	 * Stack:  ... value => ... typeCode */
+    {"tryCvtToBoolean",	 1,	+1,	  0,	{OPERAND_NONE}},
+	/* Try converting stktop to boolean if possible. No errors.
+	 * Stack:  ... value => ... value isStrictBool */
+    {"strclass",	 2,	0,	  1,	{OPERAND_SCLS1}},
+	/* See if all the characters of the given string are a member of the
+	 * specified (by opnd) character class. Note that an empty string will
+	 * satisfy the class check (standard definition of "all").
+	 * Stack:  ... stringValue => ... boolean */
+
     {NULL, 0, 0, 0, {OPERAND_NONE}}
 };
 
@@ -5085,6 +5097,11 @@ FormatInstruction(
 	    }
 	    Tcl_AppendPrintfToObj(bufferObj, "%%v%u ", (unsigned) opnd);
 	    break;
+	case OPERAND_SCLS1:
+	    opnd = TclGetUInt1AtPtr(pc+numBytes); numBytes++;
+	    Tcl_AppendPrintfToObj(bufferObj, "%s ",
+		    tclStringClassTable[opnd].name);
+	    break;
 	case OPERAND_NONE:
 	default:
 	    break;
diff --git a/generic/tclCompile.h b/generic/tclCompile.h
index 7994e2c..5665ca9 100644
--- a/generic/tclCompile.h
+++ b/generic/tclCompile.h
@@ -795,8 +795,12 @@ typedef struct ByteCode {
 
 #define INST_YIELD_TO_INVOKE		181
 
+#define INST_NUM_TYPE			182
+#define INST_TRY_CVT_TO_BOOLEAN		183
+#define INST_STR_CLASS			184
+
 /* The last opcode */
-#define LAST_INST_OPCODE		181
+#define LAST_INST_OPCODE		184
 
 /*
  * Table describing the Tcl bytecode instructions: their name (for displaying
@@ -821,8 +825,9 @@ typedef enum InstOperandType {
 				 * variable table. */
     OPERAND_LVT4,		/* Four byte unsigned index into the local
 				 * variable table. */
-    OPERAND_AUX4		/* Four byte unsigned index into the aux data
+    OPERAND_AUX4,		/* Four byte unsigned index into the aux data
 				 * table. */
+    OPERAND_SCLS1		/* Index into tclStringClassTable. */
 } InstOperandType;
 
 typedef struct InstructionDesc {
@@ -841,6 +846,40 @@ typedef struct InstructionDesc {
 MODULE_SCOPE InstructionDesc const tclInstructionTable[];
 
 /*
+ * Constants used by INST_STRING_CLASS to indicate character classes. These
+ * correspond closely by name with what [string is] can support, but there is
+ * no requirement to keep the values the same.
+ */
+
+typedef enum InstStringClassType {
+    STR_CLASS_ALNUM,		/* Unicode alphabet or digit characters. */
+    STR_CLASS_ALPHA,		/* Unicode alphabet characters. */
+    STR_CLASS_ASCII,		/* Characters in range U+000000..U+00007F. */
+    STR_CLASS_CONTROL,		/* Unicode control characters. */
+    STR_CLASS_DIGIT,		/* Unicode digit characters. */
+    STR_CLASS_GRAPH,		/* Unicode printing characters, excluding
+				 * space. */
+    STR_CLASS_LOWER,		/* Unicode lower-case alphabet characters. */
+    STR_CLASS_PRINT,		/* Unicode printing characters, including
+				 * spaces. */
+    STR_CLASS_PUNCT,		/* Unicode punctuation characters. */
+    STR_CLASS_SPACE,		/* Unicode space characters. */
+    STR_CLASS_UPPER,		/* Unicode upper-case alphabet characters. */
+    STR_CLASS_WORD,		/* Unicode word (alphabetic, digit, connector
+				 * punctuation) characters. */
+    STR_CLASS_XDIGIT		/* Characters that can be used as digits in
+				 * hexadecimal numbers ([0-9A-Fa-f]). */
+} InstStringClassType;
+
+typedef struct StringClassDesc {
+    const char *name;		/* Name of the class. */
+    int (*comparator)(int);	/* Function to test if a single unicode
+				 * character is a member of the class. */
+} StringClassDesc;
+
+MODULE_SCOPE StringClassDesc const tclStringClassTable[];
+
+/*
  * Compilation of some Tcl constructs such as if commands and the logical or
  * (||) and logical and (&&) operators in expressions requires the generation
  * of forward jumps. Since the PC target of these jumps isn't known when the
diff --git a/generic/tclExecute.c b/generic/tclExecute.c
index e8bc903..ac78370 100644
--- a/generic/tclExecute.c
+++ b/generic/tclExecute.c
@@ -5810,6 +5810,25 @@ TEBCresume(
 
 	TclNewIntObj(objResultPtr, match);
 	NEXT_INST_F(1, 2, 1);
+
+    case INST_STR_CLASS:
+	opnd = TclGetInt1AtPtr(pc+1);
+	valuePtr = OBJ_AT_TOS;
+	TRACE(("%s \"%.30s\" => ", tclStringClassTable[opnd].name,
+		O2S(valuePtr)));
+	ustring1 = Tcl_GetUnicodeFromObj(valuePtr, &length);
+	match = 1;
+	if (length > 0) {
+	    end = ustring1 + length;
+	    for (p=ustring1 ; p<end ; p++) {
+		if (!tclStringClassTable[opnd].comparator(*p)) {
+		    match = 0;
+		    break;
+		}
+	    }
+	}
+	TRACE_APPEND(("%d\n", match));
+	JUMP_PEEPHOLE_F(match, 2, 1);
     }
 
     case INST_STR_MATCH:
@@ -5944,6 +5963,14 @@ TEBCresume(
 	int type1, type2;
 	long l1, l2, lResult;
 
+    case INST_NUM_TYPE:
+	if (GetNumberFromObj(NULL, OBJ_AT_TOS, &ptr1, &type1) != TCL_OK) {
+	    type1 = 0;
+	}
+	TclNewIntObj(objResultPtr, type1);
+	TRACE(("\"%.20s\" => %d\n", O2S(OBJ_AT_TOS), type1));
+	NEXT_INST_F(1, 1, 1);
+
     case INST_EQ:
     case INST_NEQ:
     case INST_LT:
@@ -6621,6 +6648,17 @@ TEBCresume(
      * -----------------------------------------------------------------
      */
 
+    case INST_TRY_CVT_TO_BOOLEAN:
+	valuePtr = OBJ_AT_TOS;
+	if (valuePtr->typePtr == &tclBooleanType) {
+	    objResultPtr = TCONST(1);
+	} else {
+	    int result = (TclSetBooleanFromAny(NULL, valuePtr) == TCL_OK);
+	    objResultPtr = TCONST(result);
+	}
+	TRACE_WITH_OBJ(("\"%.30s\" => ", O2S(valuePtr)), objResultPtr);
+	NEXT_INST_F(1, 0, 1);
+
     case INST_BREAK:
 	/*
 	DECACHE_STACK_INFO();
diff --git a/generic/tclInt.h b/generic/tclInt.h
index 6ddb015..a9f4c16 100644
--- a/generic/tclInt.h
+++ b/generic/tclInt.h
@@ -3622,6 +3622,9 @@ MODULE_SCOPE int	TclCompileStringFirstCmd(Tcl_Interp *interp,
 MODULE_SCOPE int	TclCompileStringIndexCmd(Tcl_Interp *interp,
 			    Tcl_Parse *parsePtr, Command *cmdPtr,
 			    struct CompileEnv *envPtr);
+MODULE_SCOPE int	TclCompileStringIsCmd(Tcl_Interp *interp,
+			    Tcl_Parse *parsePtr, Command *cmdPtr,
+			    struct CompileEnv *envPtr);
 MODULE_SCOPE int	TclCompileStringLastCmd(Tcl_Interp *interp,
 			    Tcl_Parse *parsePtr, Command *cmdPtr,
 			    struct CompileEnv *envPtr);