Compile [string is] with character classes in a non-awful way. Needs more work to make resulting bytecode disassemble nicely.

1 files changed, 36 insertions, 1 deletions

diff --git a/generic/tclCompile.h b/generic/tclCompile.h
index d6d515d..502a2e6 100644
--- a/generic/tclCompile.h
+++ b/generic/tclCompile.h
@@ -797,9 +797,10 @@ typedef struct ByteCode {
 
 #define INST_NUM_TYPE			182
 #define INST_TRY_CVT_TO_BOOLEAN		183
+#define INST_STR_CLASS			184
 
 /* The last opcode */
-#define LAST_INST_OPCODE		183
+#define LAST_INST_OPCODE		184
 
 /*
  * Table describing the Tcl bytecode instructions: their name (for displaying
@@ -844,6 +845,40 @@ 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