]>
git.saurik.com Git - apple/icu.git/blob - icuSources/i18n/regeximp.h
2 // Copyright (C) 2002-2012 International Business Machines Corporation
3 // and others. All rights reserved.
7 // ICU Regular Expressions,
8 // Definitions of constant values used in the compiled form of
9 // a regular expression pattern.
15 #include "unicode/utypes.h"
16 #include "unicode/uobject.h"
17 #include "unicode/uniset.h"
18 #include "unicode/utext.h"
25 // For debugging, define REGEX_DEBUG
26 // To define with configure,
27 // ./runConfigureICU --enable-debug --disable-release Linux CPPFLAGS="-DREGEX_DEBUG"
31 // debugging options. Enable one or more of the three #defines immediately following
34 //#define REGEX_SCAN_DEBUG
35 #define REGEX_DUMP_DEBUG
36 #define REGEX_RUN_DEBUG
38 // End of #defines inteded to be directly set.
43 #ifdef REGEX_SCAN_DEBUG
44 #define REGEX_SCAN_DEBUG_PRINTF(a) printf a
46 #define REGEX_SCAN_DEBUG_PRINTF(a)
49 #ifdef REGEX_DUMP_DEBUG
50 #define REGEX_DUMP_DEBUG_PRINTF(a) printf a
52 #define REGEX_DUMP_DEBUG_PRINTF(a)
55 #ifdef REGEX_RUN_DEBUG
56 #define REGEX_RUN_DEBUG_PRINTF(a) printf a
57 #define REGEX_DUMP_DEBUG_PRINTF(a) printf a
59 #define REGEX_RUN_DEBUG_PRINTF(a)
64 // Opcode types In the compiled form of the regexp, these are the type, or opcodes,
68 URX_RESERVED_OP
= 0, // For multi-operand ops, most non-first words.
69 URX_RESERVED_OP_N
= 255, // For multi-operand ops, negative operand values.
70 URX_BACKTRACK
= 1, // Force a backtrack, as if a match test had failed.
72 URX_ONECHAR
= 3, // Value field is the 21 bit unicode char to match
73 URX_STRING
= 4, // Value field is index of string start
74 URX_STRING_LEN
= 5, // Value field is string length (code units)
75 URX_STATE_SAVE
= 6, // Value field is pattern position to push
77 URX_START_CAPTURE
= 8, // Value field is capture group number.
78 URX_END_CAPTURE
= 9, // Value field is capture group number
79 URX_STATIC_SETREF
= 10, // Value field is index of set in array of sets.
80 URX_SETREF
= 11, // Value field is index of set in array of sets.
82 URX_JMP
= 13, // Value field is destination position in
84 URX_FAIL
= 14, // Stop match operation, No match.
86 URX_JMP_SAV
= 15, // Operand: JMP destination location
87 URX_BACKSLASH_B
= 16, // Value field: 0: \b 1: \B
89 URX_JMP_SAV_X
= 18, // Conditional JMP_SAV,
90 // Used in (x)+, breaks loop on zero length match.
91 // Operand: Jmp destination.
93 URX_BACKSLASH_Z
= 20, // \z Unconditional end of line.
95 URX_DOTANY_ALL
= 21, // ., in the . matches any mode.
96 URX_BACKSLASH_D
= 22, // Value field: 0: \d 1: \D
97 URX_CARET
= 23, // Value field: 1: multi-line mode.
98 URX_DOLLAR
= 24, // Also for \Z
100 URX_CTR_INIT
= 25, // Counter Inits for {Interval} loops.
101 URX_CTR_INIT_NG
= 26, // 2 kinds, normal and non-greedy.
102 // These are 4 word opcodes. See description.
103 // First Operand: Data loc of counter variable
104 // 2nd Operand: Pat loc of the URX_CTR_LOOPx
105 // at the end of the loop.
106 // 3rd Operand: Minimum count.
107 // 4th Operand: Max count, -1 for unbounded.
109 URX_DOTANY_UNIX
= 27, // '.' operator in UNIX_LINES mode, only \n marks end of line.
111 URX_CTR_LOOP
= 28, // Loop Ops for {interval} loops.
112 URX_CTR_LOOP_NG
= 29, // Also in three flavors.
113 // Operand is loc of corresponding CTR_INIT.
115 URX_CARET_M_UNIX
= 30, // '^' operator, test for start of line in multi-line
116 // plus UNIX_LINES mode.
118 URX_RELOC_OPRND
= 31, // Operand value in multi-operand ops that refers
119 // back into compiled pattern code, and thus must
120 // be relocated when inserting/deleting ops in code.
122 URX_STO_SP
= 32, // Store the stack ptr. Operand is location within
123 // matcher data (not stack data) to store it.
124 URX_LD_SP
= 33, // Load the stack pointer. Operand is location
126 URX_BACKREF
= 34, // Back Reference. Parameter is the index of the
127 // capture group variables in the state stack frame.
128 URX_STO_INP_LOC
= 35, // Store the input location. Operand is location
129 // within the matcher stack frame.
130 URX_JMPX
= 36, // Conditional JMP.
131 // First Operand: JMP target location.
132 // Second Operand: Data location containing an
133 // input position. If current input position ==
134 // saved input position, FAIL rather than taking
136 URX_LA_START
= 37, // Starting a LookAround expression.
137 // Save InputPos and SP in static data.
138 // Operand: Static data offset for the save
139 URX_LA_END
= 38, // Ending a Lookaround expression.
140 // Restore InputPos and Stack to saved values.
141 // Operand: Static data offset for saved data.
142 URX_ONECHAR_I
= 39, // Test for case-insensitive match of a literal character.
143 // Operand: the literal char.
144 URX_STRING_I
= 40, // Case insensitive string compare.
145 // First Operand: Index of start of string in string literals
146 // Second Operand (next word in compiled code):
147 // the length of the string.
148 URX_BACKREF_I
= 41, // Case insensitive back reference.
149 // Parameter is the index of the
150 // capture group variables in the state stack frame.
151 URX_DOLLAR_M
= 42, // $ in multi-line mode.
152 URX_CARET_M
= 43, // ^ in multi-line mode.
153 URX_LB_START
= 44, // LookBehind Start.
154 // Paramater is data location
155 URX_LB_CONT
= 45, // LookBehind Continue.
156 // Param 0: the data location
157 // Param 1: The minimum length of the look-behind match
158 // Param 2: The max length of the look-behind match
159 URX_LB_END
= 46, // LookBehind End.
160 // Parameter is the data location.
161 // Check that match ended at the right spot,
162 // Restore original input string len.
163 URX_LBN_CONT
= 47, // Negative LookBehind Continue
164 // Param 0: the data location
165 // Param 1: The minimum length of the look-behind match
166 // Param 2: The max length of the look-behind match
167 // Param 3: The pattern loc following the look-behind block.
168 URX_LBN_END
= 48, // Negative LookBehind end
169 // Parameter is the data location.
170 // Check that the match ended at the right spot.
171 URX_STAT_SETREF_N
= 49, // Reference to a prebuilt set (e.g. \w), negated
172 // Operand is index of set in array of sets.
173 URX_LOOP_SR_I
= 50, // Init a [set]* loop.
174 // Operand is the sets index in array of user sets.
175 URX_LOOP_C
= 51, // Continue a [set]* or OneChar* loop.
176 // Operand is a matcher static data location.
177 // Must always immediately follow LOOP_x_I instruction.
178 URX_LOOP_DOT_I
= 52, // .*, initialization of the optimized loop.
181 // 0: Normal (. doesn't match new-line) mode.
182 // 1: . matches new-line mode.
183 // bit 1: controls what new-lines are recognized by this operation.
184 // 0: All Unicode New-lines
185 // 1: UNIX_LINES, \u000a only.
186 URX_BACKSLASH_BU
= 53, // \b or \B in UREGEX_UWORD mode, using Unicode style
188 URX_DOLLAR_D
= 54, // $ end of input test, in UNIX_LINES mode.
189 URX_DOLLAR_MD
= 55 // $ end of input test, in MULTI_LINE and UNIX_LINES mode.
193 // Keep this list of opcode names in sync with the above enum
194 // Used for debug printing only.
195 #define URX_OPCODE_NAMES \
206 "URX_STATIC_SETREF", \
226 "URX_CARET_M_UNIX", \
255 // Convenience macros for assembling and disassembling a compiled operation.
257 #define URX_BUILD(type, val) (int32_t)((type << 24) | (val))
258 #define URX_TYPE(x) ((uint32_t)(x) >> 24)
259 #define URX_VAL(x) ((x) & 0xffffff)
263 // Access to Unicode Sets composite character properties
264 // The sets are accessed by the match engine for things like \w (word boundary)
272 URX_GC_NORMAL
, // Sets for finding grapheme cluster boundaries.
283 URX_NEG_SET
= 0x800000 // Flag bit to reverse sense of set
289 // Match Engine State Stack Frame Layout.
291 struct REStackFrame
{
293 int64_t fInputIdx
; // Position of next character in the input string
294 int64_t fPatIdx
; // Position of next Op in the compiled pattern
295 // (int64_t for UVector64, values fit in an int32_t)
297 int64_t fExtra
[1]; // Extra state, for capture group start/ends
298 // atomic parentheses, repeat counts, etc.
299 // Locations assigned at pattern compile time.
300 // Variable-length array.
302 // number of UVector elements in the header
303 #define RESTACKFRAME_HDRCOUNT 2
306 // Start-Of-Match type. Used by find() to quickly scan to positions where a
307 // match might start before firing up the full match engine.
310 START_NO_INFO
, // No hint available.
311 START_CHAR
, // Match starts with a literal code point.
312 START_SET
, // Match starts with something matching a set.
313 START_START
, // Match starts at start of buffer only (^ or \A)
314 START_LINE
, // Match starts with ^ in multi-line mode.
315 START_STRING
// Match starts with a literal string.
318 #define START_OF_MATCH_STR(v) ((v)==START_NO_INFO? "START_NO_INFO" : \
319 (v)==START_CHAR? "START_CHAR" : \
320 (v)==START_SET? "START_SET" : \
321 (v)==START_START? "START_START" : \
322 (v)==START_LINE? "START_LINE" : \
323 (v)==START_STRING? "START_STRING" : \
327 // 8 bit set, to fast-path latin-1 set membership tests.
329 struct Regex8BitSet
: public UMemory
{
330 inline Regex8BitSet();
331 inline void operator = (const Regex8BitSet
&s
);
332 inline void init(const UnicodeSet
*src
);
333 inline UBool
contains(UChar32 c
);
334 inline void add(UChar32 c
);
338 inline Regex8BitSet::Regex8BitSet() {
339 uprv_memset(d
, 0, sizeof(d
));
342 inline UBool
Regex8BitSet::contains(UChar32 c
) {
343 // No bounds checking! This is deliberate.
344 return ((d
[c
>>3] & 1 <<(c
&7)) != 0);
347 inline void Regex8BitSet::add(UChar32 c
) {
348 d
[c
>>3] |= 1 << (c
&7);
351 inline void Regex8BitSet::init(const UnicodeSet
*s
) {
353 for (int32_t i
=0; i
<=255; i
++) {
354 if (s
->contains(i
)) {
361 inline void Regex8BitSet::operator = (const Regex8BitSet
&s
) {
362 uprv_memcpy(d
, s
.d
, sizeof(d
));
366 // Case folded UText Iterator helper class.
367 // Wraps a UText, provides a case-folded enumeration over its contents.
368 // Used in implementing case insensitive matching constructs.
369 // Implementation in rematch.cpp
371 class CaseFoldingUTextIterator
: public UMemory
{
373 CaseFoldingUTextIterator(UText
&text
);
374 ~CaseFoldingUTextIterator();
376 UChar32
next(); // Next case folded character
378 UBool
inExpansion(); // True if last char returned from next() and the
379 // next to be returned both originated from a string
380 // folding of the same code point from the orignal UText.
383 const UCaseProps
*fcsp
;
384 const UChar
*fFoldChars
;
391 // Case folded UChar * string iterator.
392 // Wraps a UChar *, provides a case-folded enumeration over its contents.
393 // Used in implementing case insensitive matching constructs.
394 // Implementation in rematch.cpp
396 class CaseFoldingUCharIterator
: public UMemory
{
398 CaseFoldingUCharIterator(const UChar
*chars
, int64_t start
, int64_t limit
);
399 ~CaseFoldingUCharIterator();
401 UChar32
next(); // Next case folded character
403 UBool
inExpansion(); // True if last char returned from next() and the
404 // next to be returned both originated from a string
405 // folding of the same code point from the orignal UText.
407 int64_t getIndex(); // Return the current input buffer index.
413 const UCaseProps
*fcsp
;
414 const UChar
*fFoldChars
;