/[pcre]/code/trunk/HACKING
ViewVC logotype

Diff of /code/trunk/HACKING

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

code/trunk/doc/Tech.Notes revision 77 by nigel, Sat Feb 24 21:40:45 2007 UTC code/trunk/HACKING revision 840 by ph10, Fri Dec 30 19:32:50 2011 UTC
# Line 1  Line 1 
1  Technical Notes about PCRE  Technical Notes about PCRE
2  --------------------------  --------------------------
3    
4    These are very rough technical notes that record potentially useful information
5    about PCRE internals. For information about testing PCRE, see the pcretest
6    documentation and the comment at the head of the RunTest file.
7    
8    
9  Historical note 1  Historical note 1
10  -----------------  -----------------
11    
# Line 13  not operate by backtracking, as the orig Line 18  not operate by backtracking, as the orig
18  Perl code does, but instead checked all possibilities simultaneously by keeping  Perl code does, but instead checked all possibilities simultaneously by keeping
19  a list of current states and checking all of them as it advanced through the  a list of current states and checking all of them as it advanced through the
20  subject string. In the terminology of Jeffrey Friedl's book, it was a "DFA  subject string. In the terminology of Jeffrey Friedl's book, it was a "DFA
21  algorithm". When the pattern was all used up, all remaining states were  algorithm", though it was not a traditional Finite State Machine (FSM). When
22  possible matches, and the one matching the longest subset of the subject string  the pattern was all used up, all remaining states were possible matches, and
23  was chosen. This did not necessarily maximize the individual wild portions of  the one matching the longest subset of the subject string was chosen. This did
24  the pattern, as is expected in Unix and Perl-style regular expressions.  not necessarily maximize the individual wild portions of the pattern, as is
25    expected in Unix and Perl-style regular expressions.
26    
27    
28  Historical note 2  Historical note 2
29  -----------------  -----------------
30    
31  By contrast, the code originally written by Henry Spencer and subsequently  By contrast, the code originally written by Henry Spencer (which was
32  heavily modified for Perl actually compiles the expression twice: once in a  subsequently heavily modified for Perl) compiles the expression twice: once in
33  dummy mode in order to find out how much store will be needed, and then for  a dummy mode in order to find out how much store will be needed, and then for
34  real. The execution function operates by backtracking and maximizing (or,  real. (The Perl version probably doesn't do this any more; I'm talking about
35  optionally, minimizing in Perl) the amount of the subject that matches  the original library.) The execution function operates by backtracking and
36  individual wild portions of the pattern. This is an "NFA algorithm" in Friedl's  maximizing (or, optionally, minimizing in Perl) the amount of the subject that
37  terminology.  matches individual wild portions of the pattern. This is an "NFA algorithm" in
38    Friedl's terminology.
39    
40    
41  OK, here's the real stuff  OK, here's the real stuff
42  -------------------------  -------------------------
# Line 37  unrelated to those mentioned above), I t Line 46  unrelated to those mentioned above), I t
46  that used an amount of store bounded by a multiple of the number of characters  that used an amount of store bounded by a multiple of the number of characters
47  in the pattern, to save on compiling time. However, because of the greater  in the pattern, to save on compiling time. However, because of the greater
48  complexity in Perl regular expressions, I couldn't do this. In any case, a  complexity in Perl regular expressions, I couldn't do this. In any case, a
49  first pass through the pattern is needed, for a number of reasons. PCRE works  first pass through the pattern is helpful for other reasons.
50  by running a very degenerate first pass to calculate a maximum store size, and  
51  then a second pass to do the real compile - which may use a bit less than the  
52  predicted amount of store. The idea is that this is going to turn out faster  Support for 16-bit data strings
53  because the first pass is degenerate and the second pass can just store stuff  -------------------------------
54  straight into the vector, which it knows is big enough. It does make the  
55  compiling functions bigger, of course, but they have got quite big anyway to  From release 8.30, PCRE supports 16-bit as well as 8-bit data strings, by being
56  handle all the Perl stuff.  compilable in either 8-bit or 16-bit modes, or both. Thus, two different
57    libraries can be created. In the description that follows, the word "short" is
58    used for a 16-bit data quantity, and the word "unit" is used for a quantity
59    that is a byte in 8-bit mode and a short in 16-bit mode. However, so as not to
60    over-complicate the text, the names of PCRE functions are given in 8-bit form
61    only.
62    
63    
64    Computing the memory requirement: how it was
65    --------------------------------------------
66    
67    Up to and including release 6.7, PCRE worked by running a very degenerate first
68    pass to calculate a maximum store size, and then a second pass to do the real
69    compile - which might use a bit less than the predicted amount of memory. The
70    idea was that this would turn out faster than the Henry Spencer code because
71    the first pass is degenerate and the second pass can just store stuff straight
72    into the vector, which it knows is big enough.
73    
74    
75    Computing the memory requirement: how it is
76    -------------------------------------------
77    
78    By the time I was working on a potential 6.8 release, the degenerate first pass
79    had become very complicated and hard to maintain. Indeed one of the early
80    things I did for 6.8 was to fix Yet Another Bug in the memory computation. Then
81    I had a flash of inspiration as to how I could run the real compile function in
82    a "fake" mode that enables it to compute how much memory it would need, while
83    actually only ever using a few hundred bytes of working memory, and without too
84    many tests of the mode that might slow it down. So I refactored the compiling
85    functions to work this way. This got rid of about 600 lines of source. It
86    should make future maintenance and development easier. As this was such a major
87    change, I never released 6.8, instead upping the number to 7.0 (other quite
88    major changes were also present in the 7.0 release).
89    
90    A side effect of this work was that the previous limit of 200 on the nesting
91    depth of parentheses was removed. However, there is a downside: pcre_compile()
92    runs more slowly than before (30% or more, depending on the pattern) because it
93    is doing a full analysis of the pattern. My hope was that this would not be a
94    big issue, and in the event, nobody has commented on it.
95    
96    
97  Traditional matching function  Traditional matching function
98  -----------------------------  -----------------------------
99    
100  The "traditional", and original, matching function is called pcre_exec(), and  The "traditional", and original, matching function is called pcre_exec(), and
101  it implements an NFA algorithm, similar to the original Henry Spencer algorithm  it implements an NFA algorithm, similar to the original Henry Spencer algorithm
102  and the way that Perl works. Not surprising, since it is intended to be as  and the way that Perl works. This is not surprising, since it is intended to be
103  compatible with Perl as possible. This is the function most users of PCRE will  as compatible with Perl as possible. This is the function most users of PCRE
104  use most of the time.  will use most of the time. From release 8.20, if PCRE is compiled with
105    just-in-time (JIT) support, and studying a compiled pattern with JIT is
106    successful, the JIT code is run instead of the normal pcre_exec() code, but the
107    result is the same.
108    
109    
110  Supplementary matching function  Supplementary matching function
111  -------------------------------  -------------------------------
# Line 63  pcre_dfa_exec(). This implements a DFA m Line 115  pcre_dfa_exec(). This implements a DFA m
115  simultaneously for all possible matches that start at one point in the subject  simultaneously for all possible matches that start at one point in the subject
116  string. (Going back to my roots: see Historical Note 1 above.) This function  string. (Going back to my roots: see Historical Note 1 above.) This function
117  intreprets the same compiled pattern data as pcre_exec(); however, not all the  intreprets the same compiled pattern data as pcre_exec(); however, not all the
118  facilities are available, and those that are don't always work in quite the  facilities are available, and those that are do not always work in quite the
119  same way. See the user documentation for details.  same way. See the user documentation for details.
120    
121    The algorithm that is used for pcre_dfa_exec() is not a traditional FSM,
122    because it may have a number of states active at one time. More work would be
123    needed at compile time to produce a traditional FSM where only one state is
124    ever active at once. I believe some other regex matchers work this way.
125    
126    
127    Changeable options
128    ------------------
129    
130    The /i, /m, or /s options (PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL) may
131    change in the middle of patterns. From PCRE 8.13, their processing is handled
132    entirely at compile time by generating different opcodes for the different
133    settings. The runtime functions do not need to keep track of an options state
134    any more.
135    
136    
137  Format of compiled patterns  Format of compiled patterns
138  ---------------------------  ---------------------------
139    
140  The compiled form of a pattern is a vector of bytes, containing items of  The compiled form of a pattern is a vector of units (bytes in 8-bit mode, or
141  variable length. The first byte in an item is an opcode, and the length of the  shorts in 16-bit mode), containing items of variable length. The first unit in
142  item is either implicit in the opcode or contained in the data bytes that  an item contains an opcode, and the length of the item is either implicit in
143  follow it.  the opcode or contained in the data that follows it.
144    
145  In many cases below "two-byte" data values are specified. This is in fact just  In many cases listed below, LINK_SIZE data values are specified for offsets
146  a default. PCRE can be compiled to use 3-byte or 4-byte values (impairing the  within the compiled pattern. LINK_SIZE always specifies a number of bytes. The
147  performance). This is necessary only when patterns whose compiled length is  default value for LINK_SIZE is 2, but PCRE can be compiled to use 3-byte or
148  greater than 64K are going to be processed. In this description, we assume the  4-byte values for these offsets, although this impairs the performance. (3-byte
149  "normal" compilation options.  LINK_SIZE values are available only in 8-bit mode.) Specifing a LINK_SIZE
150    larger than 2 is necessary only when patterns whose compiled length is greater
151  A list of all the opcodes follows:  than 64K are going to be processed. In this description, we assume the "normal"
152    compilation options. Data values that are counts (e.g. for quantifiers) are
153    always just two bytes long (one short in 16-bit mode).
154    
155  Opcodes with no following data  Opcodes with no following data
156  ------------------------------  ------------------------------
157    
158  These items are all just one byte long  These items are all just one unit long
159    
160    OP_END                 end of pattern    OP_END                 end of pattern
161    OP_ANY                 match any character    OP_ANY                 match any one character other than newline
162      OP_ALLANY              match any one character, including newline
163    OP_ANYBYTE             match any single byte, even in UTF-8 mode    OP_ANYBYTE             match any single byte, even in UTF-8 mode
164    OP_SOD                 match start of data: \A    OP_SOD                 match start of data: \A
165    OP_SOM,                start of match (subject + offset): \G    OP_SOM,                start of match (subject + offset): \G
166    OP_CIRC                ^ (start of data, or after \n in multiline)    OP_SET_SOM,            set start of match (\K)
167      OP_CIRC                ^ (start of data)
168      OP_CIRCM               ^ multiline mode (start of data or after newline)
169    OP_NOT_WORD_BOUNDARY   \W    OP_NOT_WORD_BOUNDARY   \W
170    OP_WORD_BOUNDARY       \w    OP_WORD_BOUNDARY       \w
171    OP_NOT_DIGIT           \D    OP_NOT_DIGIT           \D
172    OP_DIGIT               \d    OP_DIGIT               \d
173      OP_NOT_HSPACE          \H
174      OP_HSPACE              \h
175    OP_NOT_WHITESPACE      \S    OP_NOT_WHITESPACE      \S
176    OP_WHITESPACE          \s    OP_WHITESPACE          \s
177      OP_NOT_VSPACE          \V
178      OP_VSPACE              \v
179    OP_NOT_WORDCHAR        \W    OP_NOT_WORDCHAR        \W
180    OP_WORDCHAR            \w    OP_WORDCHAR            \w
181    OP_EODN                match end of data or \n at end: \Z    OP_EODN                match end of data or \n at end: \Z
182    OP_EOD                 match end of data: \z    OP_EOD                 match end of data: \z
183    OP_DOLL                $ (end of data, or before \n in multiline)    OP_DOLL                $ (end of data, or before final newline)
184      OP_DOLLM               $ multiline mode (end of data or before newline)
185    OP_EXTUNI              match an extended Unicode character    OP_EXTUNI              match an extended Unicode character
186      OP_ANYNL               match any Unicode newline sequence
187    
188      OP_ACCEPT              ) These are Perl 5.10's "backtracking control
189      OP_COMMIT              ) verbs". If OP_ACCEPT is inside capturing
190      OP_FAIL                ) parentheses, it may be preceded by one or more
191      OP_PRUNE               ) OP_CLOSE, followed by a 2-byte number,
192      OP_SKIP                ) indicating which parentheses must be closed.
193    
194    
195    Backtracking control verbs with (optional) data
196    -----------------------------------------------
197    
198    (*THEN) without an argument generates the opcode OP_THEN and no following data.
199    OP_MARK is followed by the mark name, preceded by a one-unit length, and
200    followed by a binary zero. For (*PRUNE), (*SKIP), and (*THEN) with arguments,
201    the opcodes OP_PRUNE_ARG, OP_SKIP_ARG, and OP_THEN_ARG are used, with the name
202    following in the same format.
203    
204    
205    Matching literal characters
206    ---------------------------
207    
208    The OP_CHAR opcode is followed by a single character that is to be matched
209    casefully. For caseless matching, OP_CHARI is used. In UTF-8 or UTF-16 modes,
210    the character may be more than one unit long.
211    
212    
213  Repeating single characters  Repeating single characters
214  ---------------------------  ---------------------------
215    
216  The common repeats (*, +, ?) when applied to a single character use the  The common repeats (*, +, ?), when applied to a single character, use the
217  following opcodes:  following opcodes, which come in caseful and caseless versions:
218    
219    OP_STAR    Caseful         Caseless
220    OP_MINSTAR    OP_STAR         OP_STARI
221    OP_PLUS    OP_MINSTAR      OP_MINSTARI
222    OP_MINPLUS    OP_POSSTAR      OP_POSSTARI
223    OP_QUERY    OP_PLUS         OP_PLUSI
224    OP_MINQUERY    OP_MINPLUS      OP_MINPLUSI
225      OP_POSPLUS      OP_POSPLUSI
226  In ASCII mode, these are two-byte items; in UTF-8 mode, the length is variable.    OP_QUERY        OP_QUERYI
227  Those with "MIN" in their name are the minimizing versions. Each is followed by    OP_MINQUERY     OP_MINQUERYI
228  the character that is to be repeated. Other repeats make use of    OP_POSQUERY     OP_POSQUERYI
229    
230    OP_UPTO  Each opcode is followed by the character that is to be repeated. In ASCII mode,
231    OP_MINUPTO  these are two-unit items; in UTF-8 or UTF-16 modes, the length is variable.
232    OP_EXACT  Those with "MIN" in their names are the minimizing versions. Those with "POS"
233    in their names are possessive versions. Other repeats make use of these
234  which are followed by a two-byte count (most significant first) and the  opcodes:
235  repeated character. OP_UPTO matches from 0 to the given number. A repeat with a  
236  non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an    Caseful         Caseless
237  OP_UPTO (or OP_MINUPTO).    OP_UPTO         OP_UPTOI
238      OP_MINUPTO      OP_MINUPTOI
239      OP_POSUPTO      OP_POSUPTOI
240      OP_EXACT        OP_EXACTI
241    
242    Each of these is followed by a two-byte (one short) count (most significant
243    byte first in 8-bit mode) and then the repeated character. OP_UPTO matches from
244    0 to the given number. A repeat with a non-zero minimum and a fixed maximum is
245    coded as an OP_EXACT followed by an OP_UPTO (or OP_MINUPTO or OPT_POSUPTO).
246    
247    
248  Repeating character types  Repeating character types
# Line 139  Repeating character types Line 250  Repeating character types
250    
251  Repeats of things like \d are done exactly as for single characters, except  Repeats of things like \d are done exactly as for single characters, except
252  that instead of a character, the opcode for the type is stored in the data  that instead of a character, the opcode for the type is stored in the data
253  byte. The opcodes are:  unit. The opcodes are:
254    
255    OP_TYPESTAR    OP_TYPESTAR
256    OP_TYPEMINSTAR    OP_TYPEMINSTAR
257      OP_TYPEPOSSTAR
258    OP_TYPEPLUS    OP_TYPEPLUS
259    OP_TYPEMINPLUS    OP_TYPEMINPLUS
260      OP_TYPEPOSPLUS
261    OP_TYPEQUERY    OP_TYPEQUERY
262    OP_TYPEMINQUERY    OP_TYPEMINQUERY
263      OP_TYPEPOSQUERY
264    OP_TYPEUPTO    OP_TYPEUPTO
265    OP_TYPEMINUPTO    OP_TYPEMINUPTO
266      OP_TYPEPOSUPTO
267    OP_TYPEEXACT    OP_TYPEEXACT
268    
269    
# Line 157  Match by Unicode property Line 272  Match by Unicode property
272    
273  OP_PROP and OP_NOTPROP are used for positive and negative matches of a  OP_PROP and OP_NOTPROP are used for positive and negative matches of a
274  character by testing its Unicode property (the \p and \P escape sequences).  character by testing its Unicode property (the \p and \P escape sequences).
275  Each is followed by a single byte that encodes the desired property value.  Each is followed by two units that encode the desired property as a type and a
276    value.
277    
278  Repeats of these items use the OP_TYPESTAR etc. set of opcodes, followed by two  Repeats of these items use the OP_TYPESTAR etc. set of opcodes, followed by
279  bytes: OP_PROP or OP_NOTPROP and then the desired property value.  three units: OP_PROP or OP_NOTPROP, and then the desired property type and
280    value.
   
 Matching literal characters  
 ---------------------------  
   
 The OP_CHAR opcode is followed by a single character that is to be matched  
 casefully. For caseless matching, OP_CHARNC is used. In UTF-8 mode, the  
 character may be more than one byte long. (Earlier versions of PCRE used  
 multi-character strings, but this was changed to allow some new features to be  
 added.)  
281    
282    
283  Character classes  Character classes
284  -----------------  -----------------
285    
286  If there is only one character, OP_CHAR or OP_CHARNC is used for a positive  If there is only one character in the class, OP_CHAR or OP_CHARI is used for a
287  class, and OP_NOT for a negative one (that is, for something like [^a]).  positive class, and OP_NOT or OP_NOTI for a negative one (that is, for
288  However, in UTF-8 mode, the use of OP_NOT applies only to characters with  something like [^a]). However, OP_NOT[I] can be used only with single-unit
289  values < 128, because OP_NOT is confined to single bytes.  characters, so in UTF-8 (UTF-16) mode, the use of OP_NOT[I] applies only to
290    characters whose code points are no greater than 127 (0xffff).
291  Another set of repeating opcodes (OP_NOTSTAR etc.) are used for a repeated,  
292  negated, single-character class. The normal ones (OP_STAR etc.) are used for a  Another set of 13 repeating opcodes (called OP_NOTSTAR etc.) are used for
293  repeated positive single-character class.  repeated, negated, single-character classes. The normal single-character
294    opcodes (OP_STAR, etc.) are used for repeated positive single-character
295  When there's more than one character in a class and all the characters are less  classes.
296  than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a negative  
297  one. In either case, the opcode is followed by a 32-byte bit map containing a 1  When there is more than one character in a class and all the characters are
298  bit for every character that is acceptable. The bits are counted from the least  less than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a
299  significant end of each byte.  negative one. In either case, the opcode is followed by a 32-byte (16-short)
300    bit map containing a 1 bit for every character that is acceptable. The bits are
301  The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8 mode,  counted from the least significant end of each unit. In caseless mode, bits for
302  subject characters with values greater than 256 can be handled correctly. For  both cases are set.
303  OP_CLASS they don't match, whereas for OP_NCLASS they do.  
304    The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8/16 mode,
305  For classes containing characters with values > 255, OP_XCLASS is used. It  subject characters with values greater than 255 can be handled correctly. For
306  optionally uses a bit map (if any characters lie within it), followed by a list  OP_CLASS they do not match, whereas for OP_NCLASS they do.
307  of pairs and single characters. There is a flag character than indicates  
308  whether it's a positive or a negative class.  For classes containing characters with values greater than 255, OP_XCLASS is
309    used. It optionally uses a bit map (if any characters lie within it), followed
310    by a list of pairs (for a range) and single characters. In caseless mode, both
311    cases are explicitly listed. There is a flag character than indicates whether
312    it is a positive or a negative class.
313    
314    
315  Back references  Back references
316  ---------------  ---------------
317    
318  OP_REF is followed by two bytes containing the reference number.  OP_REF (caseful) or OP_REFI (caseless) is followed by two bytes (one short)
319    containing the reference number.
320    
321    
322  Repeating character classes and back references  Repeating character classes and back references
323  -----------------------------------------------  -----------------------------------------------
324    
325  Single-character classes are handled specially (see above). This applies to  Single-character classes are handled specially (see above). This section
326  OP_CLASS and OP_REF. In both cases, the repeat information follows the base  applies to OP_CLASS and OP_REF[I]. In both cases, the repeat information
327  item. The matching code looks at the following opcode to see if it is one of  follows the base item. The matching code looks at the following opcode to see
328    if it is one of
329    
330    OP_CRSTAR    OP_CRSTAR
331    OP_CRMINSTAR    OP_CRMINSTAR
# Line 223  item. The matching code looks at the fol Line 336  item. The matching code looks at the fol
336    OP_CRRANGE    OP_CRRANGE
337    OP_CRMINRANGE    OP_CRMINRANGE
338    
339  All but the last two are just single-byte items. The others are followed by  All but the last two are just single-unit items. The others are followed by
340  four bytes of data, comprising the minimum and maximum repeat counts.  four bytes (two shorts) of data, comprising the minimum and maximum repeat
341    counts. There are no special possessive opcodes for these repeats; a possessive
342    repeat is compiled into an atomic group.
343    
344    
345  Brackets and alternation  Brackets and alternation
# Line 233  Brackets and alternation Line 348  Brackets and alternation
348  A pair of non-capturing (round) brackets is wrapped round each expression at  A pair of non-capturing (round) brackets is wrapped round each expression at
349  compile time, so alternation always happens in the context of brackets.  compile time, so alternation always happens in the context of brackets.
350    
351  Non-capturing brackets use the opcode OP_BRA, while capturing brackets use  [Note for North Americans: "bracket" to some English speakers, including
352  OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English  myself, can be round, square, curly, or pointy. Hence this usage rather than
353  speakers, including myself, can be round, square, curly, or pointy. Hence this  "parentheses".]
354  usage.]  
355    Non-capturing brackets use the opcode OP_BRA. Originally PCRE was limited to 99
356  Originally PCRE was limited to 99 capturing brackets (so as not to use up all  capturing brackets and it used a different opcode for each one. From release
357  the opcodes). From release 3.5, there is no limit. What happens is that the  3.5, the limit was removed by putting the bracket number into the data for
358  first ones, up to EXTRACT_BASIC_MAX are handled with separate opcodes, as  higher-numbered brackets. From release 7.0 all capturing brackets are handled
359  above. If there are more, the opcode is set to EXTRACT_BASIC_MAX+1, and the  this way, using the single opcode OP_CBRA.
 first operation in the bracket is OP_BRANUMBER, followed by a 2-byte bracket  
 number. This opcode is ignored while matching, but is fished out when handling  
 the bracket itself. (They could have all been done like this, but I was making  
 minimal changes.)  
360    
361  A bracket opcode is followed by LINK_SIZE bytes which give the offset to the  A bracket opcode is followed by LINK_SIZE bytes which give the offset to the
362  next alternative OP_ALT or, if there aren't any branches, to the matching  next alternative OP_ALT or, if there aren't any branches, to the matching
363  OP_KET opcode. Each OP_ALT is followed by LINK_SIZE bytes giving the offset to  OP_KET opcode. Each OP_ALT is followed by LINK_SIZE bytes giving the offset to
364  the next one, or to the OP_KET opcode.  the next one, or to the OP_KET opcode. For capturing brackets, the bracket
365    number immediately follows the offset, always as a 2-byte (one short) item.
366    
367  OP_KET is used for subpatterns that do not repeat indefinitely, while  OP_KET is used for subpatterns that do not repeat indefinitely, and
368  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
369  maximally respectively. All three are followed by LINK_SIZE bytes giving (as a  maximally respectively (see below for possessive repetitions). All three are
370  positive number) the offset back to the matching OP_BRA opcode.  followed by LINK_SIZE bytes giving (as a positive number) the offset back to
371    the matching bracket opcode.
372    
373  If a subpattern is quantified such that it is permitted to match zero times, it  If a subpattern is quantified such that it is permitted to match zero times, it
374  is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte  is preceded by one of OP_BRAZERO, OP_BRAMINZERO, or OP_SKIPZERO. These are
375  opcodes which tell the matcher that skipping this subpattern entirely is a  single-unit opcodes that tell the matcher that skipping the following
376  valid branch.  subpattern entirely is a valid branch. In the case of the first two, not
377    skipping the pattern is also valid (greedy and non-greedy). The third is used
378    when a pattern has the quantifier {0,0}. It cannot be entirely discarded,
379    because it may be called as a subroutine from elsewhere in the regex.
380    
381  A subpattern with an indefinite maximum repetition is replicated in the  A subpattern with an indefinite maximum repetition is replicated in the
382  compiled data its minimum number of times (or once with OP_BRAZERO if the  compiled data its minimum number of times (or once with OP_BRAZERO if the
# Line 270  as appropriate. Line 386  as appropriate.
386  A subpattern with a bounded maximum repetition is replicated in a nested  A subpattern with a bounded maximum repetition is replicated in a nested
387  fashion up to the maximum number of times, with OP_BRAZERO or OP_BRAMINZERO  fashion up to the maximum number of times, with OP_BRAZERO or OP_BRAMINZERO
388  before each replication after the minimum, so that, for example, (abc){2,5} is  before each replication after the minimum, so that, for example, (abc){2,5} is
389  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?.  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?, except that each bracketed group
390    has the same number.
391    
392    When a repeated subpattern has an unbounded upper limit, it is checked to see
393    whether it could match an empty string. If this is the case, the opcode in the
394    final replication is changed to OP_SBRA or OP_SCBRA. This tells the matcher
395    that it needs to check for matching an empty string when it hits OP_KETRMIN or
396    OP_KETRMAX, and if so, to break the loop.
397    
398    Possessive brackets
399    -------------------
400    
401    When a repeated group (capturing or non-capturing) is marked as possessive by
402    the "+" notation, e.g. (abc)++, different opcodes are used. Their names all
403    have POS on the end, e.g. OP_BRAPOS instead of OP_BRA and OP_SCPBRPOS instead
404    of OP_SCBRA. The end of such a group is marked by OP_KETRPOS. If the minimum
405    repetition is zero, the group is preceded by OP_BRAPOSZERO.
406    
407    
408  Assertions  Assertions
# Line 279  Assertions Line 411  Assertions
411  Forward assertions are just like other subpatterns, but starting with one of  Forward assertions are just like other subpatterns, but starting with one of
412  the opcodes OP_ASSERT or OP_ASSERT_NOT. Backward assertions use the opcodes  the opcodes OP_ASSERT or OP_ASSERT_NOT. Backward assertions use the opcodes
413  OP_ASSERTBACK and OP_ASSERTBACK_NOT, and the first opcode inside the assertion  OP_ASSERTBACK and OP_ASSERTBACK_NOT, and the first opcode inside the assertion
414  is OP_REVERSE, followed by a two byte count of the number of characters to move  is OP_REVERSE, followed by a two byte (one short) count of the number of
415  back the pointer in the subject string. When operating in UTF-8 mode, the count  characters to move back the pointer in the subject string. In ASCII mode, the
416  is a character count rather than a byte count. A separate count is present in  count is a number of units, but in UTF-8/16 mode each character may occupy more
417  each alternative of a lookbehind assertion, allowing them to have different  than one unit. A separate count is present in each alternative of a lookbehind
418  fixed lengths.  assertion, allowing them to have different fixed lengths.
419    
420    
421  Once-only subpatterns  Once-only (atomic) subpatterns
422  ---------------------  ------------------------------
423    
424  These are also just like other subpatterns, but they start with the opcode  These are also just like other subpatterns, but they start with the opcode
425  OP_ONCE.  OP_ONCE. The check for matching an empty string in an unbounded repeat is
426    handled entirely at runtime, so there is just this one opcode.
427    
428    
429  Conditional subpatterns  Conditional subpatterns
430  -----------------------  -----------------------
431    
432  These are like other subpatterns, but they start with the opcode OP_COND. If  These are like other subpatterns, but they start with the opcode OP_COND, or
433    OP_SCOND for one that might match an empty string in an unbounded repeat. If
434  the condition is a back reference, this is stored at the start of the  the condition is a back reference, this is stored at the start of the
435  subpattern using the opcode OP_CREF followed by two bytes containing the  subpattern using the opcode OP_CREF followed by two bytes (one short)
436  reference number. If the condition is "in recursion" (coded as "(?(R)"), the  containing the reference number. OP_NCREF is used instead if the reference was
437  same scheme is used, with a "reference number" of 0xffff. Otherwise, a  generated by name (so that the runtime code knows to check for duplicate
438  conditional subpattern always starts with one of the assertions.  names).
439    
440    If the condition is "in recursion" (coded as "(?(R)"), or "in recursion of
441    group x" (coded as "(?(Rx)"), the group number is stored at the start of the
442    subpattern using the opcode OP_RREF or OP_NRREF (cf OP_NCREF), and a value of
443    zero for "the whole pattern". For a DEFINE condition, just the single unit
444    OP_DEF is used (it has no associated data). Otherwise, a conditional subpattern
445    always starts with one of the assertions.
446    
447    
448  Recursion  Recursion
# Line 309  Recursion Line 450  Recursion
450    
451  Recursion either matches the current regex, or some subexpression. The opcode  Recursion either matches the current regex, or some subexpression. The opcode
452  OP_RECURSE is followed by an value which is the offset to the starting bracket  OP_RECURSE is followed by an value which is the offset to the starting bracket
453  from the start of the whole pattern.  from the start of the whole pattern. From release 6.5, OP_RECURSE is
454    automatically wrapped inside OP_ONCE brackets (because otherwise some patterns
455    broke it). OP_RECURSE is also used for "subroutine" calls, even though they
456    are not strictly a recursion.
457    
458    
459  Callout  Callout
460  -------  -------
461    
462  OP_CALLOUT is followed by one byte of data that holds a callout number in the  OP_CALLOUT is followed by one unit of data that holds a callout number in the
463  range 0 to 254 for manual callouts, or 255 for an automatic callout. In both  range 0 to 254 for manual callouts, or 255 for an automatic callout. In both
464  cases there follows a two-byte value giving the offset in the pattern to the  cases there follows a two-byte (one short) value giving the offset in the
465  start of the following item, and another two-byte item giving the length of the  pattern to the start of the following item, and another two-byte (one short)
466  next item.  item giving the length of the next item.
467    
   
 Changing options  
 ----------------  
   
 If any of the /i, /m, or /s options are changed within a pattern, an OP_OPT  
 opcode is compiled, followed by one byte containing the new settings of these  
 flags. If there are several alternatives, there is an occurrence of OP_OPT at  
 the start of all those following the first options change, to set appropriate  
 options for the start of the alternative. Immediately after the end of the  
 group there is another such item to reset the flags to their previous values. A  
 change of flag right at the very start of the pattern can be handled entirely  
 at compile time, and so does not cause anything to be put into the compiled  
 data.  
468    
469  Philip Hazel  Philip Hazel
470  March 2005  December 2011

Legend:
Removed from v.77  
changed lines
  Added in v.840

  ViewVC Help
Powered by ViewVC 1.1.5