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

Diff of /code/trunk/HACKING

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

revision 181 by ph10, Wed Jun 13 14:55:18 2007 UTC revision 604 by ph10, Thu Jun 2 19:04:54 2011 UTC
# Line 4  Technical Notes about PCRE Line 4  Technical Notes about PCRE
4  These are very rough technical notes that record potentially useful information  These are very rough technical notes that record potentially useful information
5  about PCRE internals.  about PCRE internals.
6    
7    
8  Historical note 1  Historical note 1
9  -----------------  -----------------
10    
# Line 22  the one matching the longest subset of t Line 23  the one matching the longest subset of t
23  not necessarily maximize the individual wild portions of the pattern, as is  not necessarily maximize the individual wild portions of the pattern, as is
24  expected in Unix and Perl-style regular expressions.  expected in Unix and Perl-style regular expressions.
25    
26    
27  Historical note 2  Historical note 2
28  -----------------  -----------------
29    
# Line 34  maximizing (or, optionally, minimizing i Line 36  maximizing (or, optionally, minimizing i
36  matches individual wild portions of the pattern. This is an "NFA algorithm" in  matches individual wild portions of the pattern. This is an "NFA algorithm" in
37  Friedl's terminology.  Friedl's terminology.
38    
39    
40  OK, here's the real stuff  OK, here's the real stuff
41  -------------------------  -------------------------
42    
# Line 44  in the pattern, to save on compiling tim Line 47  in the pattern, to save on compiling tim
47  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
48  first pass through the pattern is helpful for other reasons.  first pass through the pattern is helpful for other reasons.
49    
50    
51  Computing the memory requirement: how it was  Computing the memory requirement: how it was
52  --------------------------------------------  --------------------------------------------
53    
# Line 54  idea was that this would turn out faster Line 58  idea was that this would turn out faster
58  the first pass is degenerate and the second pass can just store stuff straight  the first pass is degenerate and the second pass can just store stuff straight
59  into the vector, which it knows is big enough.  into the vector, which it knows is big enough.
60    
61    
62  Computing the memory requirement: how it is  Computing the memory requirement: how it is
63  -------------------------------------------  -------------------------------------------
64    
# Line 63  things I did for 6.8 was to fix Yet Anot Line 68  things I did for 6.8 was to fix Yet Anot
68  I had a flash of inspiration as to how I could run the real compile function in  I had a flash of inspiration as to how I could run the real compile function in
69  a "fake" mode that enables it to compute how much memory it would need, while  a "fake" mode that enables it to compute how much memory it would need, while
70  actually only ever using a few hundred bytes of working memory, and without too  actually only ever using a few hundred bytes of working memory, and without too
71  many tests of the mode that might slow it down. So I re-factored the compiling  many tests of the mode that might slow it down. So I refactored the compiling
72  functions to work this way. This got rid of about 600 lines of source. It  functions to work this way. This got rid of about 600 lines of source. It
73  should make future maintenance and development easier. As this was such a major  should make future maintenance and development easier. As this was such a major
74  change, I never released 6.8, instead upping the number to 7.0 (other quite  change, I never released 6.8, instead upping the number to 7.0 (other quite
75  major changes are also present in the 7.0 release).  major changes were also present in the 7.0 release).
76    
77  A side effect of this work is that the previous limit of 200 on the nesting  A side effect of this work was that the previous limit of 200 on the nesting
78  depth of parentheses was removed. However, there is a downside: pcre_compile()  depth of parentheses was removed. However, there is a downside: pcre_compile()
79  runs more slowly than before (30% or more, depending on the pattern) because it  runs more slowly than before (30% or more, depending on the pattern) because it
80  is doing a full analysis of the pattern. My hope is that this is not a big  is doing a full analysis of the pattern. My hope was that this would not be a
81  issue.  big issue, and in the event, nobody has commented on it.
82    
83    
84  Traditional matching function  Traditional matching function
85  -----------------------------  -----------------------------
86    
87  The "traditional", and original, matching function is called pcre_exec(), and  The "traditional", and original, matching function is called pcre_exec(), and
88  it implements an NFA algorithm, similar to the original Henry Spencer algorithm  it implements an NFA algorithm, similar to the original Henry Spencer algorithm
89  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
90  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
91  use most of the time.  will use most of the time.
92    
93    
94  Supplementary matching function  Supplementary matching function
95  -------------------------------  -------------------------------
# Line 101  needed at compile time to produce a trad Line 108  needed at compile time to produce a trad
108  ever active at once. I believe some other regex matchers work this way.  ever active at once. I believe some other regex matchers work this way.
109    
110    
111    Changeable options
112    ------------------
113    
114    The /i, /m, or /s options (PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL) may
115    change in the middle of patterns. From PCRE 8.13, their processing is handled
116    entirely at compile time by generating different opcodes for the different
117    settings. The runtime functions do not need to keep track of an options state
118    any more.
119    
120    
121  Format of compiled patterns  Format of compiled patterns
122  ---------------------------  ---------------------------
123    
# Line 109  variable length. The first byte in an it Line 126  variable length. The first byte in an it
126  item is either implicit in the opcode or contained in the data bytes that  item is either implicit in the opcode or contained in the data bytes that
127  follow it.  follow it.
128    
129  In many cases below "two-byte" data values are specified. This is in fact just  In many cases below LINK_SIZE data values are specified for offsets within the
130  a default when the number is an offset within the compiled pattern. PCRE can be  compiled pattern. The default value for LINK_SIZE is 2, but PCRE can be
131  compiled to use 3-byte or 4-byte values for these offsets (impairing the  compiled to use 3-byte or 4-byte values for these offsets (impairing the
132  performance). This is necessary only when patterns whose compiled length is  performance). This is necessary only when patterns whose compiled length is
133  greater than 64K are going to be processed. In this description, we assume the  greater than 64K are going to be processed. In this description, we assume the
134  "normal" compilation options. "Two-byte" data values that are counts (e.g. for  "normal" compilation options. Data values that are counts (e.g. for
135  quantifiers) are always just two bytes.  quantifiers) are always just two bytes long.
   
 A list of all the opcodes follows:  
136    
137  Opcodes with no following data  Opcodes with no following data
138  ------------------------------  ------------------------------
# Line 125  Opcodes with no following data Line 140  Opcodes with no following data
140  These items are all just one byte long  These items are all just one byte long
141    
142    OP_END                 end of pattern    OP_END                 end of pattern
143    OP_ANY                 match any character    OP_ANY                 match any one character other than newline
144      OP_ALLANY              match any one character, including newline
145    OP_ANYBYTE             match any single byte, even in UTF-8 mode    OP_ANYBYTE             match any single byte, even in UTF-8 mode
146    OP_SOD                 match start of data: \A    OP_SOD                 match start of data: \A
147    OP_SOM,                start of match (subject + offset): \G    OP_SOM,                start of match (subject + offset): \G
148    OP_SET_SOM,            set start of match (\K)    OP_SET_SOM,            set start of match (\K)
149    OP_CIRC                ^ (start of data, or after \n in multiline)    OP_CIRC                ^ (start of data)
150      OP_CIRCM               ^ multiline mode (start of data or after newline)
151    OP_NOT_WORD_BOUNDARY   \W    OP_NOT_WORD_BOUNDARY   \W
152    OP_WORD_BOUNDARY       \w    OP_WORD_BOUNDARY       \w
153    OP_NOT_DIGIT           \D    OP_NOT_DIGIT           \D
# Line 145  These items are all just one byte long Line 162  These items are all just one byte long
162    OP_WORDCHAR            \w    OP_WORDCHAR            \w
163    OP_EODN                match end of data or \n at end: \Z    OP_EODN                match end of data or \n at end: \Z
164    OP_EOD                 match end of data: \z    OP_EOD                 match end of data: \z
165    OP_DOLL                $ (end of data, or before \n in multiline)    OP_DOLL                $ (end of data, or before final newline)
166      OP_DOLLM               $ multiline mode (end of data or before newline)
167    OP_EXTUNI              match an extended Unicode character    OP_EXTUNI              match an extended Unicode character
168    OP_ANYNL               match any Unicode newline sequence    OP_ANYNL               match any Unicode newline sequence
169    
170      OP_ACCEPT              ) These are Perl 5.10's "backtracking control
171      OP_COMMIT              ) verbs". If OP_ACCEPT is inside capturing
172      OP_FAIL                ) parentheses, it may be preceded by one or more
173      OP_PRUNE               ) OP_CLOSE, followed by a 2-byte number,
174      OP_SKIP                ) indicating which parentheses must be closed.
175    
176    
177    Backtracking control verbs with data
178    ------------------------------------
179    
180    OP_THEN is followed by a LINK_SIZE offset, which is the distance back to the
181    start of the current branch.
182    
183    OP_MARK is followed by the mark name, preceded by a one-byte length, and
184    followed by a binary zero. For (*PRUNE), (*SKIP), and (*THEN) with arguments,
185    the opcodes OP_PRUNE_ARG, OP_SKIP_ARG, and OP_THEN_ARG are used. For the first
186    two, the name follows immediately; for OP_THEN_ARG, it follows the LINK_SIZE
187    offset value.
188    
189    
190    Matching literal characters
191    ---------------------------
192    
193    The OP_CHAR opcode is followed by a single character that is to be matched
194    casefully. For caseless matching, OP_CHARI is used. In UTF-8 mode, the
195    character may be more than one byte long. (Earlier versions of PCRE used
196    multi-character strings, but this was changed to allow some new features to be
197    added.)
198    
199    
200  Repeating single characters  Repeating single characters
201  ---------------------------  ---------------------------
202    
203  The common repeats (*, +, ?) when applied to a single character use the  The common repeats (*, +, ?) when applied to a single character use the
204  following opcodes:  following opcodes, which come in caseful and caseless versions:
205    
206    OP_STAR    Caseful         Caseless
207    OP_MINSTAR    OP_STAR         OP_STARI
208    OP_POSSTAR    OP_MINSTAR      OP_MINSTARI
209    OP_PLUS    OP_POSSTAR      OP_POSSTARI
210    OP_MINPLUS    OP_PLUS         OP_PLUSI
211    OP_POSPLUS    OP_MINPLUS      OP_MINPLUSI
212    OP_QUERY    OP_POSPLUS      OP_POSPLUSI
213    OP_MINQUERY    OP_QUERY        OP_QUERYI
214    OP_POSQUERY    OP_MINQUERY     OP_MINQUERYI
215      OP_POSQUERY     OP_POSQUERYI
216    
217  In ASCII mode, these are two-byte items; in UTF-8 mode, the length is variable.  In ASCII mode, these are two-byte items; in UTF-8 mode, the length is variable.
218  Those with "MIN" in their name are the minimizing versions. Those with "POS" in  Those with "MIN" in their name are the minimizing versions. Those with "POS" in
219  their names are possessive versions. Each is followed by the character that is  their names are possessive versions. Each is followed by the character that is
220  to be repeated. Other repeats make use of  to be repeated. Other repeats make use of these opcodes:
221    
222    OP_UPTO    Caseful         Caseless
223    OP_MINUPTO    OP_UPTO         OP_UPTOI
224    OP_POSUPTO    OP_MINUPTO      OP_MINUPTOI
225    OP_EXACT    OP_POSUPTO      OP_POSUPTOI
226      OP_EXACT        OP_EXACTI
227    
228  which are followed by a two-byte count (most significant first) and the  Each of these is followed by a two-byte count (most significant first) and the
229  repeated character. OP_UPTO matches from 0 to the given number. A repeat with a  repeated character. OP_UPTO matches from 0 to the given number. A repeat with a
230  non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an  non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an
231  OP_UPTO (or OP_MINUPTO or OPT_POSUPTO).  OP_UPTO (or OP_MINUPTO or OPT_POSUPTO).
# Line 217  three bytes: OP_PROP or OP_NOTPROP and t Line 266  three bytes: OP_PROP or OP_NOTPROP and t
266  value.  value.
267    
268    
 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.)  
   
   
269  Character classes  Character classes
270  -----------------  -----------------
271    
272  If there is only one character, OP_CHAR or OP_CHARNC is used for a positive  If there is only one character, OP_CHAR or OP_CHARI is used for a positive
273  class, and OP_NOT for a negative one (that is, for something like [^a]).  class, and OP_NOT or OP_NOTI for a negative one (that is, for something like
274  However, in UTF-8 mode, the use of OP_NOT applies only to characters with  [^a]). However, in UTF-8 mode, the use of OP_NOT[I] applies only to characters
275  values < 128, because OP_NOT is confined to single bytes.  with values < 128, because OP_NOT[I] is confined to single bytes.
276    
277  Another set of repeating opcodes (OP_NOTSTAR etc.) are used for a repeated,  Another set of 13 repeating opcodes (called OP_NOTSTAR etc.) are used for a
278  negated, single-character class. The normal ones (OP_STAR etc.) are used for a  repeated, negated, single-character class. The normal single-character opcodes
279  repeated positive single-character class.  (OP_STAR, etc.) are used for a repeated positive single-character class.
280    
281  When there's more than one character in a class and all the characters are less  When there is more than one character in a class and all the characters are
282  than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a negative  less than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a
283  one. In either case, the opcode is followed by a 32-byte bit map containing a 1  negative one. In either case, the opcode is followed by a 32-byte bit map
284  bit for every character that is acceptable. The bits are counted from the least  containing a 1 bit for every character that is acceptable. The bits are counted
285  significant end of each byte.  from the least significant end of each byte. In caseless mode, bits for both
286    cases are set.
287    
288  The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8 mode,  The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8 mode,
289  subject characters with values greater than 256 can be handled correctly. For  subject characters with values greater than 256 can be handled correctly. For
290  OP_CLASS they don't match, whereas for OP_NCLASS they do.  OP_CLASS they do not match, whereas for OP_NCLASS they do.
291    
292  For classes containing characters with values > 255, OP_XCLASS is used. It  For classes containing characters with values > 255, OP_XCLASS is used. It
293  optionally uses a bit map (if any characters lie within it), followed by a list  optionally uses a bit map (if any characters lie within it), followed by a list
294  of pairs and single characters. There is a flag character than indicates  of pairs (for a range) and single characters. In caseless mode, both cases are
295  whether it's a positive or a negative class.  explicitly listed. There is a flag character than indicates whether it is a
296    positive or a negative class.
297    
298    
299  Back references  Back references
300  ---------------  ---------------
301    
302  OP_REF is followed by two bytes containing the reference number.  OP_REF (caseful) or OP_REFI (caseless) is followed by two bytes containing the
303    reference number.
304    
305    
306  Repeating character classes and back references  Repeating character classes and back references
307  -----------------------------------------------  -----------------------------------------------
308    
309  Single-character classes are handled specially (see above). This section  Single-character classes are handled specially (see above). This section
310  applies to OP_CLASS and OP_REF. In both cases, the repeat information follows  applies to OP_CLASS and OP_REF[I]. In both cases, the repeat information
311  the base item. The matching code looks at the following opcode to see if it is  follows the base item. The matching code looks at the following opcode to see
312  one of  if it is one of
313    
314    OP_CRSTAR    OP_CRSTAR
315    OP_CRMINSTAR    OP_CRMINSTAR
# Line 307  number immediately follows the offset, a Line 349  number immediately follows the offset, a
349    
350  OP_KET is used for subpatterns that do not repeat indefinitely, while  OP_KET is used for subpatterns that do not repeat indefinitely, while
351  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
352  maximally respectively. All three are followed by LINK_SIZE bytes giving (as a  maximally respectively (see below for possessive repetitions). All three are
353  positive number) the offset back to the matching bracket opcode.  followed by LINK_SIZE bytes giving (as a positive number) the offset back to
354    the matching bracket opcode.
355    
356  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
357  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
358  opcodes which tell the matcher that skipping this subpattern entirely is a  single-byte opcodes that tell the matcher that skipping the following
359  valid branch.  subpattern entirely is a valid branch. In the case of the first two, not
360    skipping the pattern is also valid (greedy and non-greedy). The third is used
361    when a pattern has the quantifier {0,0}. It cannot be entirely discarded,
362    because it may be called as a subroutine from elsewhere in the regex.
363    
364  A subpattern with an indefinite maximum repetition is replicated in the  A subpattern with an indefinite maximum repetition is replicated in the
365  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 332  final replication is changed to OP_SBRA Line 378  final replication is changed to OP_SBRA
378  that it needs to check for matching an empty string when it hits OP_KETRMIN or  that it needs to check for matching an empty string when it hits OP_KETRMIN or
379  OP_KETRMAX, and if so, to break the loop.  OP_KETRMAX, and if so, to break the loop.
380    
381    Possessive brackets
382    -------------------
383    
384    When a repeated group (capturing or non-capturing) is marked as possessive by
385    the "+" notation, e.g. (abc)++, different opcodes are used. Their names all
386    have POS on the end, e.g. OP_BRAPOS instead of OP_BRA and OP_SCPBRPOS instead
387    of OP_SCBRA. The end of such a group is marked by OP_KETRPOS. If the minimum
388    repetition is zero, the group is preceded by OP_BRAPOSZERO.
389    
390    
391  Assertions  Assertions
392  ----------  ----------
# Line 361  These are like other subpatterns, but th Line 416  These are like other subpatterns, but th
416  OP_SCOND for one that might match an empty string in an unbounded repeat. If  OP_SCOND for one that might match an empty string in an unbounded repeat. If
417  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
418  subpattern using the opcode OP_CREF followed by two bytes containing the  subpattern using the opcode OP_CREF followed by two bytes containing the
419  reference number. If the condition is "in recursion" (coded as "(?(R)"), or "in  reference number. OP_NCREF is used instead if the reference was generated by
420  recursion of group x" (coded as "(?(Rx)"), the group number is stored at the  name (so that the runtime code knows to check for duplicate names).
421  start of the subpattern using the opcode OP_RREF, and a value of zero for "the  
422  whole pattern". For a DEFINE condition, just the single byte OP_DEF is used (it  If the condition is "in recursion" (coded as "(?(R)"), or "in recursion of
423  has no associated data). Otherwise, a conditional subpattern always starts with  group x" (coded as "(?(Rx)"), the group number is stored at the start of the
424  one of the assertions.  subpattern using the opcode OP_RREF or OP_NRREF (cf OP_NCREF), and a value of
425    zero for "the whole pattern". For a DEFINE condition, just the single byte
426    OP_DEF is used (it has no associated data). Otherwise, a conditional subpattern
427    always starts with one of the assertions.
428    
429    
430  Recursion  Recursion
# Line 390  start of the following item, and another Line 448  start of the following item, and another
448  next item.  next item.
449    
450    
 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.  
   
451  Philip Hazel  Philip Hazel
452  June 2007  May 2011

Legend:
Removed from v.181  
changed lines
  Added in v.604

  ViewVC Help
Powered by ViewVC 1.1.5