/[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 75 by nigel, Sat Feb 24 21:40:37 2007 UTC code/trunk/HACKING revision 212 by ph10, Thu Aug 9 11:16:34 2007 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.
6    
7  Historical note 1  Historical note 1
8  -----------------  -----------------
9    
# Line 13  not operate by backtracking, as the orig Line 16  not operate by backtracking, as the orig
16  Perl code does, but instead checked all possibilities simultaneously by keeping  Perl code does, but instead checked all possibilities simultaneously by keeping
17  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
18  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
19  algorithm". When the pattern was all used up, all remaining states were  algorithm", though it was not a traditional Finite State Machine (FSM). When
20  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
21  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
22  the pattern, as is expected in Unix and Perl-style regular expressions.  not necessarily maximize the individual wild portions of the pattern, as is
23    expected in Unix and Perl-style regular expressions.
24    
25  Historical note 2  Historical note 2
26  -----------------  -----------------
27    
28  By contrast, the code originally written by Henry Spencer and subsequently  By contrast, the code originally written by Henry Spencer (which was
29  heavily modified for Perl actually compiles the expression twice: once in a  subsequently heavily modified for Perl) compiles the expression twice: once in
30  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
31  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
32  optionally, minimizing in Perl) the amount of the subject that matches  the original library.) The execution function operates by backtracking and
33  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
34  terminology.  matches individual wild portions of the pattern. This is an "NFA algorithm" in
35    Friedl's terminology.
36    
37  OK, here's the real stuff  OK, here's the real stuff
38  -------------------------  -------------------------
39    
40  For the set of functions that forms PCRE (which are unrelated to those  For the set of functions that form the "basic" PCRE library (which are
41  mentioned above), I tried at first to invent an algorithm that used an amount  unrelated to those mentioned above), I tried at first to invent an algorithm
42  of store bounded by a multiple of the number of characters in the pattern, to  that used an amount of store bounded by a multiple of the number of characters
43  save on compiling time. However, because of the greater complexity in Perl  in the pattern, to save on compiling time. However, because of the greater
44  regular expressions, I couldn't do this. In any case, a first pass through the  complexity in Perl regular expressions, I couldn't do this. In any case, a
45  pattern is needed, for a number of reasons. PCRE works by running a very  first pass through the pattern is helpful for other reasons.
46  degenerate first pass to calculate a maximum store size, and then a second pass  
47  to do the real compile - which may use a bit less than the predicted amount of  Computing the memory requirement: how it was
48  store. The idea is that this is going to turn out faster because the first pass  --------------------------------------------
49  is degenerate and the second pass can just store stuff straight into the  
50  vector. It does make the compiling functions bigger, of course, but they have  Up to and including release 6.7, PCRE worked by running a very degenerate first
51  got quite big anyway to handle all the Perl stuff.  pass to calculate a maximum store size, and then a second pass to do the real
52    compile - which might use a bit less than the predicted amount of memory. The
53    idea was that this would turn out faster than the Henry Spencer code because
54    the first pass is degenerate and the second pass can just store stuff straight
55    into the vector, which it knows is big enough.
56    
57    Computing the memory requirement: how it is
58    -------------------------------------------
59    
60    By the time I was working on a potential 6.8 release, the degenerate first pass
61    had become very complicated and hard to maintain. Indeed one of the early
62    things I did for 6.8 was to fix Yet Another Bug in the memory computation. Then
63    I had a flash of inspiration as to how I could run the real compile function in
64    a "fake" mode that enables it to compute how much memory it would need, while
65    actually only ever using a few hundred bytes of working memory, and without too
66    many tests of the mode that might slow it down. So I re-factored the compiling
67    functions to work this way. This got rid of about 600 lines of source. It
68    should make future maintenance and development easier. As this was such a major
69    change, I never released 6.8, instead upping the number to 7.0 (other quite
70    major changes are also present in the 7.0 release).
71    
72    A side effect of this work is that the previous limit of 200 on the nesting
73    depth of parentheses was removed. However, there is a downside: pcre_compile()
74    runs more slowly than before (30% or more, depending on the pattern) because it
75    is doing a full analysis of the pattern. My hope is that this is not a big
76    issue.
77    
78    Traditional matching function
79    -----------------------------
80    
81    The "traditional", and original, matching function is called pcre_exec(), and
82    it implements an NFA algorithm, similar to the original Henry Spencer algorithm
83    and the way that Perl works. Not surprising, since it is intended to be as
84    compatible with Perl as possible. This is the function most users of PCRE will
85    use most of the time.
86    
87    Supplementary matching function
88    -------------------------------
89    
90    From PCRE 6.0, there is also a supplementary matching function called
91    pcre_dfa_exec(). This implements a DFA matching algorithm that searches
92    simultaneously for all possible matches that start at one point in the subject
93    string. (Going back to my roots: see Historical Note 1 above.) This function
94    intreprets the same compiled pattern data as pcre_exec(); however, not all the
95    facilities are available, and those that are do not always work in quite the
96    same way. See the user documentation for details.
97    
98    The algorithm that is used for pcre_dfa_exec() is not a traditional FSM,
99    because it may have a number of states active at one time. More work would be
100    needed at compile time to produce a traditional FSM where only one state is
101    ever active at once. I believe some other regex matchers work this way.
102    
103    
104    Format of compiled patterns
105    ---------------------------
106    
107  The compiled form of a pattern is a vector of bytes, containing items of  The compiled form of a pattern is a vector of bytes, containing items of
108  variable length. The first byte in an item is an opcode, and the length of the  variable length. The first byte in an item is an opcode, and the length of the
109  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
110  follow it.  follow it.
111    
112  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
113  a default. PCRE can be compiled to use 3-byte or 4-byte values (impairing the  compiled pattern. The default value for LINK_SIZE is 2, but PCRE can be
114    compiled to use 3-byte or 4-byte values for these offsets (impairing the
115  performance). This is necessary only when patterns whose compiled length is  performance). This is necessary only when patterns whose compiled length is
116  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
117  "normal" compilation options.  "normal" compilation options. Data values that are counts (e.g. for
118    quantifiers) are always just two bytes long.
119    
120  A list of all the opcodes follows:  A list of the opcodes follows:
121    
122  Opcodes with no following data  Opcodes with no following data
123  ------------------------------  ------------------------------
# Line 68  These items are all just one byte long Line 129  These items are all just one byte long
129    OP_ANYBYTE             match any single byte, even in UTF-8 mode    OP_ANYBYTE             match any single byte, even in UTF-8 mode
130    OP_SOD                 match start of data: \A    OP_SOD                 match start of data: \A
131    OP_SOM,                start of match (subject + offset): \G    OP_SOM,                start of match (subject + offset): \G
132      OP_SET_SOM,            set start of match (\K)
133    OP_CIRC                ^ (start of data, or after \n in multiline)    OP_CIRC                ^ (start of data, or after \n in multiline)
134    OP_NOT_WORD_BOUNDARY   \W    OP_NOT_WORD_BOUNDARY   \W
135    OP_WORD_BOUNDARY       \w    OP_WORD_BOUNDARY       \w
136    OP_NOT_DIGIT           \D    OP_NOT_DIGIT           \D
137    OP_DIGIT               \d    OP_DIGIT               \d
138      OP_NOT_HSPACE          \H
139      OP_HSPACE              \h
140    OP_NOT_WHITESPACE      \S    OP_NOT_WHITESPACE      \S
141    OP_WHITESPACE          \s    OP_WHITESPACE          \s
142      OP_NOT_VSPACE          \V
143      OP_VSPACE              \v
144    OP_NOT_WORDCHAR        \W    OP_NOT_WORDCHAR        \W
145    OP_WORDCHAR            \w    OP_WORDCHAR            \w
146    OP_EODN                match end of data or \n at end: \Z    OP_EODN                match end of data or \n at end: \Z
147    OP_EOD                 match end of data: \z    OP_EOD                 match end of data: \z
148    OP_DOLL                $ (end of data, or before \n in multiline)    OP_DOLL                $ (end of data, or before \n in multiline)
149    OP_EXTUNI              match an extended Unicode character    OP_EXTUNI              match an extended Unicode character
150      OP_ANYNL               match any Unicode newline sequence
151    
152      OP_ACCEPT              )
153      OP_COMMIT              )
154      OP_FAIL                ) These are Perl 5.10's "backtracking
155      OP_PRUNE               ) control verbs".
156      OP_SKIP                )
157      OP_THEN                )
158    
159    
160  Repeating single characters  Repeating single characters
# Line 91  following opcodes: Line 165  following opcodes:
165    
166    OP_STAR    OP_STAR
167    OP_MINSTAR    OP_MINSTAR
168      OP_POSSTAR
169    OP_PLUS    OP_PLUS
170    OP_MINPLUS    OP_MINPLUS
171      OP_POSPLUS
172    OP_QUERY    OP_QUERY
173    OP_MINQUERY    OP_MINQUERY
174      OP_POSQUERY
175    
176  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.
177  Those with "MIN" in their name are the minimizing versions. Each is followed by  Those with "MIN" in their name are the minimizing versions. Those with "POS" in
178  the character that is to be repeated. Other repeats make use of  their names are possessive versions. Each is followed by the character that is
179    to be repeated. Other repeats make use of
180    
181    OP_UPTO    OP_UPTO
182    OP_MINUPTO    OP_MINUPTO
183      OP_POSUPTO
184    OP_EXACT    OP_EXACT
185    
186  which are followed by a two-byte count (most significant first) and the  which are followed by a two-byte count (most significant first) and the
187  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
188  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
189  OP_UPTO (or OP_MINUPTO).  OP_UPTO (or OP_MINUPTO or OPT_POSUPTO).
190    
191    
192  Repeating character types  Repeating character types
# Line 119  byte. The opcodes are: Line 198  byte. The opcodes are:
198    
199    OP_TYPESTAR    OP_TYPESTAR
200    OP_TYPEMINSTAR    OP_TYPEMINSTAR
201      OP_TYPEPOSSTAR
202    OP_TYPEPLUS    OP_TYPEPLUS
203    OP_TYPEMINPLUS    OP_TYPEMINPLUS
204      OP_TYPEPOSPLUS
205    OP_TYPEQUERY    OP_TYPEQUERY
206    OP_TYPEMINQUERY    OP_TYPEMINQUERY
207      OP_TYPEPOSQUERY
208    OP_TYPEUPTO    OP_TYPEUPTO
209    OP_TYPEMINUPTO    OP_TYPEMINUPTO
210      OP_TYPEPOSUPTO
211    OP_TYPEEXACT    OP_TYPEEXACT
212    
213    
# Line 133  Match by Unicode property Line 216  Match by Unicode property
216    
217  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
218  character by testing its Unicode property (the \p and \P escape sequences).  character by testing its Unicode property (the \p and \P escape sequences).
219  Each is followed by a single byte that encodes the desired property value.  Each is followed by two bytes that encode the desired property as a type and a
220    value.
221    
222  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
223  bytes: OP_PROP or OP_NOTPROP and then the desired property value.  three bytes: OP_PROP or OP_NOTPROP and then the desired property type and
224    value.
225    
226    
227  Matching literal characters  Matching literal characters
# Line 186  OP_REF is followed by two bytes containi Line 271  OP_REF is followed by two bytes containi
271  Repeating character classes and back references  Repeating character classes and back references
272  -----------------------------------------------  -----------------------------------------------
273    
274  Single-character classes are handled specially (see above). This applies to  Single-character classes are handled specially (see above). This section
275  OP_CLASS and OP_REF. In both cases, the repeat information follows the base  applies to OP_CLASS and OP_REF. In both cases, the repeat information follows
276  item. The matching code looks at the following opcode to see if it is one of  the base item. The matching code looks at the following opcode to see if it is
277    one of
278    
279    OP_CRSTAR    OP_CRSTAR
280    OP_CRMINSTAR    OP_CRMINSTAR
# Line 200  item. The matching code looks at the fol Line 286  item. The matching code looks at the fol
286    OP_CRMINRANGE    OP_CRMINRANGE
287    
288  All but the last two are just single-byte items. The others are followed by  All but the last two are just single-byte items. The others are followed by
289  four bytes of data, comprising the minimum and maximum repeat counts.  four bytes of data, comprising the minimum and maximum repeat counts. There are
290    no special possessive opcodes for these repeats; a possessive repeat is
291    compiled into an atomic group.
292    
293    
294  Brackets and alternation  Brackets and alternation
# Line 209  Brackets and alternation Line 297  Brackets and alternation
297  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
298  compile time, so alternation always happens in the context of brackets.  compile time, so alternation always happens in the context of brackets.
299    
300  Non-capturing brackets use the opcode OP_BRA, while capturing brackets use  [Note for North Americans: "bracket" to some English speakers, including
301  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.]
302  speakers, including myself, can be round, square, curly, or pointy. Hence this  
303  usage.]  Non-capturing brackets use the opcode OP_BRA. Originally PCRE was limited to 99
304    capturing brackets and it used a different opcode for each one. From release
305  Originally PCRE was limited to 99 capturing brackets (so as not to use up all  3.5, the limit was removed by putting the bracket number into the data for
306  the opcodes). From release 3.5, there is no limit. What happens is that the  higher-numbered brackets. From release 7.0 all capturing brackets are handled
307  first ones, up to EXTRACT_BASIC_MAX are handled with separate opcodes, as  this way, using the single opcode OP_CBRA.
308  above. If there are more, the opcode is set to EXTRACT_BASIC_MAX+1, and the  
309  first operation in the bracket is OP_BRANUMBER, followed by a 2-byte bracket  A bracket opcode is followed by LINK_SIZE bytes which give the offset to the
310  number. This opcode is ignored while matching, but is fished out when handling  next alternative OP_ALT or, if there aren't any branches, to the matching
311  the bracket itself. (They could have all been done like this, but I was making  OP_KET opcode. Each OP_ALT is followed by LINK_SIZE bytes giving the offset to
312  minimal changes.)  the next one, or to the OP_KET opcode. For capturing brackets, the bracket
313    number immediately follows the offset, always as a 2-byte item.
 A bracket opcode is followed by two bytes which give the offset to the next  
 alternative OP_ALT or, if there aren't any branches, to the matching OP_KET  
 opcode. Each OP_ALT is followed by two bytes giving the offset to the next one,  
 or to the OP_KET opcode.  
314    
315  OP_KET is used for subpatterns that do not repeat indefinitely, while  OP_KET is used for subpatterns that do not repeat indefinitely, while
316  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or  OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
317  maximally respectively. All three are followed by two bytes giving (as a  maximally respectively. All three are followed by LINK_SIZE bytes giving (as a
318  positive number) the offset back to the matching OP_BRA opcode.  positive number) the offset back to the matching bracket opcode.
319    
320  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
321  is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte  is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte
# Line 246  as appropriate. Line 330  as appropriate.
330  A subpattern with a bounded maximum repetition is replicated in a nested  A subpattern with a bounded maximum repetition is replicated in a nested
331  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
332  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
333  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?.  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?, except that each bracketed group
334    has the same number.
335    
336    When a repeated subpattern has an unbounded upper limit, it is checked to see
337    whether it could match an empty string. If this is the case, the opcode in the
338    final replication is changed to OP_SBRA or OP_SCBRA. This tells the matcher
339    that it needs to check for matching an empty string when it hits OP_KETRMIN or
340    OP_KETRMAX, and if so, to break the loop.
341    
342    
343  Assertions  Assertions
# Line 262  each alternative of a lookbehind asserti Line 353  each alternative of a lookbehind asserti
353  fixed lengths.  fixed lengths.
354    
355    
356  Once-only subpatterns  Once-only (atomic) subpatterns
357  ---------------------  ------------------------------
358    
359  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
360  OP_ONCE.  OP_ONCE. The check for matching an empty string in an unbounded repeat is
361    handled entirely at runtime, so there is just this one opcode.
362    
363    
364  Conditional subpatterns  Conditional subpatterns
365  -----------------------  -----------------------
366    
367  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
368    OP_SCOND for one that might match an empty string in an unbounded repeat. If
369  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
370  subpattern using the opcode OP_CREF followed by two bytes containing the  subpattern using the opcode OP_CREF followed by two bytes containing the
371  reference number. If the condition is "in recursion" (coded as "(?(R)"), the  reference number. If the condition is "in recursion" (coded as "(?(R)"), or "in
372  same scheme is used, with a "reference number" of 0xffff. Otherwise, a  recursion of group x" (coded as "(?(Rx)"), the group number is stored at the
373  conditional subpattern always starts with one of the assertions.  start of the subpattern using the opcode OP_RREF, and a value of zero for "the
374    whole pattern". For a DEFINE condition, just the single byte OP_DEF is used (it
375    has no associated data). Otherwise, a conditional subpattern always starts with
376    one of the assertions.
377    
378    
379  Recursion  Recursion
# Line 285  Recursion Line 381  Recursion
381    
382  Recursion either matches the current regex, or some subexpression. The opcode  Recursion either matches the current regex, or some subexpression. The opcode
383  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
384  from the start of the whole pattern.  from the start of the whole pattern. From release 6.5, OP_RECURSE is
385    automatically wrapped inside OP_ONCE brackets (because otherwise some patterns
386    broke it). OP_RECURSE is also used for "subroutine" calls, even though they
387    are not strictly a recursion.
388    
389    
390  Callout  Callout
# Line 312  at compile time, and so does not cause a Line 411  at compile time, and so does not cause a
411  data.  data.
412    
413  Philip Hazel  Philip Hazel
414  September 2004  August 2007

Legend:
Removed from v.75  
changed lines
  Added in v.212

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12