/[pcre]/code/trunk/maintain/Tech.Notes
ViewVC logotype

Diff of /code/trunk/maintain/Tech.Notes

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

revision 41 by nigel, Sat Feb 24 21:39:17 2007 UTC revision 53 by nigel, Sat Feb 24 21:39:42 2007 UTC
# Line 23  optionally, minimizing in Perl) the amou Line 23  optionally, minimizing in Perl) the amou
23  individual wild portions of the pattern. This is an "NFA algorithm" in Friedl's  individual wild portions of the pattern. This is an "NFA algorithm" in Friedl's
24  terminology.  terminology.
25    
26  For this set of functions that forms PCRE, I tried at first to invent an  For the set of functions that forms PCRE (which are unrelated to those
27  algorithm that used an amount of store bounded by a multiple of the number of  mentioned above), I tried at first to invent an algorithm that used an amount
28  characters in the pattern, to save on compiling time. However, because of the  of store bounded by a multiple of the number of characters in the pattern, to
29  greater complexity in Perl regular expressions, I couldn't do this. In any  save on compiling time. However, because of the greater complexity in Perl
30  case, a first pass through the pattern is needed, in order to find internal  regular expressions, I couldn't do this. In any case, a first pass through the
31  flag settings like (?i) at top level. So it works by running a very degenerate  pattern is needed, in order to find internal flag settings like (?i) at top
32  first pass to calculate a maximum store size, and then a second pass to do the  level. So PCRE works by running a very degenerate first pass to calculate a
33  real compile - which may use a bit less than the predicted amount of store. The  maximum store size, and then a second pass to do the real compile - which may
34  idea is that this is going to turn out faster because the first pass is  use a bit less than the predicted amount of store. The idea is that this is
35  degenerate and the second can just store stuff straight into the vector. It  going to turn out faster because the first pass is degenerate and the second
36  does make the compiling functions bigger, of course, but they have got quite  pass can just store stuff straight into the vector. It does make the compiling
37  big anyway to handle all the Perl stuff.  functions bigger, of course, but they have got quite big anyway to handle all
38    the Perl stuff.
39    
40  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
41  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
# Line 61  These items are all just one byte long Line 62  These items are all just one byte long
62    OP_EODN                match end of data or \n at end: \Z    OP_EODN                match end of data or \n at end: \Z
63    OP_EOD                 match end of data: \z    OP_EOD                 match end of data: \z
64    OP_DOLL                $ (end of data, or before \n in multiline)    OP_DOLL                $ (end of data, or before \n in multiline)
65      OP_RECURSE             match the pattern recursively
66    
67    
68  Repeating single characters  Repeating single characters
# Line 125  positive class, and OP_NOT for a negativ Line 127  positive class, and OP_NOT for a negativ
127  repeated, negated, single-character class. The normal ones (OP_STAR etc.) are  repeated, negated, single-character class. The normal ones (OP_STAR etc.) are
128  used for a repeated positive single-character class.  used for a repeated positive single-character class.
129    
130  OP_CLASS is followed by a 32-byte bit map containing a 1  OP_CLASS is followed by a 32-byte bit map containing a 1 bit for every
131  bit for every character that is acceptable. The bits are counted from the least  character that is acceptable. The bits are counted from the least significant
132  significant end of each byte.  end of each byte.
133    
134    
135  Back references  Back references
136  ---------------  ---------------
137    
138  OP_REF is followed by a single byte containing the reference number.  OP_REF is followed by two bytes containing the reference number.
139    
140    
141  Repeating character classes and back references  Repeating character classes and back references
# Line 159  four bytes of data, comprising the minim Line 161  four bytes of data, comprising the minim
161  Brackets and alternation  Brackets and alternation
162  ------------------------  ------------------------
163    
164  A pair of non-identifying (round) brackets is wrapped round each expression at  A pair of non-capturing (round) brackets is wrapped round each expression at
165  compile time, so alternation always happens in the context of brackets.  compile time, so alternation always happens in the context of brackets.
166  Non-identifying brackets use the opcode OP_BRA, while identifying brackets use  
167    Non-capturing brackets use the opcode OP_BRA, while capturing brackets use
168  OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English  OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English
169  speakers, including myself, can be round, square, or curly. Hence this usage.]  speakers, including myself, can be round, square, curly, or pointy. Hence this
170    usage.]
171    
172    Originally PCRE was limited to 99 capturing brackets (so as not to use up all
173    the opcodes). From release 3.5, there is no limit. What happens is that the
174    first ones, up to EXTRACT_BASIC_MAX are handled with separate opcodes, as
175    above. If there are more, the opcode is set to EXTRACT_BASIC_MAX+1, and the
176    first operation in the bracket is OP_BRANUMBER, followed by a 2-byte bracket
177    number. This opcode is ignored while matching, but is fished out when handling
178    the bracket itself. (They could have all been done like this, but I was making
179    minimal changes.)
180    
181  A bracket opcode is followed by two bytes which give the offset to the next  A bracket opcode is followed by two bytes which give the offset to the next
182  alternative OP_ALT or, if there aren't any branches, to the matching KET  alternative OP_ALT or, if there aren't any branches, to the matching KET
# Line 188  appropriate. Line 201  appropriate.
201  A subpattern with a bounded maximum repetition is replicated in a nested  A subpattern with a bounded maximum repetition is replicated in a nested
202  fashion up to the maximum number of times, with BRAZERO or BRAMINZERO before  fashion up to the maximum number of times, with BRAZERO or BRAMINZERO before
203  each replication after the minimum, so that, for example, (abc){2,5} is  each replication after the minimum, so that, for example, (abc){2,5} is
204  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?. The 200-bracket limit does not  compiled as (abc)(abc)((abc)((abc)(abc)?)?)?. The 99 and 200 bracket limits do
205  apply to these internally generated brackets.  not apply to these internally generated brackets.
206    
207    
208  Assertions  Assertions
# Line 199  Forward assertions are just like other s Line 212  Forward assertions are just like other s
212  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
213  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
214  is OP_REVERSE, followed by a two byte count of the number of characters to move  is OP_REVERSE, followed by a two byte count of the number of characters to move
215  back the pointer in the subject string. A separate count is present in each  back the pointer in the subject string. When operating in UTF-8 mode, the count
216  alternative of a lookbehind assertion, allowing them to have different fixed  is a character count rather than a byte count. A separate count is present in
217  lengths.  each alternative of a lookbehind assertion, allowing them to have different
218    fixed lengths.
219    
220    
221  Once-only subpatterns  Once-only subpatterns
# Line 216  Conditional subpatterns Line 230  Conditional subpatterns
230    
231  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. If
232  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
233  subpattern using the opcode OP_CREF followed by one byte containing the  subpattern using the opcode OP_CREF followed by two bytes containing the
234  reference number. Otherwise, a conditional subpattern will always start with  reference number. Otherwise, a conditional subpattern will always start with
235  one of the assertions.  one of the assertions.
236    
# Line 236  the compiled data. Line 250  the compiled data.
250    
251    
252  Philip Hazel  Philip Hazel
253  January 1999  August 2001

Legend:
Removed from v.41  
changed lines
  Added in v.53

  ViewVC Help
Powered by ViewVC 1.1.5