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

Contents of /code/trunk/Tech.Notes

Parent Directory Parent Directory | Revision Log Revision Log


Revision 3 - (hide annotations) (download)
Sat Feb 24 21:38:01 2007 UTC (7 years, 9 months ago) by nigel
File size: 7401 byte(s)
Load pcre-1.00 into code/trunk.

1 nigel 3 Technical Notes about PCRE
2     --------------------------
3    
4     Many years ago I implemented some regular expression functions to an algorithm
5     suggested by Martin Richards. These were not Unix-like in form, and were quite
6     restricted in what they could do by comparison with Perl. The interesting part
7     about the algorithm was that the amount of space required to hold the compiled
8     form of an expression was known in advance. The code to apply an expression did
9     not operate by backtracking, as the Henry Spencer and Perl code does, but
10     instead checked all possibilities simultaneously by keeping a list of current
11     states and checking all of them as it advanced through the subject string. (In
12     the terminology of Jeffrey Friedl's book, it was a "DFA algorithm".) When the
13     pattern was all used up, all remaining states were possible matches, and the
14     one matching the longest subset of the subject string was chosen. This did not
15     necessarily maximize the individual wild portions of the pattern, as is
16     expected in Unix and Perl-style regular expressions.
17    
18     By contrast, the code originally written by Henry Spencer and subsequently
19     heavily modified for Perl actually compiles the expression twice: once in a
20     dummy mode in order to find out how much store will be needed, and then for
21     real. The execution function operates by backtracking and maximizing (or
22     minimizing in Perl) the amount of the subject that matches individual wild
23     portions of the pattern. This is a "NFA algorithm".
24    
25     For this set of functions, I tried at first to invent an algorithm that used an
26     amount of store bounded by a multiple of the number of characters in the
27     pattern, to save on compiling time. However, because of the greater complexity
28     in Perl regular expressions, I couldn't do this. In any case, a first pass
29     through the pattern is needed, in order to find internal flag settings like
30     (?i). So it works by running a very degenerate first pass to calculate a
31     maximum store size, and then a second pass to do the real compile - which may
32     use a bit less than the predicted amount of store. The idea is that this is
33     going to turn out faster because the first pass is degenerate and the second
34     can just store stuff straight into the vector. It does make the compiling
35     functions bigger, of course, but they have got quite big anyway to handle all
36     the Perl stuff.
37    
38     The compiled form of a pattern is a vector of bytes, containing items of
39     variable length. The first byte in an item is an opcode, and the length of the
40     item is either implicit in the opcode or contained in the data bytes which
41     follow it. A list of all the opcodes follows:
42    
43     Opcodes with no following data
44     ------------------------------
45    
46     These items are all just one byte long
47    
48     OP_END end of pattern
49     OP_ANY match any character
50     OP_SOD match start of data: \A
51     OP_CIRC ^ (start of data, or after \n in multiline)
52     OP_NOT_WORD_BOUNDARY \W
53     OP_WORD_BOUNDARY \w
54     OP_NOT_DIGIT \D
55     OP_DIGIT \d
56     OP_NOT_WHITESPACE \S
57     OP_WHITESPACE \s
58     OP_NOT_WORDCHAR \W
59     OP_WORDCHAR \w
60     OP_CUT analogue of Prolog's "cut"
61     OP_EOD match end of data: \Z
62     OP_DOLL $ (end of data, or before \n in multiline)
63    
64    
65     Repeating single characters
66     ---------------------------
67    
68     The common repeats (*, +, ?) when applied to a single character appear as
69     two-byte items using the following opcodes:
70    
71     OP_STAR
72     OP_MINSTAR
73     OP_PLUS
74     OP_MINPLUS
75     OP_QUERY
76     OP_MINQUERY
77    
78     Those with "MIN" in their name are the minimizing versions. Each is followed by
79     the character that is to be repeated. Other repeats make use of
80    
81     OP_UPTO
82     OP_MINUPTO
83     OP_EXACT
84    
85     which are followed by a two-byte count (most significant first) and the
86     repeated character. OP_UPTO matches from 0 to the given number. A repeat with a
87     non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an
88     OP_UPTO (or OP_MINUPTO).
89    
90    
91     Repeating character types
92     -------------------------
93    
94     Repeats of things like \d are done exactly as for single characters, except
95     that instead of a character, the opcode for the type is stored in the data
96     byte. The opcodes are:
97    
98     OP_TYPESTAR
99     OP_TYPEMINSTAR
100     OP_TYPEPLUS
101     OP_TYPEMINPLUS
102     OP_TYPEQUERY
103     OP_TYPEMINQUERY
104     OP_TYPEUPTO
105     OP_TYPEMINUPTO
106     OP_TYPEEXACT
107    
108    
109     Matching a character string
110     ---------------------------
111    
112     The OP_CHARS opcode is followed by a one-byte count and then that number of
113     characters. If there are more than 255 characters in sequence, successive
114     instances of OP_CHARS are used.
115    
116    
117     Character classes
118     -----------------
119    
120     OP_CLASS is used for a character class. It is followed by a 32-byte bit map
121     containing a 1 bit for every character that is acceptable. The bits are counted
122     from the least significant end of each byte.
123    
124    
125     Back references
126     ---------------
127    
128     OP_REF is followed by a single byte containing the reference number.
129    
130    
131     Repeating character classes and back references
132     -----------------------------------------------
133    
134     In both cases, the repeat information follows the base item. The matching code
135     looks at the following opcode to see if it is one of
136    
137     OP_CRSTAR
138     OP_CRMINSTAR
139     OP_CRPLUS
140     OP_CRMINPLUS
141     OP_CRQUERY
142     OP_CRMINQUERY
143     OP_CRRANGE
144     OP_CRMINRANGE
145    
146     All but the last two are just single-byte items. The others are followed by
147     four bytes of data, comprising the minimum and maximum repeat counts.
148    
149    
150     Brackets and alternation
151     ------------------------
152    
153     A pair of non-identifying (round) brackets is wrapped round each expression at
154     compile time, so alternation always happens in the context of brackets.
155     Non-identifying brackets use the opcode OP_BRA, while identifying brackets use
156     OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English
157     speakers, including myself, can be round, square, or curly. Hence this usage.]
158    
159     A bracket opcode is followed by two bytes which give the offset to the next
160     alternative OP_ALT or, if there aren't any branches, to the matching KET
161     opcode. Each OP_ALT is followed by two bytes giving the offset to the next one,
162     or to the KET opcode.
163    
164     OP_KET is used for subpatterns that do not repeat indefinitely, while
165     OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
166     maximally respectively. All three are followed by two bytes giving (as a
167     positive number) the offset back to the matching BRA opcode.
168    
169     If a subpattern is quantified such that it is permitted to match zero times, it
170     is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte
171     opcodes which tell the matcher that skipping this subpattern entirely is a
172     valid branch.
173    
174     A subpattern with an indefinite maximum repetition is replicated in the
175     compiled data its minimum number of times (or once with a BRAZERO if the
176     minimum is zero), with the final copy terminating with a KETRMIN or KETRMAX as
177     appropriate.
178    
179     A subpattern with a bounded maximum repetition is replicated up to the maximum
180     number of times, with BRAZERO or BRAMINZERO before each replication after the
181     minimum. In effect, (abc){2,5} becomes (abc)(abc)(abc)?(abc)?(abc)?.
182    
183    
184     Assertions
185     ----------
186    
187     Assertions are just like other subpatterns, but starting with one of the
188     opcodes OP_ASSERT or OP_ASSERT_NOT.
189    
190    
191     Once-only subpatterns
192     ---------------------
193    
194     These are also just like other subpatterns, but they start with the opcode
195     OP_ONCE.
196    
197    
198     Philip Hazel
199     October 1997

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12