/[pcre]/code/trunk/doc/pcretest.txt
ViewVC logotype

Diff of /code/trunk/doc/pcretest.txt

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

revision 47 by nigel, Sat Feb 24 21:39:29 2007 UTC revision 1194 by ph10, Wed Oct 31 17:42:29 2012 UTC
# Line 1  Line 1 
1  The pcretest program  PCRETEST(1)                                                        PCRETEST(1)
 --------------------  
2    
 This program is intended for testing PCRE, but it can also be used for  
 experimenting with regular expressions.  
3    
4  If it is given two filename arguments, it reads from the first and writes to  NAME
5  the second. If it is given only one filename argument, it reads from that file         pcretest - a program for testing Perl-compatible regular expressions.
 and writes to stdout. Otherwise, it reads from stdin and writes to stdout, and  
 prompts for each line of input, using "re>" to prompt for regular expressions,  
 and "data>" to prompt for data lines.  
   
 The program handles any number of sets of input on a single input file. Each  
 set starts with a regular expression, and continues with any number of data  
 lines to be matched against the pattern. An empty line signals the end of the  
 data lines, at which point a new regular expression is read. The regular  
 expressions are given enclosed in any non-alphameric delimiters other than  
 backslash, for example  
   
   /(a|bc)x+yz/  
   
 White space before the initial delimiter is ignored. A regular expression may  
 be continued over several input lines, in which case the newline characters are  
 included within it. See the test input files in the testdata directory for many  
 examples. It is possible to include the delimiter within the pattern by  
 escaping it, for example  
   
   /abc\/def/  
   
 If you do so, the escape and the delimiter form part of the pattern, but since  
 delimiters are always non-alphameric, this does not affect its interpretation.  
 If the terminating delimiter is immediately followed by a backslash, for  
 example,  
   
   /abc/\  
   
 then a backslash is added to the end of the pattern. This is done to provide a  
 way of testing the error condition that arises if a pattern finishes with a  
 backslash, because  
   
   /abc\/  
   
 is interpreted as the first line of a pattern that starts with "abc/", causing  
 pcretest to read the next line as a continuation of the regular expression.  
   
 The pattern may be followed by i, m, s, or x to set the PCRE_CASELESS,  
 PCRE_MULTILINE, PCRE_DOTALL, or PCRE_EXTENDED options, respectively. For  
 example:  
   
   /caseless/i  
   
 These modifier letters have the same effect as they do in Perl. There are  
 others which set PCRE options that do not correspond to anything in Perl: /A,  
 /E, and /X set PCRE_ANCHORED, PCRE_DOLLAR_ENDONLY, and PCRE_EXTRA respectively.  
   
 Searching for all possible matches within each subject string can be requested  
 by the /g or /G modifier. After finding a match, PCRE is called again to search  
 the remainder of the subject string. The difference between /g and /G is that  
 the former uses the startoffset argument to pcre_exec() to start searching at  
 a new point within the entire string (which is in effect what Perl does),  
 whereas the latter passes over a shortened substring. This makes a difference  
 to the matching process if the pattern begins with a lookbehind assertion  
 (including \b or \B).  
   
 If any call to pcre_exec() in a /g or /G sequence matches an empty string, the  
 next call is done with the PCRE_NOTEMPTY and PCRE_ANCHORED flags set in order  
 to search for another, non-empty, match at the same point. If this second match  
 fails, the start offset is advanced by one, and the normal match is retried.  
 This imitates the way Perl handles such cases when using the /g modifier or the  
 split() function.  
   
 There are a number of other modifiers for controlling the way pcretest  
 operates.  
   
 The /+ modifier requests that as well as outputting the substring that matched  
 the entire pattern, pcretest should in addition output the remainder of the  
 subject string. This is useful for tests where the subject contains multiple  
 copies of the same substring.  
   
 The /L modifier must be followed directly by the name of a locale, for example,  
   
   /pattern/Lfr  
   
 For this reason, it must be the last modifier letter. The given locale is set,  
 pcre_maketables() is called to build a set of character tables for the locale,  
 and this is then passed to pcre_compile() when compiling the regular  
 expression. Without an /L modifier, NULL is passed as the tables pointer; that  
 is, /L applies only to the expression on which it appears.  
   
 The /I modifier requests that pcretest output information about the compiled  
 expression (whether it is anchored, has a fixed first character, and so on). It  
 does this by calling pcre_fullinfo() after compiling an expression, and  
 outputting the information it gets back. If the pattern is studied, the results  
 of that are also output.  
   
 The /D modifier is a PCRE debugging feature, which also assumes /I. It causes  
 the internal form of compiled regular expressions to be output after  
 compilation.  
   
 The /S modifier causes pcre_study() to be called after the expression has been  
 compiled, and the results used when the expression is matched.  
   
 The /M modifier causes the size of memory block used to hold the compiled  
 pattern to be output.  
   
 Finally, the /P modifier causes pcretest to call PCRE via the POSIX wrapper API  
 rather than its native API. When this is done, all other modifiers except /i,  
 /m, and /+ are ignored. REG_ICASE is set if /i is present, and REG_NEWLINE is  
 set if /m is present. The wrapper functions force PCRE_DOLLAR_ENDONLY always,  
 and PCRE_DOTALL unless REG_NEWLINE is set.  
   
 Before each data line is passed to pcre_exec(), leading and trailing whitespace  
 is removed, and it is then scanned for \ escapes. The following are recognized:  
   
   \a     alarm (= BEL)  
   \b     backspace  
   \e     escape  
   \f     formfeed  
   \n     newline  
   \r     carriage return  
   \t     tab  
   \v     vertical tab  
   \nnn   octal character (up to 3 octal digits)  
   \xhh   hexadecimal character (up to 2 hex digits)  
   
   \A     pass the PCRE_ANCHORED option to pcre_exec()  
   \B     pass the PCRE_NOTBOL option to pcre_exec()  
   \Cdd   call pcre_copy_substring() for substring dd after a successful match  
            (any decimal number less than 32)  
   \Gdd   call pcre_get_substring() for substring dd after a successful match  
            (any decimal number less than 32)  
   \L     call pcre_get_substringlist() after a successful match  
   \N     pass the PCRE_NOTEMPTY option to pcre_exec()  
   \Odd   set the size of the output vector passed to pcre_exec() to dd  
            (any number of decimal digits)  
   \Z     pass the PCRE_NOTEOL option to pcre_exec()  
   
 A backslash followed by anything else just escapes the anything else. If the  
 very last character is a backslash, it is ignored. This gives a way of passing  
 an empty line as data, since a real empty line terminates the data input.  
   
 If /P was present on the regex, causing the POSIX wrapper API to be used, only  
 \B, and \Z have any effect, causing REG_NOTBOL and REG_NOTEOL to be passed to  
 regexec() respectively.  
   
 When a match succeeds, pcretest outputs the list of captured substrings that  
 pcre_exec() returns, starting with number 0 for the string that matched the  
 whole pattern. Here is an example of an interactive pcretest run.  
   
   $ pcretest  
   PCRE version 2.06 08-Jun-1999  
   
     re> /^abc(\d+)/  
   data> abc123  
    0: abc123  
    1: 123  
   data> xyz  
   No match  
   
 If the strings contain any non-printing characters, they are output as \0x  
 escapes. If the pattern has the /+ modifier, then the output for substring 0 is  
 followed by the the rest of the subject string, identified by "0+" like this:  
   
     re> /cat/+  
   data> cataract  
    0: cat  
    0+ aract  
   
 If the pattern has the /g or /G modifier, the results of successive matching  
 attempts are output in sequence, like this:  
   
     re> /\Bi(\w\w)/g  
   data> Mississippi  
    0: iss  
    1: ss  
    0: iss  
    1: ss  
    0: ipp  
    1: pp  
   
 "No match" is output only if the first match attempt fails.  
   
 If any of \C, \G, or \L are present in a data line that is successfully  
 matched, the substrings extracted by the convenience functions are output with  
 C, G, or L after the string number instead of a colon. This is in addition to  
 the normal full list. The string length (that is, the return from the  
 extraction function) is given in parentheses after each string for \C and \G.  
   
 Note that while patterns can be continued over several lines (a plain ">"  
 prompt is used for continuations), data lines may not. However newlines can be  
 included in data by means of the \n escape.  
   
 If the -p option is given to pcretest, it is equivalent to adding /P to each  
 regular expression: the POSIX wrapper API is used to call PCRE. None of the  
 following flags has any effect in this case.  
   
 If the option -d is given to pcretest, it is equivalent to adding /D to each  
 regular expression: the internal form is output after compilation.  
   
 If the option -i is given to pcretest, it is equivalent to adding /I to each  
 regular expression: information about the compiled pattern is given after  
 compilation.  
   
 If the option -m is given to pcretest, it outputs the size of each compiled  
 pattern after it has been compiled. It is equivalent to adding /M to each  
 regular expression. For compatibility with earlier versions of pcretest, -s is  
 a synonym for -m.  
   
 If the -t option is given, each compile, study, and match is run 20000 times  
 while being timed, and the resulting time per compile or match is output in  
 milliseconds. Do not set -t with -s, because you will then get the size output  
 20000 times and the timing will be distorted. If you want to change the number  
 of repetitions used for timing, edit the definition of LOOPREPEAT at the top of  
 pcretest.c  
6    
7  Philip Hazel <ph10@cam.ac.uk>  
8  January 2000  SYNOPSIS
9    
10           pcretest [options] [input file [output file]]
11    
12           pcretest  was written as a test program for the PCRE regular expression
13           library itself, but it can also be used for experimenting with  regular
14           expressions.  This document describes the features of the test program;
15           for details of the regular expressions themselves, see the  pcrepattern
16           documentation. For details of the PCRE library function calls and their
17           options, see the pcreapi , pcre16 and pcre32 documentation.
18    
19           The input for pcretest is a sequence of regular expression patterns and
20           strings  to be matched, as described below. The output shows the result
21           of each match. Options on the command line  and  the  patterns  control
22           PCRE options and exactly what is output.
23    
24           As  PCRE has evolved, it has acquired many different features, and as a
25           result, pcretest now has rather a lot of obscure  options  for  testing
26           every possible feature. Some of these options are specifically designed
27           for use in conjunction with the test script and  data  files  that  are
28           distributed  as  part of PCRE, and are unlikely to be of use otherwise.
29           They are all documented here, but without much justification.
30    
31    
32    PCRE's 8-BIT, 16-BIT AND 32-BIT LIBRARIES
33    
34           From release 8.30, two separate PCRE libraries can be built. The origi-
35           nal  one  supports  8-bit  character  strings, whereas the newer 16-bit
36           library supports  character  strings  encoded  in  16-bit  units.  From
37           release  8.32,  a  third  library  can  be  built, supporting character
38           strings encoded in 32-bit units.  The pcretest program can be  used  to
39           test all three libraries. However, it is itself still an 8-bit program,
40           reading 8-bit input and writing 8-bit output. When testing  the  16-bit
41           or  32-bit  library, the patterns and data strings are converted to 16-
42           or 32-bit format before being passed to  the  PCRE  library  functions.
43           Results are converted to 8-bit for output.
44    
45           References to functions and structures of the form pcre[16|32]_xx below
46           mean "pcre_xx when using the 8-bit library or pcre16_xx when using  the
47           16-bit library".
48    
49    
50    COMMAND LINE OPTIONS
51    
52           -8        If  both the 8-bit library has been built, this option causes
53                     the 8-bit library to be used (which is the default);  if  the
54                     8-bit  library  has  not  been  built,  this option causes an
55                     error.
56    
57           -16       If both the 8-bit or the 32-bit,  and  the  16-bit  libraries
58                     have  been built, this option causes the 16-bit library to be
59                     used. If only the 16-bit library has been built, this is  the
60                     default  (so  has no effect). If only the 8-bit or the 32-bit
61                     library has been built, this option causes an error.
62    
63           -32       If both the 8-bit or the 16-bit,  and  the  32-bit  libraries
64                     have  been built, this option causes the 32-bit library to be
65                     used. If only the 32-bit library has been built, this is  the
66                     default  (so  has no effect). If only the 8-bit or the 16-bit
67                     library has been built, this option causes an error.
68    
69           -b        Behave as if each pattern has the /B (show byte  code)  modi-
70                     fier; the internal form is output after compilation.
71    
72           -C        Output the version number of the PCRE library, and all avail-
73                     able  information  about  the  optional  features  that   are
74                     included, and then exit. All other options are ignored.
75    
76           -C option Output  information  about a specific build-time option, then
77                     exit. This functionality is intended for use in scripts  such
78                     as RunTest. The following options output the value indicated:
79    
80                       ebcdic-nl  the code for LF (= NL) in an EBCDIC environment:
81                                    0x15 or 0x25
82                                    0 if used in an ASCII environment
83                       linksize   the internal link size (2, 3, or 4)
84                       newline    the default newline setting:
85                                    CR, LF, CRLF, ANYCRLF, or ANY
86    
87                     The following options output 1 for true or zero for false:
88    
89                       ebcdic     compiled for an EBCDIC environment
90                       jit        just-in-time support is available
91                       pcre16     the 16-bit library was built
92                       pcre32     the 32-bit library was built
93                       pcre8      the 8-bit library was built
94                       ucp        Unicode property support is available
95                       utf         UTF-8  and/or  UTF-16  and/or UTF-32 support is
96                     available
97    
98           -d        Behave as if each pattern has the /D  (debug)  modifier;  the
99                     internal  form  and information about the compiled pattern is
100                     output after compilation; -d is equivalent to -b -i.
101    
102           -dfa      Behave as if each data line contains the \D escape  sequence;
103                     this    causes    the    alternative    matching    function,
104                     pcre[16|32]_dfa_exec(), to be used instead  of  the  standard
105                     pcre[16|32]_exec() function (more detail is given below).
106    
107           -help     Output a brief summary these options and then exit.
108    
109           -i        Behave  as  if  each pattern has the /I modifier; information
110                     about the compiled pattern is given after compilation.
111    
112           -M        Behave as if each data line contains the \M escape  sequence;
113                     this  causes  PCRE  to  discover  the minimum MATCH_LIMIT and
114                     MATCH_LIMIT_RECURSION settings by calling  pcre[16|32]_exec()
115                     repeatedly with different limits.
116    
117           -m        Output  the  size  of each compiled pattern after it has been
118                     compiled. This is equivalent to adding  /M  to  each  regular
119                     expression. The size is given in bytes for both libraries.
120    
121           -o osize  Set  the number of elements in the output vector that is used
122                     when calling pcre[16|32]_exec() or pcre[16|32]_dfa_exec()  to
123                     be  osize.  The  default  value is 45, which is enough for 14
124                     capturing subexpressions for pcre[16|32]_exec() or 22 differ-
125                     ent  matches for pcre[16|32]_dfa_exec().  The vector size can
126                     be changed for individual matching calls by including  \O  in
127                     the data line (see below).
128    
129           -p        Behave  as  if  each  pattern  has the /P modifier; the POSIX
130                     wrapper API is used to call PCRE. None of the  other  options
131                     has  any  effect when -p is set. This option can be used only
132                     with the 8-bit library.
133    
134           -q        Do not output the version number of pcretest at the start  of
135                     execution.
136    
137           -S size   On  Unix-like  systems, set the size of the run-time stack to
138                     size megabytes.
139    
140           -s or -s+ Behave as if each pattern  has  the  /S  modifier;  in  other
141                     words,  force each pattern to be studied. If -s+ is used, all
142                     the JIT compile options are  passed  to  pcre[16|32]_study(),
143                     causing  just-in-time  optimization  to  be  set  up if it is
144                     available, for both full and partial matching.  Specific  JIT
145                     compile options can be selected by following -s+ with a digit
146                     in the range 1 to 7, which selects the JIT compile  modes  as
147                     follows:
148    
149                       1  normal match only
150                       2  soft partial match only
151                       3  normal match and soft partial match
152                       4  hard partial match only
153                       6  soft and hard partial match
154                       7  all three modes (default)
155    
156                     If  -s++  is used instead of -s+ (with or without a following
157                     digit), the text "(JIT)" is added to the  first  output  line
158                     after a match or no match when JIT-compiled code was actually
159                     used.
160    
161                     Note that there are pattern options  that  can  override  -s,
162                     either specifying no studying at all, or suppressing JIT com-
163                     pilation.
164    
165                     If the /I or /D option is present on  a  pattern  (requesting
166                     output  about  the  compiled  pattern), information about the
167                     result of studying is not included when  studying  is  caused
168                     only  by  -s  and neither -i nor -d is present on the command
169                     line. This behaviour means that the output  from  tests  that
170                     are  run with and without -s should be identical, except when
171                     options that output information about the actual running of a
172                     match are set.
173    
174                     The  -M,  -t,  and  -tm options, which give information about
175                     resources used, are likely to produce different  output  with
176                     and  without  -s.  Output may also differ if the /C option is
177                     present on an individual pattern. This uses callouts to trace
178                     the  the  matching process, and this may be different between
179                     studied and non-studied patterns.  If  the  pattern  contains
180                     (*MARK)  items  there  may  also be differences, for the same
181                     reason. The -s command line option can be overridden for spe-
182                     cific  patterns that should never be studied (see the /S pat-
183                     tern modifier below).
184    
185           -t        Run each compile, study, and match many times with  a  timer,
186                     and  output resulting time per compile or match (in millisec-
187                     onds). Do not set -m with -t, because you will then  get  the
188                     size  output  a  zillion  times,  and the timing will be dis-
189                     torted. You can control the number  of  iterations  that  are
190                     used  for timing by following -t with a number (as a separate
191                     item on the command line). For example, "-t 1000" would iter-
192                     ate 1000 times. The default is to iterate 500000 times.
193    
194           -tm       This is like -t except that it times only the matching phase,
195                     not the compile or study phases.
196    
197    
198    DESCRIPTION
199    
200           If pcretest is given two filename arguments, it reads  from  the  first
201           and writes to the second. If it is given only one filename argument, it
202           reads from that file and writes to stdout.  Otherwise,  it  reads  from
203           stdin  and  writes to stdout, and prompts for each line of input, using
204           "re>" to prompt for regular expressions, and "data>" to prompt for data
205           lines.
206    
207           When  pcretest  is  built,  a  configuration option can specify that it
208           should be linked with the libreadline library. When this  is  done,  if
209           the input is from a terminal, it is read using the readline() function.
210           This provides line-editing and history facilities. The output from  the
211           -help option states whether or not readline() will be used.
212    
213           The program handles any number of sets of input on a single input file.
214           Each set starts with a regular expression, and continues with any  num-
215           ber of data lines to be matched against the pattern.
216    
217           Each  data line is matched separately and independently. If you want to
218           do multi-line matches, you have to use the \n escape sequence (or \r or
219           \r\n, etc., depending on the newline setting) in a single line of input
220           to encode the newline sequences. There is no limit  on  the  length  of
221           data  lines;  the  input  buffer is automatically extended if it is too
222           small.
223    
224           An empty line signals the end of the data lines, at which point  a  new
225           regular  expression is read. The regular expressions are given enclosed
226           in any non-alphanumeric delimiters other than backslash, for example:
227    
228             /(a|bc)x+yz/
229    
230           White space before the initial delimiter is ignored. A regular  expres-
231           sion  may be continued over several input lines, in which case the new-
232           line characters are included within it. It is possible to  include  the
233           delimiter within the pattern by escaping it, for example
234    
235             /abc\/def/
236    
237           If  you  do  so, the escape and the delimiter form part of the pattern,
238           but since delimiters are always non-alphanumeric, this does not  affect
239           its  interpretation.   If the terminating delimiter is immediately fol-
240           lowed by a backslash, for example,
241    
242             /abc/\
243    
244           then a backslash is added to the end of the pattern. This  is  done  to
245           provide  a  way of testing the error condition that arises if a pattern
246           finishes with a backslash, because
247    
248             /abc\/
249    
250           is interpreted as the first line of a pattern that starts with  "abc/",
251           causing pcretest to read the next line as a continuation of the regular
252           expression.
253    
254    
255    PATTERN MODIFIERS
256    
257           A pattern may be followed by any number of modifiers, which are  mostly
258           single  characters,  though  some  of these can be qualified by further
259           characters.  Following Perl usage, these are referred to below as,  for
260           example,  "the  /i  modifier", even though the delimiter of the pattern
261           need not always be a slash, and no slash is  used  when  writing  modi-
262           fiers.  White  space may appear between the final pattern delimiter and
263           the first modifier, and between the modifiers  themselves.  For  refer-
264           ence,  here  is  a  complete  list of modifiers. They fall into several
265           groups that are described in detail in the following sections.
266    
267             /8              set UTF mode
268             /?              disable UTF validity check
269             /+              show remainder of subject after match
270             /=              show all captures (not just those that are set)
271    
272             /A              set PCRE_ANCHORED
273             /B              show compiled code
274             /C              set PCRE_AUTO_CALLOUT
275             /D              same as /B plus /I
276             /E              set PCRE_DOLLAR_ENDONLY
277             /F              flip byte order in compiled pattern
278             /f              set PCRE_FIRSTLINE
279             /G              find all matches (shorten string)
280             /g              find all matches (use startoffset)
281             /I              show information about pattern
282             /i              set PCRE_CASELESS
283             /J              set PCRE_DUPNAMES
284             /K              show backtracking control names
285             /L              set locale
286             /M              show compiled memory size
287             /m              set PCRE_MULTILINE
288             /N              set PCRE_NO_AUTO_CAPTURE
289             /P              use the POSIX wrapper
290             /S              study the pattern after compilation
291             /s              set PCRE_DOTALL
292             /T              select character tables
293             /U              set PCRE_UNGREEDY
294             /W              set PCRE_UCP
295             /X              set PCRE_EXTRA
296             /x              set PCRE_EXTENDED
297             /Y              set PCRE_NO_START_OPTIMIZE
298             /Z              don't show lengths in /B output
299    
300             /<any>          set PCRE_NEWLINE_ANY
301             /<anycrlf>      set PCRE_NEWLINE_ANYCRLF
302             /<cr>           set PCRE_NEWLINE_CR
303             /<crlf>         set PCRE_NEWLINE_CRLF
304             /<lf>           set PCRE_NEWLINE_LF
305             /<bsr_anycrlf>  set PCRE_BSR_ANYCRLF
306             /<bsr_unicode>  set PCRE_BSR_UNICODE
307             /<JS>           set PCRE_JAVASCRIPT_COMPAT
308    
309    
310       Perl-compatible modifiers
311    
312           The /i, /m, /s, and /x modifiers set the PCRE_CASELESS, PCRE_MULTILINE,
313           PCRE_DOTALL,    or    PCRE_EXTENDED    options,    respectively,   when
314           pcre[16|32]_compile() is called. These four modifier letters  have  the
315           same effect as they do in Perl. For example:
316    
317             /caseless/i
318    
319    
320       Modifiers for other PCRE options
321    
322           The  following  table  shows additional modifiers for setting PCRE com-
323           pile-time options that do not correspond to anything in Perl:
324    
325             /8              PCRE_UTF8           ) when using the 8-bit
326             /?              PCRE_NO_UTF8_CHECK  )   library
327    
328             /8              PCRE_UTF16          ) when using the 16-bit
329             /?              PCRE_NO_UTF16_CHECK )   library
330    
331             /8              PCRE_UTF32          ) when using the 32-bit
332             /?              PCRE_NO_UTF32_CHECK )   library
333    
334             /A              PCRE_ANCHORED
335             /C              PCRE_AUTO_CALLOUT
336             /E              PCRE_DOLLAR_ENDONLY
337             /f              PCRE_FIRSTLINE
338             /J              PCRE_DUPNAMES
339             /N              PCRE_NO_AUTO_CAPTURE
340             /U              PCRE_UNGREEDY
341             /W              PCRE_UCP
342             /X              PCRE_EXTRA
343             /Y              PCRE_NO_START_OPTIMIZE
344             /<any>          PCRE_NEWLINE_ANY
345             /<anycrlf>      PCRE_NEWLINE_ANYCRLF
346             /<cr>           PCRE_NEWLINE_CR
347             /<crlf>         PCRE_NEWLINE_CRLF
348             /<lf>           PCRE_NEWLINE_LF
349             /<bsr_anycrlf>  PCRE_BSR_ANYCRLF
350             /<bsr_unicode>  PCRE_BSR_UNICODE
351             /<JS>           PCRE_JAVASCRIPT_COMPAT
352    
353           The modifiers that are enclosed in angle brackets are  literal  strings
354           as  shown,  including the angle brackets, but the letters within can be
355           in either case.  This example sets multiline matching with CRLF as  the
356           line ending sequence:
357    
358             /^abc/m<CRLF>
359    
360           As  well  as  turning  on  the  PCRE_UTF8/16/32 option, the /8 modifier
361           causes all non-printing characters in  output  strings  to  be  printed
362           using the \x{hh...} notation. Otherwise, those less than 0x100 are out-
363           put in hex without the curly brackets.
364    
365           Full details of the PCRE options are given in  the  pcreapi  documenta-
366           tion.
367    
368       Finding all matches in a string
369    
370           Searching  for  all  possible matches within each subject string can be
371           requested by the /g or /G modifier. After  finding  a  match,  PCRE  is
372           called again to search the remainder of the subject string. The differ-
373           ence between /g and /G is that the former uses the startoffset argument
374           to  pcre[16|32]_exec()  to  start  searching  at a new point within the
375           entire string (which is in effect what Perl does), whereas  the  latter
376           passes  over  a  shortened  substring.  This  makes a difference to the
377           matching process if the pattern  begins  with  a  lookbehind  assertion
378           (including \b or \B).
379    
380           If  any  call  to  pcre[16|32]_exec() in a /g or /G sequence matches an
381           empty string, the next call is done with the PCRE_NOTEMPTY_ATSTART  and
382           PCRE_ANCHORED  flags  set  in  order  to search for another, non-empty,
383           match at the same point. If this second match fails, the  start  offset
384           is  advanced,  and  the  normal match is retried. This imitates the way
385           Perl handles such cases when using the /g modifier or the split() func-
386           tion.  Normally,  the start offset is advanced by one character, but if
387           the newline convention recognizes CRLF as a newline,  and  the  current
388           character is CR followed by LF, an advance of two is used.
389    
390       Other modifiers
391    
392           There are yet more modifiers for controlling the way pcretest operates.
393    
394           The  /+ modifier requests that as well as outputting the substring that
395           matched the entire pattern, pcretest  should  in  addition  output  the
396           remainder  of  the  subject  string. This is useful for tests where the
397           subject contains multiple copies of the same substring. If the +  modi-
398           fier  appears  twice, the same action is taken for captured substrings.
399           In each case the remainder is output on the following line with a  plus
400           character  following  the  capture number. Note that this modifier must
401           not immediately follow the /S modifier because /S+ and /S++ have  other
402           meanings.
403    
404           The  /=  modifier  requests  that  the values of all potential captured
405           parentheses be output after a match. By default, only those up  to  the
406           highest one actually used in the match are output (corresponding to the
407           return code from pcre[16|32]_exec()). Values in the offsets vector cor-
408           responding  to higher numbers should be set to -1, and these are output
409           as "<unset>". This modifier gives a way of checking that this  is  hap-
410           pening.
411    
412           The  /B modifier is a debugging feature. It requests that pcretest out-
413           put a representation of the compiled code after  compilation.  Normally
414           this  information  contains length and offset values; however, if /Z is
415           also present, this data is replaced by spaces. This is a  special  fea-
416           ture  for  use  in the automatic test scripts; it ensures that the same
417           output is generated for different internal link sizes.
418    
419           The /D modifier is a PCRE debugging feature, and is equivalent to  /BI,
420           that is, both the /B and the /I modifiers.
421    
422           The  /F  modifier  causes pcretest to flip the byte order of the 2-byte
423           and 4-byte fields in the compiled pattern. This facility is for testing
424           the  feature  in PCRE that allows it to execute patterns that were com-
425           piled on a host with a different endianness. This feature is not avail-
426           able  when the POSIX interface to PCRE is being used, that is, when the
427           /P pattern modifier is specified. See also the section about saving and
428           reloading compiled patterns below.
429    
430           The  /I  modifier  requests  that pcretest output information about the
431           compiled pattern (whether it is anchored, has a fixed first  character,
432           and  so  on). It does this by calling pcre[16|32]_fullinfo() after com-
433           piling a pattern. If the pattern is studied, the results  of  that  are
434           also output.
435    
436           The  /K modifier requests pcretest to show names from backtracking con-
437           trol verbs that are  returned  from  calls  to  pcre[16|32]_exec().  It
438           causes  pcretest  to  create  a  pcre[16|32]_extra block if one has not
439           already been created by a call to pcre[16|32]_study(), and to  set  the
440           PCRE_EXTRA_MARK  flag  and  the  mark  field within it, every time that
441           pcre[16|32]_exec() is called. If  the  variable  that  the  mark  field
442           points  to  is  non-NULL  for  a  match,  non-match,  or partial match,
443           pcretest prints the string to which it points. For  a  match,  this  is
444           shown  on  a  line  by itself, tagged with "MK:". For a non-match it is
445           added to the message.
446    
447           The /L modifier must be followed directly by the name of a locale,  for
448           example,
449    
450             /pattern/Lfr_FR
451    
452           For this reason, it must be the last modifier. The given locale is set,
453           pcre[16|32]_maketables() is called to build a set of  character  tables
454           for  the  locale, and this is then passed to pcre[16|32]_compile() when
455           compiling the regular expression. Without an /L (or /T) modifier,  NULL
456           is  passed  as  the  tables  pointer;  that  is, /L applies only to the
457           expression on which it appears.
458    
459           The /M modifier causes the size in bytes of the memory  block  used  to
460           hold  the compiled pattern to be output. This does not include the size
461           of the pcre[16|32] block; it is just the actual compiled data.  If  the
462           pattern is successfully studied with the PCRE_STUDY_JIT_COMPILE option,
463           the size of the JIT compiled code is also output.
464    
465           The /S modifier causes  pcre[16|32]_study()  to  be  called  after  the
466           expression  has been compiled, and the results used when the expression
467           is matched. There are a number of qualifying characters that may follow
468           /S.  They may appear in any order.
469    
470           If  S is followed by an exclamation mark, pcre[16|32]_study() is called
471           with the PCRE_STUDY_EXTRA_NEEDED option, causing it always to return  a
472           pcre_extra block, even when studying discovers no useful information.
473    
474           If /S is followed by a second S character, it suppresses studying, even
475           if it was requested externally by the  -s  command  line  option.  This
476           makes  it possible to specify that certain patterns are always studied,
477           and others are never studied, independently of -s. This feature is used
478           in the test files in a few cases where the output is different when the
479           pattern is studied.
480    
481           If the  /S  modifier  is  followed  by  a  +  character,  the  call  to
482           pcre[16|32]_study()  is made with all the JIT study options, requesting
483           just-in-time optimization support if it is available, for  both  normal
484           and  partial matching. If you want to restrict the JIT compiling modes,
485           you can follow /S+ with a digit in the range 1 to 7:
486    
487             1  normal match only
488             2  soft partial match only
489             3  normal match and soft partial match
490             4  hard partial match only
491             6  soft and hard partial match
492             7  all three modes (default)
493    
494           If /S++ is used instead of /S+ (with or without a following digit), the
495           text  "(JIT)"  is  added  to  the first output line after a match or no
496           match when JIT-compiled code was actually used.
497    
498           Note that there is also an independent /+  modifier;  it  must  not  be
499           given immediately after /S or /S+ because this will be misinterpreted.
500    
501           If JIT studying is successful, the compiled JIT code will automatically
502           be used when pcre[16|32]_exec() is run, except when  incompatible  run-
503           time  options are specified. For more details, see the pcrejit documen-
504           tation. See also the \J escape sequence below for a way of setting  the
505           size of the JIT stack.
506    
507           Finally,  if  /S  is  followed by a minus character, JIT compilation is
508           suppressed, even if it was requested externally by the -s command  line
509           option.  This makes it possible to specify that JIT is never to be used
510           for certain patterns.
511    
512           The /T modifier must be followed by a single digit. It  causes  a  spe-
513           cific set of built-in character tables to be passed to pcre[16|32]_com-
514           pile(). It is used in the standard PCRE tests to check  behaviour  with
515           different character tables. The digit specifies the tables as follows:
516    
517             0   the default ASCII tables, as distributed in
518                   pcre_chartables.c.dist
519             1   a set of tables defining ISO 8859 characters
520    
521           In  table 1, some characters whose codes are greater than 128 are iden-
522           tified as letters, digits, spaces, etc.
523    
524       Using the POSIX wrapper API
525    
526           The /P modifier causes pcretest to call PCRE via the POSIX wrapper  API
527           rather  than its native API. This supports only the 8-bit library. When
528           /P is set, the following modifiers set options for the regcomp()  func-
529           tion:
530    
531             /i    REG_ICASE
532             /m    REG_NEWLINE
533             /N    REG_NOSUB
534             /s    REG_DOTALL     )
535             /U    REG_UNGREEDY   ) These options are not part of
536             /W    REG_UCP        )   the POSIX standard
537             /8    REG_UTF8       )
538    
539           The  /+  modifier  works  as  described  above. All other modifiers are
540           ignored.
541    
542    
543    DATA LINES
544    
545           Before each data line is  passed  to  pcre[16|32]_exec(),  leading  and
546           trailing  white space is removed, and it is then scanned for \ escapes.
547           Some of these are pretty esoteric features, intended for  checking  out
548           some  of the more complicated features of PCRE. If you are just testing
549           "ordinary" regular expressions, you probably don't need any  of  these.
550           The following escapes are recognized:
551    
552             \a         alarm (BEL, \x07)
553             \b         backspace (\x08)
554             \e         escape (\x27)
555             \f         form feed (\x0c)
556             \n         newline (\x0a)
557             \qdd       set the PCRE_MATCH_LIMIT limit to dd
558                          (any number of digits)
559             \r         carriage return (\x0d)
560             \t         tab (\x09)
561             \v         vertical tab (\x0b)
562             \nnn       octal character (up to 3 octal digits); always
563                          a byte unless > 255 in UTF-8 or 16-bit or 32-bit mode
564             \xhh       hexadecimal byte (up to 2 hex digits)
565             \x{hh...}  hexadecimal character (any number of hex digits)
566             \A         pass the PCRE_ANCHORED option to pcre[16|32]_exec()
567                          or pcre[16|32]_dfa_exec()
568             \B         pass the PCRE_NOTBOL option to pcre[16|32]_exec()
569                          or pcre[16|32]_dfa_exec()
570             \Cdd       call pcre[16|32]_copy_substring() for substring dd
571                          after a successful match (number less than 32)
572             \Cname     call pcre[16|32]_copy_named_substring() for substring
573                          "name" after a successful match (name termin-
574                          ated by next non alphanumeric character)
575             \C+        show the current captured substrings at callout
576                          time
577             \C-        do not supply a callout function
578             \C!n       return 1 instead of 0 when callout number n is
579                          reached
580             \C!n!m     return 1 instead of 0 when callout number n is
581                          reached for the nth time
582             \C*n       pass the number n (may be negative) as callout
583                          data; this is used as the callout return value
584             \D         use the pcre[16|32]_dfa_exec() match function
585             \F         only shortest match for pcre[16|32]_dfa_exec()
586             \Gdd       call pcre[16|32]_get_substring() for substring dd
587                          after a successful match (number less than 32)
588             \Gname     call pcre[16|32]_get_named_substring() for substring
589                          "name" after a successful match (name termin-
590                          ated by next non-alphanumeric character)
591             \Jdd       set up a JIT stack of dd kilobytes maximum (any
592                          number of digits)
593             \L         call pcre[16|32]_get_substringlist() after a
594                          successful match
595             \M         discover the minimum MATCH_LIMIT and
596                          MATCH_LIMIT_RECURSION settings
597             \N         pass the PCRE_NOTEMPTY option to pcre[16|32]_exec()
598                          or pcre[16|32]_dfa_exec(); if used twice, pass the
599                          PCRE_NOTEMPTY_ATSTART option
600             \Odd       set the size of the output vector passed to
601                          pcre[16|32]_exec() to dd (any number of digits)
602             \P         pass the PCRE_PARTIAL_SOFT option to pcre[16|32]_exec()
603                          or pcre[16|32]_dfa_exec(); if used twice, pass the
604                          PCRE_PARTIAL_HARD option
605             \Qdd       set the PCRE_MATCH_LIMIT_RECURSION limit to dd
606                          (any number of digits)
607             \R         pass the PCRE_DFA_RESTART option to pcre[16|32]_dfa_exec()
608             \S         output details of memory get/free calls during matching
609             \Y             pass     the    PCRE_NO_START_OPTIMIZE    option    to
610           pcre[16|32]_exec()
611                          or pcre[16|32]_dfa_exec()
612             \Z         pass the PCRE_NOTEOL option to pcre[16|32]_exec()
613                          or pcre[16|32]_dfa_exec()
614             \?         pass the PCRE_NO_UTF[8|16|32]_CHECK option to
615                          pcre[16|32]_exec() or pcre[16|32]_dfa_exec()
616             \>dd       start the match at offset dd (optional "-"; then
617                          any number of digits); this sets the startoffset
618                          argument        for        pcre[16|32]_exec()         or
619           pcre[16|32]_dfa_exec()
620             \<cr>      pass the PCRE_NEWLINE_CR option to pcre[16|32]_exec()
621                          or pcre[16|32]_dfa_exec()
622             \<lf>      pass the PCRE_NEWLINE_LF option to pcre[16|32]_exec()
623                          or pcre[16|32]_dfa_exec()
624             \<crlf>    pass the PCRE_NEWLINE_CRLF option to pcre[16|32]_exec()
625                          or pcre[16|32]_dfa_exec()
626             \<anycrlf> pass the PCRE_NEWLINE_ANYCRLF option to pcre[16|32]_exec()
627                          or pcre[16|32]_dfa_exec()
628             \<any>     pass the PCRE_NEWLINE_ANY option to pcre[16|32]_exec()
629                          or pcre[16|32]_dfa_exec()
630    
631           The  use of \x{hh...} is not dependent on the use of the /8 modifier on
632           the pattern. It is recognized always. There may be any number of  hexa-
633           decimal  digits  inside  the  braces; invalid values provoke error mes-
634           sages.
635    
636           Note that \xhh specifies one byte rather than one  character  in  UTF-8
637           mode;  this  makes it possible to construct invalid UTF-8 sequences for
638           testing purposes. On the other hand, \x{hh} is interpreted as  a  UTF-8
639           character  in UTF-8 mode, generating more than one byte if the value is
640           greater than 127.  When testing the 8-bit library not  in  UTF-8  mode,
641           \x{hh} generates one byte for values less than 256, and causes an error
642           for greater values.
643    
644           In UTF-16 mode, all 4-digit \x{hhhh} values are accepted. This makes it
645           possible to construct invalid UTF-16 sequences for testing purposes.
646    
647           In  UTF-32  mode,  all  4- to 8-digit \x{...} values are accepted. This
648           makes it possible to construct invalid  UTF-32  sequences  for  testing
649           purposes.
650    
651           The  escapes  that  specify  line ending sequences are literal strings,
652           exactly as shown. No more than one newline setting should be present in
653           any data line.
654    
655           A  backslash  followed by anything else just escapes the anything else.
656           If the very last character is a backslash, it is ignored. This gives  a
657           way  of  passing  an empty line as data, since a real empty line termi-
658           nates the data input.
659    
660           The \J escape provides a way of setting the maximum stack size that  is
661           used  by the just-in-time optimization code. It is ignored if JIT opti-
662           mization is not being used. Providing a stack that is larger  than  the
663           default 32K is necessary only for very complicated patterns.
664    
665           If \M is present, pcretest calls pcre[16|32]_exec() several times, with
666           different values in the match_limit and match_limit_recursion fields of
667           the  pcre[16|32]_extra  data structure, until it finds the minimum num-
668           bers for each parameter that allow pcre[16|32]_exec() to complete with-
669           out  error.  Because  this  is testing a specific feature of the normal
670           interpretive pcre[16|32]_exec() execution, the use of any JIT optimiza-
671           tion  that might have been set up by the /S+ qualifier of -s+ option is
672           disabled.
673    
674           The match_limit number is a measure of the amount of backtracking  that
675           takes  place,  and  checking it out can be instructive. For most simple
676           matches, the number is quite small, but for patterns  with  very  large
677           numbers  of  matching  possibilities,  it can become large very quickly
678           with increasing length of  subject  string.  The  match_limit_recursion
679           number  is  a  measure  of how much stack (or, if PCRE is compiled with
680           NO_RECURSE, how much heap) memory  is  needed  to  complete  the  match
681           attempt.
682    
683           When  \O  is  used, the value specified may be higher or lower than the
684           size set by the -O command line option (or defaulted to 45); \O applies
685           only  to  the  call  of  pcre[16|32]_exec()  for  the  line in which it
686           appears.
687    
688           If the /P modifier was present on the pattern, causing the POSIX  wrap-
689           per  API  to  be  used, the only option-setting sequences that have any
690           effect are \B,  \N,  and  \Z,  causing  REG_NOTBOL,  REG_NOTEMPTY,  and
691           REG_NOTEOL, respectively, to be passed to regexec().
692    
693    
694    THE ALTERNATIVE MATCHING FUNCTION
695    
696           By   default,  pcretest  uses  the  standard  PCRE  matching  function,
697           pcre[16|32]_exec() to match each  data  line.  PCRE  also  supports  an
698           alternative  matching  function, pcre[16|32]_dfa_test(), which operates
699           in a different way, and has some restrictions. The differences  between
700           the two functions are described in the pcrematching documentation.
701    
702           If  a data line contains the \D escape sequence, or if the command line
703           contains the -dfa option, the alternative matching  function  is  used.
704           This function finds all possible matches at a given point. If, however,
705           the \F escape sequence is present in the data line, it stops after  the
706           first match is found. This is always the shortest possible match.
707    
708    
709    DEFAULT OUTPUT FROM PCRETEST
710    
711           This  section  describes  the output when the normal matching function,
712           pcre[16|32]_exec(), is being used.
713    
714           When a match succeeds, pcretest outputs the list of captured substrings
715           that  pcre[16|32]_exec() returns, starting with number 0 for the string
716           that matched the whole pattern. Otherwise, it outputs "No  match"  when
717           the  return is PCRE_ERROR_NOMATCH, and "Partial match:" followed by the
718           partially   matching   substring   when   pcre[16|32]_exec()    returns
719           PCRE_ERROR_PARTIAL.  (Note  that  this is the entire substring that was
720           inspected during the partial match; it may  include  characters  before
721           the  actual  match  start  if a lookbehind assertion, \K, \b, or \B was
722           involved.) For any other return, pcretest  outputs  the  PCRE  negative
723           error  number  and a short descriptive phrase. If the error is a failed
724           UTF string check, the offset of the start of the failing character  and
725           the  reason  code are also output, provided that the size of the output
726           vector is at least two. Here is an example of an  interactive  pcretest
727           run.
728    
729             $ pcretest
730             PCRE version 8.13 2011-04-30
731    
732               re> /^abc(\d+)/
733             data> abc123
734              0: abc123
735              1: 123
736             data> xyz
737             No match
738    
739           Unset capturing substrings that are not followed by one that is set are
740           not returned by pcre[16|32]_exec(), and are not shown by  pcretest.  In
741           the following example, there are two capturing substrings, but when the
742           first data line is matched, the second, unset substring is  not  shown.
743           An  "internal" unset substring is shown as "<unset>", as for the second
744           data line.
745    
746               re> /(a)|(b)/
747             data> a
748              0: a
749              1: a
750             data> b
751              0: b
752              1: <unset>
753              2: b
754    
755           If the strings contain any non-printing characters, they are output  as
756           \xhh  escapes  if  the  value is less than 256 and UTF mode is not set.
757           Otherwise they are output as \x{hh...} escapes. See below for the defi-
758           nition  of non-printing characters. If the pattern has the /+ modifier,
759           the output for substring 0 is followed by the the rest of  the  subject
760           string, identified by "0+" like this:
761    
762               re> /cat/+
763             data> cataract
764              0: cat
765              0+ aract
766    
767           If  the  pattern  has  the /g or /G modifier, the results of successive
768           matching attempts are output in sequence, like this:
769    
770               re> /\Bi(\w\w)/g
771             data> Mississippi
772              0: iss
773              1: ss
774              0: iss
775              1: ss
776              0: ipp
777              1: pp
778    
779           "No match" is output only if the first match attempt fails. Here is  an
780           example  of a failure message (the offset 4 that is specified by \>4 is
781           past the end of the subject string):
782    
783               re> /xyz/
784             data> xyz\>4
785             Error -24 (bad offset value)
786    
787           If any of the sequences \C, \G, or \L are present in a data  line  that
788           is  successfully  matched,  the substrings extracted by the convenience
789           functions are output with C, G, or L after the string number instead of
790           a colon. This is in addition to the normal full list. The string length
791           (that is, the return from the extraction function) is given  in  paren-
792           theses after each string for \C and \G.
793    
794           Note that whereas patterns can be continued over several lines (a plain
795           ">" prompt is used for continuations), data lines may not. However new-
796           lines  can  be included in data by means of the \n escape (or \r, \r\n,
797           etc., depending on the newline sequence setting).
798    
799    
800    OUTPUT FROM THE ALTERNATIVE MATCHING FUNCTION
801    
802           When the alternative matching function, pcre[16|32]_dfa_exec(), is used
803           (by  means  of the \D escape sequence or the -dfa command line option),
804           the output consists of a list of all the  matches  that  start  at  the
805           first point in the subject where there is at least one match. For exam-
806           ple:
807    
808               re> /(tang|tangerine|tan)/
809             data> yellow tangerine\D
810              0: tangerine
811              1: tang
812              2: tan
813    
814           (Using the normal matching function on this data  finds  only  "tang".)
815           The  longest matching string is always given first (and numbered zero).
816           After a PCRE_ERROR_PARTIAL return, the output is "Partial match:", fol-
817           lowed  by  the  partially  matching  substring.  (Note that this is the
818           entire substring that was inspected during the partial  match;  it  may
819           include characters before the actual match start if a lookbehind asser-
820           tion, \K, \b, or \B was involved.)
821    
822           If /g is present on the pattern, the search for further matches resumes
823           at the end of the longest match. For example:
824    
825               re> /(tang|tangerine|tan)/g
826             data> yellow tangerine and tangy sultana\D
827              0: tangerine
828              1: tang
829              2: tan
830              0: tang
831              1: tan
832              0: tan
833    
834           Since  the  matching  function  does not support substring capture, the
835           escape sequences that are concerned with captured  substrings  are  not
836           relevant.
837    
838    
839    RESTARTING AFTER A PARTIAL MATCH
840    
841           When the alternative matching function has given the PCRE_ERROR_PARTIAL
842           return, indicating that the subject partially matched the pattern,  you
843           can  restart  the match with additional subject data by means of the \R
844           escape sequence. For example:
845    
846               re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
847             data> 23ja\P\D
848             Partial match: 23ja
849             data> n05\R\D
850              0: n05
851    
852           For further information about partial  matching,  see  the  pcrepartial
853           documentation.
854    
855    
856    CALLOUTS
857    
858           If  the pattern contains any callout requests, pcretest's callout func-
859           tion is called during matching. This works  with  both  matching  func-
860           tions. By default, the called function displays the callout number, the
861           start and current positions in the text at the callout  time,  and  the
862           next pattern item to be tested. For example:
863    
864             --->pqrabcdef
865               0    ^  ^     \d
866    
867           This  output  indicates  that  callout  number  0  occurred for a match
868           attempt starting at the fourth character of the  subject  string,  when
869           the pointer was at the seventh character of the data, and when the next
870           pattern item was \d. Just one circumflex is output  if  the  start  and
871           current positions are the same.
872    
873           Callouts numbered 255 are assumed to be automatic callouts, inserted as
874           a result of the /C pattern modifier. In this case, instead  of  showing
875           the  callout  number, the offset in the pattern, preceded by a plus, is
876           output. For example:
877    
878               re> /\d?[A-E]\*/C
879             data> E*
880             --->E*
881              +0 ^      \d?
882              +3 ^      [A-E]
883              +8 ^^     \*
884             +10 ^ ^
885              0: E*
886    
887           If a pattern contains (*MARK) items, an additional line is output when-
888           ever  a  change  of  latest mark is passed to the callout function. For
889           example:
890    
891               re> /a(*MARK:X)bc/C
892             data> abc
893             --->abc
894              +0 ^       a
895              +1 ^^      (*MARK:X)
896             +10 ^^      b
897             Latest Mark: X
898             +11 ^ ^     c
899             +12 ^  ^
900              0: abc
901    
902           The mark changes between matching "a" and "b", but stays the  same  for
903           the  rest  of  the match, so nothing more is output. If, as a result of
904           backtracking, the mark reverts to being unset, the  text  "<unset>"  is
905           output.
906    
907           The  callout  function  in pcretest returns zero (carry on matching) by
908           default, but you can use a \C item in a data line (as described  above)
909           to change this and other parameters of the callout.
910    
911           Inserting  callouts can be helpful when using pcretest to check compli-
912           cated regular expressions. For further information about callouts,  see
913           the pcrecallout documentation.
914    
915    
916    NON-PRINTING CHARACTERS
917    
918           When  pcretest is outputting text in the compiled version of a pattern,
919           bytes other than 32-126 are always treated as  non-printing  characters
920           are are therefore shown as hex escapes.
921    
922           When  pcretest  is  outputting text that is a matched part of a subject
923           string, it behaves in the same way, unless a different locale has  been
924           set  for  the  pattern  (using  the  /L  modifier).  In  this case, the
925           isprint() function to distinguish printing and non-printing characters.
926    
927    
928    SAVING AND RELOADING COMPILED PATTERNS
929    
930           The facilities described in this section are  not  available  when  the
931           POSIX  interface  to  PCRE  is being used, that is, when the /P pattern
932           modifier is specified.
933    
934           When the POSIX interface is not in use, you can cause pcretest to write
935           a  compiled  pattern to a file, by following the modifiers with > and a
936           file name.  For example:
937    
938             /pattern/im >/some/file
939    
940           See the pcreprecompile documentation for a discussion about saving  and
941           re-using  compiled patterns.  Note that if the pattern was successfully
942           studied with JIT optimization, the JIT data cannot be saved.
943    
944           The data that is written is binary.  The  first  eight  bytes  are  the
945           length  of  the  compiled  pattern  data  followed by the length of the
946           optional study data, each written as four  bytes  in  big-endian  order
947           (most  significant  byte  first). If there is no study data (either the
948           pattern was not studied, or studying did not return any data), the sec-
949           ond  length  is  zero. The lengths are followed by an exact copy of the
950           compiled pattern. If there is additional study  data,  this  (excluding
951           any  JIT  data)  follows  immediately after the compiled pattern. After
952           writing the file, pcretest expects to read a new pattern.
953    
954           A saved pattern can be reloaded into pcretest by  specifying  <  and  a
955           file name instead of a pattern. The name of the file must not contain a
956           < character, as otherwise pcretest will interpret the line as a pattern
957           delimited by < characters.  For example:
958    
959              re> </some/file
960             Compiled pattern loaded from /some/file
961             No study data
962    
963           If  the  pattern  was previously studied with the JIT optimization, the
964           JIT information cannot be saved and restored, and so is lost. When  the
965           pattern  has  been  loaded, pcretest proceeds to read data lines in the
966           usual way.
967    
968           You can copy a file written by pcretest to a different host and  reload
969           it  there,  even  if the new host has opposite endianness to the one on
970           which the pattern was compiled. For example, you can compile on an  i86
971           machine  and  run  on  a SPARC machine. When a pattern is reloaded on a
972           host with different endianness, the confirmation message is changed to:
973    
974             Compiled pattern (byte-inverted) loaded from /some/file
975    
976           The test suite contains some saved pre-compiled patterns with different
977           endianness.  These  are  reloaded  using "<!" instead of just "<". This
978           suppresses the "(byte-inverted)" text so that the output is the same on
979           all  hosts.  It  also forces debugging output once the pattern has been
980           reloaded.
981    
982           File names for saving and reloading can be absolute  or  relative,  but
983           note  that the shell facility of expanding a file name that starts with
984           a tilde (~) is not available.
985    
986           The ability to save and reload files in pcretest is intended for  test-
987           ing  and experimentation. It is not intended for production use because
988           only a single pattern can be written to a file. Furthermore,  there  is
989           no  facility  for  supplying  custom  character  tables  for use with a
990           reloaded pattern. If the original  pattern  was  compiled  with  custom
991           tables,  an  attempt to match a subject string using a reloaded pattern
992           is likely to cause pcretest to crash.  Finally, if you attempt to  load
993           a file that is not in the correct format, the result is undefined.
994    
995    
996    SEE ALSO
997    
998           pcre(3),  pcre16(3),  pcre32(3),  pcreapi(3),  pcrecallout(3), pcrejit,
999           pcrematching(3), pcrepartial(d), pcrepattern(3), pcreprecompile(3).
1000    
1001    
1002    AUTHOR
1003    
1004           Philip Hazel
1005           University Computing Service
1006           Cambridge CB2 3QH, England.
1007    
1008    
1009    REVISION
1010    
1011           Last updated: 10 September 2012
1012           Copyright (c) 1997-2012 University of Cambridge.

Legend:
Removed from v.47  
changed lines
  Added in v.1194

  ViewVC Help
Powered by ViewVC 1.1.5