mux/src/pcre.cpp

Go to the documentation of this file.

/*************************************************
*      Perl-Compatible Regular Expressions       *
*************************************************/


/* This is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language. See
the file Tech.Notes for some information on the internals.

Written by: Philip Hazel <ph10@cam.ac.uk>

           Copyright (c) 1997-2003 University of Cambridge

-----------------------------------------------------------------------------
Permission is granted to anyone to use this software for any purpose on any
computer system, and to redistribute it freely, subject to the following
restrictions:

1. This software is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

2. The origin of this software must not be misrepresented, either by
   explicit claim or by omission.

3. Altered versions must be plainly marked as such, and must not be
   misrepresented as being the original software.

4. If PCRE is embedded in any software that is released under the GNU
   General Purpose Licence (GPL), then the terms of that licence shall
   supersede any condition above with which it is incompatible.
-----------------------------------------------------------------------------
*/

/* Modified by Shawn Wagner for MUX to fit in one file and remove
   things we don't use, like a bunch of API functions and utf-8
   support. If you want the full thing, see http://www.pcre.org.

   Patched by Alierak to protect against integer overflow in repeat
   counts.
 */

#include "autoconf.h"
#include "config.h"

#include <limits.h>
#include <string.h>
#include <ctype.h>
#include <stdlib.h>
#include <stddef.h>
#include "pcre.h"

#include "externs.h"
#include "timeutil.h"

/* Bits of PCRE's config.h */
#define LINK_SIZE 2
#define MATCH_LIMIT 100000
#define NEWLINE '\n'

/* Bits of internal.h */
/* This header contains definitions that are shared between the different
modules, but which are not relevant to the outside. */


/* PCRE keeps offsets in its compiled code as 2-byte quantities by default.
These are used, for example, to link from the start of a subpattern to its
alternatives and its end. The use of 2 bytes per offset limits the size of the
compiled regex to around 64K, which is big enough for almost everybody.
However, I received a request for an even bigger limit. For this reason, and
also to make the code easier to maintain, the storing and loading of offsets
from the byte string is now handled by the macros that are defined here.

The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
the config.h file, but can be overridden by using -D on the command line. This
is automated on Unix systems via the "configure" command. */

#define PUT(a,n,d)   \
  (a[n] = (d) >> 8), \
  (a[(n)+1] = (d) & 255)

#define GET(a,n) \
  (((a)[n] << 8) | (a)[(n)+1])

#define MAX_PATTERN_SIZE (1 << 16)


/* Convenience macro defined in terms of the others */

#define PUTINC(a,n,d)   PUT(a,n,d), a += LINK_SIZE


/* PCRE uses some other 2-byte quantities that do not change when the size of
offsets changes. There are used for repeat counts and for other things such as
capturing parenthesis numbers in back references. */

#define PUT2(a,n,d)   \
  a[n] = (d) >> 8; \
  a[(n)+1] = (d) & 255

#define GET2(a,n) \
  (((a)[n] << 8) | (a)[(n)+1])

#define PUT2INC(a,n,d)  PUT2(a,n,d), a += 2

/* These are the public options that can change during matching. */

#define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)

/* Private options flags start at the most significant end of the four bytes,
but skip the top bit so we can use ints for convenience without getting tangled
with negative values. The public options defined in pcre.h start at the least
significant end. Make sure they don't overlap, though now that we have expanded
to four bytes there is plenty of space. */

#define PCRE_FIRSTSET      0x40000000  /* first_byte is set */
#define PCRE_REQCHSET      0x20000000  /* req_byte is set */
#define PCRE_STARTLINE     0x10000000  /* start after \n for multiline */
#define PCRE_ICHANGED      0x08000000  /* i option changes within regex */

/* Options for the "extra" block produced by pcre_study(). */

#define PCRE_STUDY_MAPPED   0x01     /* a map of starting chars exists */

/* Masks for identifying the public options which are permitted at compile
time, run time or study time, respectively. */

#define PUBLIC_OPTIONS \
  (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
   PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
   PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK)

#define PUBLIC_EXEC_OPTIONS \
  (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK)

#define PUBLIC_STUDY_OPTIONS 0   /* None defined */

/* Magic number to provide a small check against being handed junk. */

#define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */

/* Negative values for the firstchar and reqchar variables */

#define REQ_UNSET (-2)
#define REQ_NONE  (-1)

/* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
variable-length repeat, or a anything other than literal characters. */

#define REQ_CASELESS 0x0100    /* indicates caselessness */
#define REQ_VARY     0x0200    /* reqbyte followed non-literal item */

/* Miscellaneous definitions */

/* Escape items that are just an encoding of a particular data value. Note that
ESC_n is defined as yet another macro, which is set in config.h to either \n
(the default) or \r (which some people want). */

#ifndef ESC_e
#define ESC_e 27
#endif

#ifndef ESC_f
#define ESC_f '\f'
#endif

#ifndef ESC_n
#define ESC_n NEWLINE
#endif

#ifndef ESC_r
#define ESC_r '\r'
#endif

/* We can't officially use ESC_t because it is a POSIX reserved identifier
(presumably because of all the others like size_t). */

#ifndef ESC_tee
#define ESC_tee '\t'
#endif

/* These are escaped items that aren't just an encoding of a particular data
value such as \n. They must have non-zero values, as check_escape() returns
their negation. Also, they must appear in the same order as in the opcode
definitions below, up to ESC_z. There's a dummy for OP_ANY because it
corresponds to "." rather than an escape sequence. The final one must be
ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two
tests in the code for an escape greater than ESC_b and less than ESC_Z to
detect the types that may be repeated. These are the types that consume a
character. If any new escapes are put in between that don't consume a
character, that code will have to change. */

enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
       ESC_w, ESC_dum1, ESC_C, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_REF };

/* Flag bits and data types for the extended class (OP_XCLASS) for classes that
contain UTF-8 characters with values greater than 255. */

#define XCL_NOT    0x01    /* Flag: this is a negative class */
#define XCL_MAP    0x02    /* Flag: a 32-byte map is present */

#define XCL_END       0    /* Marks end of individual items */
#define XCL_SINGLE    1    /* Single item (one multibyte char) follows */
#define XCL_RANGE     2    /* A range (two multibyte chars) follows */


/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
OP_EOD must correspond in order to the list of escapes immediately above.
Note that whenever this list is updated, the two macro definitions that follow
must also be updated to match. */

enum {
  OP_END,            /* 0 End of pattern */

  /* Values corresponding to backslashed metacharacters */

  OP_SOD,            /* 1 Start of data: \A */
  OP_SOM,            /* 2 Start of match (subject + offset): \G */
  OP_NOT_WORD_BOUNDARY,  /*  3 \B */
  OP_WORD_BOUNDARY,      /*  4 \b */
  OP_NOT_DIGIT,          /*  5 \D */
  OP_DIGIT,              /*  6 \d */
  OP_NOT_WHITESPACE,     /*  7 \S */
  OP_WHITESPACE,         /*  8 \s */
  OP_NOT_WORDCHAR,       /*  9 \W */
  OP_WORDCHAR,           /* 10 \w */
  OP_ANY,            /* 11 Match any character */
  OP_ANYBYTE,        /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
  OP_EODN,           /* 13 End of data or \n at end of data: \Z. */
  OP_EOD,            /* 14 End of data: \z */

  OP_OPT,            /* 15 Set runtime options */
  OP_CIRC,           /* 16 Start of line - varies with multiline switch */
  OP_DOLL,           /* 17 End of line - varies with multiline switch */
  OP_CHARS,          /* 18 Match string of characters */
  OP_NOT,            /* 19 Match anything but the following char */

  OP_STAR,           /* 20 The maximizing and minimizing versions of */
  OP_MINSTAR,        /* 21 all these opcodes must come in pairs, with */
  OP_PLUS,           /* 22 the minimizing one second. */
  OP_MINPLUS,        /* 23 This first set applies to single characters */
  OP_QUERY,          /* 24 */
  OP_MINQUERY,       /* 25 */
  OP_UPTO,           /* 26 From 0 to n matches */
  OP_MINUPTO,        /* 27 */
  OP_EXACT,          /* 28 Exactly n matches */

  OP_NOTSTAR,        /* 29 The maximizing and minimizing versions of */
  OP_NOTMINSTAR,     /* 30 all these opcodes must come in pairs, with */
  OP_NOTPLUS,        /* 31 the minimizing one second. */
  OP_NOTMINPLUS,     /* 32 This set applies to "not" single characters */
  OP_NOTQUERY,       /* 33 */
  OP_NOTMINQUERY,    /* 34 */
  OP_NOTUPTO,        /* 35 From 0 to n matches */
  OP_NOTMINUPTO,     /* 36 */
  OP_NOTEXACT,       /* 37 Exactly n matches */

  OP_TYPESTAR,       /* 38 The maximizing and minimizing versions of */
  OP_TYPEMINSTAR,    /* 39 all these opcodes must come in pairs, with */
  OP_TYPEPLUS,       /* 40 the minimizing one second. These codes must */
  OP_TYPEMINPLUS,    /* 41 be in exactly the same order as those above. */
  OP_TYPEQUERY,      /* 42 This set applies to character types such as \d */
  OP_TYPEMINQUERY,   /* 43 */
  OP_TYPEUPTO,       /* 44 From 0 to n matches */
  OP_TYPEMINUPTO,    /* 45 */
  OP_TYPEEXACT,      /* 46 Exactly n matches */

  OP_CRSTAR,         /* 47 The maximizing and minimizing versions of */
  OP_CRMINSTAR,      /* 48 all these opcodes must come in pairs, with */
  OP_CRPLUS,         /* 49 the minimizing one second. These codes must */
  OP_CRMINPLUS,      /* 50 be in exactly the same order as those above. */
  OP_CRQUERY,        /* 51 These are for character classes and back refs */
  OP_CRMINQUERY,     /* 52 */
  OP_CRRANGE,        /* 53 These are different to the three seta above. */
  OP_CRMINRANGE,     /* 54 */

  OP_CLASS,          /* 55 Match a character class, chars < 256 only */
  OP_NCLASS,         /* 56 Same, but the bitmap was created from a negative
                           class - the difference is relevant only when a UTF-8
                           character > 255 is encountered. */

  OP_XCLASS,         /* 57 Extended class for handling UTF-8 chars within the
                           class. This does both positive and negative. */

  OP_REF,            /* 58 Match a back reference */
  OP_RECURSE,        /* 59 Match a numbered subpattern (possibly recursive) */
  OP_CALLOUT,        /* 60 Call out to external function if provided */

  OP_ALT,            /* 61 Start of alternation */
  OP_KET,            /* 62 End of group that doesn't have an unbounded repeat */
  OP_KETRMAX,        /* 63 These two must remain together and in this */
  OP_KETRMIN,        /* 64 order. They are for groups the repeat for ever. */

  /* The assertions must come before ONCE and COND */

  OP_ASSERT,         /* 65 Positive lookahead */
  OP_ASSERT_NOT,     /* 66 Negative lookahead */
  OP_ASSERTBACK,     /* 67 Positive lookbehind */
  OP_ASSERTBACK_NOT, /* 68 Negative lookbehind */
  OP_REVERSE,        /* 69 Move pointer back - used in lookbehind assertions */

  /* ONCE and COND must come after the assertions, with ONCE first, as there's
  a test for >= ONCE for a subpattern that isn't an assertion. */

  OP_ONCE,           /* 70 Once matched, don't back up into the subpattern */
  OP_COND,           /* 71 Conditional group */
  OP_CREF,           /* 72 Used to hold an extraction string number (cond ref) */

  OP_BRAZERO,        /* 73 These two must remain together and in this */
  OP_BRAMINZERO,     /* 74 order. */

  OP_BRANUMBER,      /* 75 Used for extracting brackets whose number is greater
                           than can fit into an opcode. */

  OP_BRA             /* 76 This and greater values are used for brackets that
                           extract substrings up to a basic limit. After that,
                           use is made of OP_BRANUMBER. */
};

/* WARNING: There is an implicit assumption in study.c that all opcodes are
less than 128 in value. This makes handling UTF-8 character sequences easier.
*/


/* This macro defines textual names for all the opcodes. There are used only
for debugging, in pcre.c when DEBUG is defined, and also in pcretest.c. The
macro is referenced only in printint.c. */

#define OP_NAME_LIST \
  "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d",                \
  "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", "\\Z", "\\z",     \
  "Opt", "^", "$", "chars", "not",                                \
  "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
  "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
  "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
  "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
  "class", "nclass", "xclass", "Ref", "Recurse", "Callout",       \
  "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not",     \
  "AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\
  "Brazero", "Braminzero", "Branumber", "Bra"


/* This macro defines the length of fixed length operations in the compiled
regex. The lengths are used when searching for specific things, and also in the
debugging printing of a compiled regex. We use a macro so that it can be
incorporated both into pcre.c and pcretest.c without being publicly exposed.

As things have been extended, some of these are no longer fixed lenths, but are
minima instead. For example, the length of a single-character repeat may vary
in UTF-8 mode. The code that uses this table must know about such things. */

#define OP_LENGTHS \
  1,                             /* End                                    */ \
  1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
  1, 1, 1, 1, 2, 1, 1,           /* Any, Anybyte, \Z, \z, Opt, ^, $        */ \
  2,                             /* Chars - the minimum length             */ \
  2,                             /* not                                    */ \
  /* Positive single-char repeats                            ** These are  */ \
  2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??      ** minima in  */ \
  4, 4, 4,                       /* upto, minupto, exact     ** UTF-8 mode */ \
  /* Negative single-char repeats - only for chars < 256                   */ \
  2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \
  4, 4, 4,                       /* NOT upto, minupto, exact               */ \
  /* Positive type repeats                                                 */ \
  2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \
  4, 4, 4,                       /* Type upto, minupto, exact              */ \
  /* Character class & ref repeats                                         */ \
  1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
  5, 5,                          /* CRRANGE, CRMINRANGE                    */ \
 33,                             /* CLASS                                  */ \
 33,                             /* NCLASS                                 */ \
  0,                             /* XCLASS - variable length               */ \
  3,                             /* REF                                    */ \
  1+LINK_SIZE,                   /* RECURSE                                */ \
  2,                             /* CALLOUT                                */ \
  1+LINK_SIZE,                   /* Alt                                    */ \
  1+LINK_SIZE,                   /* Ket                                    */ \
  1+LINK_SIZE,                   /* KetRmax                                */ \
  1+LINK_SIZE,                   /* KetRmin                                */ \
  1+LINK_SIZE,                   /* Assert                                 */ \
  1+LINK_SIZE,                   /* Assert not                             */ \
  1+LINK_SIZE,                   /* Assert behind                          */ \
  1+LINK_SIZE,                   /* Assert behind not                      */ \
  1+LINK_SIZE,                   /* Reverse                                */ \
  1+LINK_SIZE,                   /* Once                                   */ \
  1+LINK_SIZE,                   /* COND                                   */ \
  3,                             /* CREF                                   */ \
  1, 1,                          /* BRAZERO, BRAMINZERO                    */ \
  3,                             /* BRANUMBER                              */ \
  1+LINK_SIZE                    /* BRA                                    */ \


/* The highest extraction number before we have to start using additional
bytes. (Originally PCRE didn't have support for extraction counts highter than
this number.) The value is limited by the number of opcodes left after OP_BRA,
i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional
opcodes. */

#define EXTRACT_BASIC_MAX  150

/* A magic value for OP_CREF to indicate the "in recursion" condition. */

#define CREF_RECURSE  0xffff

/* The texts of compile-time error messages are defined as macros here so that
they can be accessed by the POSIX wrapper and converted into error codes.  Yes,
I could have used error codes in the first place, but didn't feel like changing
just to accommodate the POSIX wrapper. */

#define ERR1  "\\ at end of pattern"
#define ERR2  "\\c at end of pattern"
#define ERR3  "unrecognized character follows \\"
#define ERR4  "numbers out of order in {} quantifier"
#define ERR5  "number too big in {} quantifier"
#define ERR6  "missing terminating ] for character class"
#define ERR7  "invalid escape sequence in character class"
#define ERR8  "range out of order in character class"
#define ERR9  "nothing to repeat"
#define ERR10 "operand of unlimited repeat could match the empty string"
#define ERR11 "internal error: unexpected repeat"
#define ERR12 "unrecognized character after (?"
#define ERR13 "POSIX named classes are supported only within a class"
#define ERR14 "missing )"
#define ERR15 "reference to non-existent subpattern"
#define ERR16 "erroffset passed as NULL"
#define ERR17 "unknown option bit(s) set"
#define ERR18 "missing ) after comment"
#define ERR19 "parentheses nested too deeply"
#define ERR20 "regular expression too large"
#define ERR21 "failed to get memory"
#define ERR22 "unmatched parentheses"
#define ERR23 "internal error: code overflow"
#define ERR24 "unrecognized character after (?<"
#define ERR25 "lookbehind assertion is not fixed length"
#define ERR26 "malformed number after (?("
#define ERR27 "conditional group contains more than two branches"
#define ERR28 "assertion expected after (?("
#define ERR29 "(?R or (?digits must be followed by )"
#define ERR30 "unknown POSIX class name"
#define ERR31 "POSIX collating elements are not supported"
#define ERR32 "this version of PCRE is not compiled with PCRE_UTF8 support"
#define ERR33 "spare error"
#define ERR34 "character value in \\x{...} sequence is too large"
#define ERR35 "invalid condition (?(0)"
#define ERR36 "\\C not allowed in lookbehind assertion"
#define ERR37 "PCRE does not support \\L, \\l, \\N, \\P, \\p, \\U, \\u, or \\X"
#define ERR38 "number after (?C is > 255"
#define ERR39 "closing ) for (?C expected"
#define ERR40 "recursive call could loop indefinitely"
#define ERR41 "unrecognized character after (?P"
#define ERR42 "syntax error after (?P"
#define ERR43 "two named groups have the same name"
#define ERR44 "invalid UTF-8 string"

/* All character handling must be done as unsigned characters. Otherwise there
are problems with top-bit-set characters and functions such as isspace().
However, we leave the interface to the outside world as char *, because that
should make things easier for callers. We define a short type for unsigned char
to save lots of typing. I tried "uchar", but it causes problems on Digital
Unix, where it is defined in sys/types, so use "uschar" instead. */

typedef unsigned char uschar;

/* The real format of the start of the pcre block; the index of names and the
code vector run on as long as necessary after the end. */

typedef struct real_pcre {
  unsigned long int magic_number;
  size_t size;                        /* Total that was malloced */
  const unsigned char *tables;        /* Pointer to tables */
  unsigned long int options;
  unsigned short int top_bracket;
  unsigned short int top_backref;
  unsigned short int first_byte;
  unsigned short int req_byte;
  unsigned short int name_entry_size; /* Size of any name items; 0 => none */
  unsigned short int name_count;      /* Number of name items */
} real_pcre;

/* The format of the block used to store data from pcre_study(). */

typedef struct pcre_study_data {
  size_t size;                        /* Total that was malloced */
  uschar options;
  uschar start_bits[32];
} pcre_study_data;

/* Structure for passing "static" information around between the functions
doing the compiling, so that they are thread-safe. */

typedef struct compile_data {
  const uschar *lcc;            /* Points to lower casing table */
  const uschar *fcc;            /* Points to case-flipping table */
  const uschar *cbits;          /* Points to character type table */
  const uschar *ctypes;         /* Points to table of type maps */
  const uschar *start_code;     /* The start of the compiled code */
  uschar *name_table;           /* The name/number table */
  int  names_found;             /* Number of entries so far */
  int  name_entry_size;         /* Size of each entry */
  int  top_backref;             /* Maximum back reference */
  unsigned int backref_map;     /* Bitmap of low back refs */
  int  req_varyopt;             /* "After variable item" flag for reqbyte */
} compile_data;

/* Structure for maintaining a chain of pointers to the currently incomplete
branches, for testing for left recursion. */

typedef struct branch_chain {
  struct branch_chain *outer;
  uschar *current;
} branch_chain;

/* Structure for items in a linked list that represents an explicit recursive
call within the pattern. */

typedef struct recursion_info {
  struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
  int group_num;                /* Number of group that was called */
  const uschar *after_call;     /* "Return value": points after the call in the expr */
  const uschar *save_start;     /* Old value of md->start_match */
  int *offset_save;             /* Pointer to start of saved offsets */
  int saved_max;                /* Number of saved offsets */
} recursion_info;

/* When compiling in a mode that doesn't use recursive calls to match(),
a structure is used to remember local variables on the heap. It is defined in
pcre.c, close to the match() function, so that it is easy to keep it in step
with any changes of local variable. However, the pointer to the current frame
must be saved in some "static" place over a longjmp(). We declare the
structure here so that we can put a pointer in the match_data structure.
NOTE: This isn't used for a "normal" compilation of pcre. */

/* Structure for passing "static" information around between the functions
doing the matching, so that they are thread-safe. */

typedef struct match_data {
  unsigned long int match_call_count; /* As it says */
  unsigned long int match_limit;/* As it says */
  int   *offset_vector;         /* Offset vector */
  int    offset_end;            /* One past the end */
  int    offset_max;            /* The maximum usable for return data */
  const uschar *lcc;            /* Points to lower casing table */
  const uschar *ctypes;         /* Points to table of type maps */
  bool   offset_overflow;       /* Set if too many extractions */
  bool   notbol;                /* NOTBOL flag */
  bool   noteol;                /* NOTEOL flag */
  bool   utf8;                  /* UTF8 flag */
  bool   endonly;               /* Dollar not before final \n */
  bool   notempty;              /* Empty string match not wanted */
  const uschar *start_code;     /* For use when recursing */
  const uschar *start_subject;  /* Start of the subject string */
  const uschar *end_subject;    /* End of the subject string */
  const uschar *start_match;    /* Start of this match attempt */
  const uschar *end_match_ptr;  /* Subject position at end match */
  int    end_offset_top;        /* Highwater mark at end of match */
  int    capture_last;          /* Most recent capture number */
  int    start_offset;          /* The start offset value */
  recursion_info *recursive;    /* Linked list of recursion data */
  void  *callout_data;          /* To pass back to callouts */
} match_data;

/* Bit definitions for entries in the pcre_ctypes table. */

#define ctype_space   0x01
#define ctype_letter  0x02
#define ctype_digit   0x04
#define ctype_xdigit  0x08
#define ctype_word    0x10   /* alphameric or '_' */
#define ctype_meta    0x80   /* regexp meta char or zero (end pattern) */

/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
of bits for a class map. Some classes are built by combining these tables. */

#define cbit_space     0      /* [:space:] or \s */
#define cbit_xdigit   32      /* [:xdigit:] */
#define cbit_digit    64      /* [:digit:] or \d */
#define cbit_upper    96      /* [:upper:] */
#define cbit_lower   128      /* [:lower:] */
#define cbit_word    160      /* [:word:] or \w */
#define cbit_graph   192      /* [:graph:] */
#define cbit_print   224      /* [:print:] */
#define cbit_punct   256      /* [:punct:] */
#define cbit_cntrl   288      /* [:cntrl:] */
#define cbit_length  320      /* Length of the cbits table */

/* Offsets of the various tables from the base tables pointer, and
total length. */

#define lcc_offset      0
#define fcc_offset    256
#define cbits_offset  512
#define ctypes_offset (cbits_offset + cbit_length)
#define tables_length (ctypes_offset + 256)

/* End of internal.h */
/* chartables.c */
/*************************************************
*      Perl-Compatible Regular Expressions       *
*************************************************/

/* This file is automatically written by the dftables auxiliary
program. If you edit it by hand, you might like to edit the Makefile to
prevent its ever being regenerated.

This file is #included in the compilation of pcre.c to build the default
character tables which are used when no tables are passed to the compile
function. */

static unsigned char pcre_default_tables[] = {

/* This table is a lower casing table. */

    0,  1,  2,  3,  4,  5,  6,  7,
    8,  9, 10, 11, 12, 13, 14, 15,
   16, 17, 18, 19, 20, 21, 22, 23,
   24, 25, 26, 27, 28, 29, 30, 31,
   32, 33, 34, 35, 36, 37, 38, 39,
   40, 41, 42, 43, 44, 45, 46, 47,
   48, 49, 50, 51, 52, 53, 54, 55,
   56, 57, 58, 59, 60, 61, 62, 63,
   64, 97, 98, 99,100,101,102,103,
  104,105,106,107,108,109,110,111,
  112,113,114,115,116,117,118,119,
  120,121,122, 91, 92, 93, 94, 95,
   96, 97, 98, 99,100,101,102,103,
  104,105,106,107,108,109,110,111,
  112,113,114,115,116,117,118,119,
  120,121,122,123,124,125,126,127,
  128,129,130,131,132,133,134,135,
  136,137,138,139,140,141,142,143,
  144,145,146,147,148,149,150,151,
  152,153,154,155,156,157,158,159,
  160,161,162,163,164,165,166,167,
  168,169,170,171,172,173,174,175,
  176,177,178,179,180,181,182,183,
  184,185,186,187,188,189,190,191,
  192,193,194,195,196,197,198,199,
  200,201,202,203,204,205,206,207,
  208,209,210,211,212,213,214,215,
  216,217,218,219,220,221,222,223,
  224,225,226,227,228,229,230,231,
  232,233,234,235,236,237,238,239,
  240,241,242,243,244,245,246,247,
  248,249,250,251,252,253,254,255,

/* This table is a case flipping table. */

    0,  1,  2,  3,  4,  5,  6,  7,
    8,  9, 10, 11, 12, 13, 14, 15,
   16, 17, 18, 19, 20, 21, 22, 23,
   24, 25, 26, 27, 28, 29, 30, 31,
   32, 33, 34, 35, 36, 37, 38, 39,
   40, 41, 42, 43, 44, 45, 46, 47,
   48, 49, 50, 51, 52, 53, 54, 55,
   56, 57, 58, 59, 60, 61, 62, 63,
   64, 97, 98, 99,100,101,102,103,
  104,105,106,107,108,109,110,111,
  112,113,114,115,116,117,118,119,
  120,121,122, 91, 92, 93, 94, 95,
   96, 65, 66, 67, 68, 69, 70, 71,
   72, 73, 74, 75, 76, 77, 78, 79,
   80, 81, 82, 83, 84, 85, 86, 87,
   88, 89, 90,123,124,125,126,127,
  128,129,130,131,132,133,134,135,
  136,137,138,139,140,141,142,143,
  144,145,146,147,148,149,150,151,
  152,153,154,155,156,157,158,159,
  160,161,162,163,164,165,166,167,
  168,169,170,171,172,173,174,175,
  176,177,178,179,180,181,182,183,
  184,185,186,187,188,189,190,191,
  192,193,194,195,196,197,198,199,
  200,201,202,203,204,205,206,207,
  208,209,210,211,212,213,214,215,
  216,217,218,219,220,221,222,223,
  224,225,226,227,228,229,230,231,
  232,233,234,235,236,237,238,239,
  240,241,242,243,244,245,246,247,
  248,249,250,251,252,253,254,255,

/* This table contains bit maps for various character classes.
Each map is 32 bytes long and the bits run from the least
significant end of each byte. The classes that have their own
maps are: space, xdigit, digit, upper, lower, word, graph
print, punct, and cntrl. Other classes are built from combinations. */

  0x00,0x3e,0x00,0x00,0x01,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
  0x7e,0x00,0x00,0x00,0x7e,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0xfe,0xff,0xff,0x07,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0x07,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
  0xfe,0xff,0xff,0x87,0xfe,0xff,0xff,0x07,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0xff,
  0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0xff,0xff,0xff,0xff,
  0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0x00,0x00,0x00,0x00,0xfe,0xff,0x00,0xfc,
  0x01,0x00,0x00,0xf8,0x01,0x00,0x00,0x78,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

  0xff,0xff,0xff,0xff,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x80,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,

/* This table identifies various classes of character by individual bits:
  0x01   white space character
  0x02   letter
  0x04   decimal digit
  0x08   hexadecimal digit
  0x10   alphanumeric or '_'
  0x80   regular expression metacharacter or binary zero
*/

  0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
  0x00,0x01,0x01,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  16- 23 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  24- 31 */
  0x01,0x00,0x00,0x00,0x80,0x00,0x00,0x00, /*    - '  */
  0x80,0x80,0x80,0x80,0x00,0x00,0x80,0x00, /*  ( - /  */
  0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /*  0 - 7  */
  0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x80, /*  8 - ?  */
  0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /*  @ - G  */
  0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /*  H - O  */
  0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /*  P - W  */
  0x12,0x12,0x12,0x80,0x00,0x00,0x80,0x10, /*  X - _  */
  0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /*  ` - g  */
  0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /*  h - o  */
  0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /*  p - w  */
  0x12,0x12,0x12,0x80,0x80,0x00,0x00,0x00, /*  x -127 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */

/* End of chartables.c */
/* get.c */
/*************************************************
*      Copy captured string to given buffer      *
*************************************************/

/* This function copies a single captured substring into a given buffer.
Note that we use memcpy() rather than strncpy() in case there are binary zeros
in the string.

Arguments:
  subject        the subject string that was matched
  ovector        pointer to the offsets table
  stringcount    the number of substrings that were captured
                   (i.e. the yield of the pcre_exec call, unless
                   that was zero, in which case it should be 1/3
                   of the offset table size)
  stringnumber   the number of the required substring
  buffer         where to put the substring
  size           the size of the buffer

Returns:         if successful:
                   the length of the copied string, not including the zero
                   that is put on the end; can be zero
                 if not successful:
                   PCRE_ERROR_NOMEMORY (-6) buffer too small
                   PCRE_ERROR_NOSUBSTRING (-7) no such captured substring
*/

int
pcre_copy_substring(const char *subject, int *ovector, int stringcount,
  int stringnumber, char *buffer, int size)
{
int yield;
if (stringnumber < 0 || stringnumber >= stringcount)
  return PCRE_ERROR_NOSUBSTRING;
stringnumber *= 2;
yield = ovector[stringnumber+1] - ovector[stringnumber];
if (size < yield + 1) return PCRE_ERROR_NOMEMORY;
memcpy(buffer, subject + ovector[stringnumber], yield);
buffer[yield] = 0;
return yield;
}

/* End of get.c */
/* maketables.c */
/*************************************************
*           Create PCRE character tables         *
*************************************************/

/* This function builds a set of character tables for use by PCRE and returns
a pointer to them. They are build using the ctype functions, and consequently
their contents will depend upon the current locale setting. When compiled as
part of the library, the store is obtained via malloc(), but when compiled
inside dftables, use malloc().

Arguments:   none
Returns:     pointer to the contiguous block of data
*/

const unsigned char *
pcre_maketables(void)
{
unsigned char *yield, *p;
int i;

yield = static_cast<unsigned char*>(malloc(tables_length));

if (yield == NULL) return NULL;
p = yield;

/* First comes the lower casing table */

for (i = 0; i < 256; i++) *p++ = tolower(i);

/* Next the case-flipping table */

for (i = 0; i < 256; i++) *p++ = islower(i)? toupper(i) : tolower(i);

/* Then the character class tables. Don't try to be clever and save effort
on exclusive ones - in some locales things may be different. Note that the
table for "space" includes everything "isspace" gives, including VT in the
default locale. This makes it work for the POSIX class [:space:]. */

memset(p, 0, cbit_length);
for (i = 0; i < 256; i++)
  {
  if (isdigit(i))
    {
    p[cbit_digit  + i/8] |= 1 << (i&7);
    p[cbit_word   + i/8] |= 1 << (i&7);
    }
  if (isupper(i))
    {
    p[cbit_upper  + i/8] |= 1 << (i&7);
    p[cbit_word   + i/8] |= 1 << (i&7);
    }
  if (islower(i))
    {
    p[cbit_lower  + i/8] |= 1 << (i&7);
    p[cbit_word   + i/8] |= 1 << (i&7);
    }
  if (i == '_')   p[cbit_word   + i/8] |= 1 << (i&7);
  if (isspace(i)) p[cbit_space  + i/8] |= 1 << (i&7);
  if (isxdigit(i))p[cbit_xdigit + i/8] |= 1 << (i&7);
  if (isgraph(i)) p[cbit_graph  + i/8] |= 1 << (i&7);
  if (isprint(i)) p[cbit_print  + i/8] |= 1 << (i&7);
  if (ispunct(i)) p[cbit_punct  + i/8] |= 1 << (i&7);
  if (iscntrl(i)) p[cbit_cntrl  + i/8] |= 1 << (i&7);
  }
p += cbit_length;

/* Finally, the character type table. In this, we exclude VT from the white
space chars, because Perl doesn't recognize it as such for \s and for comments
within regexes. */

for (i = 0; i < 256; i++)
  {
  int x = 0;
  if (i != 0x0b && isspace(i)) x += ctype_space;
  if (isalpha(i)) x += ctype_letter;
  if (isdigit(i)) x += ctype_digit;
  if (isxdigit(i)) x += ctype_xdigit;
  if (isalnum(i) || i == '_') x += ctype_word;
  if (strchr("*+?{^.$|()[", i) != 0) x += ctype_meta;
  *p++ = x;
  }

return yield;
}

/* End of maketables.c */
/* study.c */

/*************************************************
*      Set a bit and maybe its alternate case    *
*************************************************/

/* Given a character, set its bit in the table, and also the bit for the other
version of a letter if we are caseless.

Arguments:
  start_bits    points to the bit map
  c             is the character
  caseless      the caseless flag
  cd            the block with char table pointers

Returns:        nothing
*/

static void
set_bit(uschar *start_bits, int c, bool caseless, compile_data *cd)
{
start_bits[c/8] |= (1 << (c&7));
if (caseless && (cd->ctypes[c] & ctype_letter) != 0)
  start_bits[cd->fcc[c]/8] |= (1 << (cd->fcc[c]&7));
}



/*************************************************
*          Create bitmap of starting chars       *
*************************************************/

/* This function scans a compiled unanchored expression and attempts to build a
bitmap of the set of initial characters. If it can't, it returns false. As time
goes by, we may be able to get more clever at doing this.

Arguments:
  code         points to an expression
  start_bits   points to a 32-byte table, initialized to 0
  caseless     the current state of the caseless flag
  utf8         true if in UTF-8 mode
  cd           the block with char table pointers

Returns:       true if table built, false otherwise
*/

static bool
set_start_bits(const uschar *code, uschar *start_bits, bool caseless,
  bool utf8, compile_data *cd)
{
register int c;

/* This next statement and the later reference to dummy are here in order to
trick the optimizer of the IBM C compiler for OS/2 into generating correct
code. Apparently IBM isn't going to fix the problem, and we would rather not
disable optimization (in this module it actually makes a big difference, and
the pcre module can use all the optimization it can get). */

volatile int dummy;

do
  {
  const uschar *tcode = code + 1 + LINK_SIZE;
  bool try_next = true;

  while (try_next)
    {
    /* If a branch starts with a bracket or a positive lookahead assertion,
    recurse to set bits from within them. That's all for this branch. */

    if ((int)*tcode >= OP_BRA || *tcode == OP_ASSERT)
      {
      if (!set_start_bits(tcode, start_bits, caseless, utf8, cd))
        return false;
      try_next = false;
      }

    else switch(*tcode)
      {
      default:
      return false;

      /* Skip over callout */

      case OP_CALLOUT:
      tcode += 2;
      break;

      /* Skip over extended extraction bracket number */

      case OP_BRANUMBER:
      tcode += 3;
      break;

      /* Skip over lookbehind and negative l