/[pcre]/code/branches/pcre16/pcre_internal.h
ViewVC logotype

Diff of /code/branches/pcre16/pcre_internal.h

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

revision 755 by ph10, Mon Nov 21 10:41:54 2011 UTC revision 756 by ph10, Mon Nov 21 10:48:42 2011 UTC
# Line 158  set, we ensure here that it has no effec Line 158  set, we ensure here that it has no effec
158  #define PCRE_CALL_CONVENTION  #define PCRE_CALL_CONVENTION
159  #endif  #endif
160    
161  /* We need to have types that specify unsigned 16-bit and 32-bit integers. We  /* We need to have types that specify unsigned 8, 16 and 32-bit integers. We
162  cannot determine these outside the compilation (e.g. by running a program as  cannot determine these outside the compilation (e.g. by running a program as
163  part of "configure") because PCRE is often cross-compiled for use on other  part of "configure") because PCRE is often cross-compiled for use on other
164  systems. Instead we make use of the maximum sizes that are available at  systems. Instead we make use of the maximum sizes that are available at
165  preprocessor time in standard C environments. */  preprocessor time in standard C environments. */
166    
167    typedef unsigned char pcre_uint8;
168    
169  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
170    typedef unsigned short pcre_uint16;    typedef unsigned short pcre_uint16;
171    typedef short pcre_int16;    typedef short pcre_int16;
# Line 207  by "configure". */ Line 209  by "configure". */
209  /* All character handling must be done as unsigned characters. Otherwise there  /* All character handling must be done as unsigned characters. Otherwise there
210  are problems with top-bit-set characters and functions such as isspace().  are problems with top-bit-set characters and functions such as isspace().
211  However, we leave the interface to the outside world as char *, because that  However, we leave the interface to the outside world as char *, because that
212  should make things easier for callers. We define a short type for unsigned char  should make things easier for callers. */
 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. */  
213    
214  typedef unsigned char uschar;  typedef unsigned char pcre_uchar;
215    
216  /* This is an unsigned int value that no character can ever have. UTF-8  /* This is an unsigned int value that no character can ever have. UTF-8
217  characters only go up to 0x7fffffff (though Unicode doesn't go beyond  characters only go up to 0x7fffffff (though Unicode doesn't go beyond
# Line 267  used for the external interface and appe Line 267  used for the external interface and appe
267  must begin with PCRE_. */  must begin with PCRE_. */
268    
269  #ifdef CUSTOM_SUBJECT_PTR  #ifdef CUSTOM_SUBJECT_PTR
270  #define PCRE_SPTR CUSTOM_SUBJECT_PTR  #define PCRE_PUCHAR CUSTOM_SUBJECT_PTR
 #define USPTR CUSTOM_SUBJECT_PTR  
271  #else  #else
272  #define PCRE_SPTR const char *  #define PCRE_PUCHAR const pcre_uchar *
 #define USPTR const unsigned char *  
273  #endif  #endif
274    
275    
# Line 1699  typedef struct real_pcre { Line 1697  typedef struct real_pcre {
1697    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
1698    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
1699    
1700    const unsigned char *tables;    /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
1701    const unsigned char *nullpad;   /* NULL padding */    const pcre_uint8 *nullpad;      /* NULL padding */
1702  } real_pcre;  } real_pcre;
1703    
1704  /* The format of the block used to store data from pcre_study(). The same  /* The format of the block used to store data from pcre_study(). The same
# Line 1709  remark (see NOTE above) about extending Line 1707  remark (see NOTE above) about extending
1707  typedef struct pcre_study_data {  typedef struct pcre_study_data {
1708    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
1709    pcre_uint32 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
1710    uschar start_bits[32];          /* Starting char bits */    pcre_uint8 start_bits[32];      /* Starting char bits */
1711    pcre_uint32 minlength;          /* Minimum subject length */    pcre_uint32 minlength;          /* Minimum subject length */
1712  } pcre_study_data;  } pcre_study_data;
1713    
# Line 1728  typedef struct open_capitem { Line 1726  typedef struct open_capitem {
1726  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
1727    
1728  typedef struct compile_data {  typedef struct compile_data {
1729    const uschar *lcc;            /* Points to lower casing table */    const pcre_uint8 *lcc;            /* Points to lower casing table */
1730    const uschar *fcc;            /* Points to case-flipping table */    const pcre_uint8 *fcc;            /* Points to case-flipping table */
1731    const uschar *cbits;          /* Points to character type table */    const pcre_uint8 *cbits;          /* Points to character type table */
1732    const uschar *ctypes;         /* Points to table of type maps */    const pcre_uint8 *ctypes;         /* Points to table of type maps */
1733    const uschar *start_workspace;/* The start of working space */    const pcre_uchar *start_workspace;/* The start of working space */
1734    const uschar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
1735    const uschar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
1736    const uschar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
1737    open_capitem *open_caps;      /* Chain of open capture items */    open_capitem *open_caps;          /* Chain of open capture items */
1738    uschar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
1739    uschar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
1740    int  names_found;             /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
1741    int  name_entry_size;         /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
1742    int  bracount;                /* Count of capturing parens as we compile */    int  bracount;                    /* Count of capturing parens as we compile */
1743    int  final_bracount;          /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
1744    int  top_backref;             /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
1745    unsigned int backref_map;     /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
1746    int  assert_depth;            /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
1747    int  external_options;        /* External (initial) options */    int  external_options;            /* External (initial) options */
1748    int  external_flags;          /* External flag bits to be set */    int  external_flags;              /* External flag bits to be set */
1749    int  req_varyopt;             /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
1750    BOOL had_accept;              /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
1751    BOOL check_lookbehind;        /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
1752    int  nltype;                  /* Newline type */    int  nltype;                      /* Newline type */
1753    int  nllen;                   /* Newline string length */    int  nllen;                       /* Newline string length */
1754    uschar nl[4];                 /* Newline string when fixed length */    pcre_uchar nl[4];                 /* Newline string when fixed length */
1755  } compile_data;  } compile_data;
1756    
1757  /* Structure for maintaining a chain of pointers to the currently incomplete  /* Structure for maintaining a chain of pointers to the currently incomplete
# Line 1761  branches, for testing for left recursion Line 1759  branches, for testing for left recursion
1759    
1760  typedef struct branch_chain {  typedef struct branch_chain {
1761    struct branch_chain *outer;    struct branch_chain *outer;
1762    uschar *current_branch;    pcre_uchar *current_branch;
1763  } branch_chain;  } branch_chain;
1764    
1765  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
# Line 1772  typedef struct recursion_info { Line 1770  typedef struct recursion_info {
1770    int group_num;                  /* Number of group that was called */    int group_num;                  /* Number of group that was called */
1771    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
1772    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
1773    USPTR subject_position;         /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
1774  } recursion_info;  } recursion_info;
1775    
1776  /* A similar structure for pcre_dfa_exec(). */  /* A similar structure for pcre_dfa_exec(). */
# Line 1780  typedef struct recursion_info { Line 1778  typedef struct recursion_info {
1778  typedef struct dfa_recursion_info {  typedef struct dfa_recursion_info {
1779    struct dfa_recursion_info *prevrec;    struct dfa_recursion_info *prevrec;
1780    int group_num;    int group_num;
1781    USPTR subject_position;    PCRE_PUCHAR subject_position;
1782  } dfa_recursion_info;  } dfa_recursion_info;
1783    
1784  /* Structure for building a chain of data for holding the values of the subject  /* Structure for building a chain of data for holding the values of the subject
# Line 1790  pcre_exec(). */ Line 1788  pcre_exec(). */
1788    
1789  typedef struct eptrblock {  typedef struct eptrblock {
1790    struct eptrblock *epb_prev;    struct eptrblock *epb_prev;
1791    USPTR epb_saved_eptr;    PCRE_PUCHAR epb_saved_eptr;
1792  } eptrblock;  } eptrblock;
1793    
1794    
# Line 1801  typedef struct match_data { Line 1799  typedef struct match_data {
1799    unsigned long int match_call_count;      /* As it says */    unsigned long int match_call_count;      /* As it says */
1800    unsigned long int match_limit;           /* As it says */    unsigned long int match_limit;           /* As it says */
1801    unsigned long int match_limit_recursion; /* As it says */    unsigned long int match_limit_recursion; /* As it says */
1802    int   *offset_vector;         /* Offset vector */    int   *offset_vector;           /* Offset vector */
1803    int    offset_end;            /* One past the end */    int    offset_end;              /* One past the end */
1804    int    offset_max;            /* The maximum usable for return data */    int    offset_max;              /* The maximum usable for return data */
1805    int    nltype;                /* Newline type */    int    nltype;                  /* Newline type */
1806    int    nllen;                 /* Newline string length */    int    nllen;                   /* Newline string length */
1807    int    name_count;            /* Number of names in name table */    int    name_count;              /* Number of names in name table */
1808    int    name_entry_size;       /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
1809    uschar *name_table;           /* Table of names */    pcre_uchar *name_table;         /* Table of names */
1810    uschar nl[4];                 /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
1811    const  uschar *lcc;           /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
1812    const  uschar *ctypes;        /* Points to table of type maps */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
1813    BOOL   offset_overflow;       /* Set if too many extractions */    BOOL   offset_overflow;         /* Set if too many extractions */
1814    BOOL   notbol;                /* NOTBOL flag */    BOOL   notbol;                  /* NOTBOL flag */
1815    BOOL   noteol;                /* NOTEOL flag */    BOOL   noteol;                  /* NOTEOL flag */
1816    BOOL   utf8;                  /* UTF8 flag */    BOOL   utf8;                    /* UTF8 flag */
1817    BOOL   jscript_compat;        /* JAVASCRIPT_COMPAT flag */    BOOL   jscript_compat;          /* JAVASCRIPT_COMPAT flag */
1818    BOOL   use_ucp;               /* PCRE_UCP flag */    BOOL   use_ucp;                 /* PCRE_UCP flag */
1819    BOOL   endonly;               /* Dollar not before final \n */    BOOL   endonly;                 /* Dollar not before final \n */
1820    BOOL   notempty;              /* Empty string match not wanted */    BOOL   notempty;                /* Empty string match not wanted */
1821    BOOL   notempty_atstart;      /* Empty string match at start not wanted */    BOOL   notempty_atstart;        /* Empty string match at start not wanted */
1822    BOOL   hitend;                /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
1823    BOOL   bsr_anycrlf;           /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
1824    BOOL   hasthen;               /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
1825    const  uschar *start_code;    /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
1826    USPTR  start_subject;         /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
1827    USPTR  end_subject;           /* End of the subject string */    PCRE_PUCHAR end_subject;        /* End of the subject string */
1828    USPTR  start_match_ptr;       /* Start of matched string */    PCRE_PUCHAR start_match_ptr;    /* Start of matched string */
1829    USPTR  end_match_ptr;         /* Subject position at end match */    PCRE_PUCHAR end_match_ptr;      /* Subject position at end match */
1830    USPTR  start_used_ptr;        /* Earliest consulted character */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
1831    int    partial;               /* PARTIAL options */    int    partial;                 /* PARTIAL options */
1832    int    end_offset_top;        /* Highwater mark at end of match */    int    end_offset_top;          /* Highwater mark at end of match */
1833    int    capture_last;          /* Most recent capture number */    int    capture_last;            /* Most recent capture number */
1834    int    start_offset;          /* The start offset value */    int    start_offset;            /* The start offset value */
1835    int    match_function_type;   /* Set for certain special calls of MATCH() */    int    match_function_type;     /* Set for certain special calls of MATCH() */
1836    eptrblock *eptrchain;         /* Chain of eptrblocks for tail recursions */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
1837    int    eptrn;                 /* Next free eptrblock */    int    eptrn;                   /* Next free eptrblock */
1838    recursion_info *recursive;    /* Linked list of recursion data */    recursion_info *recursive;      /* Linked list of recursion data */
1839    void  *callout_data;          /* To pass back to callouts */    void  *callout_data;            /* To pass back to callouts */
1840    const  uschar *mark;          /* Mark pointer to pass back */    const  pcre_uchar *mark;        /* Mark pointer to pass back */
1841    const  uschar *once_target;   /* Where to back up to for atomic groups */    const  pcre_uchar *once_target; /* Where to back up to for atomic groups */
1842  } match_data;  } match_data;
1843    
1844  /* A similar structure is used for the same purpose by the DFA matching  /* A similar structure is used for the same purpose by the DFA matching
1845  functions. */  functions. */
1846    
1847  typedef struct dfa_match_data {  typedef struct dfa_match_data {
1848    const uschar *start_code;      /* Start of the compiled pattern */    const pcre_uchar *start_code;     /* Start of the compiled pattern */
1849    const uschar *start_subject;   /* Start of the subject string */    const pcre_uchar *start_subject ; /* Start of the subject string */
1850    const uschar *end_subject;     /* End of subject string */    const pcre_uchar *end_subject;    /* End of subject string */
1851    const uschar *start_used_ptr;  /* Earliest consulted character */    const pcre_uchar *start_used_ptr; /* Earliest consulted character */
1852    const uschar *tables;          /* Character tables */    const pcre_uint8 *tables;         /* Character tables */
1853    int   start_offset;            /* The start offset value */    int   start_offset;               /* The start offset value */
1854    int   moptions;                /* Match options */    int   moptions;                   /* Match options */
1855    int   poptions;                /* Pattern options */    int   poptions;                   /* Pattern options */
1856    int    nltype;                 /* Newline type */    int   nltype;                     /* Newline type */
1857    int    nllen;                  /* Newline string length */    int   nllen;                      /* Newline string length */
1858    uschar nl[4];                  /* Newline string when fixed */    pcre_uchar nl[4];                 /* Newline string when fixed */
1859    void  *callout_data;           /* To pass back to callouts */    void *callout_data;               /* To pass back to callouts */
1860    dfa_recursion_info *recursive; /* Linked list of recursion data */    dfa_recursion_info *recursive;    /* Linked list of recursion data */
1861  } dfa_match_data;  } dfa_match_data;
1862    
1863  /* Bit definitions for entries in the pcre_ctypes table. */  /* Bit definitions for entries in the pcre_ctypes table. */
# Line 1912  of the exported public functions. They h Line 1910  of the exported public functions. They h
1910  but are not part of the PCRE public API. The data for these tables is in the  but are not part of the PCRE public API. The data for these tables is in the
1911  pcre_tables.c module. */  pcre_tables.c module. */
1912    
1913  extern const int    _pcre_utf8_table1[];  extern const int            _pcre_utf8_table1[];
1914  extern const int    _pcre_utf8_table2[];  extern const int            _pcre_utf8_table2[];
1915  extern const int    _pcre_utf8_table3[];  extern const int            _pcre_utf8_table3[];
1916  extern const uschar _pcre_utf8_table4[];  extern const pcre_uint8     _pcre_utf8_table4[];
   
 #ifdef SUPPORT_JIT  
 extern const uschar _pcre_utf8_char_sizes[];  
 #endif  
1917    
1918  extern const int    _pcre_utf8_table1_size;  extern const int            _pcre_utf8_table1_size;
1919    
1920  extern const char   _pcre_utt_names[];  extern const char           _pcre_utt_names[];
1921  extern const ucp_type_table _pcre_utt[];  extern const ucp_type_table _pcre_utt[];
1922  extern const int _pcre_utt_size;  extern const int            _pcre_utt_size;
1923    
1924  extern const uschar _pcre_default_tables[];  extern const pcre_uint8     _pcre_default_tables[];
1925    
1926  extern const uschar _pcre_OP_lengths[];  extern const pcre_uint8     _pcre_OP_lengths[];
1927    
1928    
1929  /* Internal shared functions. These are functions that are used by more than  /* Internal shared functions. These are functions that are used by more than
1930  one of the exported public functions. They have to be "external" in the C  one of the exported public functions. They have to be "external" in the C
1931  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
1932    
1933  extern const uschar *_pcre_find_bracket(const uschar *, BOOL, int);  extern const pcre_uchar *_pcre_find_bracket(const pcre_uchar *, BOOL, int);
1934  extern BOOL          _pcre_is_newline(USPTR, int, USPTR, int *, BOOL);  extern BOOL              _pcre_is_newline(PCRE_PUCHAR, int, PCRE_PUCHAR,
1935  extern int           _pcre_ord2utf8(int, uschar *);                             int *, BOOL);
1936  extern real_pcre    *_pcre_try_flipped(const real_pcre *, real_pcre *,  extern int               _pcre_ord2utf8(int, pcre_uint8 *);
1937                         const pcre_study_data *, pcre_study_data *);  extern real_pcre        *_pcre_try_flipped(const real_pcre *, real_pcre *,
1938  extern int           _pcre_valid_utf8(USPTR, int, int *);                             const pcre_study_data *, pcre_study_data *);
1939  extern BOOL          _pcre_was_newline(USPTR, int, USPTR, int *, BOOL);  extern int               _pcre_valid_utf8(PCRE_PUCHAR, int, int *);
1940  extern BOOL          _pcre_xclass(int, const uschar *);  extern BOOL              _pcre_was_newline(PCRE_PUCHAR, int, PCRE_PUCHAR,
1941                               int *, BOOL);
1942    extern BOOL              _pcre_xclass(int, const pcre_uchar *);
1943    
1944  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
1945  extern void          _pcre_jit_compile(const real_pcre *, pcre_extra *);  extern void              _pcre_jit_compile(const real_pcre *, pcre_extra *);
1946  extern int           _pcre_jit_exec(const real_pcre *, void *, PCRE_SPTR,  extern int               _pcre_jit_exec(const real_pcre *, void *, PCRE_SPTR,
1947                          int, int, int, int, int *, int);                             int, int, int, int, int *, int);
1948  extern void          _pcre_jit_free(void *);  extern void              _pcre_jit_free(void *);
1949  #endif  #endif
1950    
1951  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
1952    
1953  typedef struct {  typedef struct {
1954    uschar script;    pcre_uint8 script;
1955    uschar chartype;    pcre_uint8 chartype;
1956    pcre_int32 other_case;    pcre_int32 other_case;
1957  } ucd_record;  } ucd_record;
1958    
1959  extern const ucd_record  _pcre_ucd_records[];  extern const ucd_record  _pcre_ucd_records[];
1960  extern const uschar      _pcre_ucd_stage1[];  extern const pcre_uint8  _pcre_ucd_stage1[];
1961  extern const pcre_uint16 _pcre_ucd_stage2[];  extern const pcre_uint16 _pcre_ucd_stage2[];
1962  extern const int         _pcre_ucp_gentype[];  extern const int         _pcre_ucp_gentype[];
1963  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
Legend:
Removed from v.755  
changed lines
  Added in v.756
  ViewVC Help
Powered by ViewVC 1.1.5