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 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 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
|
/* parser.h
*
* Copyright (c) 2006, 2007, 2009, 2010, 2011 Larry Wall and others
*
* You may distribute under the terms of either the GNU General Public
* License or the Artistic License, as specified in the README file.
*
* This file defines the layout of the parser object used by the parser
* and lexer (perly.c, toke.c).
*/
#define YYEMPTY (-2)
typedef struct {
YYSTYPE val; /* semantic value */
short state;
I32 savestack_ix; /* size of savestack at this state */
CV *compcv; /* value of PL_compcv when this value was created */
#ifdef DEBUGGING
const char *name; /* token/rule name for -Dpv */
#endif
} yy_stack_frame;
/* Fields that need to be shared with (i.e., visible to) inner lex-
ing scopes. */
typedef struct yy_lexshared {
struct yy_lexshared *ls_prev;
SV *ls_linestr; /* mirrors PL_parser->linestr */
char *ls_bufptr; /* mirrors PL_parser->bufptr */
char *re_eval_start; /* start of "(?{..." text */
SV *re_eval_str; /* "(?{...})" text */
} LEXSHARED;
typedef struct yy_parser {
/* parser state */
struct yy_parser *old_parser; /* previous value of PL_parser */
YYSTYPE yylval; /* value of lookahead symbol, set by yylex() */
int yychar; /* The lookahead symbol. */
/* Number of tokens to shift before error messages enabled. */
int yyerrstatus;
int yylen; /* length of active reduction */
yy_stack_frame *stack; /* base of stack */
yy_stack_frame *stack_max1;/* (top-1)th element of allocated stack */
yy_stack_frame *ps; /* current stack frame */
/* lexer state */
I32 lex_brackets; /* square and curly bracket count */
I32 lex_casemods; /* casemod count */
char *lex_brackstack;/* what kind of brackets to pop */
char *lex_casestack; /* what kind of case mods in effect */
U8 lex_defer; /* state after determined token */
U8 lex_dojoin; /* doing an array interpolation
1 = @{...} 2 = ->@ */
U8 expect; /* how to interpret ambiguous tokens */
bool preambled;
bool sub_no_recover; /* can't recover from a sublex error */
I32 lex_formbrack; /* bracket count at outer format level */
OP *lex_inpat; /* in pattern $) and $| are special */
OP *lex_op; /* extra info to pass back on op */
SV *lex_repl; /* runtime replacement from s/// */
U16 lex_inwhat; /* what kind of quoting are we in */
OPCODE last_lop_op; /* last named list or unary operator */
I32 lex_starts; /* how many interps done on level */
SV *lex_stuff; /* runtime pattern from m// or s/// */
I32 multi_start; /* 1st line of multi-line string */
I32 multi_end; /* last line of multi-line string */
UV multi_open; /* delimiter of said string */
UV multi_close; /* delimiter of said string */
bool lex_re_reparsing; /* we're doing G_RE_REPARSING */
U8 lex_super_state;/* lexer state to save */
U16 lex_sub_inwhat; /* "lex_inwhat" to use in sublex_push */
I32 lex_allbrackets;/* (), [], {}, ?: bracket count */
OP *lex_sub_op; /* current op in y/// or pattern */
SV *lex_sub_repl; /* repl of s/// used in sublex_push */
LEXSHARED *lex_shared;
SV *linestr; /* current chunk of src text */
char *bufptr; /* carries the cursor (current parsing
position) from one invocation of yylex
to the next */
char *oldbufptr; /* in yylex, beginning of current token */
char *oldoldbufptr; /* in yylex, beginning of previous token */
char *bufend;
char *linestart; /* beginning of most recently read line */
char *last_uni; /* position of last named-unary op */
char *last_lop; /* position of last list operator */
/* copline is used to pass a specific line number to newSTATEOP. It
is a one-time line number, as newSTATEOP invalidates it (sets it to
NOLINE) after using it. The purpose of this is to report line num-
bers in multiline constructs using the number of the first line. */
line_t copline;
U16 in_my; /* we're compiling a "my"/"our" declaration */
U8 lex_state; /* next token is determined */
U8 error_count; /* how many compile errors so far, max 10 */
U8 sub_error_count; /* the number of errors before sublexing */
HV *in_my_stash; /* declared class of this "my" declaration */
PerlIO *rsfp; /* current source file pointer */
AV *rsfp_filters; /* holds chain of active source filters */
YYSTYPE nextval[5]; /* value of next token, if any */
I32 nexttype[5]; /* type of next token */
U8 nexttoke;
U8 form_lex_state; /* remember lex_state when parsing fmt */
U8 lex_fakeeof; /* precedence at which to fake EOF */
U8 lex_flags;
COP *saved_curcop; /* the previous PL_curcop */
char tokenbuf[256];
line_t herelines; /* number of lines in here-doc */
line_t preambling; /* line # when processing $ENV{PERL5DB} */
bool sig_seen; /* the currently parsing sub has a signature */
/* these are valid while parsing a subroutine signature */
IV sig_elems; /* number of signature elements seen so far */
IV sig_optelems; /* number of optional signature elems seen */
char sig_slurpy; /* the sigil of the slurpy var (or null) */
bool recheck_utf8_validity;
PERL_BITFIELD16 in_pod:1; /* lexer is within a =pod section */
PERL_BITFIELD16 filtered:1; /* source filters in evalbytes */
PERL_BITFIELD16 saw_infix_sigil:1; /* saw & or * or % operator */
PERL_BITFIELD16 parsed_sub:1; /* last thing parsed was a sub */
} yy_parser;
/* flags for lexer API */
#define LEX_STUFF_UTF8 0x00000001
#define LEX_KEEP_PREVIOUS 0x00000002
#ifdef PERL_CORE
# define LEX_START_SAME_FILTER 0x00000001
# define LEX_IGNORE_UTF8_HINTS 0x00000002
# define LEX_EVALBYTES 0x00000004
# define LEX_START_COPIED 0x00000008
# define LEX_DONT_CLOSE_RSFP 0x00000010
# define LEX_START_FLAGS \
(LEX_START_SAME_FILTER|LEX_START_COPIED \
|LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES|LEX_DONT_CLOSE_RSFP)
#endif
/* flags for parser API */
#define PARSE_OPTIONAL 0x00000001
/* values for lex_fakeeof */
enum {
LEX_FAKEEOF_NEVER, /* don't fake EOF */
LEX_FAKEEOF_CLOSING, /* fake EOF at unmatched closing punctuation */
LEX_FAKEEOF_NONEXPR, /* ... and at token that can't be in expression */
LEX_FAKEEOF_LOWLOGIC, /* ... and at low-precedence logic operator */
LEX_FAKEEOF_COMMA, /* ... and at comma */
LEX_FAKEEOF_ASSIGN, /* ... and at assignment operator */
LEX_FAKEEOF_IFELSE, /* ... and at ?: operator */
LEX_FAKEEOF_RANGE, /* ... and at range operator */
LEX_FAKEEOF_LOGIC, /* ... and at logic operator */
LEX_FAKEEOF_BITWISE, /* ... and at bitwise operator */
LEX_FAKEEOF_COMPARE, /* ... and at comparison operator */
LEX_FAKEEOF_MAX
};
/*
* ex: set ts=8 sts=4 sw=4 et:
*/
|