2 /* flexdef - definitions file for flex */
4 /*  Copyright (c) 1990 The Regents of the University of California. */
10 /*  The United States Government has rights in this work pursuant */
11 /*  to contract no. DE-AC03-76SF00098 between the United States */
90 /* Always be prepared to generate an 8-bit scanner. */
93 /* Size of input alphabet - should be size of ASCII set. */
108 #define ABS(x) ((x) < 0 ? -(x) : (x))
112 #define is_power_of_2(n) ((n) > 0 && ((n) & ((n) - 1)) == 0)
114 #define unspecified -1
116 /* Special chk[] values marking the slots taking by end-of-buffer and action
119 #define EOB_POSITION -1
120 #define ACTION_POSITION -2
122 /* Number of data items per line for -f output. */
125 /* Number of lines of data in -f output before inserting a blank line for
136 /* Returns true if an nfa state has an epsilon out-transition slot
144 /* Returns true if an nfa state has an epsilon out-transition character
151 /* Maximum number of NFA states that can comprise a DFA state.  It's real
176 #define MAX_RULE (YY_TRAILING_MASK - 1)
180  * (it marks the representative of a given e.c.) will be unidentifiable.
184 #define JAM -1			/* to mark a missing DFA transition */
186 #define UNIQUE -1		/* marks a symbol as an e.c. representative */
187 #define INFINITE_REPEAT -1		/* for x{5,} constructions */
199 #define INITIAL_MNS 2000	/* default maximum number of nfa states */
202 #define INITIAL_MAX_DFAS 1000	/* default maximum number of dfa states */
205 #define JAMSTATE -32766		/* marks a reference to the state that always jams */
207 /* Maximum number of NFA states. */
216 /* Maximum number of nxt/chk pairs for non-templates. */
229 #define ONE_STACK_SIZE 500	/* stack of states with only one out-transition */
230 #define SAME_TRANS -1		/* transition is the same as "default" entry for state */
234  * The percentage the number of out-transitions a state must be of the
240 /* The percentage the number of homogeneous out-transitions of a state
241  * must be of the number of total out-transitions of the state in order
243  * template of the most common out-transition instead of with the first
250  * number of out-transitions of the state in order to keep the first
257  * of out-transitions to use the proto as an acceptable close match.
261 /* The percentage the number of homogeneous out-transitions of a state
262  * must be of the number of total out-transitions of the state in order
269  * of out-transitions to create a new proto from the state.
273 /* The percentage the total number of out-transitions of a state must be
286 /* Maximum number of out-transitions a state can have that we'll rummage
300 #define BAD_SUBSCRIPT -32767
312  * printstats - if true (-v), dump statistics
313  * syntaxerror - true if a syntax error has been found
314  * eofseen - true if we've seen an eof in the input file
315  * ddebug - if true (-d), make a "debug" scanner
316  * trace - if true (-T), trace processing
317  * nowarn - if true (-w), do not generate warnings
318  * spprdflt - if true (-s), suppress the default rule
319  * interactive - if true (-I), generate an interactive scanner
320  * lex_compat - if true (-l), maximize compatibility with AT&T lex
321  * posix_compat - if true (-X), maximize compatibility with POSIX lex
322  * do_yylineno - if true, generate code to maintain yylineno
323  * useecs - if true (-Ce flag), use equivalence classes
324  * fulltbl - if true (-Cf flag), don't compress the DFA state table
325  * usemecs - if true (-Cm flag), use meta-equivalence classes
326  * fullspd - if true (-F flag), use Jacobson method of table representation
327  * gen_line_dirs - if true (i.e., no -L flag), generate #line directives
328  * performance_report - if > 0 (i.e., -p flag), generate a report relating
329  *   to scanner performance; if > 1 (-p -p), report on minor performance
331  * backing_up_report - if true (i.e., -b flag), generate "lex.backup" file
332  *   listing backing-up states
333  * C_plus_plus - if true (i.e., -+ flag), generate a C++ scanner class;
334  *   otherwise, a standard C scanner
335  * reentrant - if true (-R), generate a reentrant C scanner.
336  * bison_bridge_lval - if true (--bison-bridge), bison pure calling convention.
337  * bison_bridge_lloc - if true (--bison-locations), bison yylloc.
338  * long_align - if true (-Ca flag), favor long-word alignment.
339  * use_read - if true (-f, -F, or -Cr) then use read() for scanner input;
341  * yytext_is_array - if true (i.e., %array directive), then declare
343  * do_yywrap - do yywrap() processing on EOF.  If false, EOF treated as
345  * csize - size of character set for the scanner we're generating;
346  *   128 for 7-bit chars and 256 for 8-bit
347  * yymore_used - if true, yymore() is used in input rules
348  * reject - if true, generate back-up tables for REJECT macro
349  * real_reject - if true, scanner really uses REJECT (as opposed to just
351  * continued_action - true if this rule's action is to "fall through" to
353  * in_rule - true if we're inside an individual rule, false if not.
354  * yymore_really_used - whether to treat yymore() as really used, regardless
356  * reject_really_used - same for REJECT
357  * trace_hex - use hexadecimal numbers in trace/debug outputs instead of octals
374  * datapos - characters on current output line
375  * dataline - number of contiguous lines of data in current data
376  * 	statement.  Used to generate readable -f output
377  * linenum - current input line number
378  * skelfile - the skeleton file
379  * skel - compiled-in skeleton array
380  * skel_ind - index into "skel" array, if skelfile is nil
381  * yyin - input file
382  * backing_up_file - file to summarize backing-up states to
383  * infilename - name of input file
384  * outfilename - name of output file
385  * headerfilename - name of the .h file to generate
386  * did_outfilename - whether outfilename was explicitly set
387  * prefix - the prefix used for externally visible names ("yy" by default)
388  * yyclass - yyFlexLexer subclass to use for YY_DECL
389  * do_stdinit - whether to initialize yyin/yyout to stdin/stdout
390  * use_stdout - the -t flag
391  * input_files - array holding names of input files
392  * num_input_files - size of input_files array
393  * program_name - name with which program was invoked
395  * action_array - array to hold the rule actions
396  * action_size - size of action_array
397  * defs1_offset - index where the user's section 1 definitions start
399  * prolog_offset - index where the prolog starts in action_array
400  * action_offset - index where the non-prolog starts in action_array
401  * action_index - index where the next action should go, with respect
422 /* Variables for stack of states having only one out-transition:
423  * onestate - state number
424  * onesym - transition symbol
425  * onenext - target state
426  * onedef - default base entry
427  * onesp - stack pointer
435  * maximum_mns - maximal number of NFA states supported by tables
436  * current_mns - current maximum on number of NFA states
437  * num_rules - number of the last accepting state; also is number of
439  * num_eof_rules - number of <<EOF>> rules
440  * default_rule - number of the default rule
441  * current_max_rules - current maximum number of rules
442  * lastnfa - last nfa state number created
443  * firstst - physically the first state of a fragment
444  * lastst - last physical state of fragment
445  * finalst - last logical state of fragment
446  * transchar - transition character
447  * trans1 - transition state
448  * trans2 - 2nd transition state for epsilons
449  * accptnum - accepting number
450  * assoc_rule - rule associated with this NFA state (or 0 if none)
451  * state_type - a STATE_xxx type identifying whether the state is part
454  * 	recognizing the text-to-be-matched to the beginning of
457  * rule_type - a RULE_xxx type identifying whether this a ho-hum
460  * rule_linenum - line number associated with rule
461  * rule_useful - true if we've determined that the rule can be matched
462  * rule_has_nl - true if rule could possibly match a newline
463  * ccl_has_nl - true if current ccl could match a newline
464  * nlch - default eol char
475 /* Different types of states; values are useful as masks, as well, for
489 /* True if the input rules include a rule with both variable-length head
496  * numtemps - number of templates created
497  * numprots - number of protos created
498  * protprev - backlink to a more-recently used proto
499  * protnext - forward link to a less-recently used proto
500  * prottbl - base/def table entry for proto
501  * protcomst - common state of proto
502  * firstprot - number of the most recently used proto
503  * lastprot - number of the least recently used proto
512  * numecs - number of equivalence classes
513  * nextecm - forward link of Equivalence Class members
514  * ecgroup - class number or backward link of EC members
515  * nummecs - number of meta-equivalence classes (used to compress
517  * tecfwd - forward link of meta-equivalence classes members
518  * tecbck - backward link of MEC's
528 /* Meta-equivalence classes are indexed starting at 1, so it's possible
530  * slots total (since the arrays are 0-based).  nextecm[] and ecgroup[]
531  * don't require the extra position since they're indexed from 1 .. CSIZE - 1.
537  * lastsc - last start condition created
538  * current_max_scs - current limit on number of start conditions
539  * scset - set of rules active in start condition
540  * scbol - set of rules active only at the beginning of line in a s.c.
541  * scxclu - true if start condition is exclusive
542  * sceof - true if start condition has EOF rule
543  * scname - start condition name
552  * current_max_dfa_size - current maximum number of NFA states in DFA
553  * current_max_xpairs - current maximum number of non-template xtion pairs
554  * current_max_template_xpairs - current maximum number of template pairs
555  * current_max_dfas - current maximum number DFA states
556  * lastdfa - last dfa state number created
557  * nxt - state to enter upon reading character
558  * chk - check value to see if "nxt" applies
559  * tnxt - internal nxt table for templates
560  * base - offset into "nxt" for given state
561  * def - where to go if "chk" disallows "nxt" entry
562  * nultrans - NUL transition for each state
563  * NUL_ec - equivalence class of the NUL character
564  * tblend - last "nxt/chk" table entry being used
565  * firstfree - first empty entry in "nxt/chk" table
566  * dss - nfa state set for each dfa
567  * dfasiz - size of nfa state set for each dfa
568  * dfaacc - accepting set for each dfa state (if using REJECT), or accepting
570  * accsiz - size of accepting set for each dfa state
571  * dhash - dfa state hash value
572  * numas - number of DFA accepting states created; note that this
575  * numsnpairs - number of state/nextstate transition pairs
576  * jambase - position in base/def where the default jam table starts
577  * jamstate - state number corresponding to "jam" state
578  * end_of_buffer_state - end-of-buffer dfa state number
595  * lastccl - ccl index of the last created ccl
596  * current_maxccls - current limit on the maximum number of unique ccl's
597  * cclmap - maps a ccl index to its set pointer
598  * ccllen - gives the length of a ccl
599  * cclng - true for a given ccl if the ccl is negated
600  * cclreuse - counts how many times a ccl is re-used
601  * current_max_ccl_tbl_size - current limit on number of characters needed
603  * ccltbl - holds the characters in each ccl - indexed by cclmap
612  * nmstr - last NAME scanned by the scanner
613  * sectnum - section number currently being parsed
614  * nummt - number of empty nxt/chk table entries
615  * hshcol - number of hash collisions detected by snstods
616  * dfaeql - number of times a newly created dfa was equal to an old one
617  * numeps - number of epsilon NFA states created
618  * eps2 - number of epsilon states which have 2 out-transitions
619  * num_reallocs - number of times it was necessary to realloc() a group
621  * tmpuses - number of DFA states that chain to templates
622  * totnst - total number of NFA states used to make DFA states
623  * peakpairs - peak number of transition pairs we had to store internally
624  * numuniq - number of unique transitions
625  * numdup - number of duplicate transitions
626  * hshsave - number of hash collisions saved by checking number of states
627  * num_backing_up - number of DFA states requiring backing up
628  * bol_needed - whether scanner needs beginning-of-line recognition
686 /* External functions that are cross-referenced among the flex source files. */
689 /* from file ccl.c */
701 /* from file dfa.c */
709 /* Construct the epsilon closure of a set of ndfa states. */
717 /* Converts a set of ndfa states into a dfa state. */
721 /* from file ecs.c */
736 /* from file gen.c */
740 /* Generate the code to keep backing-up information. */
754 /* Generate the code to find the next compressed-table state. */
781 /* from file main.c */
788 /* from file misc.c */
806 extern void check_char(int c);
808 /* Replace upper-case letter to lower-case. */
875 /* Generate a data statment for a two-dimensional array. */
886 /* Output a (possibly-formatted) string to the generated scanner. */
899  * 8-bit.
909 /* Only needed when using certain broken versions of bison to build parse.c. */
913 /* from file nfa.c */
983 /* The Flex-generated scanner for flex. */
990 /* from file sym.c */
1009 /* from file tblcmp.c */
1025  * only one out-transition.
1032 /* Save states with only one out-transition to be processed later. */
1036 /* from file yylex.c */
1040 /* A growable array. See buf.c. */
1061 extern struct Buf userdef_buf; /* a string buffer for #define's generated by user-options on cmd li…
1066 extern bool no_section3_escape; /* True if the undocumented option --unsafe-no-m4-sect3-escape was …
1072 /* For setjmp/longjmp (instead of calling exit(2)). Linkage in main.c */
1081 #define b_isalnum(c) (isalnum(c)?true:false)  argument
1082 #define b_isalpha(c) (isalpha(c)?true:false)  argument
1083 #define b_isascii(c) (isascii(c)?true:false)  argument
1084 #define b_isblank(c) (isblank(c)?true:false)  argument
1085 #define b_iscntrl(c) (iscntrl(c)?true:false)  argument
1086 #define b_isdigit(c) (isdigit(c)?true:false)  argument
1087 #define b_isgraph(c) (isgraph(c)?true:false)  argument
1088 #define b_islower(c) (islower(c)?true:false)  argument
1089 #define b_isprint(c) (isprint(c)?true:false)  argument
1090 #define b_ispunct(c) (ispunct(c)?true:false)  argument
1091 #define b_isspace(c) (isspace(c)?true:false)  argument
1092 #define b_isupper(c) (isupper(c)?true:false)  argument
1093 #define b_isxdigit(c) (isxdigit(c)?true:false)  argument
1096 bool has_case(int c);
1099 int reverse_case(int c);
1101 /* return false if [c1-c2] is ambiguous for a caseless scanner. */
1105  *  From "filter.c"
1116 	const char ** argv;   /**< arg vector, \0-terminated */
1133  * From "regex.c"