X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/c476f4254aaf06aafc9c81d98ebc185b60699523..d11389730686d2620d5490fa96eb210ea5837773:/regcomp.sym diff --git a/regcomp.sym b/regcomp.sym index f3f7164..02b278c 100644 --- a/regcomp.sym +++ b/regcomp.sym @@ -2,182 +2,235 @@ # # File has two sections, divided by a line of dashes '-'. # -# Empty rows after #-comment are removed from input are ignored +# Lines beginning with # are ignored, except for those that start with #* +# which are included in pod/perldebguts.pod. # within a line may be part +# of a description. # -# First section is for regops, second sectionis for regmatch-states +# First section is for regops, second section is for regmatch-states # # Note that the order in this file is important. # # Format for first section: -# NAME \t TYPE, arg-description [num-args] [longjump-len] \t DESCRIPTION +# NAME \s+ TYPE, arg-description [num-args] [flags] [longjump-len] ; DESCRIPTION +# flag means is REGNODE_SIMPLE; flag means is REGNODE_VARIES # # # run perl regen.pl after editing this file -#* Exit points (0,1) - -END END, no End of program. -SUCCEED END, no Return from a subroutine, basically. - -#* Anchors: (2..13) - -BOL BOL, no Match "" at beginning of line. -MBOL BOL, no Same, assuming multiline. -SBOL BOL, no Same, assuming singleline. -EOS EOL, no Match "" at end of string. -EOL EOL, no Match "" at end of line. -MEOL EOL, no Same, assuming multiline. -SEOL EOL, no Same, assuming singleline. -BOUND BOUND, no Match "" at any word boundary -BOUNDL BOUND, no Match "" at any word boundary -NBOUND NBOUND, no Match "" at any word non-boundary -NBOUNDL NBOUND, no Match "" at any word non-boundary -GPOS GPOS, no Matches where last m//g left off. - -#* [Special] alternatives: (14..30) - -REG_ANY REG_ANY, no Match any one character (except newline). -SANY REG_ANY, no Match any one character. -CANY REG_ANY, no Match any one byte. -ANYOF ANYOF, sv Match character in (or not in) this class. -ALNUM ALNUM, no Match any alphanumeric character -ALNUML ALNUM, no Match any alphanumeric char in locale -NALNUM NALNUM, no Match any non-alphanumeric character -NALNUML NALNUM, no Match any non-alphanumeric char in locale -SPACE SPACE, no Match any whitespace character -SPACEL SPACE, no Match any whitespace char in locale -NSPACE NSPACE, no Match any non-whitespace character -NSPACEL NSPACE, no Match any non-whitespace char in locale -DIGIT DIGIT, no Match any numeric character -DIGITL DIGIT, no Match any numeric character in locale -NDIGIT NDIGIT, no Match any non-numeric character -NDIGITL NDIGIT, no Match any non-numeric character in locale -CLUMP CLUMP, no Match any combining character sequence - -#* Alternation (31) - -# BRANCH The set of branches constituting a single choice are hooked -# together with their "next" pointers, since precedence prevents -# anything being concatenated to any individual branch. The -# "next" pointer of the last BRANCH in a choice points to the -# thing following the whole choice. This is also where the -# final "next" pointer of each individual branch points; each -# branch starts with the operand node of a BRANCH node. -# -BRANCH BRANCH, node Match this alternative, or the next... - -#*Back pointer (32) - -# BACK Normal "next" pointers all implicitly point forward; BACK -# exists to make loop structures possible. -# not used -BACK BACK, no Match "", "next" ptr points backward. - -#*Literals (33..35) - -EXACT EXACT, sv Match this string (preceded by length). -EXACTF EXACT, sv Match this string, folded (prec. by length). -EXACTFL EXACT, sv Match this string, folded in locale (w/len). - -#*Do nothing types (36..37) - -NOTHING NOTHING,no Match empty string. -# A variant of above which delimits a group, thus stops optimizations -TAIL NOTHING,no Match empty string. Can jump here from outside. - -#*Loops (38..44) - -# STAR,PLUS '?', and complex '*' and '+', are implemented as circular -# BRANCH structures using BACK. Simple cases (one character -# per match) are implemented with STAR and PLUS for speed -# and to minimize recursive plunges. -# -STAR STAR, node Match this (simple) thing 0 or more times. -PLUS PLUS, node Match this (simple) thing 1 or more times. - -CURLY CURLY, sv 2 Match this simple thing {n,m} times. -CURLYN CURLY, no 2 Match next-after-this simple thing -# {n,m} times, set parenths. -CURLYM CURLY, no 2 Match this medium-complex thing {n,m} times. -CURLYX CURLY, sv 2 Match this complex thing {n,m} times. +#* Exit points + +END END, no ; End of program. +SUCCEED END, no ; Return from a subroutine, basically. + +#* Line Start Anchors: +#Note flags field for SBOL indicates if it is a /^/ or a /\A/ +SBOL BOL, no ; Match "" at beginning of line: /^/, /\A/ +MBOL BOL, no ; Same, assuming multiline: /^/m + +#* Line End Anchors: +SEOL EOL, no ; Match "" at end of line: /$/ +MEOL EOL, no ; Same, assuming multiline: /$/m +EOS EOL, no ; Match "" at end of string: /\z/ + +#* Match Start Anchors: +GPOS GPOS, no ; Matches where last m//g left off. + +#* Word Boundary Opcodes: +# The regops that have varieties that vary depending on the character set regex +# modifiers have to ordered thusly: /d, /l, /u, /a, /aa. This is because code +# in regcomp.c uses the enum value of the modifier as an offset from the /d +# version. The complements must come after the non-complements. +# BOUND, POSIX and their complements are affected, as well as EXACTF. +BOUND BOUND, no ; Match "" at any word boundary using native charset rules for non-utf8 +BOUNDL BOUND, no ; Match "" at any locale word boundary +BOUNDU BOUND, no ; Match "" at any word boundary using Unicode rules +BOUNDA BOUND, no ; Match "" at any word boundary using ASCII rules +# All NBOUND nodes are required by code in regexec.c to be greater than all BOUND ones +NBOUND NBOUND, no ; Match "" at any word non-boundary using native charset rules for non-utf8 +NBOUNDL NBOUND, no ; Match "" at any locale word non-boundary +NBOUNDU NBOUND, no ; Match "" at any word non-boundary using Unicode rules +NBOUNDA NBOUND, no ; Match "" at any word non-boundary using ASCII rules + +#* [Special] alternatives: +REG_ANY REG_ANY, no 0 S ; Match any one character (except newline). +SANY REG_ANY, no 0 S ; Match any one character. +CANY REG_ANY, no 0 S ; Match any one byte. +ANYOF ANYOF, sv 1 S ; Match character in (or not in) this class, single char match only + +#* POSIX Character Classes: +# Order of the below is important. See ordering comment above. +POSIXD POSIXD, none 0 S ; Some [[:class:]] under /d; the FLAGS field gives which one +POSIXL POSIXD, none 0 S ; Some [[:class:]] under /l; the FLAGS field gives which one +POSIXU POSIXD, none 0 S ; Some [[:class:]] under /u; the FLAGS field gives which one +POSIXA POSIXD, none 0 S ; Some [[:class:]] under /a; the FLAGS field gives which one +NPOSIXD NPOSIXD, none 0 S ; complement of POSIXD, [[:^class:]] +NPOSIXL NPOSIXD, none 0 S ; complement of POSIXL, [[:^class:]] +NPOSIXU NPOSIXD, none 0 S ; complement of POSIXU, [[:^class:]] +NPOSIXA NPOSIXD, none 0 S ; complement of POSIXA, [[:^class:]] +# End of order is important + +CLUMP CLUMP, no 0 V ; Match any extended grapheme cluster sequence + +#* Alternation + +#* BRANCH The set of branches constituting a single choice are +#* hooked together with their "next" pointers, since +#* precedence prevents anything being concatenated to +#* any individual branch. The "next" pointer of the last +#* BRANCH in a choice points to the thing following the +#* whole choice. This is also where the final "next" +#* pointer of each individual branch points; each branch +#* starts with the operand node of a BRANCH node. +#* +BRANCH BRANCH, node 0 V ; Match this alternative, or the next... + +#*Literals +# NOTE: the relative ordering of these types is important do not change it + +EXACT EXACT, str ; Match this string (preceded by length). +EXACTF EXACT, str ; Match this non-UTF-8 string (not guaranteed to be folded) using /id rules (w/len). +EXACTFL EXACT, str ; Match this string (not guaranteed to be folded) using /il rules (w/len). +EXACTFU EXACT, str ; Match this string (folded iff in UTF-8, length in folding doesn't change if not in UTF-8) using /iu rules (w/len). +EXACTFA EXACT, str ; Match this string (not guaranteed to be folded) using /iaa rules (w/len). +EXACTFU_SS EXACT, str ; Match this string (folded iff in UTF-8, length in folding may change even if not in UTF-8) using /iu rules (w/len). +EXACTFA_NO_TRIE EXACT, str ; Match this string (which is not trie-able; not guaranteed to be folded) using /iaa rules (w/len). + +#*Do nothing types + +NOTHING NOTHING, no ; Match empty string. +#*A variant of above which delimits a group, thus stops optimizations +TAIL NOTHING, no ; Match empty string. Can jump here from outside. + +#*Loops + +#* STAR,PLUS '?', and complex '*' and '+', are implemented as +#* circular BRANCH structures. Simple cases +#* (one character per match) are implemented with STAR +#* and PLUS for speed and to minimize recursive plunges. +#* +STAR STAR, node 0 V ; Match this (simple) thing 0 or more times. +PLUS PLUS, node 0 V ; Match this (simple) thing 1 or more times. + +CURLY CURLY, sv 2 V ; Match this simple thing {n,m} times. +CURLYN CURLY, no 2 V ; Capture next-after-this simple thing +CURLYM CURLY, no 2 V ; Capture this medium-complex thing {n,m} times. +CURLYX CURLY, sv 2 V ; Match this complex thing {n,m} times. + +#*This terminator creates a loop structure for CURLYX +WHILEM WHILEM, no 0 V ; Do curly processing and see if rest matches. + +#*Buffer related + +#*OPEN,CLOSE,GROUPP ...are numbered at compile time. +OPEN OPEN, num 1 ; Mark this point in input as start of #n. +CLOSE CLOSE, num 1 ; Analogous to OPEN. + +REF REF, num 1 V ; Match some already matched string +REFF REF, num 1 V ; Match already matched string, folded using native charset rules for non-utf8 +REFFL REF, num 1 V ; Match already matched string, folded in loc. +# N?REFF[AU] could have been implemented using the FLAGS field of the +# regnode, but by having a separate node type, we can use the existing switch +# statement to avoid some tests +REFFU REF, num 1 V ; Match already matched string, folded using unicode rules for non-utf8 +REFFA REF, num 1 V ; Match already matched string, folded using unicode rules for non-utf8, no mixing ASCII, non-ASCII + +#*Named references. Code in regcomp.c assumes that these all are after +#*the numbered references +NREF REF, no-sv 1 V ; Match some already matched string +NREFF REF, no-sv 1 V ; Match already matched string, folded using native charset rules for non-utf8 +NREFFL REF, no-sv 1 V ; Match already matched string, folded in loc. +NREFFU REF, num 1 V ; Match already matched string, folded using unicode rules for non-utf8 +NREFFA REF, num 1 V ; Match already matched string, folded using unicode rules for non-utf8, no mixing ASCII, non-ASCII + +#*Support for long RE +LONGJMP LONGJMP, off 1 . 1 ; Jump far away. +BRANCHJ BRANCHJ, off 1 V 1 ; BRANCH with long offset. + +#*Special Case Regops +IFMATCH BRANCHJ, off 1 . 2 ; Succeeds if the following matches. +UNLESSM BRANCHJ, off 1 . 2 ; Fails if the following matches. +SUSPEND BRANCHJ, off 1 V 1 ; "Independent" sub-RE. +IFTHEN BRANCHJ, off 1 V 1 ; Switch, should be preceded by switcher. +GROUPP GROUPP, num 1 ; Whether the group matched. + + +#*The heavy worker + +EVAL EVAL, evl/flags 2L ; Execute some Perl code. + +#*Modifiers + +MINMOD MINMOD, no ; Next operator is not greedy. +LOGICAL LOGICAL, no ; Next opcode should set the flag only. + +#*This is not used yet +RENUM BRANCHJ, off 1 . 1 ; Group with independently numbered parens. -# This terminator creates a loop structure for CURLYX -WHILEM WHILEM, no Do curly processing and see if rest matches. +#*Trie Related -#*Buffer related (45..49) +#* Behave the same as A|LIST|OF|WORDS would. The '..C' variants +#* have inline charclass data (ascii only), the 'C' store it in the +#* structure. +# NOTE: the relative order of the TRIE-like regops is significant + +TRIE TRIE, trie 1 ; Match many EXACT(F[ALU]?)? at once. flags==type +TRIEC TRIE,trie charclass ; Same as TRIE, but with embedded charclass data -# OPEN,CLOSE,GROUPP ...are numbered at compile time. -OPEN OPEN, num 1 Mark this point in input as start of #n. -CLOSE CLOSE, num 1 Analogous to OPEN. - -REF REF, num 1 Match some already matched string -REFF REF, num 1 Match already matched string, folded -REFFL REF, num 1 Match already matched string, folded in loc. - -#*Grouping assertions (50..54) - -IFMATCH BRANCHJ,off 1 2 Succeeds if the following matches. -UNLESSM BRANCHJ,off 1 2 Fails if the following matches. -SUSPEND BRANCHJ,off 1 1 "Independent" sub-RE. -IFTHEN BRANCHJ,off 1 1 Switch, should be preceeded by switcher . -GROUPP GROUPP, num 1 Whether the group matched. - -#*Support for long RE (55..56) - -LONGJMP LONGJMP,off 1 1 Jump far away. -BRANCHJ BRANCHJ,off 1 1 BRANCH with long offset. - -#*The heavy worker (57..58) - -EVAL EVAL, evl 1 Execute some Perl code. - -#*Modifiers (59..60) +# For start classes, contains an added fail table. +AHOCORASICK TRIE, trie 1 ; Aho Corasick stclass. flags==type +AHOCORASICKC TRIE,trie charclass ; Same as AHOCORASICK, but with embedded charclass data -MINMOD MINMOD, no Next operator is not greedy. -LOGICAL LOGICAL,no Next opcode should set the flag only. +#*Regex Subroutines +GOSUB GOSUB, num/ofs 2L ; recurse to paren arg1 at (signed) ofs arg2 +GOSTART GOSTART, no ; recurse to start of pattern -# This is not used yet (61) -RENUM BRANCHJ,off 1 1 Group with independently numbered parens. +#*Special conditionals +NGROUPP NGROUPP, no-sv 1 ; Whether the group matched. +INSUBP INSUBP, num 1 ; Whether we are in a specific recurse. +DEFINEP DEFINEP, none 1 ; Never execute directly. -#*Trie Related (62..64) +#*Backtracking Verbs +ENDLIKE ENDLIKE, none ; Used only for the type field of verbs +OPFAIL ENDLIKE, none ; Same as (?!) +ACCEPT ENDLIKE, parno 1 ; Accepts the current matched string. -# Behave the same as A|LIST|OF|WORDS would. The '..C' variants have -# inline charclass data (ascii only), the 'C' store it in the structure. -# NOTE: the relative order of the TRIE-like regops is signifigant -TRIE TRIE, trie 1 Match many EXACT(FL?)? at once. flags==type -TRIEC TRIE, trie charclass Same as TRIE, but with embedded charclass data +#*Verbs With Arguments +VERB VERB, no-sv 1 ; Used only for the type field of verbs +PRUNE VERB, no-sv 1 ; Pattern fails at this startpoint if no-backtracking through this +MARKPOINT VERB, no-sv 1 ; Push the current location for rollback by cut. +SKIP VERB, no-sv 1 ; On failure skip forward (to the mark) before retrying +COMMIT VERB, no-sv 1 ; Pattern fails outright if backtracking through this +CUTGROUP VERB, no-sv 1 ; On failure go to the next alternation in the group -# For start classes, contains an added fail table. -AHOCORASICK TRIE, trie 1 Aho Corasick stclass. flags==type -AHOCORASICKC TRIE, trie charclass Same as AHOCORASICK, but with embedded charclass data +#*Control what to keep in $&. +KEEPS KEEPS, no ; $& begins here. -#*Recursion (65) -RECURSE RECURSE, num/ofs 2L recurse to paren arg1 at (signed) ofs arg2 -SRECURSE RECURSE, no recurse to start of pattern +#*New charclass like patterns +LNBREAK LNBREAK, none ; generic newline pattern -# NEW STUFF ABOVE THIS LINE -- Please update counts below. +# NEW STUFF SOMEWHERE ABOVE THIS LINE ################################################################################ -#*SPECIAL REGOPS (65, 66) +#*SPECIAL REGOPS -# This is not really a node, but an optimized away piece of a "long" node. -# To simplify debugging output, we mark it as if it were a node -OPTIMIZED NOTHING,off Placeholder for dump. +#* This is not really a node, but an optimized away piece of a "long" +#* node. To simplify debugging output, we mark it as if it were a node +OPTIMIZED NOTHING, off ; Placeholder for dump. -# Special opcode with the property that no opcode in a compiled program -# will ever be of this type. Thus it can be used as a flag value that -# no other opcode has been seen. END is used similarly, in that an END -# node cant be optimized. So END implies "unoptimizable" and PSEUDO mean -# "not seen anything to optimize yet". -PSEUDO PSEUDO,off Pseudo opcode for internal use. +#* Special opcode with the property that no opcode in a compiled program +#* will ever be of this type. Thus it can be used as a flag value that +#* no other opcode has been seen. END is used similarly, in that an END +#* node cant be optimized. So END implies "unoptimizable" and PSEUDO +#* mean "not seen anything to optimize yet". +PSEUDO PSEUDO, off ; Pseudo opcode for internal use. ------------------------------------------------------------------------------- # Format for second section: -# REGOP \t typelist [ \t typelist] [# Comment] +# REGOP \t typelist [ \t typelist] # typelist= namelist # = namelist:FAIL # = name:count @@ -185,12 +238,16 @@ PSEUDO PSEUDO,off Pseudo opcode for internal use. # Anything below is a state # # -TRIE next:FAIL -EVAL AB:FAIL -CURLYX end:FAIL -WHILEM A_pre,A_min,A_max,B_min,B_max:FAIL -BRANCH next:FAIL -CURLYM A,B:FAIL -IFMATCH A:FAIL -CURLY B_min_known,B_min,B_max:FAIL - +TRIE next:FAIL +EVAL AB:FAIL +CURLYX end:FAIL +WHILEM A_pre,A_min,A_max,B_min,B_max:FAIL +BRANCH next:FAIL +CURLYM A,B:FAIL +IFMATCH A:FAIL +CURLY B_min_known,B_min,B_max:FAIL +COMMIT next:FAIL +MARKPOINT next:FAIL +SKIP next:FAIL +CUTGROUP next:FAIL +KEEPS next:FAIL