File:  [gforth] / gforth / regexp.fs
Revision 1.1: download - view: text, annotated - select for diffs
Sat Nov 5 23:26:49 2005 UTC (18 years, 4 months ago) by pazsan
Branches: MAIN
CVS tags: HEAD
Added regexp stuff

    1: \ Regexp compile
    2: 
    3: \ The idea of the parser is the following:
    4: \ As long as there's a match, continue
    5: \ On a mismatch, LEAVE.
    6: \ Insert appropriate control structures on alternative branches
    7: \ Keep the old pointer (backtracking) on the stack
    8: \ I try to keep the syntax as close to a real regexp system as possible
    9: \ All regexp stuff is compiled into one function as forward branching
   10: \ state machine
   11: 
   12: \ bulk-postponing
   13: 
   14: : FORK ( compilation -- orig ; run-time f -- ) \ core
   15:     POSTPONE call >mark ; immediate restrict
   16: : JOIN ( orig -- )  postpone THEN ; immediate restrict
   17: 
   18: \ Charclasses
   19: 
   20: : +bit ( addr n -- )  + 1 swap c! ;
   21: : -bit ( addr n -- )  + 0 swap c! ;
   22: : @+ ( addr -- n addr' )  dup @ swap cell+ ;
   23: 
   24: 0 Value cur-class
   25: : charclass ( -- )  Create here dup to cur-class $100 dup allot erase ;
   26: : +char ( char -- )  cur-class swap +bit ;
   27: : -char ( char -- )  cur-class swap -bit ;
   28: : ..char ( start end -- )  1+ swap ?DO  I +char  LOOP ;
   29: : or! ( n addr -- )  dup @ rot or swap ! ;
   30: : and! ( n addr -- )  dup @ rot and swap ! ;
   31: : +class ( class -- )  $100 0 ?DO  @+ swap
   32:         cur-class I + or!  cell +LOOP  drop ;
   33: : -class ( class -- )  $100 0 ?DO  @+ swap invert
   34:         cur-class I + and!  cell +LOOP  drop ;
   35: 
   36: : char? ( addr class -- addr' flag )
   37:     >r count r> + c@ ;
   38: 
   39: \ Charclass tests
   40: 
   41: : c? ( addr class -- )   ]] char? 0= ?LEAVE [[ ; immediate
   42: : -c? ( addr class -- )  ]] char?    ?LEAVE [[ ; immediate
   43: 
   44: charclass digit  '0 '9 ..char
   45: charclass blanks 0 bl ..char
   46: \ bl +char #tab +char #cr +char #lf +char ctrl L +char
   47: charclass letter 'a 'z ..char 'A 'Z ..char
   48: charclass any    0 $FF ..char #lf -char
   49: 
   50: : \d ( addr -- addr' )   ]] digit c?        [[ ; immediate
   51: : \s ( addr -- addr' )   ]] blanks c?       [[ ; immediate
   52: : .? ( addr -- addr' )   ]] any c?          [[ ; immediate
   53: : -\d ( addr -- addr' )  ]] digit -c?       [[ ; immediate
   54: : -\s ( addr -- addr' )  ]] blanks -c?      [[ ; immediate
   55: : ` ( -- )
   56:     ]] count [[  char ]] Literal <> ?LEAVE [[ ;  immediate
   57: 
   58: \ A word for string comparison
   59: 
   60: : $= ( addr1 addr2 u -- f )  tuck compare ;
   61: : ,=" ( addr u -- ) tuck ]] dup SLiteral $= ?LEAVE Literal + noop [[ ;
   62: : =" ( <string>" -- )  '" parse ,=" ; immediate
   63: 
   64: \ loop stack
   65: 
   66: Variable loops  $40 3 * cells allot
   67: : 3@ ( addr -- a b c )  dup >r 2 cells + @ r> 2@ ;
   68: : 3! ( a b c addr -- )  dup >r 2! r> 2 cells + ! ;
   69: : loops> ( -- addr ) -3 loops +!  loops @+ swap cells + 3@ ;
   70: : >loops ( addr -- ) loops @+ swap cells + 3! 3 loops +! ;
   71: : BEGIN, ( -- )  ]] BEGIN [[ >loops ;
   72: : DONE, ( -- )  loops @ IF  loops> ]] DONE [[ THEN ]] noop [[ ;
   73: 
   74: \ variables
   75: 
   76: Variable vars   &18 cells allot
   77: Variable varstack 9 cells allot
   78: Variable varsmax
   79: : >var ( -- addr ) vars @+ swap 2* cells +
   80:     vars @ varstack @+ swap cells + !
   81:     1 vars +! 1 varstack +! ;
   82: : var> ( -- addr ) -1 varstack +!
   83:     varstack @+ swap cells + @
   84:     1+ 2* cells vars + ;
   85: 
   86: \ start end
   87: 
   88: 0 Value end$
   89: 0 Value start$
   90: : !end ( addr u -- addr )  over + to end$ dup to start$ ;
   91: : $? ( addr -- addr flag ) dup end$ u< ;
   92: : ^? ( addr -- addr flag ) dup start$ u> ;
   93: : ?end ( addr -- addr ) ]] dup end$ u> ?LEAVE [[ ; immediate
   94: 
   95: \ start and end
   96: 
   97: : \^ ( addr -- addr )
   98:     ]] ^? ?LEAVE [[ ; immediate
   99: : \$ ( addr -- addr )
  100:     ]] $? ?LEAVE [[ ; immediate
  101: 
  102: \ regexp block
  103: 
  104: \ FORK/JOIN are like AHEAD THEN, but producing a call on AHEAD
  105: \ instead of a jump.
  106: 
  107: : (( ( addr u -- )  vars off varsmax off loops off
  108:     ]] FORK  AHEAD BUT JOIN !end [[ BEGIN, ; immediate
  109: : )) ( -- addr f )
  110:     ]] ?end drop true EXIT [[
  111:     DONE, ]] drop false EXIT THEN [[ ; immediate
  112: 
  113: \ greedy loops
  114: 
  115: \ Idea: scan as many characters as possible, try the rest of the pattern
  116: \ and then back off one pattern at a time
  117: 
  118: : drops ( n -- ) 1+ cells sp@ + sp! ;
  119: 
  120: : {** ( addr -- addr addr )
  121:     0 ]] Literal >r BEGIN dup [[ BEGIN, ; immediate
  122: ' {** Alias {++ ( addr -- addr addr ) immediate
  123: : n*} ( sys n -- )  >r ]] r> 1+ >r $? 0= UNTIL dup [[ DONE, ]] drop [[
  124:     r@ IF r@ ]] r@ Literal u< IF  r> 1+ drops false  EXIT  THEN [[ THEN
  125:     r@ ]] r> 1+ Literal U+DO FORK BUT [[
  126:     ]] IF  I' I - [[ r@ 1- ]] Literal + drops true UNLOOP EXIT  THEN  LOOP [[
  127:     r@ IF  r@ ]] Literal drops [[ THEN
  128:     rdrop ]] false  EXIT  JOIN [[ ; immediate
  129: : **}  0 postpone n*} ; immediate
  130: : ++}  1 postpone n*} ; immediate
  131: 
  132: \ non-greedy loops
  133: 
  134: \ Idea: Try to match rest of the regexp, and if that fails, try match
  135: \ first expr and then try again rest of regexp.
  136: 
  137: : {+ ( addr -- addr addr )
  138:     ]] BEGIN  [[ BEGIN, ; immediate
  139: : {* ( addr -- addr addr )
  140:     ]] {+ dup FORK BUT  IF  drop true  EXIT THEN [[ ; immediate
  141: : *} ( addr addr' -- addr' )
  142:     ]] dup end$ u>  UNTIL [[
  143:     DONE, ]] drop false  EXIT  JOIN [[ ; immediate
  144: : +} ( addr addr' -- addr' )
  145:     ]] dup FORK BUT  IF  drop true  EXIT [[
  146:     DONE, ]] drop false  EXIT  THEN *} [[ ; immediate
  147: 
  148: : // ( -- ) ]] {* 1+ *} [[ ; immediate
  149: 
  150: \ alternatives
  151: 
  152: \ idea: try to match one alternative and then the rest of regexp.
  153: \ if that fails, jump back to second alternative
  154: 
  155: : THENs ( sys -- )  BEGIN  dup  WHILE  ]] THEN [[  REPEAT  drop ;
  156: 
  157: : {{ ( addr -- addr addr )  0 ]] dup BEGIN [[  vars @ ; immediate
  158: : || ( addr addr -- addr addr ) vars @ varsmax @ max varsmax !
  159:     ]] nip AHEAD [[ >r >r >r vars !
  160:     ]] DONE drop dup [[ r> r> r> ]] BEGIN [[ vars @ ; immediate
  161: : }} ( addr addr -- addr addr ) vars @ varsmax @ max vars !
  162:     ]] nip AHEAD [[ >r >r >r drop
  163:     ]] DONE drop LEAVE [[ r> r> r> THENs ; immediate
  164: 
  165: \ match variables
  166: 
  167: : \( ( addr -- addr )  ]] dup [[
  168:     >var ]] ALiteral ! [[ ; immediate
  169: : \) ( addr -- addr )  ]] dup [[
  170:     var> ]] ALiteral ! [[ ; immediate
  171: : \0 ( -- addr u )  start$ end$ over - ;
  172: : \: ( i -- )
  173:     Create 2* 1+ cells vars + ,
  174:   DOES> ( -- addr u ) @ 2@ tuck - ;
  175: : \:s ( n -- ) 0 ?DO  I \:  LOOP ;
  176: 9 \:s \1 \2 \3 \4 \5 \6 \7 \8 \9

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>