?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
* doc/install.texi (*-*-aix): Update explanation of XLC bootstrap.
[official-gcc.git] / gcc / ada / s-regpat.adb
blob68d915f8ad0f5417d2ec58262dec9237c6b79880
1 ------------------------------------------------------------------------------
2 -- --
3 -- GNAT LIBRARY COMPONENTS --
4 -- --
5 -- G N A T . R E G P A T --
6 -- --
7 -- B o d y --
8 -- --
9 -- Copyright (C) 1986 by University of Toronto. --
10 -- Copyright (C) 1999-2008, AdaCore --
11 -- --
12 -- GNAT is free software; you can redistribute it and/or modify it under --
13 -- terms of the GNU General Public License as published by the Free Soft- --
14 -- ware Foundation; either version 2, or (at your option) any later ver- --
15 -- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
16 -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
17 -- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License --
18 -- for more details. You should have received a copy of the GNU General --
19 -- Public License distributed with GNAT; see file COPYING. If not, write --
20 -- to the Free Software Foundation, 51 Franklin Street, Fifth Floor, --
21 -- Boston, MA 02110-1301, USA. --
22 -- --
23 -- As a special exception, if other files instantiate generics from this --
24 -- unit, or you link this unit with other files to produce an executable, --
25 -- this unit does not by itself cause the resulting executable to be --
26 -- covered by the GNU General Public License. This exception does not --
27 -- however invalidate any other reasons why the executable file might be --
28 -- covered by the GNU Public License. --
29 -- --
30 -- GNAT was originally developed by the GNAT team at New York University. --
31 -- Extensive contributions were provided by Ada Core Technologies Inc. --
32 -- --
33 ------------------------------------------------------------------------------
34
35 -- This is an altered Ada 95 version of the original V8 style regular
36 -- expression library written in C by Henry Spencer. Apart from the
37 -- translation to Ada, the interface has been considerably changed to
38 -- use the Ada String type instead of C-style nul-terminated strings.
39
40 -- Beware that some of this code is subtly aware of the way operator
41 -- precedence is structured in regular expressions. Serious changes in
42 -- regular-expression syntax might require a total rethink.
43
44 with System.IO; use System.IO;
45 with Ada.Characters.Handling; use Ada.Characters.Handling;
46 with Ada.Unchecked_Conversion;
47
48 package body System.Regpat is
49
50 MAGIC : constant Character := Character'Val (10#0234#);
51 -- The first byte of the regexp internal "program" is actually
52 -- this magic number; the start node begins in the second byte.
53 --
54 -- This is used to make sure that a regular expression was correctly
55 -- compiled.
56
57 ----------------------------
58 -- Implementation details --
59 ----------------------------
60
61 -- This is essentially a linear encoding of a nondeterministic
62 -- finite-state machine, also known as syntax charts or
63 -- "railroad normal form" in parsing technology.
64
65 -- Each node is an opcode plus a "next" pointer, possibly plus an
66 -- operand. "Next" pointers of all nodes except BRANCH implement
67 -- concatenation; a "next" pointer with a BRANCH on both ends of it
68 -- is connecting two alternatives.
69
70 -- The operand of some types of node is a literal string; for others,
71 -- it is a node leading into a sub-FSM. In particular, the operand of
72 -- a BRANCH node is the first node of the branch.
73 -- (NB this is *not* a tree structure: the tail of the branch connects
74 -- to the thing following the set of BRANCHes).
75
76 -- You can see the exact byte-compiled version by using the Dump
77 -- subprogram. However, here are a few examples:
78
79 -- (a|b): 1 : MAGIC
80 -- 2 : BRANCH (next at 10)
81 -- 5 : EXACT (next at 18) operand=a
82 -- 10 : BRANCH (next at 18)
83 -- 13 : EXACT (next at 18) operand=b
84 -- 18 : EOP (next at 0)
85 --
86 -- (ab)*: 1 : MAGIC
87 -- 2 : CURLYX (next at 26) { 0, 32767}
88 -- 9 : OPEN 1 (next at 13)
89 -- 13 : EXACT (next at 19) operand=ab
90 -- 19 : CLOSE 1 (next at 23)
91 -- 23 : WHILEM (next at 0)
92 -- 26 : NOTHING (next at 29)
93 -- 29 : EOP (next at 0)
94
95 -- The opcodes are:
96
97 type Opcode is
98
99 -- Name Operand? Meaning
100
101 (EOP, -- no End of program
102 MINMOD, -- no Next operator is not greedy
103
104 -- Classes of characters
105
106 ANY, -- no Match any one character except newline
107 SANY, -- no Match any character, including new line
108 ANYOF, -- class Match any character in this class
109 EXACT, -- str Match this string exactly
110 EXACTF, -- str Match this string (case-folding is one)
111 NOTHING, -- no Match empty string
112 SPACE, -- no Match any whitespace character
113 NSPACE, -- no Match any non-whitespace character
114 DIGIT, -- no Match any numeric character
115 NDIGIT, -- no Match any non-numeric character
116 ALNUM, -- no Match any alphanumeric character
117 NALNUM, -- no Match any non-alphanumeric character
118
119 -- Branches
120
121 BRANCH, -- node Match this alternative, or the next
122
123 -- Simple loops (when the following node is one character in length)
124
125 STAR, -- node Match this simple thing 0 or more times
126 PLUS, -- node Match this simple thing 1 or more times
127 CURLY, -- 2num node Match this simple thing between n and m times.
128
129 -- Complex loops
130
131 CURLYX, -- 2num node Match this complex thing {n,m} times
132 -- The nums are coded on two characters each
133
134 WHILEM, -- no Do curly processing and see if rest matches
135
136 -- Matches after or before a word
137
138 BOL, -- no Match "" at beginning of line
139 MBOL, -- no Same, assuming multiline (match after \n)
140 SBOL, -- no Same, assuming single line (don't match at \n)
141 EOL, -- no Match "" at end of line
142 MEOL, -- no Same, assuming multiline (match before \n)
143 SEOL, -- no Same, assuming single line (don't match at \n)
144
145 BOUND, -- no Match "" at any word boundary
146 NBOUND, -- no Match "" at any word non-boundary
147
148 -- Parenthesis groups handling
149
150 REFF, -- num Match some already matched string, folded
151 OPEN, -- num Mark this point in input as start of #n
152 CLOSE); -- num Analogous to OPEN
153
154 for Opcode'Size use 8;
155
156 -- Opcode notes:
157
158 -- BRANCH
159 -- The set of branches constituting a single choice are hooked
160 -- together with their "next" pointers, since precedence prevents
161 -- anything being concatenated to any individual branch. The
162 -- "next" pointer of the last BRANCH in a choice points to the
163 -- thing following the whole choice. This is also where the
164 -- final "next" pointer of each individual branch points; each
165 -- branch starts with the operand node of a BRANCH node.
166
167 -- STAR,PLUS
168 -- '?', and complex '*' and '+', are implemented with CURLYX.
169 -- branches. Simple cases (one character per match) are implemented with
170 -- STAR and PLUS for speed and to minimize recursive plunges.
171
172 -- OPEN,CLOSE
173 -- ...are numbered at compile time.
174
175 -- EXACT, EXACTF
176 -- There are in fact two arguments, the first one is the length (minus
177 -- one of the string argument), coded on one character, the second
178 -- argument is the string itself, coded on length + 1 characters.
179
180 -- A node is one char of opcode followed by two chars of "next" pointer.
181 -- "Next" pointers are stored as two 8-bit pieces, high order first. The
182 -- value is a positive offset from the opcode of the node containing it.
183 -- An operand, if any, simply follows the node. (Note that much of the
184 -- code generation knows about this implicit relationship.)
185
186 -- Using two bytes for the "next" pointer is vast overkill for most
187 -- things, but allows patterns to get big without disasters.
188
189 -----------------------
190 -- Character classes --
191 -----------------------
192 -- This is the implementation for character classes ([...]) in the
193 -- syntax for regular expressions. Each character (0..256) has an
194 -- entry into the table. This makes for a very fast matching
195 -- algorithm.
196
197 type Class_Byte is mod 256;
198 type Character_Class is array (Class_Byte range 0 .. 31) of Class_Byte;
199
200 type Bit_Conversion_Array is array (Class_Byte range 0 .. 7) of Class_Byte;
201 Bit_Conversion : constant Bit_Conversion_Array :=
202 (1, 2, 4, 8, 16, 32, 64, 128);
203
204 type Std_Class is (ANYOF_NONE,
205 ANYOF_ALNUM, -- Alphanumeric class [a-zA-Z0-9]
206 ANYOF_NALNUM,
207 ANYOF_SPACE, -- Space class [ \t\n\r\f]
208 ANYOF_NSPACE,
209 ANYOF_DIGIT, -- Digit class [0-9]
210 ANYOF_NDIGIT,
211 ANYOF_ALNUMC, -- Alphanumeric class [a-zA-Z0-9]
212 ANYOF_NALNUMC,
213 ANYOF_ALPHA, -- Alpha class [a-zA-Z]
214 ANYOF_NALPHA,
215 ANYOF_ASCII, -- Ascii class (7 bits) 0..127
216 ANYOF_NASCII,
217 ANYOF_CNTRL, -- Control class
218 ANYOF_NCNTRL,
219 ANYOF_GRAPH, -- Graphic class
220 ANYOF_NGRAPH,
221 ANYOF_LOWER, -- Lower case class [a-z]
222 ANYOF_NLOWER,
223 ANYOF_PRINT, -- printable class
224 ANYOF_NPRINT,
225 ANYOF_PUNCT, --
226 ANYOF_NPUNCT,
227 ANYOF_UPPER, -- Upper case class [A-Z]
228 ANYOF_NUPPER,
229 ANYOF_XDIGIT, -- Hexadecimal digit
230 ANYOF_NXDIGIT
231 );
232
233 procedure Set_In_Class
234 (Bitmap : in out Character_Class;
235 C : Character);
236 -- Set the entry to True for C in the class Bitmap
237
238 function Get_From_Class
239 (Bitmap : Character_Class;
240 C : Character) return Boolean;
241 -- Return True if the entry is set for C in the class Bitmap
242
243 procedure Reset_Class (Bitmap : out Character_Class);
244 -- Clear all the entries in the class Bitmap
245
246 pragma Inline (Set_In_Class);
247 pragma Inline (Get_From_Class);
248 pragma Inline (Reset_Class);
249
250 -----------------------
251 -- Local Subprograms --
252 -----------------------
253
254 function "=" (Left : Character; Right : Opcode) return Boolean;
255
256 function Is_Alnum (C : Character) return Boolean;
257 -- Return True if C is an alphanum character or an underscore ('_')
258
259 function Is_White_Space (C : Character) return Boolean;
260 -- Return True if C is a whitespace character
261
262 function Is_Printable (C : Character) return Boolean;
263 -- Return True if C is a printable character
264
265 function Operand (P : Pointer) return Pointer;
266 -- Return a pointer to the first operand of the node at P
267
268 function String_Length
269 (Program : Program_Data;
270 P : Pointer) return Program_Size;
271 -- Return the length of the string argument of the node at P
272
273 function String_Operand (P : Pointer) return Pointer;
274 -- Return a pointer to the string argument of the node at P
275
276 procedure Bitmap_Operand
277 (Program : Program_Data;
278 P : Pointer;
279 Op : out Character_Class);
280 -- Return a pointer to the string argument of the node at P
281
282 function Get_Next_Offset
283 (Program : Program_Data;
284 IP : Pointer) return Pointer;
285 -- Get the offset field of a node. Used by Get_Next
286
287 function Get_Next
288 (Program : Program_Data;
289 IP : Pointer) return Pointer;
290 -- Dig the next instruction pointer out of a node
291
292 procedure Optimize (Self : in out Pattern_Matcher);
293 -- Optimize a Pattern_Matcher by noting certain special cases
294
295 function Read_Natural
296 (Program : Program_Data;
297 IP : Pointer) return Natural;
298 -- Return the 2-byte natural coded at position IP
299
300 -- All of the subprograms above are tiny and should be inlined
301
302 pragma Inline ("=");
303 pragma Inline (Is_Alnum);
304 pragma Inline (Is_White_Space);
305 pragma Inline (Get_Next);
306 pragma Inline (Get_Next_Offset);
307 pragma Inline (Operand);
308 pragma Inline (Read_Natural);
309 pragma Inline (String_Length);
310 pragma Inline (String_Operand);
311
312 type Expression_Flags is record
313 Has_Width, -- Known never to match null string
314 Simple, -- Simple enough to be STAR/PLUS operand
315 SP_Start : Boolean; -- Starts with * or +
316 end record;
317
318 Worst_Expression : constant Expression_Flags := (others => False);
319 -- Worst case
320
321 ---------
322 -- "=" --
323 ---------
324
325 function "=" (Left : Character; Right : Opcode) return Boolean is
326 begin
327 return Character'Pos (Left) = Opcode'Pos (Right);
328 end "=";
329
330 --------------------
331 -- Bitmap_Operand --
332 --------------------
333
334 procedure Bitmap_Operand
335 (Program : Program_Data;
336 P : Pointer;
337 Op : out Character_Class)
338 is
339 function Convert is new Ada.Unchecked_Conversion
340 (Program_Data, Character_Class);
341
342 begin
343 Op (0 .. 31) := Convert (Program (P + 3 .. P + 34));
344 end Bitmap_Operand;
345
346 -------------
347 -- Compile --
348 -------------
349
350 procedure Compile
351 (Matcher : out Pattern_Matcher;
352 Expression : String;
353 Final_Code_Size : out Program_Size;
354 Flags : Regexp_Flags := No_Flags)
355 is
356 -- We can't allocate space until we know how big the compiled form
357 -- will be, but we can't compile it (and thus know how big it is)
358 -- until we've got a place to put the code. So we cheat: we compile
359 -- it twice, once with code generation turned off and size counting
360 -- turned on, and once "for real".
361
362 -- This also means that we don't allocate space until we are sure
363 -- that the thing really will compile successfully, and we never
364 -- have to move the code and thus invalidate pointers into it.
365
366 -- Beware that the optimization-preparation code in here knows
367 -- about some of the structure of the compiled regexp.
368
369 PM : Pattern_Matcher renames Matcher;
370 Program : Program_Data renames PM.Program;
371
372 Emit_Code : constant Boolean := PM.Size > 0;
373 Emit_Ptr : Pointer := Program_First;
374
375 Parse_Pos : Natural := Expression'First; -- Input-scan pointer
376 Parse_End : constant Natural := Expression'Last;
377
378 ----------------------------
379 -- Subprograms for Create --
380 ----------------------------
381
382 procedure Emit (B : Character);
383 -- Output the Character B to the Program. If code-generation is
384 -- disabled, simply increments the program counter.
385
386 function Emit_Node (Op : Opcode) return Pointer;
387 -- If code-generation is enabled, Emit_Node outputs the
388 -- opcode Op and reserves space for a pointer to the next node.
389 -- Return value is the location of new opcode, i.e. old Emit_Ptr.
390
391 procedure Emit_Natural (IP : Pointer; N : Natural);
392 -- Split N on two characters at position IP
393
394 procedure Emit_Class (Bitmap : Character_Class);
395 -- Emits a character class
396
397 procedure Case_Emit (C : Character);
398 -- Emit C, after converting is to lower-case if the regular
399 -- expression is case insensitive.
400
401 procedure Parse
402 (Parenthesized : Boolean;
403 Flags : out Expression_Flags;
404 IP : out Pointer);
405 -- Parse regular expression, i.e. main body or parenthesized thing
406 -- Caller must absorb opening parenthesis.
407
408 procedure Parse_Branch
409 (Flags : out Expression_Flags;
410 First : Boolean;
411 IP : out Pointer);
412 -- Implements the concatenation operator and handles '|'
413 -- First should be true if this is the first item of the alternative.
414
415 procedure Parse_Piece
416 (Expr_Flags : out Expression_Flags;
417 IP : out Pointer);
418 -- Parse something followed by possible [*+?]
419
420 procedure Parse_Atom
421 (Expr_Flags : out Expression_Flags;
422 IP : out Pointer);
423 -- Parse_Atom is the lowest level parse procedure.
424 -- Optimization: gobbles an entire sequence of ordinary characters
425 -- so that it can turn them into a single node, which is smaller to
426 -- store and faster to run. Backslashed characters are exceptions,
427 -- each becoming a separate node; the code is simpler that way and
428 -- it's not worth fixing.
429
430 procedure Insert_Operator
431 (Op : Opcode;
432 Operand : Pointer;
433 Greedy : Boolean := True);
434 -- Insert_Operator inserts an operator in front of an
435 -- already-emitted operand and relocates the operand.
436 -- This applies to PLUS and STAR.
437 -- If Minmod is True, then the operator is non-greedy.
438
439 procedure Insert_Curly_Operator
440 (Op : Opcode;
441 Min : Natural;
442 Max : Natural;
443 Operand : Pointer;
444 Greedy : Boolean := True);
445 -- Insert an operator for CURLY ({Min}, {Min,} or {Min,Max}).
446 -- If Minmod is True, then the operator is non-greedy.
447
448 procedure Link_Tail (P, Val : Pointer);
449 -- Link_Tail sets the next-pointer at the end of a node chain
450
451 procedure Link_Operand_Tail (P, Val : Pointer);
452 -- Link_Tail on operand of first argument; noop if operand-less
453
454 function Next_Instruction (P : Pointer) return Pointer;
455 -- Dig the "next" pointer out of a node
456
457 procedure Fail (M : String);
458 pragma No_Return (Fail);
459 -- Fail with a diagnostic message, if possible
460
461 function Is_Curly_Operator (IP : Natural) return Boolean;
462 -- Return True if IP is looking at a '{' that is the beginning
463 -- of a curly operator, i.e. it matches {\d+,?\d*}
464
465 function Is_Mult (IP : Natural) return Boolean;
466 -- Return True if C is a regexp multiplier: '+', '*' or '?'
467
468 procedure Get_Curly_Arguments
469 (IP : Natural;
470 Min : out Natural;
471 Max : out Natural;
472 Greedy : out Boolean);
473 -- Parse the argument list for a curly operator.
474 -- It is assumed that IP is indeed pointing at a valid operator.
475 -- So what is IP and how come IP is not referenced in the body ???
476
477 procedure Parse_Character_Class (IP : out Pointer);
478 -- Parse a character class.
479 -- The calling subprogram should consume the opening '[' before.
480
481 procedure Parse_Literal
482 (Expr_Flags : out Expression_Flags;
483 IP : out Pointer);
484 -- Parse_Literal encodes a string of characters to be matched exactly
485
486 function Parse_Posix_Character_Class return Std_Class;
487 -- Parse a posix character class, like [:alpha:] or [:^alpha:].
488 -- The caller is supposed to absorb the opening [.
489
490 pragma Inline (Is_Mult);
491 pragma Inline (Emit_Natural);
492 pragma Inline (Parse_Character_Class); -- since used only once
493
494 ---------------
495 -- Case_Emit --
496 ---------------
497
498 procedure Case_Emit (C : Character) is
499 begin
500 if (Flags and Case_Insensitive) /= 0 then
501 Emit (To_Lower (C));
502
503 else
504 -- Dump current character
505
506 Emit (C);
507 end if;
508 end Case_Emit;
509
510 ----------
511 -- Emit --
512 ----------
513
514 procedure Emit (B : Character) is
515 begin
516 if Emit_Code then
517 Program (Emit_Ptr) := B;
518 end if;
519
520 Emit_Ptr := Emit_Ptr + 1;
521 end Emit;
522
523 ----------------
524 -- Emit_Class --
525 ----------------
526
527 procedure Emit_Class (Bitmap : Character_Class) is
528 subtype Program31 is Program_Data (0 .. 31);
529
530 function Convert is new Ada.Unchecked_Conversion
531 (Character_Class, Program31);
532
533 begin
534 if Emit_Code then
535 Program (Emit_Ptr .. Emit_Ptr + 31) := Convert (Bitmap);
536 end if;
537
538 Emit_Ptr := Emit_Ptr + 32;
539 end Emit_Class;
540
541 ------------------
542 -- Emit_Natural --
543 ------------------
544
545 procedure Emit_Natural (IP : Pointer; N : Natural) is
546 begin
547 if Emit_Code then
548 Program (IP + 1) := Character'Val (N / 256);
549 Program (IP) := Character'Val (N mod 256);
550 end if;
551 end Emit_Natural;
552
553 ---------------
554 -- Emit_Node --
555 ---------------
556
557 function Emit_Node (Op : Opcode) return Pointer is
558 Result : constant Pointer := Emit_Ptr;
559
560 begin
561 if Emit_Code then
562 Program (Emit_Ptr) := Character'Val (Opcode'Pos (Op));
563 Program (Emit_Ptr + 1) := ASCII.NUL;
564 Program (Emit_Ptr + 2) := ASCII.NUL;
565 end if;
566
567 Emit_Ptr := Emit_Ptr + 3;
568 return Result;
569 end Emit_Node;
570
571 ----------
572 -- Fail --
573 ----------
574
575 procedure Fail (M : String) is
576 begin
577 raise Expression_Error with M;
578 end Fail;
579
580 -------------------------
581 -- Get_Curly_Arguments --
582 -------------------------
583
584 procedure Get_Curly_Arguments
585 (IP : Natural;
586 Min : out Natural;
587 Max : out Natural;
588 Greedy : out Boolean)
589 is
590 pragma Unreferenced (IP);
591
592 Save_Pos : Natural := Parse_Pos + 1;
593
594 begin
595 Min := 0;
596 Max := Max_Curly_Repeat;
597
598 while Expression (Parse_Pos) /= '}'
599 and then Expression (Parse_Pos) /= ','
600 loop
601 Parse_Pos := Parse_Pos + 1;
602 end loop;
603
604 Min := Natural'Value (Expression (Save_Pos .. Parse_Pos - 1));
605
606 if Expression (Parse_Pos) = ',' then
607 Save_Pos := Parse_Pos + 1;
608 while Expression (Parse_Pos) /= '}' loop
609 Parse_Pos := Parse_Pos + 1;
610 end loop;
611
612 if Save_Pos /= Parse_Pos then
613 Max := Natural'Value (Expression (Save_Pos .. Parse_Pos - 1));
614 end if;
615
616 else
617 Max := Min;
618 end if;
619
620 if Parse_Pos < Expression'Last
621 and then Expression (Parse_Pos + 1) = '?'
622 then
623 Greedy := False;
624 Parse_Pos := Parse_Pos + 1;
625
626 else
627 Greedy := True;
628 end if;
629 end Get_Curly_Arguments;
630
631 ---------------------------
632 -- Insert_Curly_Operator --
633 ---------------------------
634
635 procedure Insert_Curly_Operator
636 (Op : Opcode;
637 Min : Natural;
638 Max : Natural;
639 Operand : Pointer;
640 Greedy : Boolean := True)
641 is
642 Dest : constant Pointer := Emit_Ptr;
643 Old : Pointer;
644 Size : Pointer := 7;
645
646 begin
647 -- If the operand is not greedy, insert an extra operand before it
648
649 if not Greedy then
650 Size := Size + 3;
651 end if;
652
653 -- Move the operand in the byte-compilation, so that we can insert
654 -- the operator before it.
655
656 if Emit_Code then
657 Program (Operand + Size .. Emit_Ptr + Size) :=
658 Program (Operand .. Emit_Ptr);
659 end if;
660
661 -- Insert the operator at the position previously occupied by the
662 -- operand.
663
664 Emit_Ptr := Operand;
665
666 if not Greedy then
667 Old := Emit_Node (MINMOD);
668 Link_Tail (Old, Old + 3);
669 end if;
670
671 Old := Emit_Node (Op);
672 Emit_Natural (Old + 3, Min);
673 Emit_Natural (Old + 5, Max);
674
675 Emit_Ptr := Dest + Size;
676 end Insert_Curly_Operator;
677
678 ---------------------
679 -- Insert_Operator --
680 ---------------------
681
682 procedure Insert_Operator
683 (Op : Opcode;
684 Operand : Pointer;
685 Greedy : Boolean := True)
686 is
687 Dest : constant Pointer := Emit_Ptr;
688 Old : Pointer;
689 Size : Pointer := 3;
690
691 Discard : Pointer;
692 pragma Warnings (Off, Discard);
693
694 begin
695 -- If not greedy, we have to emit another opcode first
696
697 if not Greedy then
698 Size := Size + 3;
699 end if;
700
701 -- Move the operand in the byte-compilation, so that we can insert
702 -- the operator before it.
703
704 if Emit_Code then
705 Program (Operand + Size .. Emit_Ptr + Size) :=
706 Program (Operand .. Emit_Ptr);
707 end if;
708
709 -- Insert the operator at the position previously occupied by the
710 -- operand.
711
712 Emit_Ptr := Operand;
713
714 if not Greedy then
715 Old := Emit_Node (MINMOD);
716 Link_Tail (Old, Old + 3);
717 end if;
718
719 Discard := Emit_Node (Op);
720 Emit_Ptr := Dest + Size;
721 end Insert_Operator;
722
723 -----------------------
724 -- Is_Curly_Operator --
725 -----------------------
726
727 function Is_Curly_Operator (IP : Natural) return Boolean is
728 Scan : Natural := IP;
729
730 begin
731 if Expression (Scan) /= '{'
732 or else Scan + 2 > Expression'Last
733 or else not Is_Digit (Expression (Scan + 1))
734 then
735 return False;
736 end if;
737
738 Scan := Scan + 1;
739
740 -- The first digit
741
742 loop
743 Scan := Scan + 1;
744
745 if Scan > Expression'Last then
746 return False;
747 end if;
748
749 exit when not Is_Digit (Expression (Scan));
750 end loop;
751
752 if Expression (Scan) = ',' then
753 loop
754 Scan := Scan + 1;
755
756 if Scan > Expression'Last then
757 return False;
758 end if;
759
760 exit when not Is_Digit (Expression (Scan));
761 end loop;
762 end if;
763
764 return Expression (Scan) = '}';
765 end Is_Curly_Operator;
766
767 -------------
768 -- Is_Mult --
769 -------------
770
771 function Is_Mult (IP : Natural) return Boolean is
772 C : constant Character := Expression (IP);
773
774 begin
775 return C = '*'
776 or else C = '+'
777 or else C = '?'
778 or else (C = '{' and then Is_Curly_Operator (IP));
779 end Is_Mult;
780
781 -----------------------
782 -- Link_Operand_Tail --
783 -----------------------
784
785 procedure Link_Operand_Tail (P, Val : Pointer) is
786 begin
787 if Emit_Code and then Program (P) = BRANCH then
788 Link_Tail (Operand (P), Val);
789 end if;
790 end Link_Operand_Tail;
791
792 ---------------
793 -- Link_Tail --
794 ---------------
795
796 procedure Link_Tail (P, Val : Pointer) is
797 Scan : Pointer;
798 Temp : Pointer;
799 Offset : Pointer;
800
801 begin
802 if not Emit_Code then
803 return;
804 end if;
805
806 -- Find last node
807
808 Scan := P;
809 loop
810 Temp := Next_Instruction (Scan);
811 exit when Temp = 0;
812 Scan := Temp;
813 end loop;
814
815 Offset := Val - Scan;
816
817 Emit_Natural (Scan + 1, Natural (Offset));
818 end Link_Tail;
819
820 ----------------------
821 -- Next_Instruction --
822 ----------------------
823
824 function Next_Instruction (P : Pointer) return Pointer is
825 Offset : Pointer;
826
827 begin
828 if not Emit_Code then
829 return 0;
830 end if;
831
832 Offset := Get_Next_Offset (Program, P);
833
834 if Offset = 0 then
835 return 0;
836 end if;
837
838 return P + Offset;
839 end Next_Instruction;
840
841 -----------
842 -- Parse --
843 -----------
844
845 -- Combining parenthesis handling with the base level
846 -- of regular expression is a trifle forced, but the
847 -- need to tie the tails of the branches to what follows
848 -- makes it hard to avoid.
849
850 procedure Parse
851 (Parenthesized : Boolean;
852 Flags : out Expression_Flags;
853 IP : out Pointer)
854 is
855 E : String renames Expression;
856 Br : Pointer;
857 Ender : Pointer;
858 Par_No : Natural;
859 New_Flags : Expression_Flags;
860 Have_Branch : Boolean := False;
861
862 begin
863 Flags := (Has_Width => True, others => False); -- Tentatively
864
865 -- Make an OPEN node, if parenthesized
866
867 if Parenthesized then
868 if Matcher.Paren_Count > Max_Paren_Count then
869 Fail ("too many ()");
870 end if;
871
872 Par_No := Matcher.Paren_Count + 1;
873 Matcher.Paren_Count := Matcher.Paren_Count + 1;
874 IP := Emit_Node (OPEN);
875 Emit (Character'Val (Par_No));
876
877 else
878 IP := 0;
879 Par_No := 0;
880 end if;
881
882 -- Pick up the branches, linking them together
883
884 Parse_Branch (New_Flags, True, Br);
885
886 if Br = 0 then
887 IP := 0;
888 return;
889 end if;
890
891 if Parse_Pos <= Parse_End
892 and then E (Parse_Pos) = '|'
893 then
894 Insert_Operator (BRANCH, Br);
895 Have_Branch := True;
896 end if;
897
898 if IP /= 0 then
899 Link_Tail (IP, Br); -- OPEN -> first
900 else
901 IP := Br;
902 end if;
903
904 if not New_Flags.Has_Width then
905 Flags.Has_Width := False;
906 end if;
907
908 Flags.SP_Start := Flags.SP_Start or New_Flags.SP_Start;
909
910 while Parse_Pos <= Parse_End
911 and then (E (Parse_Pos) = '|')
912 loop
913 Parse_Pos := Parse_Pos + 1;
914 Parse_Branch (New_Flags, False, Br);
915
916 if Br = 0 then
917 IP := 0;
918 return;
919 end if;
920
921 Link_Tail (IP, Br); -- BRANCH -> BRANCH
922
923 if not New_Flags.Has_Width then
924 Flags.Has_Width := False;
925 end if;
926
927 Flags.SP_Start := Flags.SP_Start or New_Flags.SP_Start;
928 end loop;
929
930 -- Make a closing node, and hook it on the end
931
932 if Parenthesized then
933 Ender := Emit_Node (CLOSE);
934 Emit (Character'Val (Par_No));
935 else
936 Ender := Emit_Node (EOP);
937 end if;
938
939 Link_Tail (IP, Ender);
940
941 if Have_Branch then
942
943 -- Hook the tails of the branches to the closing node
944
945 Br := IP;
946 loop
947 exit when Br = 0;
948 Link_Operand_Tail (Br, Ender);
949 Br := Next_Instruction (Br);
950 end loop;
951 end if;
952
953 -- Check for proper termination
954
955 if Parenthesized then
956 if Parse_Pos > Parse_End or else E (Parse_Pos) /= ')' then
957 Fail ("unmatched ()");
958 end if;
959
960 Parse_Pos := Parse_Pos + 1;
961
962 elsif Parse_Pos <= Parse_End then
963 if E (Parse_Pos) = ')' then
964 Fail ("unmatched ()");
965 else
966 Fail ("junk on end"); -- "Can't happen"
967 end if;
968 end if;
969 end Parse;
970
971 ----------------
972 -- Parse_Atom --
973 ----------------
974
975 procedure Parse_Atom
976 (Expr_Flags : out Expression_Flags;
977 IP : out Pointer)
978 is
979 C : Character;
980
981 begin
982 -- Tentatively set worst expression case
983
984 Expr_Flags := Worst_Expression;
985
986 C := Expression (Parse_Pos);
987 Parse_Pos := Parse_Pos + 1;
988
989 case (C) is
990 when '^' =>
991 if (Flags and Multiple_Lines) /= 0 then
992 IP := Emit_Node (MBOL);
993 elsif (Flags and Single_Line) /= 0 then
994 IP := Emit_Node (SBOL);
995 else
996 IP := Emit_Node (BOL);
997 end if;
998
999 when '$' =>
1000 if (Flags and Multiple_Lines) /= 0 then
1001 IP := Emit_Node (MEOL);
1002 elsif (Flags and Single_Line) /= 0 then
1003 IP := Emit_Node (SEOL);
1004 else
1005 IP := Emit_Node (EOL);
1006 end if;
1007
1008 when '.' =>
1009 if (Flags and Single_Line) /= 0 then
1010 IP := Emit_Node (SANY);
1011 else
1012 IP := Emit_Node (ANY);
1013 end if;
1014
1015 Expr_Flags.Has_Width := True;
1016 Expr_Flags.Simple := True;
1017
1018 when '[' =>
1019 Parse_Character_Class (IP);
1020 Expr_Flags.Has_Width := True;
1021 Expr_Flags.Simple := True;
1022
1023 when '(' =>
1024 declare
1025 New_Flags : Expression_Flags;
1026
1027 begin
1028 Parse (True, New_Flags, IP);
1029
1030 if IP = 0 then
1031 return;
1032 end if;
1033
1034 Expr_Flags.Has_Width :=
1035 Expr_Flags.Has_Width or New_Flags.Has_Width;
1036 Expr_Flags.SP_Start :=
1037 Expr_Flags.SP_Start or New_Flags.SP_Start;
1038 end;
1039
1040 when '|' | ASCII.LF | ')' =>
1041 Fail ("internal urp"); -- Supposed to be caught earlier
1042
1043 when '?' | '+' | '*' =>
1044 Fail (C & " follows nothing");
1045
1046 when '{' =>
1047 if Is_Curly_Operator (Parse_Pos - 1) then
1048 Fail (C & " follows nothing");
1049 else
1050 Parse_Literal (Expr_Flags, IP);
1051 end if;
1052
1053 when '\' =>
1054 if Parse_Pos > Parse_End then
1055 Fail ("trailing \");
1056 end if;
1057
1058 Parse_Pos := Parse_Pos + 1;
1059
1060 case Expression (Parse_Pos - 1) is
1061 when 'b' =>
1062 IP := Emit_Node (BOUND);
1063
1064 when 'B' =>
1065 IP := Emit_Node (NBOUND);
1066
1067 when 's' =>
1068 IP := Emit_Node (SPACE);
1069 Expr_Flags.Simple := True;
1070 Expr_Flags.Has_Width := True;
1071
1072 when 'S' =>
1073 IP := Emit_Node (NSPACE);
1074 Expr_Flags.Simple := True;
1075 Expr_Flags.Has_Width := True;
1076
1077 when 'd' =>
1078 IP := Emit_Node (DIGIT);
1079 Expr_Flags.Simple := True;
1080 Expr_Flags.Has_Width := True;
1081
1082 when 'D' =>
1083 IP := Emit_Node (NDIGIT);
1084 Expr_Flags.Simple := True;
1085 Expr_Flags.Has_Width := True;
1086
1087 when 'w' =>
1088 IP := Emit_Node (ALNUM);
1089 Expr_Flags.Simple := True;
1090 Expr_Flags.Has_Width := True;
1091
1092 when 'W' =>
1093 IP := Emit_Node (NALNUM);
1094 Expr_Flags.Simple := True;
1095 Expr_Flags.Has_Width := True;
1096
1097 when 'A' =>
1098 IP := Emit_Node (SBOL);
1099
1100 when 'G' =>
1101 IP := Emit_Node (SEOL);
1102
1103 when '0' .. '9' =>
1104 IP := Emit_Node (REFF);
1105
1106 declare
1107 Save : constant Natural := Parse_Pos - 1;
1108
1109 begin
1110 while Parse_Pos <= Expression'Last
1111 and then Is_Digit (Expression (Parse_Pos))
1112 loop
1113 Parse_Pos := Parse_Pos + 1;
1114 end loop;
1115
1116 Emit (Character'Val (Natural'Value
1117 (Expression (Save .. Parse_Pos - 1))));
1118 end;
1119
1120 when others =>
1121 Parse_Pos := Parse_Pos - 1;
1122 Parse_Literal (Expr_Flags, IP);
1123 end case;
1124
1125 when others =>
1126 Parse_Literal (Expr_Flags, IP);
1127 end case;
1128 end Parse_Atom;
1129
1130 ------------------
1131 -- Parse_Branch --
1132 ------------------
1133
1134 procedure Parse_Branch
1135 (Flags : out Expression_Flags;
1136 First : Boolean;
1137 IP : out Pointer)
1138 is
1139 E : String renames Expression;
1140 Chain : Pointer;
1141 Last : Pointer;
1142 New_Flags : Expression_Flags;
1143
1144 Discard : Pointer;
1145 pragma Warnings (Off, Discard);
1146
1147 begin
1148 Flags := Worst_Expression; -- Tentatively
1149
1150 if First then
1151 IP := Emit_Ptr;
1152 else
1153 IP := Emit_Node (BRANCH);
1154 end if;
1155
1156 Chain := 0;
1157
1158 while Parse_Pos <= Parse_End
1159 and then E (Parse_Pos) /= ')'
1160 and then E (Parse_Pos) /= ASCII.LF
1161 and then E (Parse_Pos) /= '|'
1162 loop
1163 Parse_Piece (New_Flags, Last);
1164
1165 if Last = 0 then
1166 IP := 0;
1167 return;
1168 end if;
1169
1170 Flags.Has_Width := Flags.Has_Width or New_Flags.Has_Width;
1171
1172 if Chain = 0 then -- First piece
1173 Flags.SP_Start := Flags.SP_Start or New_Flags.SP_Start;
1174 else
1175 Link_Tail (Chain, Last);
1176 end if;
1177
1178 Chain := Last;
1179 end loop;
1180
1181 -- Case where loop ran zero CURLY
1182
1183 if Chain = 0 then
1184 Discard := Emit_Node (NOTHING);
1185 end if;
1186 end Parse_Branch;
1187
1188 ---------------------------
1189 -- Parse_Character_Class --
1190 ---------------------------
1191
1192 procedure Parse_Character_Class (IP : out Pointer) is
1193 Bitmap : Character_Class;
1194 Invert : Boolean := False;
1195 In_Range : Boolean := False;
1196 Named_Class : Std_Class := ANYOF_NONE;
1197 Value : Character;
1198 Last_Value : Character := ASCII.NUL;
1199
1200 begin
1201 Reset_Class (Bitmap);
1202
1203 -- Do we have an invert character class ?
1204
1205 if Parse_Pos <= Parse_End
1206 and then Expression (Parse_Pos) = '^'
1207 then
1208 Invert := True;
1209 Parse_Pos := Parse_Pos + 1;
1210 end if;
1211
1212 -- First character can be ] or - without closing the class
1213
1214 if Parse_Pos <= Parse_End
1215 and then (Expression (Parse_Pos) = ']'
1216 or else Expression (Parse_Pos) = '-')
1217 then
1218 Set_In_Class (Bitmap, Expression (Parse_Pos));
1219 Parse_Pos := Parse_Pos + 1;
1220 end if;
1221
1222 -- While we don't have the end of the class
1223
1224 while Parse_Pos <= Parse_End
1225 and then Expression (Parse_Pos) /= ']'
1226 loop
1227 Named_Class := ANYOF_NONE;
1228 Value := Expression (Parse_Pos);
1229 Parse_Pos := Parse_Pos + 1;
1230
1231 -- Do we have a Posix character class
1232 if Value = '[' then
1233 Named_Class := Parse_Posix_Character_Class;
1234
1235 elsif Value = '\' then
1236 if Parse_Pos = Parse_End then
1237 Fail ("Trailing \");
1238 end if;
1239 Value := Expression (Parse_Pos);
1240 Parse_Pos := Parse_Pos + 1;
1241
1242 case Value is
1243 when 'w' => Named_Class := ANYOF_ALNUM;
1244 when 'W' => Named_Class := ANYOF_NALNUM;
1245 when 's' => Named_Class := ANYOF_SPACE;
1246 when 'S' => Named_Class := ANYOF_NSPACE;
1247 when 'd' => Named_Class := ANYOF_DIGIT;
1248 when 'D' => Named_Class := ANYOF_NDIGIT;
1249 when 'n' => Value := ASCII.LF;
1250 when 'r' => Value := ASCII.CR;
1251 when 't' => Value := ASCII.HT;
1252 when 'f' => Value := ASCII.FF;
1253 when 'e' => Value := ASCII.ESC;
1254 when 'a' => Value := ASCII.BEL;
1255
1256 -- when 'x' => ??? hexadecimal value
1257 -- when 'c' => ??? control character
1258 -- when '0'..'9' => ??? octal character
1259
1260 when others => null;
1261 end case;
1262 end if;
1263
1264 -- Do we have a character class?
1265
1266 if Named_Class /= ANYOF_NONE then
1267
1268 -- A range like 'a-\d' or 'a-[:digit:] is not a range
1269
1270 if In_Range then
1271 Set_In_Class (Bitmap, Last_Value);
1272 Set_In_Class (Bitmap, '-');
1273 In_Range := False;
1274 end if;
1275
1276 -- Expand the range
1277
1278 case Named_Class is
1279 when ANYOF_NONE => null;
1280
1281 when ANYOF_ALNUM | ANYOF_ALNUMC =>
1282 for Value in Class_Byte'Range loop
1283 if Is_Alnum (Character'Val (Value)) then
1284 Set_In_Class (Bitmap, Character'Val (Value));
1285 end if;
1286 end loop;
1287
1288 when ANYOF_NALNUM | ANYOF_NALNUMC =>
1289 for Value in Class_Byte'Range loop
1290 if not Is_Alnum (Character'Val (Value)) then
1291 Set_In_Class (Bitmap, Character'Val (Value));
1292 end if;
1293 end loop;
1294
1295 when ANYOF_SPACE =>
1296 for Value in Class_Byte'Range loop
1297 if Is_White_Space (Character'Val (Value)) then
1298 Set_In_Class (Bitmap, Character'Val (Value));
1299 end if;
1300 end loop;
1301
1302 when ANYOF_NSPACE =>
1303 for Value in Class_Byte'Range loop
1304 if not Is_White_Space (Character'Val (Value)) then
1305 Set_In_Class (Bitmap, Character'Val (Value));
1306 end if;
1307 end loop;
1308
1309 when ANYOF_DIGIT =>
1310 for Value in Class_Byte'Range loop
1311 if Is_Digit (Character'Val (Value)) then
1312 Set_In_Class (Bitmap, Character'Val (Value));
1313 end if;
1314 end loop;
1315
1316 when ANYOF_NDIGIT =>
1317 for Value in Class_Byte'Range loop
1318 if not Is_Digit (Character'Val (Value)) then
1319 Set_In_Class (Bitmap, Character'Val (Value));
1320 end if;
1321 end loop;
1322
1323 when ANYOF_ALPHA =>
1324 for Value in Class_Byte'Range loop
1325 if Is_Letter (Character'Val (Value)) then
1326 Set_In_Class (Bitmap, Character'Val (Value));
1327 end if;
1328 end loop;
1329
1330 when ANYOF_NALPHA =>
1331 for Value in Class_Byte'Range loop
1332 if not Is_Letter (Character'Val (Value)) then
1333 Set_In_Class (Bitmap, Character'Val (Value));
1334 end if;
1335 end loop;
1336
1337 when ANYOF_ASCII =>
1338 for Value in 0 .. 127 loop
1339 Set_In_Class (Bitmap, Character'Val (Value));
1340 end loop;
1341
1342 when ANYOF_NASCII =>
1343 for Value in 128 .. 255 loop
1344 Set_In_Class (Bitmap, Character'Val (Value));
1345 end loop;
1346
1347 when ANYOF_CNTRL =>
1348 for Value in Class_Byte'Range loop
1349 if Is_Control (Character'Val (Value)) then
1350 Set_In_Class (Bitmap, Character'Val (Value));
1351 end if;
1352 end loop;
1353
1354 when ANYOF_NCNTRL =>
1355 for Value in Class_Byte'Range loop
1356 if not Is_Control (Character'Val (Value)) then
1357 Set_In_Class (Bitmap, Character'Val (Value));
1358 end if;
1359 end loop;
1360
1361 when ANYOF_GRAPH =>
1362 for Value in Class_Byte'Range loop
1363 if Is_Graphic (Character'Val (Value)) then
1364 Set_In_Class (Bitmap, Character'Val (Value));
1365 end if;
1366 end loop;
1367
1368 when ANYOF_NGRAPH =>
1369 for Value in Class_Byte'Range loop
1370 if not Is_Graphic (Character'Val (Value)) then
1371 Set_In_Class (Bitmap, Character'Val (Value));
1372 end if;
1373 end loop;
1374
1375 when ANYOF_LOWER =>
1376 for Value in Class_Byte'Range loop
1377 if Is_Lower (Character'Val (Value)) then
1378 Set_In_Class (Bitmap, Character'Val (Value));
1379 end if;
1380 end loop;
1381
1382 when ANYOF_NLOWER =>
1383 for Value in Class_Byte'Range loop
1384 if not Is_Lower (Character'Val (Value)) then
1385 Set_In_Class (Bitmap, Character'Val (Value));
1386 end if;
1387 end loop;
1388
1389 when ANYOF_PRINT =>
1390 for Value in Class_Byte'Range loop
1391 if Is_Printable (Character'Val (Value)) then
1392 Set_In_Class (Bitmap, Character'Val (Value));
1393 end if;
1394 end loop;
1395
1396 when ANYOF_NPRINT =>
1397 for Value in Class_Byte'Range loop
1398 if not Is_Printable (Character'Val (Value)) then
1399 Set_In_Class (Bitmap, Character'Val (Value));
1400 end if;
1401 end loop;
1402
1403 when ANYOF_PUNCT =>
1404 for Value in Class_Byte'Range loop
1405 if Is_Printable (Character'Val (Value))
1406 and then not Is_White_Space (Character'Val (Value))
1407 and then not Is_Alnum (Character'Val (Value))
1408 then
1409 Set_In_Class (Bitmap, Character'Val (Value));
1410 end if;
1411 end loop;
1412
1413 when ANYOF_NPUNCT =>
1414 for Value in Class_Byte'Range loop
1415 if not Is_Printable (Character'Val (Value))
1416 or else Is_White_Space (Character'Val (Value))
1417 or else Is_Alnum (Character'Val (Value))
1418 then
1419 Set_In_Class (Bitmap, Character'Val (Value));
1420 end if;
1421 end loop;
1422
1423 when ANYOF_UPPER =>
1424 for Value in Class_Byte'Range loop
1425 if Is_Upper (Character'Val (Value)) then
1426 Set_In_Class (Bitmap, Character'Val (Value));
1427 end if;
1428 end loop;
1429
1430 when ANYOF_NUPPER =>
1431 for Value in Class_Byte'Range loop
1432 if not Is_Upper (Character'Val (Value)) then
1433 Set_In_Class (Bitmap, Character'Val (Value));
1434 end if;
1435 end loop;
1436
1437 when ANYOF_XDIGIT =>
1438 for Value in Class_Byte'Range loop
1439 if Is_Hexadecimal_Digit (Character'Val (Value)) then
1440 Set_In_Class (Bitmap, Character'Val (Value));
1441 end if;
1442 end loop;
1443
1444 when ANYOF_NXDIGIT =>
1445 for Value in Class_Byte'Range loop
1446 if not Is_Hexadecimal_Digit
1447 (Character'Val (Value))
1448 then
1449 Set_In_Class (Bitmap, Character'Val (Value));
1450 end if;
1451 end loop;
1452
1453 end case;
1454
1455 -- Not a character range
1456
1457 elsif not In_Range then
1458 Last_Value := Value;
1459
1460 if Parse_Pos > Expression'Last then
1461 Fail ("Empty character class []");
1462 end if;
1463
1464 if Expression (Parse_Pos) = '-'
1465 and then Parse_Pos < Parse_End
1466 and then Expression (Parse_Pos + 1) /= ']'
1467 then
1468 Parse_Pos := Parse_Pos + 1;
1469
1470 -- Do we have a range like '\d-a' and '[:space:]-a'
1471 -- which is not a real range
1472
1473 if Named_Class /= ANYOF_NONE then
1474 Set_In_Class (Bitmap, '-');
1475 else
1476 In_Range := True;
1477 end if;
1478
1479 else
1480 Set_In_Class (Bitmap, Value);
1481
1482 end if;
1483
1484 -- Else in a character range
1485
1486 else
1487 if Last_Value > Value then
1488 Fail ("Invalid Range [" & Last_Value'Img
1489 & "-" & Value'Img & "]");
1490 end if;
1491
1492 while Last_Value <= Value loop
1493 Set_In_Class (Bitmap, Last_Value);
1494 Last_Value := Character'Succ (Last_Value);
1495 end loop;
1496
1497 In_Range := False;
1498
1499 end if;
1500
1501 end loop;
1502
1503 -- Optimize case-insensitive ranges (put the upper case or lower
1504 -- case character into the bitmap)
1505
1506 if (Flags and Case_Insensitive) /= 0 then
1507 for C in Character'Range loop
1508 if Get_From_Class (Bitmap, C) then
1509 Set_In_Class (Bitmap, To_Lower (C));
1510 Set_In_Class (Bitmap, To_Upper (C));
1511 end if;
1512 end loop;
1513 end if;
1514
1515 -- Optimize inverted classes
1516
1517 if Invert then
1518 for J in Bitmap'Range loop
1519 Bitmap (J) := not Bitmap (J);
1520 end loop;
1521 end if;
1522
1523 Parse_Pos := Parse_Pos + 1;
1524
1525 -- Emit the class
1526
1527 IP := Emit_Node (ANYOF);
1528 Emit_Class (Bitmap);
1529 end Parse_Character_Class;
1530
1531 -------------------
1532 -- Parse_Literal --
1533 -------------------
1534
1535 -- This is a bit tricky due to quoted chars and due to
1536 -- the multiplier characters '*', '+', and '?' that
1537 -- take the SINGLE char previous as their operand.
1538
1539 -- On entry, the character at Parse_Pos - 1 is going to go
1540 -- into the string, no matter what it is. It could be
1541 -- following a \ if Parse_Atom was entered from the '\' case.
1542
1543 -- Basic idea is to pick up a good char in C and examine
1544 -- the next char. If Is_Mult (C) then twiddle, if it's a \
1545 -- then frozzle and if it's another magic char then push C and
1546 -- terminate the string. If none of the above, push C on the
1547 -- string and go around again.
1548
1549 -- Start_Pos is used to remember where "the current character"
1550 -- starts in the string, if due to an Is_Mult we need to back
1551 -- up and put the current char in a separate 1-character string.
1552 -- When Start_Pos is 0, C is the only char in the string;
1553 -- this is used in Is_Mult handling, and in setting the SIMPLE
1554 -- flag at the end.
1555
1556 procedure Parse_Literal
1557 (Expr_Flags : out Expression_Flags;
1558 IP : out Pointer)
1559 is
1560 Start_Pos : Natural := 0;
1561 C : Character;
1562 Length_Ptr : Pointer;
1563
1564 Has_Special_Operator : Boolean := False;
1565
1566 begin
1567 Parse_Pos := Parse_Pos - 1; -- Look at current character
1568
1569 if (Flags and Case_Insensitive) /= 0 then
1570 IP := Emit_Node (EXACTF);
1571 else
1572 IP := Emit_Node (EXACT);
1573 end if;
1574
1575 Length_Ptr := Emit_Ptr;
1576 Emit_Ptr := String_Operand (IP);
1577
1578 Parse_Loop :
1579 loop
1580 C := Expression (Parse_Pos); -- Get current character
1581
1582 case C is
1583 when '.' | '[' | '(' | ')' | '|' | ASCII.LF | '$' | '^' =>
1584
1585 if Start_Pos = 0 then
1586 Start_Pos := Parse_Pos;
1587 Emit (C); -- First character is always emitted
1588 else
1589 exit Parse_Loop; -- Else we are done
1590 end if;
1591
1592 when '?' | '+' | '*' | '{' =>
1593
1594 if Start_Pos = 0 then
1595 Start_Pos := Parse_Pos;
1596 Emit (C); -- First character is always emitted
1597
1598 -- Are we looking at an operator, or is this
1599 -- simply a normal character ?
1600
1601 elsif not Is_Mult (Parse_Pos) then
1602 Start_Pos := Parse_Pos;
1603 Case_Emit (C);
1604
1605 else
1606 -- We've got something like "abc?d". Mark this as a
1607 -- special case. What we want to emit is a first
1608 -- constant string for "ab", then one for "c" that will
1609 -- ultimately be transformed with a CURLY operator, A
1610 -- special case has to be handled for "a?", since there
1611 -- is no initial string to emit.
1612
1613 Has_Special_Operator := True;
1614 exit Parse_Loop;
1615 end if;
1616
1617 when '\' =>
1618 Start_Pos := Parse_Pos;
1619
1620 if Parse_Pos = Parse_End then
1621 Fail ("Trailing \");
1622
1623 else
1624 case Expression (Parse_Pos + 1) is
1625 when 'b' | 'B' | 's' | 'S' | 'd' | 'D'
1626 | 'w' | 'W' | '0' .. '9' | 'G' | 'A'
1627 => exit Parse_Loop;
1628 when 'n' => Emit (ASCII.LF);
1629 when 't' => Emit (ASCII.HT);
1630 when 'r' => Emit (ASCII.CR);
1631 when 'f' => Emit (ASCII.FF);
1632 when 'e' => Emit (ASCII.ESC);
1633 when 'a' => Emit (ASCII.BEL);
1634 when others => Emit (Expression (Parse_Pos + 1));
1635 end case;
1636
1637 Parse_Pos := Parse_Pos + 1;
1638 end if;
1639
1640 when others =>
1641 Start_Pos := Parse_Pos;
1642 Case_Emit (C);
1643 end case;
1644
1645 exit Parse_Loop when Emit_Ptr - Length_Ptr = 254;
1646
1647 Parse_Pos := Parse_Pos + 1;
1648
1649 exit Parse_Loop when Parse_Pos > Parse_End;
1650 end loop Parse_Loop;
1651
1652 -- Is the string followed by a '*+?{' operator ? If yes, and if there
1653 -- is an initial string to emit, do it now.
1654
1655 if Has_Special_Operator
1656 and then Emit_Ptr >= Length_Ptr + 3
1657 then
1658 Emit_Ptr := Emit_Ptr - 1;
1659 Parse_Pos := Start_Pos;
1660 end if;
1661
1662 if Emit_Code then
1663 Program (Length_Ptr) := Character'Val (Emit_Ptr - Length_Ptr - 2);
1664 end if;
1665
1666 Expr_Flags.Has_Width := True;
1667
1668 -- Slight optimization when there is a single character
1669
1670 if Emit_Ptr = Length_Ptr + 2 then
1671 Expr_Flags.Simple := True;
1672 end if;
1673 end Parse_Literal;
1674
1675 -----------------
1676 -- Parse_Piece --
1677 -----------------
1678
1679 -- Note that the branching code sequences used for '?' and the
1680 -- general cases of '*' and + are somewhat optimized: they use
1681 -- the same NOTHING node as both the endmarker for their branch
1682 -- list and the body of the last branch. It might seem that
1683 -- this node could be dispensed with entirely, but the endmarker
1684 -- role is not redundant.
1685
1686 procedure Parse_Piece
1687 (Expr_Flags : out Expression_Flags;
1688 IP : out Pointer)
1689 is
1690 Op : Character;
1691 New_Flags : Expression_Flags;
1692 Greedy : Boolean := True;
1693
1694 begin
1695 Parse_Atom (New_Flags, IP);
1696
1697 if IP = 0 then
1698 return;
1699 end if;
1700
1701 if Parse_Pos > Parse_End
1702 or else not Is_Mult (Parse_Pos)
1703 then
1704 Expr_Flags := New_Flags;
1705 return;
1706 end if;
1707
1708 Op := Expression (Parse_Pos);
1709
1710 if Op /= '+' then
1711 Expr_Flags := (SP_Start => True, others => False);
1712 else
1713 Expr_Flags := (Has_Width => True, others => False);
1714 end if;
1715
1716 -- Detect non greedy operators in the easy cases
1717
1718 if Op /= '{'
1719 and then Parse_Pos + 1 <= Parse_End
1720 and then Expression (Parse_Pos + 1) = '?'
1721 then
1722 Greedy := False;
1723 Parse_Pos := Parse_Pos + 1;
1724 end if;
1725
1726 -- Generate the byte code
1727
1728 case Op is
1729 when '*' =>
1730
1731 if New_Flags.Simple then
1732 Insert_Operator (STAR, IP, Greedy);
1733 else
1734 Link_Tail (IP, Emit_Node (WHILEM));
1735 Insert_Curly_Operator
1736 (CURLYX, 0, Max_Curly_Repeat, IP, Greedy);
1737 Link_Tail (IP, Emit_Node (NOTHING));
1738 end if;
1739
1740 when '+' =>
1741
1742 if New_Flags.Simple then
1743 Insert_Operator (PLUS, IP, Greedy);
1744 else
1745 Link_Tail (IP, Emit_Node (WHILEM));
1746 Insert_Curly_Operator
1747 (CURLYX, 1, Max_Curly_Repeat, IP, Greedy);
1748 Link_Tail (IP, Emit_Node (NOTHING));
1749 end if;
1750
1751 when '?' =>
1752 if New_Flags.Simple then
1753 Insert_Curly_Operator (CURLY, 0, 1, IP, Greedy);
1754 else
1755 Link_Tail (IP, Emit_Node (WHILEM));
1756 Insert_Curly_Operator (CURLYX, 0, 1, IP, Greedy);
1757 Link_Tail (IP, Emit_Node (NOTHING));
1758 end if;
1759
1760 when '{' =>
1761 declare
1762 Min, Max : Natural;
1763
1764 begin
1765 Get_Curly_Arguments (Parse_Pos, Min, Max, Greedy);
1766
1767 if New_Flags.Simple then
1768 Insert_Curly_Operator (CURLY, Min, Max, IP, Greedy);
1769 else
1770 Link_Tail (IP, Emit_Node (WHILEM));
1771 Insert_Curly_Operator (CURLYX, Min, Max, IP, Greedy);
1772 Link_Tail (IP, Emit_Node (NOTHING));
1773 end if;
1774 end;
1775
1776 when others =>
1777 null;
1778 end case;
1779
1780 Parse_Pos := Parse_Pos + 1;
1781
1782 if Parse_Pos <= Parse_End
1783 and then Is_Mult (Parse_Pos)
1784 then
1785 Fail ("nested *+{");
1786 end if;
1787 end Parse_Piece;
1788
1789 ---------------------------------
1790 -- Parse_Posix_Character_Class --
1791 ---------------------------------
1792
1793 function Parse_Posix_Character_Class return Std_Class is
1794 Invert : Boolean := False;
1795 Class : Std_Class := ANYOF_NONE;
1796 E : String renames Expression;
1797
1798 -- Class names. Note that code assumes that the length of all
1799 -- classes starting with the same letter have the same length.
1800
1801 Alnum : constant String := "alnum:]";
1802 Alpha : constant String := "alpha:]";
1803 Ascii_C : constant String := "ascii:]";
1804 Cntrl : constant String := "cntrl:]";
1805 Digit : constant String := "digit:]";
1806 Graph : constant String := "graph:]";
1807 Lower : constant String := "lower:]";
1808 Print : constant String := "print:]";
1809 Punct : constant String := "punct:]";
1810 Space : constant String := "space:]";
1811 Upper : constant String := "upper:]";
1812 Word : constant String := "word:]";
1813 Xdigit : constant String := "xdigit:]";
1814
1815 begin
1816 -- Case of character class specified
1817
1818 if Parse_Pos <= Parse_End
1819 and then Expression (Parse_Pos) = ':'
1820 then
1821 Parse_Pos := Parse_Pos + 1;
1822
1823 -- Do we have something like: [[:^alpha:]]
1824
1825 if Parse_Pos <= Parse_End
1826 and then Expression (Parse_Pos) = '^'
1827 then
1828 Invert := True;
1829 Parse_Pos := Parse_Pos + 1;
1830 end if;
1831
1832 -- Check for class names based on first letter
1833
1834 case Expression (Parse_Pos) is
1835 when 'a' =>
1836
1837 -- All 'a' classes have the same length (Alnum'Length)
1838
1839 if Parse_Pos + Alnum'Length - 1 <= Parse_End then
1840 if
1841 E (Parse_Pos .. Parse_Pos + Alnum'Length - 1) = Alnum
1842 then
1843 if Invert then
1844 Class := ANYOF_NALNUMC;
1845 else
1846 Class := ANYOF_ALNUMC;
1847 end if;
1848
1849 Parse_Pos := Parse_Pos + Alnum'Length;
1850
1851 elsif
1852 E (Parse_Pos .. Parse_Pos + Alpha'Length - 1) = Alpha
1853 then
1854 if Invert then
1855 Class := ANYOF_NALPHA;
1856 else
1857 Class := ANYOF_ALPHA;
1858 end if;
1859
1860 Parse_Pos := Parse_Pos + Alpha'Length;
1861
1862 elsif E (Parse_Pos .. Parse_Pos + Ascii_C'Length - 1) =
1863 Ascii_C
1864 then
1865 if Invert then
1866 Class := ANYOF_NASCII;
1867 else
1868 Class := ANYOF_ASCII;
1869 end if;
1870
1871 Parse_Pos := Parse_Pos + Ascii_C'Length;
1872
1873 else
1874 Fail ("Invalid character class: " & E);
1875 end if;
1876
1877 else
1878 Fail ("Invalid character class: " & E);
1879 end if;
1880
1881 when 'c' =>
1882 if Parse_Pos + Cntrl'Length - 1 <= Parse_End
1883 and then
1884 E (Parse_Pos .. Parse_Pos + Cntrl'Length - 1) = Cntrl
1885 then
1886 if Invert then
1887 Class := ANYOF_NCNTRL;
1888 else
1889 Class := ANYOF_CNTRL;
1890 end if;
1891
1892 Parse_Pos := Parse_Pos + Cntrl'Length;
1893
1894 else
1895 Fail ("Invalid character class: " & E);
1896 end if;
1897
1898 when 'd' =>
1899 if Parse_Pos + Digit'Length - 1 <= Parse_End
1900 and then
1901 E (Parse_Pos .. Parse_Pos + Digit'Length - 1) = Digit
1902 then
1903 if Invert then
1904 Class := ANYOF_NDIGIT;
1905 else
1906 Class := ANYOF_DIGIT;
1907 end if;
1908
1909 Parse_Pos := Parse_Pos + Digit'Length;
1910 end if;
1911
1912 when 'g' =>
1913 if Parse_Pos + Graph'Length - 1 <= Parse_End
1914 and then
1915 E (Parse_Pos .. Parse_Pos + Graph'Length - 1) = Graph
1916 then
1917 if Invert then
1918 Class := ANYOF_NGRAPH;
1919 else
1920 Class := ANYOF_GRAPH;
1921 end if;
1922
1923 Parse_Pos := Parse_Pos + Graph'Length;
1924
1925 else
1926 Fail ("Invalid character class: " & E);
1927 end if;
1928
1929 when 'l' =>
1930 if Parse_Pos + Lower'Length - 1 <= Parse_End
1931 and then
1932 E (Parse_Pos .. Parse_Pos + Lower'Length - 1) = Lower
1933 then
1934 if Invert then
1935 Class := ANYOF_NLOWER;
1936 else
1937 Class := ANYOF_LOWER;
1938 end if;
1939
1940 Parse_Pos := Parse_Pos + Lower'Length;
1941
1942 else
1943 Fail ("Invalid character class: " & E);
1944 end if;
1945
1946 when 'p' =>
1947
1948 -- All 'p' classes have the same length
1949
1950 if Parse_Pos + Print'Length - 1 <= Parse_End then
1951 if
1952 E (Parse_Pos .. Parse_Pos + Print'Length - 1) = Print
1953 then
1954 if Invert then
1955 Class := ANYOF_NPRINT;
1956 else
1957 Class := ANYOF_PRINT;
1958 end if;
1959
1960 Parse_Pos := Parse_Pos + Print'Length;
1961
1962 elsif
1963 E (Parse_Pos .. Parse_Pos + Punct'Length - 1) = Punct
1964 then
1965 if Invert then
1966 Class := ANYOF_NPUNCT;
1967 else
1968 Class := ANYOF_PUNCT;
1969 end if;
1970
1971 Parse_Pos := Parse_Pos + Punct'Length;
1972
1973 else
1974 Fail ("Invalid character class: " & E);
1975 end if;
1976
1977 else
1978 Fail ("Invalid character class: " & E);
1979 end if;
1980
1981 when 's' =>
1982 if Parse_Pos + Space'Length - 1 <= Parse_End
1983 and then
1984 E (Parse_Pos .. Parse_Pos + Space'Length - 1) = Space
1985 then
1986 if Invert then
1987 Class := ANYOF_NSPACE;
1988 else
1989 Class := ANYOF_SPACE;
1990 end if;
1991
1992 Parse_Pos := Parse_Pos + Space'Length;
1993
1994 else
1995 Fail ("Invalid character class: " & E);
1996 end if;
1997
1998 when 'u' =>
1999 if Parse_Pos + Upper'Length - 1 <= Parse_End
2000 and then
2001 E (Parse_Pos .. Parse_Pos + Upper'Length - 1) = Upper
2002 then
2003 if Invert then
2004 Class := ANYOF_NUPPER;
2005 else
2006 Class := ANYOF_UPPER;
2007 end if;
2008
2009 Parse_Pos := Parse_Pos + Upper'Length;
2010
2011 else
2012 Fail ("Invalid character class: " & E);
2013 end if;
2014
2015 when 'w' =>
2016 if Parse_Pos + Word'Length - 1 <= Parse_End
2017 and then
2018 E (Parse_Pos .. Parse_Pos + Word'Length - 1) = Word
2019 then
2020 if Invert then
2021 Class := ANYOF_NALNUM;
2022 else
2023 Class := ANYOF_ALNUM;
2024 end if;
2025
2026 Parse_Pos := Parse_Pos + Word'Length;
2027
2028 else
2029 Fail ("Invalid character class: " & E);
2030 end if;
2031
2032 when 'x' =>
2033 if Parse_Pos + Xdigit'Length - 1 <= Parse_End
2034 and then
2035 E (Parse_Pos .. Parse_Pos + Xdigit'Length - 1) = Xdigit
2036 then
2037 if Invert then
2038 Class := ANYOF_NXDIGIT;
2039 else
2040 Class := ANYOF_XDIGIT;
2041 end if;
2042
2043 Parse_Pos := Parse_Pos + Xdigit'Length;
2044
2045 else
2046 Fail ("Invalid character class: " & E);
2047 end if;
2048
2049 when others =>
2050 Fail ("Invalid character class: " & E);
2051 end case;
2052
2053 -- Character class not specified
2054
2055 else
2056 return ANYOF_NONE;
2057 end if;
2058
2059 return Class;
2060 end Parse_Posix_Character_Class;
2061
2062 -- Local Declarations
2063
2064 Result : Pointer;
2065
2066 Expr_Flags : Expression_Flags;
2067 pragma Unreferenced (Expr_Flags);
2068
2069 -- Start of processing for Compile
2070
2071 begin
2072 Emit (MAGIC);
2073 Parse (False, Expr_Flags, Result);
2074
2075 if Result = 0 then
2076 Fail ("Couldn't compile expression");
2077 end if;
2078
2079 Final_Code_Size := Emit_Ptr - 1;
2080
2081 -- Do we want to actually compile the expression, or simply get the
2082 -- code size ???
2083
2084 if Emit_Code then
2085 Optimize (PM);
2086 end if;
2087
2088 PM.Flags := Flags;
2089 end Compile;
2090
2091 function Compile
2092 (Expression : String;
2093 Flags : Regexp_Flags := No_Flags) return Pattern_Matcher
2094 is
2095 Size : Program_Size;
2096 Dummy : Pattern_Matcher (0);
2097 pragma Unreferenced (Dummy);
2098
2099 begin
2100 Compile (Dummy, Expression, Size, Flags);
2101
2102 declare
2103 Result : Pattern_Matcher (Size);
2104 begin
2105 Compile (Result, Expression, Size, Flags);
2106 return Result;
2107 end;
2108 end Compile;
2109
2110 procedure Compile
2111 (Matcher : out Pattern_Matcher;
2112 Expression : String;
2113 Flags : Regexp_Flags := No_Flags)
2114 is
2115 Size : Program_Size;
2116 pragma Unreferenced (Size);
2117 begin
2118 Compile (Matcher, Expression, Size, Flags);
2119 end Compile;
2120
2121 ----------
2122 -- Dump --
2123 ----------
2124
2125 procedure Dump (Self : Pattern_Matcher) is
2126 Op : Opcode;
2127 Program : Program_Data renames Self.Program;
2128
2129 procedure Dump_Until
2130 (Start : Pointer;
2131 Till : Pointer;
2132 Indent : Natural := 0);
2133 -- Dump the program until the node Till (not included) is met.
2134 -- Every line is indented with Index spaces at the beginning
2135 -- Dumps till the end if Till is 0.
2136
2137 ----------------
2138 -- Dump_Until --
2139 ----------------
2140
2141 procedure Dump_Until
2142 (Start : Pointer;
2143 Till : Pointer;
2144 Indent : Natural := 0)
2145 is
2146 Next : Pointer;
2147 Index : Pointer;
2148 Local_Indent : Natural := Indent;
2149 Length : Pointer;
2150
2151 begin
2152 Index := Start;
2153 while Index < Till loop
2154 Op := Opcode'Val (Character'Pos ((Self.Program (Index))));
2155
2156 if Op = CLOSE then
2157 Local_Indent := Local_Indent - 3;
2158 end if;
2159
2160 declare
2161 Point : constant String := Pointer'Image (Index);
2162
2163 begin
2164 for J in 1 .. 6 - Point'Length loop
2165 Put (' ');
2166 end loop;
2167
2168 Put (Point
2169 & " : "
2170 & (1 .. Local_Indent => ' ')
2171 & Opcode'Image (Op));
2172 end;
2173
2174 -- Print the parenthesis number
2175
2176 if Op = OPEN or else Op = CLOSE or else Op = REFF then
2177 Put (Natural'Image (Character'Pos (Program (Index + 3))));
2178 end if;
2179
2180 Next := Index + Get_Next_Offset (Program, Index);
2181
2182 if Next = Index then
2183 Put (" (next at 0)");
2184 else
2185 Put (" (next at " & Pointer'Image (Next) & ")");
2186 end if;
2187
2188 case Op is
2189
2190 -- Character class operand
2191
2192 when ANYOF => null;
2193 declare
2194 Bitmap : Character_Class;
2195 Last : Character := ASCII.NUL;
2196 Current : Natural := 0;
2197
2198 Current_Char : Character;
2199
2200 begin
2201 Bitmap_Operand (Program, Index, Bitmap);
2202 Put (" operand=");
2203
2204 while Current <= 255 loop
2205 Current_Char := Character'Val (Current);
2206
2207 -- First item in a range
2208
2209 if Get_From_Class (Bitmap, Current_Char) then
2210 Last := Current_Char;
2211
2212 -- Search for the last item in the range
2213
2214 loop
2215 Current := Current + 1;
2216 exit when Current > 255;
2217 Current_Char := Character'Val (Current);
2218 exit when
2219 not Get_From_Class (Bitmap, Current_Char);
2220
2221 end loop;
2222
2223 if Last <= ' ' then
2224 Put (Last'Img);
2225 else
2226 Put (Last);
2227 end if;
2228
2229 if Character'Succ (Last) /= Current_Char then
2230 Put ("-" & Character'Pred (Current_Char));
2231 end if;
2232
2233 else
2234 Current := Current + 1;
2235 end if;
2236 end loop;
2237
2238 New_Line;
2239 Index := Index + 3 + Bitmap'Length;
2240 end;
2241
2242 -- string operand
2243
2244 when EXACT | EXACTF =>
2245 Length := String_Length (Program, Index);
2246 Put (" operand (length:" & Program_Size'Image (Length + 1)
2247 & ") ="
2248 & String (Program (String_Operand (Index)
2249 .. String_Operand (Index)
2250 + Length)));
2251 Index := String_Operand (Index) + Length + 1;
2252 New_Line;
2253
2254 -- Node operand
2255
2256 when BRANCH =>
2257 New_Line;
2258 Dump_Until (Index + 3, Next, Local_Indent + 3);
2259 Index := Next;
2260
2261 when STAR | PLUS =>
2262 New_Line;
2263
2264 -- Only one instruction
2265
2266 Dump_Until (Index + 3, Index + 4, Local_Indent + 3);
2267 Index := Next;
2268
2269 when CURLY | CURLYX =>
2270 Put (" {"
2271 & Natural'Image (Read_Natural (Program, Index + 3))
2272 & ","
2273 & Natural'Image (Read_Natural (Program, Index + 5))
2274 & "}");
2275 New_Line;
2276 Dump_Until (Index + 7, Next, Local_Indent + 3);
2277 Index := Next;
2278
2279 when OPEN =>
2280 New_Line;
2281 Index := Index + 4;
2282 Local_Indent := Local_Indent + 3;
2283
2284 when CLOSE | REFF =>
2285 New_Line;
2286 Index := Index + 4;
2287
2288 when EOP =>
2289 Index := Index + 3;
2290 New_Line;
2291 exit;
2292
2293 -- No operand
2294
2295 when others =>
2296 Index := Index + 3;
2297 New_Line;
2298 end case;
2299 end loop;
2300 end Dump_Until;
2301
2302 -- Start of processing for Dump
2303
2304 begin
2305 pragma Assert (Self.Program (Program_First) = MAGIC,
2306 "Corrupted Pattern_Matcher");
2307
2308 Put_Line ("Must start with (Self.First) = "
2309 & Character'Image (Self.First));
2310
2311 if (Self.Flags and Case_Insensitive) /= 0 then
2312 Put_Line (" Case_Insensitive mode");
2313 end if;
2314
2315 if (Self.Flags and Single_Line) /= 0 then
2316 Put_Line (" Single_Line mode");
2317 end if;
2318
2319 if (Self.Flags and Multiple_Lines) /= 0 then
2320 Put_Line (" Multiple_Lines mode");
2321 end if;
2322
2323 Put_Line (" 1 : MAGIC");
2324 Dump_Until (Program_First + 1, Self.Program'Last + 1);
2325 end Dump;
2326
2327 --------------------
2328 -- Get_From_Class --
2329 --------------------
2330
2331 function Get_From_Class
2332 (Bitmap : Character_Class;
2333 C : Character) return Boolean
2334 is
2335 Value : constant Class_Byte := Character'Pos (C);
2336 begin
2337 return
2338 (Bitmap (Value / 8) and Bit_Conversion (Value mod 8)) /= 0;
2339 end Get_From_Class;
2340
2341 --------------
2342 -- Get_Next --
2343 --------------
2344
2345 function Get_Next (Program : Program_Data; IP : Pointer) return Pointer is
2346 Offset : constant Pointer := Get_Next_Offset (Program, IP);
2347 begin
2348 if Offset = 0 then
2349 return 0;
2350 else
2351 return IP + Offset;
2352 end if;
2353 end Get_Next;
2354
2355 ---------------------
2356 -- Get_Next_Offset --
2357 ---------------------
2358
2359 function Get_Next_Offset
2360 (Program : Program_Data;
2361 IP : Pointer) return Pointer
2362 is
2363 begin
2364 return Pointer (Read_Natural (Program, IP + 1));
2365 end Get_Next_Offset;
2366
2367 --------------
2368 -- Is_Alnum --
2369 --------------
2370
2371 function Is_Alnum (C : Character) return Boolean is
2372 begin
2373 return Is_Alphanumeric (C) or else C = '_';
2374 end Is_Alnum;
2375
2376 ------------------
2377 -- Is_Printable --
2378 ------------------
2379
2380 function Is_Printable (C : Character) return Boolean is
2381 begin
2382 -- Printable if space or graphic character or other whitespace
2383 -- Other white space includes (HT/LF/VT/FF/CR = codes 9-13)
2384
2385 return C in Character'Val (32) .. Character'Val (126)
2386 or else C in ASCII.HT .. ASCII.CR;
2387 end Is_Printable;
2388
2389 --------------------
2390 -- Is_White_Space --
2391 --------------------
2392
2393 function Is_White_Space (C : Character) return Boolean is
2394 begin
2395 -- Note: HT = 9, LF = 10, VT = 11, FF = 12, CR = 13
2396
2397 return C = ' ' or else C in ASCII.HT .. ASCII.CR;
2398 end Is_White_Space;
2399
2400 -----------
2401 -- Match --
2402 -----------
2403
2404 procedure Match
2405 (Self : Pattern_Matcher;
2406 Data : String;
2407 Matches : out Match_Array;
2408 Data_First : Integer := -1;
2409 Data_Last : Positive := Positive'Last)
2410 is
2411 Program : Program_Data renames Self.Program; -- Shorter notation
2412
2413 First_In_Data : constant Integer := Integer'Max (Data_First, Data'First);
2414 Last_In_Data : constant Integer := Integer'Min (Data_Last, Data'Last);
2415
2416 -- Global work variables
2417
2418 Input_Pos : Natural; -- String-input pointer
2419 BOL_Pos : Natural; -- Beginning of input, for ^ check
2420 Matched : Boolean := False; -- Until proven True
2421
2422 Matches_Full : Match_Array (0 .. Natural'Max (Self.Paren_Count,
2423 Matches'Last));
2424 -- Stores the value of all the parenthesis pairs.
2425 -- We do not use directly Matches, so that we can also use back
2426 -- references (REFF) even if Matches is too small.
2427
2428 type Natural_Array is array (Match_Count range <>) of Natural;
2429 Matches_Tmp : Natural_Array (Matches_Full'Range);
2430 -- Save the opening position of parenthesis
2431
2432 Last_Paren : Natural := 0;
2433 -- Last parenthesis seen
2434
2435 Greedy : Boolean := True;
2436 -- True if the next operator should be greedy
2437
2438 type Current_Curly_Record;
2439 type Current_Curly_Access is access all Current_Curly_Record;
2440 type Current_Curly_Record is record
2441 Paren_Floor : Natural; -- How far back to strip parenthesis data
2442 Cur : Integer; -- How many instances of scan we've matched
2443 Min : Natural; -- Minimal number of scans to match
2444 Max : Natural; -- Maximal number of scans to match
2445 Greedy : Boolean; -- Whether to work our way up or down
2446 Scan : Pointer; -- The thing to match
2447 Next : Pointer; -- What has to match after it
2448 Lastloc : Natural; -- Where we started matching this scan
2449 Old_Cc : Current_Curly_Access; -- Before we started this one
2450 end record;
2451 -- Data used to handle the curly operator and the plus and star
2452 -- operators for complex expressions.
2453
2454 Current_Curly : Current_Curly_Access := null;
2455 -- The curly currently being processed
2456
2457 -----------------------
2458 -- Local Subprograms --
2459 -----------------------
2460
2461 function Index (Start : Positive; C : Character) return Natural;
2462 -- Find character C in Data starting at Start and return position
2463
2464 function Repeat
2465 (IP : Pointer;
2466 Max : Natural := Natural'Last) return Natural;
2467 -- Repeatedly match something simple, report how many
2468 -- It only matches on things of length 1.
2469 -- Starting from Input_Pos, it matches at most Max CURLY.
2470
2471 function Try (Pos : Positive) return Boolean;
2472 -- Try to match at specific point
2473
2474 function Match (IP : Pointer) return Boolean;
2475 -- This is the main matching routine. Conceptually the strategy
2476 -- is simple: check to see whether the current node matches,
2477 -- call self recursively to see whether the rest matches,
2478 -- and then act accordingly.
2479 --
2480 -- In practice Match makes some effort to avoid recursion, in
2481 -- particular by going through "ordinary" nodes (that don't
2482 -- need to know whether the rest of the match failed) by
2483 -- using a loop instead of recursion.
2484 -- Why is the above comment part of the spec rather than body ???
2485
2486 function Match_Whilem (IP : Pointer) return Boolean;
2487 -- Return True if a WHILEM matches
2488 -- How come IP is unreferenced in the body ???
2489
2490 function Recurse_Match (IP : Pointer; From : Natural) return Boolean;
2491 pragma Inline (Recurse_Match);
2492 -- Calls Match recursively. It saves and restores the parenthesis
2493 -- status and location in the input stream correctly, so that
2494 -- backtracking is possible
2495
2496 function Match_Simple_Operator
2497 (Op : Opcode;
2498 Scan : Pointer;
2499 Next : Pointer;
2500 Greedy : Boolean) return Boolean;
2501 -- Return True it the simple operator (possibly non-greedy) matches
2502
2503 pragma Inline (Index);
2504 pragma Inline (Repeat);
2505
2506 -- These are two complex functions, but used only once
2507
2508 pragma Inline (Match_Whilem);
2509 pragma Inline (Match_Simple_Operator);
2510
2511 -----------
2512 -- Index --
2513 -----------
2514
2515 function Index (Start : Positive; C : Character) return Natural is
2516 begin
2517 for J in Start .. Last_In_Data loop
2518 if Data (J) = C then
2519 return J;
2520 end if;
2521 end loop;
2522
2523 return 0;
2524 end Index;
2525
2526 -------------------
2527 -- Recurse_Match --
2528 -------------------
2529
2530 function Recurse_Match (IP : Pointer; From : Natural) return Boolean is
2531 L : constant Natural := Last_Paren;
2532
2533 Tmp_F : constant Match_Array :=
2534 Matches_Full (From + 1 .. Matches_Full'Last);
2535
2536 Start : constant Natural_Array :=
2537 Matches_Tmp (From + 1 .. Matches_Tmp'Last);
2538 Input : constant Natural := Input_Pos;
2539
2540 begin
2541 if Match (IP) then
2542 return True;
2543 end if;
2544
2545 Last_Paren := L;
2546 Matches_Full (Tmp_F'Range) := Tmp_F;
2547 Matches_Tmp (Start'Range) := Start;
2548 Input_Pos := Input;
2549 return False;
2550 end Recurse_Match;
2551
2552 -----------
2553 -- Match --
2554 -----------
2555
2556 function Match (IP : Pointer) return Boolean is
2557 Scan : Pointer := IP;
2558 Next : Pointer;
2559 Op : Opcode;
2560
2561 begin
2562 State_Machine :
2563 loop
2564 pragma Assert (Scan /= 0);
2565
2566 -- Determine current opcode and count its usage in debug mode
2567
2568 Op := Opcode'Val (Character'Pos (Program (Scan)));
2569
2570 -- Calculate offset of next instruction.
2571 -- Second character is most significant in Program_Data.
2572
2573 Next := Get_Next (Program, Scan);
2574
2575 case Op is
2576 when EOP =>
2577 return True; -- Success !
2578
2579 when BRANCH =>
2580 if Program (Next) /= BRANCH then
2581 Next := Operand (Scan); -- No choice, avoid recursion
2582
2583 else
2584 loop
2585 if Recurse_Match (Operand (Scan), 0) then
2586 return True;
2587 end if;
2588
2589 Scan := Get_Next (Program, Scan);
2590 exit when Scan = 0 or else Program (Scan) /= BRANCH;
2591 end loop;
2592
2593 exit State_Machine;
2594 end if;
2595
2596 when NOTHING =>
2597 null;
2598
2599 when BOL =>
2600 exit State_Machine when Input_Pos /= BOL_Pos
2601 and then ((Self.Flags and Multiple_Lines) = 0
2602 or else Data (Input_Pos - 1) /= ASCII.LF);
2603
2604 when MBOL =>
2605 exit State_Machine when Input_Pos /= BOL_Pos
2606 and then Data (Input_Pos - 1) /= ASCII.LF;
2607
2608 when SBOL =>
2609 exit State_Machine when Input_Pos /= BOL_Pos;
2610
2611 when EOL =>
2612 exit State_Machine when Input_Pos <= Data'Last
2613 and then ((Self.Flags and Multiple_Lines) = 0
2614 or else Data (Input_Pos) /= ASCII.LF);
2615
2616 when MEOL =>
2617 exit State_Machine when Input_Pos <= Data'Last
2618 and then Data (Input_Pos) /= ASCII.LF;
2619
2620 when SEOL =>
2621 exit State_Machine when Input_Pos <= Data'Last;
2622
2623 when BOUND | NBOUND =>
2624
2625 -- Was last char in word ?
2626
2627 declare
2628 N : Boolean := False;
2629 Ln : Boolean := False;
2630
2631 begin
2632 if Input_Pos /= First_In_Data then
2633 N := Is_Alnum (Data (Input_Pos - 1));
2634 end if;
2635
2636 if Input_Pos > Last_In_Data then
2637 Ln := False;
2638 else
2639 Ln := Is_Alnum (Data (Input_Pos));
2640 end if;
2641
2642 if Op = BOUND then
2643 if N = Ln then
2644 exit State_Machine;
2645 end if;
2646 else
2647 if N /= Ln then
2648 exit State_Machine;
2649 end if;
2650 end if;
2651 end;
2652
2653 when SPACE =>
2654 exit State_Machine when Input_Pos > Last_In_Data
2655 or else not Is_White_Space (Data (Input_Pos));
2656 Input_Pos := Input_Pos + 1;
2657
2658 when NSPACE =>
2659 exit State_Machine when Input_Pos > Last_In_Data
2660 or else Is_White_Space (Data (Input_Pos));
2661 Input_Pos := Input_Pos + 1;
2662
2663 when DIGIT =>
2664 exit State_Machine when Input_Pos > Last_In_Data
2665 or else not Is_Digit (Data (Input_Pos));
2666 Input_Pos := Input_Pos + 1;
2667
2668 when NDIGIT =>
2669 exit State_Machine when Input_Pos > Last_In_Data
2670 or else Is_Digit (Data (Input_Pos));
2671 Input_Pos := Input_Pos + 1;
2672
2673 when ALNUM =>
2674 exit State_Machine when Input_Pos > Last_In_Data
2675 or else not Is_Alnum (Data (Input_Pos));
2676 Input_Pos := Input_Pos + 1;
2677
2678 when NALNUM =>
2679 exit State_Machine when Input_Pos > Last_In_Data
2680 or else Is_Alnum (Data (Input_Pos));
2681 Input_Pos := Input_Pos + 1;
2682
2683 when ANY =>
2684 exit State_Machine when Input_Pos > Last_In_Data
2685 or else Data (Input_Pos) = ASCII.LF;
2686 Input_Pos := Input_Pos + 1;
2687
2688 when SANY =>
2689 exit State_Machine when Input_Pos > Last_In_Data;
2690 Input_Pos := Input_Pos + 1;
2691
2692 when EXACT =>
2693 declare
2694 Opnd : Pointer := String_Operand (Scan);
2695 Current : Positive := Input_Pos;
2696
2697 Last : constant Pointer :=
2698 Opnd + String_Length (Program, Scan);
2699
2700 begin
2701 while Opnd <= Last loop
2702 exit State_Machine when Current > Last_In_Data
2703 or else Program (Opnd) /= Data (Current);
2704 Current := Current + 1;
2705 Opnd := Opnd + 1;
2706 end loop;
2707
2708 Input_Pos := Current;
2709 end;
2710
2711 when EXACTF =>
2712 declare
2713 Opnd : Pointer := String_Operand (Scan);
2714 Current : Positive := Input_Pos;
2715
2716 Last : constant Pointer :=
2717 Opnd + String_Length (Program, Scan);
2718
2719 begin
2720 while Opnd <= Last loop
2721 exit State_Machine when Current > Last_In_Data
2722 or else Program (Opnd) /= To_Lower (Data (Current));
2723 Current := Current + 1;
2724 Opnd := Opnd + 1;
2725 end loop;
2726
2727 Input_Pos := Current;
2728 end;
2729
2730 when ANYOF =>
2731 declare
2732 Bitmap : Character_Class;
2733 begin
2734 Bitmap_Operand (Program, Scan, Bitmap);
2735 exit State_Machine when Input_Pos > Last_In_Data
2736 or else not Get_From_Class (Bitmap, Data (Input_Pos));
2737 Input_Pos := Input_Pos + 1;
2738 end;
2739
2740 when OPEN =>
2741 declare
2742 No : constant Natural :=
2743 Character'Pos (Program (Operand (Scan)));
2744 begin
2745 Matches_Tmp (No) := Input_Pos;
2746 end;
2747
2748 when CLOSE =>
2749 declare
2750 No : constant Natural :=
2751 Character'Pos (Program (Operand (Scan)));
2752
2753 begin
2754 Matches_Full (No) := (Matches_Tmp (No), Input_Pos - 1);
2755
2756 if Last_Paren < No then
2757 Last_Paren := No;
2758 end if;
2759 end;
2760
2761 when REFF =>
2762 declare
2763 No : constant Natural :=
2764 Character'Pos (Program (Operand (Scan)));
2765
2766 Data_Pos : Natural;
2767
2768 begin
2769 -- If we haven't seen that parenthesis yet
2770
2771 if Last_Paren < No then
2772 return False;
2773 end if;
2774
2775 Data_Pos := Matches_Full (No).First;
2776
2777 while Data_Pos <= Matches_Full (No).Last loop
2778 if Input_Pos > Last_In_Data
2779 or else Data (Input_Pos) /= Data (Data_Pos)
2780 then
2781 return False;
2782 end if;
2783
2784 Input_Pos := Input_Pos + 1;
2785 Data_Pos := Data_Pos + 1;
2786 end loop;
2787 end;
2788
2789 when MINMOD =>
2790 Greedy := False;
2791
2792 when STAR | PLUS | CURLY =>
2793 declare
2794 Greed : constant Boolean := Greedy;
2795 begin
2796 Greedy := True;
2797 return Match_Simple_Operator (Op, Scan, Next, Greed);
2798 end;
2799
2800 when CURLYX =>
2801
2802 -- Looking at something like:
2803
2804 -- 1: CURLYX {n,m} (->4)
2805 -- 2: code for complex thing (->3)
2806 -- 3: WHILEM (->0)
2807 -- 4: NOTHING
2808
2809 declare
2810 Min : constant Natural :=
2811 Read_Natural (Program, Scan + 3);
2812 Max : constant Natural :=
2813 Read_Natural (Program, Scan + 5);
2814 Cc : aliased Current_Curly_Record;
2815
2816 Has_Match : Boolean;
2817
2818 begin
2819 Cc := (Paren_Floor => Last_Paren,
2820 Cur => -1,
2821 Min => Min,
2822 Max => Max,
2823 Greedy => Greedy,
2824 Scan => Scan + 7,
2825 Next => Next,
2826 Lastloc => 0,
2827 Old_Cc => Current_Curly);
2828 Current_Curly := Cc'Unchecked_Access;
2829
2830 Has_Match := Match (Next - 3);
2831
2832 -- Start on the WHILEM
2833
2834 Current_Curly := Cc.Old_Cc;
2835 return Has_Match;
2836 end;
2837
2838 when WHILEM =>
2839 return Match_Whilem (IP);
2840 end case;
2841
2842 Scan := Next;
2843 end loop State_Machine;
2844
2845 -- If we get here, there is no match.
2846 -- For successful matches when EOP is the terminating point.
2847
2848 return False;
2849 end Match;
2850
2851 ---------------------------
2852 -- Match_Simple_Operator --
2853 ---------------------------
2854
2855 function Match_Simple_Operator
2856 (Op : Opcode;
2857 Scan : Pointer;
2858 Next : Pointer;
2859 Greedy : Boolean) return Boolean
2860 is
2861 Next_Char : Character := ASCII.NUL;
2862 Next_Char_Known : Boolean := False;
2863 No : Integer; -- Can be negative
2864 Min : Natural;
2865 Max : Natural := Natural'Last;
2866 Operand_Code : Pointer;
2867 Old : Natural;
2868 Last_Pos : Natural;
2869 Save : constant Natural := Input_Pos;
2870
2871 begin
2872 -- Lookahead to avoid useless match attempts
2873 -- when we know what character comes next.
2874
2875 if Program (Next) = EXACT then
2876 Next_Char := Program (String_Operand (Next));
2877 Next_Char_Known := True;
2878 end if;
2879
2880 -- Find the minimal and maximal values for the operator
2881
2882 case Op is
2883 when STAR =>
2884 Min := 0;
2885 Operand_Code := Operand (Scan);
2886
2887 when PLUS =>
2888 Min := 1;
2889 Operand_Code := Operand (Scan);
2890
2891 when others =>
2892 Min := Read_Natural (Program, Scan + 3);
2893 Max := Read_Natural (Program, Scan + 5);
2894 Operand_Code := Scan + 7;
2895 end case;
2896
2897 -- Non greedy operators
2898
2899 if not Greedy then
2900
2901 -- Test the minimal repetitions
2902
2903 if Min /= 0
2904 and then Repeat (Operand_Code, Min) < Min
2905 then
2906 return False;
2907 end if;
2908
2909 Old := Input_Pos;
2910
2911 -- Find the place where 'next' could work
2912
2913 if Next_Char_Known then
2914 -- Last position to check
2915
2916 if Max = Natural'Last then
2917 Last_Pos := Last_In_Data;
2918 else
2919 Last_Pos := Input_Pos + Max;
2920
2921 if Last_Pos > Last_In_Data then
2922 Last_Pos := Last_In_Data;
2923 end if;
2924 end if;
2925
2926 -- Look for the first possible opportunity
2927
2928 loop
2929 -- Find the next possible position
2930
2931 while Input_Pos <= Last_Pos
2932 and then Data (Input_Pos) /= Next_Char
2933 loop
2934 Input_Pos := Input_Pos + 1;
2935 end loop;
2936
2937 if Input_Pos > Last_Pos then
2938 return False;
2939 end if;
2940
2941 -- Check that we still match if we stop
2942 -- at the position we just found.
2943
2944 declare
2945 Num : constant Natural := Input_Pos - Old;
2946
2947 begin
2948 Input_Pos := Old;
2949
2950 if Repeat (Operand_Code, Num) < Num then
2951 return False;
2952 end if;
2953 end;
2954
2955 -- Input_Pos now points to the new position
2956
2957 if Match (Get_Next (Program, Scan)) then
2958 return True;
2959 end if;
2960
2961 Old := Input_Pos;
2962 Input_Pos := Input_Pos + 1;
2963 end loop;
2964
2965 -- We know what the next character is
2966
2967 else
2968 while Max >= Min loop
2969
2970 -- If the next character matches
2971
2972 if Match (Next) then
2973 return True;
2974 end if;
2975
2976 Input_Pos := Save + Min;
2977
2978 -- Could not or did not match -- move forward
2979
2980 if Repeat (Operand_Code, 1) /= 0 then
2981 Min := Min + 1;
2982 else
2983 return False;
2984 end if;
2985 end loop;
2986 end if;
2987
2988 return False;
2989
2990 -- Greedy operators
2991
2992 else
2993 No := Repeat (Operand_Code, Max);
2994
2995 -- ??? Perl has some special code here in case the
2996 -- next instruction is of type EOL, since $ and \Z
2997 -- can match before *and* after newline at the end.
2998
2999 -- ??? Perl has some special code here in case (paren)
3000 -- is True.
3001
3002 -- Else, if we don't have any parenthesis
3003
3004 while No >= Min loop
3005 if not Next_Char_Known
3006 or else (Input_Pos <= Last_In_Data
3007 and then Data (Input_Pos) = Next_Char)
3008 then
3009 if Match (Next) then
3010 return True;
3011 end if;
3012 end if;
3013
3014 -- Could not or did not work, we back up
3015
3016 No := No - 1;
3017 Input_Pos := Save + No;
3018 end loop;
3019
3020 return False;
3021 end if;
3022 end Match_Simple_Operator;
3023
3024 ------------------
3025 -- Match_Whilem --
3026 ------------------
3027
3028 -- This is really hard to understand, because after we match what we
3029 -- are trying to match, we must make sure the rest of the REx is going
3030 -- to match for sure, and to do that we have to go back UP the parse
3031 -- tree by recursing ever deeper. And if it fails, we have to reset
3032 -- our parent's current state that we can try again after backing off.
3033
3034 function Match_Whilem (IP : Pointer) return Boolean is
3035 pragma Unreferenced (IP);
3036
3037 Cc : constant Current_Curly_Access := Current_Curly;
3038 N : constant Natural := Cc.Cur + 1;
3039 Ln : Natural := 0;
3040
3041 Lastloc : constant Natural := Cc.Lastloc;
3042 -- Detection of 0-len
3043
3044 begin
3045 -- If degenerate scan matches "", assume scan done
3046
3047 if Input_Pos = Cc.Lastloc
3048 and then N >= Cc.Min
3049 then
3050 -- Temporarily restore the old context, and check that we
3051 -- match was comes after CURLYX.
3052
3053 Current_Curly := Cc.Old_Cc;
3054
3055 if Current_Curly /= null then
3056 Ln := Current_Curly.Cur;
3057 end if;
3058
3059 if Match (Cc.Next) then
3060 return True;
3061 end if;
3062
3063 if Current_Curly /= null then
3064 Current_Curly.Cur := Ln;
3065 end if;
3066
3067 Current_Curly := Cc;
3068 return False;
3069 end if;
3070
3071 -- First, just match a string of min scans
3072
3073 if N < Cc.Min then
3074 Cc.Cur := N;
3075 Cc.Lastloc := Input_Pos;
3076
3077 if Match (Cc.Scan) then
3078 return True;
3079 end if;
3080
3081 Cc.Cur := N - 1;
3082 Cc.Lastloc := Lastloc;
3083 return False;
3084 end if;
3085
3086 -- Prefer next over scan for minimal matching
3087
3088 if not Cc.Greedy then
3089 Current_Curly := Cc.Old_Cc;
3090
3091 if Current_Curly /= null then
3092 Ln := Current_Curly.Cur;
3093 end if;
3094
3095 if Recurse_Match (Cc.Next, Cc.Paren_Floor) then
3096 return True;
3097 end if;
3098
3099 if Current_Curly /= null then
3100 Current_Curly.Cur := Ln;
3101 end if;
3102
3103 Current_Curly := Cc;
3104
3105 -- Maximum greed exceeded ?
3106
3107 if N >= Cc.Max then
3108 return False;
3109 end if;
3110
3111 -- Try scanning more and see if it helps
3112 Cc.Cur := N;
3113 Cc.Lastloc := Input_Pos;
3114
3115 if Recurse_Match (Cc.Scan, Cc.Paren_Floor) then
3116 return True;
3117 end if;
3118
3119 Cc.Cur := N - 1;
3120 Cc.Lastloc := Lastloc;
3121 return False;
3122 end if;
3123
3124 -- Prefer scan over next for maximal matching
3125
3126 if N < Cc.Max then -- more greed allowed ?
3127 Cc.Cur := N;
3128 Cc.Lastloc := Input_Pos;
3129
3130 if Recurse_Match (Cc.Scan, Cc.Paren_Floor) then
3131 return True;
3132 end if;
3133 end if;
3134
3135 -- Failed deeper matches of scan, so see if this one works
3136
3137 Current_Curly := Cc.Old_Cc;
3138
3139 if Current_Curly /= null then
3140 Ln := Current_Curly.Cur;
3141 end if;
3142
3143 if Match (Cc.Next) then
3144 return True;
3145 end if;
3146
3147 if Current_Curly /= null then
3148 Current_Curly.Cur := Ln;
3149 end if;
3150
3151 Current_Curly := Cc;
3152 Cc.Cur := N - 1;
3153 Cc.Lastloc := Lastloc;
3154 return False;
3155 end Match_Whilem;
3156
3157 ------------
3158 -- Repeat --
3159 ------------
3160
3161 function Repeat
3162 (IP : Pointer;
3163 Max : Natural := Natural'Last) return Natural
3164 is
3165 Scan : Natural := Input_Pos;
3166 Last : Natural;
3167 Op : constant Opcode := Opcode'Val (Character'Pos (Program (IP)));
3168 Count : Natural;
3169 C : Character;
3170 Is_First : Boolean := True;
3171 Bitmap : Character_Class;
3172
3173 begin
3174 if Max = Natural'Last or else Scan + Max - 1 > Last_In_Data then
3175 Last := Last_In_Data;
3176 else
3177 Last := Scan + Max - 1;
3178 end if;
3179
3180 case Op is
3181 when ANY =>
3182 while Scan <= Last
3183 and then Data (Scan) /= ASCII.LF
3184 loop
3185 Scan := Scan + 1;
3186 end loop;
3187
3188 when SANY =>
3189 Scan := Last + 1;
3190
3191 when EXACT =>
3192
3193 -- The string has only one character if Repeat was called
3194
3195 C := Program (String_Operand (IP));
3196 while Scan <= Last
3197 and then C = Data (Scan)
3198 loop
3199 Scan := Scan + 1;
3200 end loop;
3201
3202 when EXACTF =>
3203
3204 -- The string has only one character if Repeat was called
3205
3206 C := Program (String_Operand (IP));
3207 while Scan <= Last
3208 and then To_Lower (C) = Data (Scan)
3209 loop
3210 Scan := Scan + 1;
3211 end loop;
3212
3213 when ANYOF =>
3214 if Is_First then
3215 Bitmap_Operand (Program, IP, Bitmap);
3216 Is_First := False;
3217 end if;
3218
3219 while Scan <= Last
3220 and then Get_From_Class (Bitmap, Data (Scan))
3221 loop
3222 Scan := Scan + 1;
3223 end loop;
3224
3225 when ALNUM =>
3226 while Scan <= Last
3227 and then Is_Alnum (Data (Scan))
3228 loop
3229 Scan := Scan + 1;
3230 end loop;
3231
3232 when NALNUM =>
3233 while Scan <= Last
3234 and then not Is_Alnum (Data (Scan))
3235 loop
3236 Scan := Scan + 1;
3237 end loop;
3238
3239 when SPACE =>
3240 while Scan <= Last
3241 and then Is_White_Space (Data (Scan))
3242 loop
3243 Scan := Scan + 1;
3244 end loop;
3245
3246 when NSPACE =>
3247 while Scan <= Last
3248 and then not Is_White_Space (Data (Scan))
3249 loop
3250 Scan := Scan + 1;
3251 end loop;
3252
3253 when DIGIT =>
3254 while Scan <= Last
3255 and then Is_Digit (Data (Scan))
3256 loop
3257 Scan := Scan + 1;
3258 end loop;
3259
3260 when NDIGIT =>
3261 while Scan <= Last
3262 and then not Is_Digit (Data (Scan))
3263 loop
3264 Scan := Scan + 1;
3265 end loop;
3266
3267 when others =>
3268 raise Program_Error;
3269 end case;
3270
3271 Count := Scan - Input_Pos;
3272 Input_Pos := Scan;
3273 return Count;
3274 end Repeat;
3275
3276 ---------
3277 -- Try --
3278 ---------
3279
3280 function Try (Pos : Positive) return Boolean is
3281 begin
3282 Input_Pos := Pos;
3283 Last_Paren := 0;
3284 Matches_Full := (others => No_Match);
3285
3286 if Match (Program_First + 1) then
3287 Matches_Full (0) := (Pos, Input_Pos - 1);
3288 return True;
3289 end if;
3290
3291 return False;
3292 end Try;
3293
3294 -- Start of processing for Match
3295
3296 begin
3297 -- Do we have the regexp Never_Match?
3298
3299 if Self.Size = 0 then
3300 Matches := (others => No_Match);
3301 return;
3302 end if;
3303
3304 -- Check validity of program
3305
3306 pragma Assert
3307 (Program (Program_First) = MAGIC,
3308 "Corrupted Pattern_Matcher");
3309
3310 -- If there is a "must appear" string, look for it
3311
3312 if Self.Must_Have_Length > 0 then
3313 declare
3314 First : constant Character := Program (Self.Must_Have);
3315 Must_First : constant Pointer := Self.Must_Have;
3316 Must_Last : constant Pointer :=
3317 Must_First + Pointer (Self.Must_Have_Length - 1);
3318 Next_Try : Natural := Index (First_In_Data, First);
3319
3320 begin
3321 while Next_Try /= 0
3322 and then Data (Next_Try .. Next_Try + Self.Must_Have_Length - 1)
3323 = String (Program (Must_First .. Must_Last))
3324 loop
3325 Next_Try := Index (Next_Try + 1, First);
3326 end loop;
3327
3328 if Next_Try = 0 then
3329 Matches := (others => No_Match);
3330 return; -- Not present
3331 end if;
3332 end;
3333 end if;
3334
3335 -- Mark beginning of line for ^
3336
3337 BOL_Pos := Data'First;
3338
3339 -- Simplest case first: an anchored match need be tried only once
3340
3341 if Self.Anchored and then (Self.Flags and Multiple_Lines) = 0 then
3342 Matched := Try (First_In_Data);
3343
3344 elsif Self.Anchored then
3345 declare
3346 Next_Try : Natural := First_In_Data;
3347 begin
3348 -- Test the first position in the buffer
3349 Matched := Try (Next_Try);
3350
3351 -- Else only test after newlines
3352
3353 if not Matched then
3354 while Next_Try <= Last_In_Data loop
3355 while Next_Try <= Last_In_Data
3356 and then Data (Next_Try) /= ASCII.LF
3357 loop
3358 Next_Try := Next_Try + 1;
3359 end loop;
3360
3361 Next_Try := Next_Try + 1;
3362
3363 if Next_Try <= Last_In_Data then
3364 Matched := Try (Next_Try);
3365 exit when Matched;
3366 end if;
3367 end loop;
3368 end if;
3369 end;
3370
3371 elsif Self.First /= ASCII.NUL then
3372 -- We know what char it must start with
3373
3374 declare
3375 Next_Try : Natural := Index (First_In_Data, Self.First);
3376
3377 begin
3378 while Next_Try /= 0 loop
3379 Matched := Try (Next_Try);
3380 exit when Matched;
3381 Next_Try := Index (Next_Try + 1, Self.First);
3382 end loop;
3383 end;
3384
3385 else
3386 -- Messy cases: try all locations (including for the empty string)
3387
3388 Matched := Try (First_In_Data);
3389
3390 if not Matched then
3391 for S in First_In_Data + 1 .. Last_In_Data loop
3392 Matched := Try (S);
3393 exit when Matched;
3394 end loop;
3395 end if;
3396 end if;
3397
3398 -- Matched has its value
3399
3400 for J in Last_Paren + 1 .. Matches'Last loop
3401 Matches_Full (J) := No_Match;
3402 end loop;
3403
3404 Matches := Matches_Full (Matches'Range);
3405 end Match;
3406
3407 -----------
3408 -- Match --
3409 -----------
3410
3411 function Match
3412 (Self : Pattern_Matcher;
3413 Data : String;
3414 Data_First : Integer := -1;
3415 Data_Last : Positive := Positive'Last) return Natural
3416 is
3417 Matches : Match_Array (0 .. 0);
3418
3419 begin
3420 Match (Self, Data, Matches, Data_First, Data_Last);
3421 if Matches (0) = No_Match then
3422 return Data'First - 1;
3423 else
3424 return Matches (0).First;
3425 end if;
3426 end Match;
3427
3428 function Match
3429 (Self : Pattern_Matcher;
3430 Data : String;
3431 Data_First : Integer := -1;
3432 Data_Last : Positive := Positive'Last) return Boolean
3433 is
3434 Matches : Match_Array (0 .. 0);
3435
3436 begin
3437 Match (Self, Data, Matches, Data_First, Data_Last);
3438 return Matches (0).First >= Data'First;
3439 end Match;
3440
3441 procedure Match
3442 (Expression : String;
3443 Data : String;
3444 Matches : out Match_Array;
3445 Size : Program_Size := Auto_Size;
3446 Data_First : Integer := -1;
3447 Data_Last : Positive := Positive'Last)
3448 is
3449 PM : Pattern_Matcher (Size);
3450 Finalize_Size : Program_Size;
3451 pragma Unreferenced (Finalize_Size);
3452 begin
3453 if Size = 0 then
3454 Match (Compile (Expression), Data, Matches, Data_First, Data_Last);
3455 else
3456 Compile (PM, Expression, Finalize_Size);
3457 Match (PM, Data, Matches, Data_First, Data_Last);
3458 end if;
3459 end Match;
3460
3461 -----------
3462 -- Match --
3463 -----------
3464
3465 function Match
3466 (Expression : String;
3467 Data : String;
3468 Size : Program_Size := Auto_Size;
3469 Data_First : Integer := -1;
3470 Data_Last : Positive := Positive'Last) return Natural
3471 is
3472 PM : Pattern_Matcher (Size);
3473 Final_Size : Program_Size;
3474 pragma Unreferenced (Final_Size);
3475 begin
3476 if Size = 0 then
3477 return Match (Compile (Expression), Data, Data_First, Data_Last);
3478 else
3479 Compile (PM, Expression, Final_Size);
3480 return Match (PM, Data, Data_First, Data_Last);
3481 end if;
3482 end Match;
3483
3484 -----------
3485 -- Match --
3486 -----------
3487
3488 function Match
3489 (Expression : String;
3490 Data : String;
3491 Size : Program_Size := Auto_Size;
3492 Data_First : Integer := -1;
3493 Data_Last : Positive := Positive'Last) return Boolean
3494 is
3495 Matches : Match_Array (0 .. 0);
3496 PM : Pattern_Matcher (Size);
3497 Final_Size : Program_Size;
3498 pragma Unreferenced (Final_Size);
3499 begin
3500 if Size = 0 then
3501 Match (Compile (Expression), Data, Matches, Data_First, Data_Last);
3502 else
3503 Compile (PM, Expression, Final_Size);
3504 Match (PM, Data, Matches, Data_First, Data_Last);
3505 end if;
3506
3507 return Matches (0).First >= Data'First;
3508 end Match;
3509
3510 -------------
3511 -- Operand --
3512 -------------
3513
3514 function Operand (P : Pointer) return Pointer is
3515 begin
3516 return P + 3;
3517 end Operand;
3518
3519 --------------
3520 -- Optimize --
3521 --------------
3522
3523 procedure Optimize (Self : in out Pattern_Matcher) is
3524 Scan : Pointer;
3525 Program : Program_Data renames Self.Program;
3526
3527 begin
3528 -- Start with safe defaults (no optimization):
3529 -- * No known first character of match
3530 -- * Does not necessarily start at beginning of line
3531 -- * No string known that has to appear in data
3532
3533 Self.First := ASCII.NUL;
3534 Self.Anchored := False;
3535 Self.Must_Have := Program'Last + 1;
3536 Self.Must_Have_Length := 0;
3537
3538 Scan := Program_First + 1; -- First instruction (can be anything)
3539
3540 if Program (Scan) = EXACT then
3541 Self.First := Program (String_Operand (Scan));
3542
3543 elsif Program (Scan) = BOL
3544 or else Program (Scan) = SBOL
3545 or else Program (Scan) = MBOL
3546 then
3547 Self.Anchored := True;
3548 end if;
3549 end Optimize;
3550
3551 -----------------
3552 -- Paren_Count --
3553 -----------------
3554
3555 function Paren_Count (Regexp : Pattern_Matcher) return Match_Count is
3556 begin
3557 return Regexp.Paren_Count;
3558 end Paren_Count;
3559
3560 -----------
3561 -- Quote --
3562 -----------
3563
3564 function Quote (Str : String) return String is
3565 S : String (1 .. Str'Length * 2);
3566 Last : Natural := 0;
3567
3568 begin
3569 for J in Str'Range loop
3570 case Str (J) is
3571 when '^' | '$' | '|' | '*' | '+' | '?' | '{' |
3572 '}' | '[' | ']' | '(' | ')' | '\' | '.' =>
3573
3574 S (Last + 1) := '\';
3575 S (Last + 2) := Str (J);
3576 Last := Last + 2;
3577
3578 when others =>
3579 S (Last + 1) := Str (J);
3580 Last := Last + 1;
3581 end case;
3582 end loop;
3583
3584 return S (1 .. Last);
3585 end Quote;
3586
3587 ------------------
3588 -- Read_Natural --
3589 ------------------
3590
3591 function Read_Natural
3592 (Program : Program_Data;
3593 IP : Pointer) return Natural
3594 is
3595 begin
3596 return Character'Pos (Program (IP)) +
3597 256 * Character'Pos (Program (IP + 1));
3598 end Read_Natural;
3599
3600 -----------------
3601 -- Reset_Class --
3602 -----------------
3603
3604 procedure Reset_Class (Bitmap : out Character_Class) is
3605 begin
3606 Bitmap := (others => 0);
3607 end Reset_Class;
3608
3609 ------------------
3610 -- Set_In_Class --
3611 ------------------
3612
3613 procedure Set_In_Class
3614 (Bitmap : in out Character_Class;
3615 C : Character)
3616 is
3617 Value : constant Class_Byte := Character'Pos (C);
3618 begin
3619 Bitmap (Value / 8) := Bitmap (Value / 8)
3620 or Bit_Conversion (Value mod 8);
3621 end Set_In_Class;
3622
3623 -------------------
3624 -- String_Length --
3625 -------------------
3626
3627 function String_Length
3628 (Program : Program_Data;
3629 P : Pointer) return Program_Size
3630 is
3631 begin
3632 pragma Assert (Program (P) = EXACT or else Program (P) = EXACTF);
3633 return Character'Pos (Program (P + 3));
3634 end String_Length;
3635
3636 --------------------
3637 -- String_Operand --
3638 --------------------
3639
3640 function String_Operand (P : Pointer) return Pointer is
3641 begin
3642 return P + 4;
3643 end String_Operand;
3644
3645 end System.Regpat;
Mirror of the offical gcc git repository hosted at gcc.gnu.org