source: proto/SymbolTable/parabix2_symtab_pbs_div.py @ 1262

Last change on this file since 1262 was 1232, checked in by vla24, 8 years ago

Updated symbol table implementation.

File size: 23.1 KB
Line 
1# -*- coding: utf-8 -*-
2#
3# parabix2_compilable.py
4#
5# Parallel XML Parsing with Bitstream Addition
6#
7# - Complete prototype for all bitstream computations in Parabix2
8# - Optimized for compilation
9# - Separate compilation
10
11# Robert D. Cameron
12# July 29, 2010
13#
14
15#import bitutil
16
17class u8 ():
18  unibyte = 0
19  prefix = 0
20  prefix2 = 0
21  prefix3 = 0
22  prefix4 = 0
23  suffix = 0
24  badprefix = 0
25  xE0 = 0
26  xED = 0
27  xF0 = 0
28  xF4 = 0
29  xA0_xBF = 0
30  x80_x9F = 0
31  x90_xBF = 0
32  x80_x8F = 0
33  xEF = 0
34  xBF = 0
35  xBE = 0
36  scope22 = 0
37  scope32 = 0
38  scope33 = 0
39  scope42 = 0
40  scope43 = 0
41  scope44 = 0
42  xE0_scope = 0
43  xED_scope = 0
44  xF0_scope = 0
45  xF4_scope = 0
46  xEF_scope = 0
47
48class Lex ():
49        CR = 0
50        LF = 0
51        HT = 0
52        SP = 0
53        CRLF = 0
54        RefStart = 0
55        Semicolon = 0
56        Colon = 0
57        LAngle = 0
58        RAngle = 0
59        LBracket = 0
60        RBracket = 0
61        Exclam = 0
62        QMark = 0
63        Hyphen = 0
64        Equals = 0
65        SQuote = 0
66        DQuote = 0
67        Slash = 0
68        Hash = 0
69        x = 0
70        ASCII_name_start = 0
71        ASCII_name_char = 0
72        NameScan = 0
73        Digit = 0
74        Hex = 0
75        WS = 0
76
77class Scope1 ():
78        RefStart = 0
79        LAngle = 0
80        Hyphen = 0
81        QMark = 0
82        RBracket = 0
83
84class CtCDPI_Callouts():
85        CD_end = 0
86        Ct_starts = 0
87        Ct_ends = 0
88        CD_starts = 0
89        CD_ends = 0
90        PI_starts = 0
91        PI_name_starts = 0
92        PI_name_ends = 0
93        PI_ends = 0
94        CtCDPI_mask = 0
95
96class Ref_Callouts():
97        GenRef_starts = 0
98        GenRef_ends = 0
99        DecRef_starts = 0
100        DecRef_ends = 0
101        HexRef_starts = 0
102        HexRef_ends = 0
103
104class Tag_Callouts():
105        ElemName_starts = 0
106        ElemName_ends = 0
107        ElemName_ends_1 = 0
108        ElemName_ends_2 = 0
109        ElemName_ends_3 = 0
110        ElemName_ends_4 = 0
111        ElemName_ends_5 = 0
112        ElemName_ends_6 = 0
113        ElemName_ends_7 = 0
114        ElemName_ends_8 = 0
115        ElemName_ends_9 = 0
116        ElemName_ends_10 = 0
117        ElemName_ends_11 = 0
118        ElemName_ends_12 = 0
119        ElemName_ends_13 = 0
120        ElemName_ends_14 = 0
121        ElemName_ends_15 = 0
122        ElemName_ends_16 = 0
123        ElemName_ends_17_and_longer = 0
124        AttName_starts = 0
125        AttName_ends = 0
126        AttVal_starts = 0
127        AttVal_ends = 0
128        AttVal_spans = 0
129        EmptyTag_marks = 0
130        EndTag_marks = 0
131
132class Basis_bits():
133        bit_0 = 0
134        bit_1 = 0
135        bit_2 = 0
136        bit_3 = 0
137        bit_4 = 0
138        bit_5 = 0
139        bit_6 = 0
140        bit_7 = 0
141
142class Check_streams():
143        misc_mask = 0
144        non_ascii_name_starts = 0
145        non_ascii_names = 0
146        tag_marks = 0
147        name_follows = 0
148        att_refs = 0
149
150class Xml_names():
151        namespace_error = 0
152
153def Classify_bytes_Validate_utf8(basis_bits, lex, u8):
154        temp1 = (basis_bits.bit_0 | basis_bits.bit_1);
155        temp2 = (basis_bits.bit_2 &~ basis_bits.bit_3);
156        temp3 = (temp2 &~ temp1);
157        temp4 = (basis_bits.bit_5 &~ basis_bits.bit_4);
158        temp5 = (basis_bits.bit_6 &~ basis_bits.bit_7);
159        temp6 = (temp4 & temp5);
160        lex.RefStart = (temp3 & temp6);
161        temp7 = (basis_bits.bit_2 & basis_bits.bit_3);
162        temp8 = (temp7 &~ temp1);
163        temp9 = (basis_bits.bit_4 &~ basis_bits.bit_5);
164        temp10 = (basis_bits.bit_6 & basis_bits.bit_7);
165        temp11 = (temp9 & temp10);
166        lex.Semicolon = (temp8 & temp11);
167        temp12 = (basis_bits.bit_4 & basis_bits.bit_5);
168        temp13 = (basis_bits.bit_6 | basis_bits.bit_7);
169        temp14 = (temp12 &~ temp13);
170        lex.LAngle = (temp8 & temp14);
171        temp15 = (temp12 & temp5);
172        lex.RAngle = (temp8 & temp15);
173        temp16 = (basis_bits.bit_1 &~ basis_bits.bit_0);
174        temp17 = (basis_bits.bit_3 &~ basis_bits.bit_2);
175        temp18 = (temp16 & temp17);
176        lex.LBracket = (temp18 & temp11);
177        temp19 = (basis_bits.bit_7 &~ basis_bits.bit_6);
178        temp20 = (temp12 & temp19);
179        lex.RBracket = (temp18 & temp20);
180        temp21 = (basis_bits.bit_4 | basis_bits.bit_5);
181        temp22 = (temp19 &~ temp21);
182        lex.Exclam = (temp3 & temp22);
183        temp23 = (temp12 & temp10);
184        lex.QMark = (temp8 & temp23);
185        lex.Hyphen = (temp3 & temp20);
186        lex.Equals = (temp8 & temp20);
187        temp24 = (temp4 & temp10);
188        lex.SQuote = (temp3 & temp24);
189        temp25 = (temp5 &~ temp21);
190        lex.DQuote = (temp3 & temp25);
191        lex.Slash = (temp3 & temp23);
192        temp26 = (temp10 &~ temp21);
193        lex.Hash = (temp3 & temp26);
194        temp27 = (temp16 & temp7);
195        temp28 = (temp9 &~ temp13);
196        lex.x = (temp27 & temp28);
197        temp29 = (temp9 & temp5);
198        lex.Colon = (temp8 & temp29);
199        temp30 = (temp18 & temp23);
200        temp31 = (temp30 | lex.Colon);
201        temp32 = (temp16 &~ basis_bits.bit_2);
202        temp33 = (basis_bits.bit_5 | temp10);
203        temp34 = (basis_bits.bit_4 & temp33);
204        temp35 = (~temp34);
205        temp36 = (temp21 | temp13);
206        temp37 = ((basis_bits.bit_3 & temp35)|(~(basis_bits.bit_3) & temp36));
207        temp38 = (temp32 & temp37);
208        temp39 = (temp31 | temp38);
209        temp40 = (temp16 & basis_bits.bit_2);
210        temp41 = (temp40 & temp37);
211        lex.ASCII_name_start = (temp39 | temp41);
212        temp42 = (temp30 | lex.Hyphen);
213        temp43 = (temp3 & temp15);
214        temp44 = (temp42 | temp43);
215        temp45 = (temp8 &~ temp34);
216        temp46 = (temp44 | temp45);
217        temp47 = (temp46 | temp38);
218        lex.ASCII_name_char = (temp47 | temp41);
219        lex.NameScan = (lex.ASCII_name_char | basis_bits.bit_0);
220        temp48 = (temp1 | basis_bits.bit_2);
221        x00_x1F = (~temp48);
222        temp49 = (basis_bits.bit_2 | basis_bits.bit_3);
223        temp50 = (temp1 | temp49);
224        lex.CR = (temp20 &~ temp50);
225        lex.LF = (temp29 &~ temp50);
226        temp51 = (temp9 & temp19);
227        lex.HT = (temp51 &~ temp50);
228        lex.SP = (temp3 &~ temp36);
229        temp52 = (temp20 | temp29);
230        temp53 = (temp52 | temp51);
231        temp54 = (temp53 &~ temp50);
232        lex.WS = (temp54 | lex.SP);
233        temp55 = (basis_bits.bit_5 | basis_bits.bit_6);
234        temp56 = (basis_bits.bit_4 & temp55);
235        lex.Digit = (temp8 &~ temp56);
236        temp57 = (temp16 &~ temp49);
237        temp58 = (temp57 &~ basis_bits.bit_4);
238        temp59 = (~temp10);
239        temp60 = ((basis_bits.bit_5 & temp59)|(~(basis_bits.bit_5) & temp13));
240        temp61 = (temp58 & temp60);
241        temp62 = (lex.Digit | temp61);
242        temp63 = (temp16 & temp2);
243        temp64 = (temp63 &~ basis_bits.bit_4);
244        temp65 = (temp64 & temp60);
245        lex.Hex = (temp62 | temp65);
246        lex_error = x00_x1F &~ lex.WS
247        if lex_error & EOF_mask:
248                error_tracker.NoteError("Error: illegal character", lex_error)
249
250
251        ### Validate_utf8(basis_bits, u8):
252        u8.unibyte = (~basis_bits.bit_0);
253        u8.suffix = 0
254        u8_error = 0
255        u8_FFFE_FFFF = 0
256        u8anyscope = 0 #local
257        if basis_bits.bit_0:
258                u8.prefix = (basis_bits.bit_0 & basis_bits.bit_1);
259                u8.prefix2 = (u8.prefix &~ basis_bits.bit_2);
260                u8.prefix3 = (u8.prefix & temp2);
261                u8.prefix4 = (u8.prefix & temp7);
262                u8.suffix = (basis_bits.bit_0 &~ basis_bits.bit_1);
263                temp66 = (u8.prefix &~ temp49);
264                temp67 = (temp21 | basis_bits.bit_6);
265                temp68 = (temp66 &~ temp67);
266                temp69 = (basis_bits.bit_5 & temp13);
267                temp70 = (basis_bits.bit_4 | temp69);
268                temp71 = (u8.prefix4 & temp70);
269                u8.badprefix = (temp68 | temp71);
270                u8_error = u8.badprefix
271                u8.scope22 = bitutil.Advance(u8.prefix2)
272                u8anyscope = u8.scope22
273                if u8.prefix3 | u8.prefix4:
274                        xE0 = (u8.prefix3 &~ temp36);
275                        xED = (u8.prefix3 & temp20);
276                        xF0 = (u8.prefix4 &~ temp36);
277                        temp72 = (temp4 &~ temp13);
278                        xF4 = (u8.prefix4 & temp72);
279                        u8.xA0_xBF = (u8.suffix & basis_bits.bit_2);
280                        u8.x80_x9F = (u8.suffix &~ basis_bits.bit_2);
281                        u8.x90_xBF = (u8.suffix & temp49);
282                        u8.x80_x8F = (u8.suffix &~ temp49);
283                        xEF = (u8.prefix3 & temp23);
284                        temp73 = (u8.suffix & temp7);
285                        u8.xBF = (temp73 & temp23);
286                        u8.xBE = (temp73 & temp15);
287                        u8.xE0_scope = bitutil.Advance(xE0);
288                        u8.xED_scope = bitutil.Advance(xED);
289                        u8.xF0_scope = bitutil.Advance(xF0);
290                        u8.xF4_scope = bitutil.Advance(xF4);
291                        u8.xEF_scope = bitutil.Advance(xEF);
292                        u8.scope32 = bitutil.Advance(u8.prefix3)
293                        u8.scope33 = bitutil.Advance(u8.scope32)
294                        u8.scope42 = bitutil.Advance(u8.prefix4)
295                        u8.scope43 = bitutil.Advance(u8.scope42)
296                        u8.scope44 = bitutil.Advance(u8.scope43)
297
298                        u8lastscope = u8.scope22 | u8.scope33 | u8.scope44
299                        u8anyscope = u8lastscope | u8.scope32 | u8.scope42 | u8.scope43
300
301                        u8error1 = u8.xE0_scope & u8.x80_x9F
302                        u8error2 = u8.xED_scope & u8.xA0_xBF
303                        u8error3 = u8.xF0_scope & u8.x80_x8F
304                        u8error4 = u8.xF4_scope & u8.x90_xBF
305
306                        u8_error |= u8error1 | u8error2 | u8error3 | u8error4
307
308                        EF_BF_pending = bitutil.Advance(u8.xEF_scope & u8.xBF)
309
310                        u8_FFFE_FFFF = (EF_BF_pending & (u8.xBE | u8.xBF))
311                u8mismatch = u8anyscope ^ u8.suffix
312                u8_error |= u8mismatch | u8_FFFE_FFFF
313                if u8_error:
314                        error_tracker.NoteError("UTF-8 error found", (u8_error))
315
316
317def Add_scope_streams(lex, scope1):
318        #scope1.LAngle = bitutil.Advance(lex.LAngle)
319        #scope1.Hyphen = bitutil.Advance(lex.Hyphen)
320        #scope1.QMark = bitutil.Advance(lex.QMark)
321        v = lex.LAngle | lex.Hyphen
322        w = lex.Hyphen | lex.QMark
323        v1 = bitutil.Advance(v)
324        w1 = bitutil.Advance(w)
325        scope1.LAngle = v1 &~ w1
326        scope1.Hyphen = v1 & w1
327        scope1.QMark = w1 &~ v1
328        scope1.RefStart = 0 # default
329
330def Parse_CtCDPI(ctCDPI_Callouts, lex, scope1, check_streams):
331        ctCDPI_Callouts.CD_end = 0
332        ctCDPI_Callouts.Ct_starts = 0
333        ctCDPI_Callouts.Ct_ends = 0
334        ctCDPI_Callouts.CD_starts = 0
335        ctCDPI_Callouts.CD_ends = 0
336        ctCDPI_Callouts.PI_starts = 0
337        ctCDPI_Callouts.PI_name_starts = 0
338        ctCDPI_Callouts.PI_name_ends = 0
339        ctCDPI_Callouts.PI_ends = 0
340        ctCDPI_Callouts.CtCDPI_mask = 0
341        ctCDPI_error = 0
342        CtCDPI_starts = 0
343        Ct_errors = 0
344
345        if lex.RBracket:
346                scope1.RBracket = bitutil.Advance(lex.RBracket)
347                ctCDPI_Callouts.CD_end = bitutil.Advance(scope1.RBracket & lex.RBracket) & lex.RAngle
348        PI_start = scope1.LAngle & lex.QMark
349        CtCD_start = scope1.LAngle & lex.Exclam
350        CtCDPI_start = PI_start | CtCD_start
351
352        DoubleHyphen = scope1.Hyphen & lex.Hyphen
353        PI_end = scope1.QMark & lex.RAngle
354
355
356        #
357        # Initiate the scan
358        CtCDPI_Cursor = bitutil.ScanToFirst(CtCDPI_start)
359        while CtCDPI_Cursor:
360                CtCDPI_starts |= CtCDPI_Cursor
361                PI_Cursor = CtCDPI_Cursor & PI_start
362                CD_Ct_Cursor = bitutil.Advance(CtCDPI_Cursor & ~PI_Cursor)
363                CD_Cursor = CD_Ct_Cursor & lex.LBracket
364                Ct_Cursor = CD_Ct_Cursor & lex.Hyphen
365                ctCDPI_Callouts.PI_starts |= PI_Cursor
366                ctCDPI_Callouts.CD_starts |= CD_Cursor
367                ctCDPI_Callouts.Ct_starts |= Ct_Cursor
368                Ct_Cursor = bitutil.Advance(Ct_Cursor)
369                Ct_errors |= Ct_Cursor & ~ lex.Hyphen
370                # Advance twice past <!--, so that we don't treat <!---
371                # as being a terminated comment.
372                Ct_Cursor = bitutil.Advance(bitutil.Advance(Ct_Cursor))
373                PI_Cursor = bitutil.Advance(PI_Cursor)
374                ctCDPI_Callouts.PI_name_starts |= PI_Cursor
375                PI_name_end = bitutil.ScanThru(PI_Cursor, lex.NameScan)
376                ctCDPI_Callouts.PI_name_ends |= PI_name_end
377                PI_Cursor = bitutil.ScanTo(PI_name_end, PI_end)
378                CD_Cursor = bitutil.ScanTo(CD_Cursor, ctCDPI_Callouts.CD_end)
379                Ct_Cursor = bitutil.Advance(bitutil.ScanTo(Ct_Cursor, DoubleHyphen))
380                ctCDPI_Callouts.PI_ends |= PI_Cursor
381                ctCDPI_Callouts.CD_ends |= CD_Cursor
382                ctCDPI_Callouts.Ct_ends |= Ct_Cursor
383                CtCDPI_Cursor = PI_Cursor | CD_Cursor | Ct_Cursor
384                CtCDPI_Cursor = bitutil.ScanTo(CtCDPI_Cursor, CtCDPI_start)
385
386                ctCDPI_Callouts.CtCDPI_mask = bitutil.Advance(ctCDPI_Callouts.CD_ends | ctCDPI_Callouts.Ct_ends | ctCDPI_Callouts.PI_ends) - CtCDPI_starts
387                #ctCDPI_Callouts.error = Ct_ends & ~lex.RAngle | Ct_starts & ~ lex.Hyphen
388                ctCDPI_error = Ct_errors | ctCDPI_Callouts.Ct_ends & ~lex.RAngle
389                ctCDPI_error |= bitutil.Advance(ctCDPI_Callouts.PI_name_ends & ~ lex.WS) & ~ PI_end
390                ctCDPI_error |= ctCDPI_Callouts.PI_name_starts & ctCDPI_Callouts.PI_name_ends
391                # If any of the Comment, CDATA or PI markups are unterminated, it is an error.
392                ctCDPI_error |= ctCDPI_Callouts.CtCDPI_mask &~ EOF_mask
393
394        if ctCDPI_error:
395                error_tracker.NoteError("Error in comment, CDATA or processing instruction syntax", ctCDPI_error)
396
397        check_streams.misc_mask = (lex.WS | lex.LAngle | (bitutil.Advance(ctCDPI_Callouts.Ct_ends | ctCDPI_Callouts.PI_ends) - (ctCDPI_Callouts.Ct_starts | ctCDPI_Callouts.PI_starts)) | CtCDPI_starts) & EOF_mask
398
399def Parse_tags(lex, scope1, ctCDPI_Callouts, tag_Callouts):
400
401
402        # Delimiters for scans.
403        DQuoteDelim = lex.DQuote | lex.LAngle
404        SQuoteDelim = lex.SQuote | lex.LAngle
405        AttListDelim = lex.Slash | lex.RAngle
406
407        # Start the parallel parsing by inspecting the character
408        # after the opening "<" of a tag.
409        LAngleFollow = scope1.LAngle &~ ctCDPI_Callouts.CtCDPI_mask
410        tag_Callouts.ElemName_starts = LAngleFollow & ~lex.Slash
411        tag_Callouts.EndTag_marks = LAngleFollow & lex.Slash
412
413        # Start Tag/Empty Element Tag Parsing
414
415        # Advance all cursors by scanning through the tag name.
416        tag_Callouts.ElemName_ends = bitutil.ScanThru(tag_Callouts.ElemName_starts, lex.NameScan)
417        # Must have at least one name character for a legal start tag.
418        # Mark any occurrences of null names as errors.
419        ParseError = tag_Callouts.ElemName_starts & tag_Callouts.ElemName_ends
420
421        # Initialize the accumulators for attribute name and value positions.
422        tag_Callouts.AttName_starts = 0
423        tag_Callouts.AttName_ends = 0
424        EqToCheck = 0
425        tag_Callouts.AttVal_starts = 0
426        AttValEnds = 0
427        tag_Callouts.AttVal_ends = 0
428
429        # After the element name, there may or may not be an attlist.
430        AfterWS = bitutil.ScanThru(tag_Callouts.ElemName_ends, lex.WS)
431        AttListEnd = AfterWS & AttListDelim
432        AttNameStart = AfterWS & ~AttListDelim
433        # At least one WS character is required between ElemNames and AttNames.
434        ParseError |= tag_Callouts.ElemName_ends & AttNameStart
435
436        #
437        # The following loop iterates through attributes within a start tag.
438        # Because all start tags are processed in parallel, the number of
439        # iterations is the maximum number of attributes found in any one
440        # start tag, plus one.
441        while AttNameStart:
442                tag_Callouts.AttName_starts |= AttNameStart
443                AttNameFollow = bitutil.ScanThru(AttNameStart, lex.NameScan)
444                tag_Callouts.AttName_ends |= AttNameFollow
445                # Scan through WS to the expected '=' delimiter.
446                # EqExpected = bitutil.ScanThru(AttNameFollow, lex.WS)
447                # But use if test to optimize.
448                if AttNameFollow & lex.WS:
449                        EqExpected = bitutil.ScanThru(AttNameFollow, lex.WS)
450                else: EqExpected = AttNameFollow
451                EqToCheck |= EqExpected
452                AttValPos = bitutil.ScanThru(EqExpected, EqExpected | lex.WS)
453                tag_Callouts.AttVal_starts |= AttValPos
454                DQuoteAttVal = AttValPos & lex.DQuote
455                SQuoteAttVal = AttValPos & lex.SQuote
456                DQuoteAttEnd = bitutil.ScanTo(DQuoteAttVal, DQuoteDelim &~ DQuoteAttVal)
457                SQuoteAttEnd = bitutil.ScanTo(SQuoteAttVal, SQuoteDelim &~ SQuoteAttVal)
458                AttValEnd = DQuoteAttEnd | SQuoteAttEnd
459                AttValEnds |= AttValEnd
460                AttValFollow = bitutil.Advance(AttValEnd)
461                tag_Callouts.AttVal_ends |= AttValFollow
462                #  AfterWS = bitutil.ScanThru(AttValFollow, lex.WS)
463                if AttValFollow & lex.WS:
464                        AfterWS = bitutil.ScanThru(AttValFollow, lex.WS)
465                else: AfterWS = AttValFollow
466                AfterWS = bitutil.ScanThru(AttValFollow, lex.WS)
467                AttListEnd |= AfterWS & AttListDelim
468                AttNameStart = AfterWS & ~AttListDelim
469
470        # No more attribute values to process when AttNameStart == 0.
471        STagEnds = AttListEnd & lex.RAngle
472        # Mark any "/" characters found as the ends of empty element tags.
473        tag_Callouts.EmptyTag_marks = bitutil.Advance(AttListEnd & lex.Slash)
474
475        # Check for errors.
476        ParseError |= tag_Callouts.AttVal_ends & tag_Callouts.AttName_starts # No intervening WS.
477        ParseError |= tag_Callouts.AttName_starts & tag_Callouts.AttName_ends # Null AttName
478        ParseError |= EqToCheck & ~lex.Equals # = not found where expected.
479        ParseError |= tag_Callouts.AttVal_starts & ~ (lex.DQuote | lex.SQuote)
480        ParseError |= AttValEnds & ~ (lex.DQuote | lex.SQuote)
481        ParseError |= tag_Callouts.EmptyTag_marks & ~lex.RAngle
482
483        # End Tag Parsing
484
485        EndTagEnds = bitutil.ScanThru(tag_Callouts.EndTag_marks, tag_Callouts.EndTag_marks | lex.NameScan)
486        if EndTagEnds & lex.WS:
487                EndTagEnds = bitutil.ScanThru(EndTagEnds, lex.WS)
488        ParseError |= EndTagEnds & ~lex.RAngle
489        if ParseError:
490                error_tracker.NoteError("Tag parsing error found", (ParseError))
491
492
493        # Attribute value spans
494        tag_Callouts.AttVal_spans = tag_Callouts.AttVal_ends - tag_Callouts.AttVal_starts
495
496def Parse_refs(lex, scope1, ctCDPI_Callouts, ref_Callouts):
497        ref_Callouts.GenRef_starts = 0
498        ref_Callouts.GenRef_ends = 0
499        ref_Callouts.DecRef_starts = 0
500        ref_Callouts.DecRef_ends = 0
501        ref_Callouts.HexRef_starts = 0
502        ref_Callouts.HexRef_ends = 0
503        ref_error = 0
504
505        Ref1 = lex.RefStart &~ ctCDPI_Callouts.CtCDPI_mask
506        # All remaining "&" must be reference start characters; parse them.
507        if Ref1:
508                scope1.RefStart = bitutil.Advance(Ref1)
509                NumRef2 = scope1.RefStart & lex.Hash
510                ref_Callouts.GenRef_starts = scope1.RefStart &~ lex.Hash
511                NumRef3 = bitutil.Advance(NumRef2)
512                HexRef3 = NumRef3 & lex.x
513                ref_Callouts.DecRef_starts = NumRef3 &~ lex.x
514                ref_Callouts.HexRef_starts = bitutil.Advance(HexRef3)
515                ref_Callouts.GenRef_ends = bitutil.ScanThru(ref_Callouts.GenRef_starts, lex.NameScan)
516                ref_Callouts.DecRef_ends = bitutil.ScanThru(ref_Callouts.DecRef_starts, lex.Digit)
517                ref_Callouts.HexRef_ends = bitutil.ScanThru(ref_Callouts.HexRef_starts, lex.Hex)
518                # Error checks
519                # At least one digit required for DecRef, one hex digit for HexRef.
520                ref_error1 = ref_Callouts.DecRef_starts &~ lex.Digit
521                ref_error2 = ref_Callouts.HexRef_starts &~ lex.Hex
522                # Semicolon terminator required (also covers unterminated at EOF).
523                ref_ends = ref_Callouts.GenRef_ends | ref_Callouts.DecRef_ends | ref_Callouts.HexRef_ends
524                ref_error3 = ref_ends &~ lex.Semicolon
525                ref_error = ref_error1 | ref_error2 | ref_error3
526                if ref_error:
527                        error_tracker.NoteError("Reference error found", (ref_error))
528
529
530
531def Validate_xml_names(ctCDPI_Callouts, ref_Callouts, tag_Callouts, lex, u8, xml_names, check_streams):
532        PI_names = ctCDPI_Callouts.PI_name_ends - ctCDPI_Callouts.PI_name_starts
533        GenRefs = ref_Callouts.GenRef_ends - ref_Callouts.GenRef_starts
534        ElemNames = tag_Callouts.ElemName_ends - tag_Callouts.ElemName_starts
535        AttNames = tag_Callouts.AttName_ends - tag_Callouts.AttName_starts
536        qname_stream =  ElemNames | AttNames
537        ncname_stream = PI_names | GenRefs
538        name_stream = qname_stream | ncname_stream
539        name_start = name_stream &~ bitutil.Advance(name_stream)
540        name_cursor = name_stream & ~bitutil.Advance(name_stream)
541        void_prefix_err = name_cursor & lex.Colon
542        namespace_sep = bitutil.ScanThru(name_cursor, lex.NameScan &~ lex.Colon) & lex.Colon
543        local_part_start = bitutil.Advance(namespace_sep)
544        local_part_err = local_part_start &~ lex.NameScan
545        colon2_err = bitutil.ScanThru(local_part_start, lex.NameScan &~ lex.Colon) & lex.Colon
546        ncname_err = ncname_stream & lex.Colon
547        namespace_error = void_prefix_err | local_part_err | colon2_err | ncname_err
548        if namespace_error:
549                error_tracker.NoteError("error found", namespace_error)
550
551
552        check_streams.non_ascii_name_starts = name_start &~lex.ASCII_name_start
553        check_streams.non_ascii_names = (name_stream &~ name_start) & ~lex.ASCII_name_char & ~u8.suffix
554
555def Do_check_streams(ctCDPI_Callouts, tag_Callouts, lex, u8, scope1, ref_Callouts, xml_names, check_streams):
556    # Ensure that no occurrence of ]]> occurs outside of markup.
557    CD_end_error = ctCDPI_Callouts.CD_end & ~(ctCDPI_Callouts.CtCDPI_mask | tag_Callouts.AttVal_spans)
558
559    # Consolidate and check for errors
560    if CD_end_error:
561                error_tracker.NoteError("Error: ]]> in text", CD_end_error)
562
563
564    check_streams.tag_marks = tag_Callouts.EmptyTag_marks | tag_Callouts.ElemName_starts | tag_Callouts.EndTag_marks | tag_Callouts.AttName_starts
565    check_streams.name_follows = tag_Callouts.ElemName_ends | tag_Callouts.AttName_ends
566    check_streams.att_refs = tag_Callouts.AttVal_spans & scope1.RefStart
567
568def Form_Length_Group_Bitstreams(tag_Callouts):
569
570    remaining_starts = tag_Callouts.ElemName_starts
571    remaining_ends = tag_Callouts.ElemName_ends
572    temp = tag_Callouts.ElemName_starts
573    temp32 = bitutil.Advance32(temp)
574
575    # Group symbols of length 1
576    tag_Callouts.ElemName_ends_1 = interpose32(temp, temp32, 1) & remaining_ends
577    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_1
578
579    # Group symbols of length 2
580    tag_Callouts.ElemName_ends_2 = interpose32(temp, temp32, 2) & remaining_ends
581    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_2
582
583    # Group symbols of length 3
584    tag_Callouts.ElemName_ends_3 = interpose32(temp, temp32, 3) & remaining_ends
585    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_3
586
587    # Group symbols of length 4
588    tag_Callouts.ElemName_ends_4 = interpose32(temp, temp32, 4) & remaining_ends
589    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_4
590
591    # Group symbols of length 5
592    tag_Callouts.ElemName_ends_5 = interpose32(temp, temp32, 5) & remaining_ends
593    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_5
594
595    # Group symbols of length 6
596    tag_Callouts.ElemName_ends_6 = interpose32(temp, temp32, 6) & remaining_ends
597    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_6
598
599    # Group symbols of length 7
600    tag_Callouts.ElemName_ends_7 = interpose32(temp, temp32, 7) & remaining_ends
601    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_7
602
603    # Group symbols of length 8
604    tag_Callouts.ElemName_ends_8 = interpose32(temp, temp32, 8) & remaining_ends
605    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_8
606
607    # Group symbols of length 9
608    tag_Callouts.ElemName_ends_9 = interpose32(temp, temp32, 9) & remaining_ends
609    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_9
610
611    # Group symbols of length 10
612    tag_Callouts.ElemName_ends_10 = interpose32(temp, temp32, 10) & remaining_ends
613    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_10
614
615    # Group symbols of length 11
616    tag_Callouts.ElemName_ends_11 = interpose32(temp, temp32, 11) & remaining_ends
617    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_11
618
619    # Group symbols of length 12
620    tag_Callouts.ElemName_ends_12 = interpose32(temp, temp32, 12) & remaining_ends
621    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_12
622
623    # Group symbols of length 13
624    tag_Callouts.ElemName_ends_13 = interpose32(temp, temp32, 13) & remaining_ends
625    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_13
626
627    # Group symbols of length 14
628    tag_Callouts.ElemName_ends_14 = interpose32(temp, temp32, 14) & remaining_ends
629    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_14
630
631    # Group symbols of length 15
632    temp15 = interpose32(temp, temp32, 15)
633    tag_Callouts.ElemName_ends_15 = temp15 & remaining_ends
634    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_15
635
636    # Group symbols of length 16
637    temp = bitutil.Advance(temp15)
638    tag_Callouts.ElemName_ends_16 = temp & remaining_ends
639    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_16
640
641    # Group symbols of length 17 and longer
642    tag_Callouts.ElemName_ends_17_and_longer = remaining_ends
643
644
645def Compute_Hash_Value_Bitstream(hash_data, basis_bits):
646    hash_data.Hash_value = basis_bits.bit_2 ^ basis_bits.bit_4 ^ basis_bits.bit_6
647    #hash_data.Hash_value = basis_bits.bit_3 ^ basis_bits.bit_5 ^ basis_bits.bit_7
648
649def Main(basis_bits, lex, u8, xml_char, scope1, ctCDPI_Callouts, ref_Callouts, tag_Callouts, masks, xml_names, check_streams, hash_data):
650
651        # Classify bytes for UTF-8 processing, whitespace and control
652        # processing and XML lexical analysis.
653        # Classify_bytes(basis_bits, lex)
654
655        # Validate UTF-8 multibyte sequences and determine the UTF-8 scope streams
656        # Validate_utf8(basis_bits, u8)
657
658        Classify_bytes_Validate_utf8(basis_bits, lex, u8)
659
660        Add_scope_streams(lex, scope1)
661
662        # Parse all comments, CDATA sections and processing instructions.
663        Parse_CtCDPI(ctCDPI_Callouts, lex, scope1, check_streams)
664
665        # All remaining '<' must be tag start characters; parse tags.
666        Parse_tags(lex, scope1, ctCDPI_Callouts, tag_Callouts)
667
668        # All remaining '&' must be reference start characters; parse them.
669        Parse_refs(lex, scope1, ctCDPI_Callouts, ref_Callouts)
670
671        # Validate XML namespaces and generate bit streams to post validate non-ascii range XML names
672        Validate_xml_names(ctCDPI_Callouts, ref_Callouts, tag_Callouts, lex, u8, xml_names, check_streams)
673
674
675        Do_check_streams(ctCDPI_Callouts, tag_Callouts, lex, u8, scope1, ref_Callouts, xml_names, check_streams)
676
677        # These methods are needed to do Paralel Bitstream Based Length Sorting
678        Form_Length_Group_Bitstreams(tag_Callouts)
679
680        Compute_Hash_Value_Bitstream(hash_data, basis_bits);
Note: See TracBrowser for help on using the repository browser.