source: proto/SymbolTable/parabix2_symtab_pbs_div.py @ 1228

Last change on this file since 1228 was 1228, checked in by vla24, 8 years ago

Integrated symbol table with xmlwf. There are various implementations for the symbol table, please read /proto/SymbolTable/README_SymbolTable for more information.

File size: 22.6 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 
48  FFFE_FFFF = 0
49  error = 0
50
51class Lex ():
52        CR = 0
53        LF = 0
54        HT = 0
55        SP = 0
56        CRLF = 0
57        RefStart = 0
58        Semicolon = 0 
59        Colon = 0
60        LAngle = 0
61        RAngle = 0
62        LBracket = 0
63        RBracket = 0
64        Exclam = 0
65        QMark = 0
66        Hyphen = 0
67        Equals = 0
68        SQuote = 0
69        DQuote = 0
70        Slash = 0
71        Hash = 0
72        x = 0
73        ASCII_name_start = 0
74        ASCII_name_char = 0
75        NameScan = 0
76        Digit = 0
77        Hex = 0
78        WS = 0
79        error = 0
80
81class Scope1 ():
82        RefStart = 0
83        LAngle = 0
84        Hyphen = 0
85        QMark = 0
86        RBracket = 0
87
88class CtCDPI_Callouts():
89        CD_end = 0
90        Ct_starts = 0
91        Ct_ends = 0
92        CD_starts = 0
93        CD_ends = 0
94        PI_starts = 0
95        PI_name_starts = 0
96        PI_name_ends = 0
97        PI_ends = 0
98        CtCDPI_mask = 0
99        error = 0
100
101class Ref_Callouts():
102        GenRef_starts = 0
103        GenRef_ends = 0
104        DecRef_starts = 0
105        DecRef_ends = 0
106        HexRef_starts = 0
107        HexRef_ends = 0
108        error = 0
109
110class Hash_data():
111        Hash_value = 0
112
113class Tag_Callouts():
114        ElemName_starts = 0
115        ElemName_ends = 0
116        ElemName_ends_1_to_4 = 0
117        ElemName_ends_5_to_8 = 0
118        ElemName_ends_9_to_12 = 0
119        ElemName_ends_13_to_16 = 0
120        ElemName_ends_17_and_longer = 0
121        AttName_starts = 0
122        AttName_ends = 0
123        AttVal_starts = 0
124        AttVal_ends = 0
125        AttVal_spans = 0
126        EmptyTag_marks = 0
127        EndTag_marks = 0
128        LAngleFollow = 0
129        error = 0
130
131class Basis_bits():     
132        bit_0 = 0
133        bit_1 = 0
134        bit_2 = 0
135        bit_3 = 0
136        bit_4 = 0
137        bit_5 = 0
138        bit_6 = 0
139        bit_7 = 0
140       
141class Check_streams():
142        misc_mask = 0
143        non_ascii_name_starts = 0
144        non_ascii_names = 0
145        tag_marks = 0 
146        name_follows = 0 
147        att_refs = 0 
148        error_mask = 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       
248        ### Validate_utf8(basis_bits, u8):
249        u8.unibyte = (~basis_bits.bit_0);
250        u8.suffix = 0
251        u8.error = 0
252        u8.FFFE_FFFF = 0
253        u8anyscope = 0 #local
254        if basis_bits.bit_0:
255                u8.prefix = (basis_bits.bit_0 & basis_bits.bit_1);
256                u8.prefix2 = (u8.prefix &~ basis_bits.bit_2);
257                u8.prefix3 = (u8.prefix & temp2);
258                u8.prefix4 = (u8.prefix & temp7);
259                u8.suffix = (basis_bits.bit_0 &~ basis_bits.bit_1);
260                temp66 = (u8.prefix &~ temp49);
261                temp67 = (temp21 | basis_bits.bit_6);
262                temp68 = (temp66 &~ temp67);
263                temp69 = (basis_bits.bit_5 & temp13);
264                temp70 = (basis_bits.bit_4 | temp69);
265                temp71 = (u8.prefix4 & temp70);
266                u8.badprefix = (temp68 | temp71);
267                u8.error = u8.badprefix
268                u8.scope22 = bitutil.Advance(u8.prefix2)
269                u8anyscope = u8.scope22
270                if u8.prefix3 | u8.prefix4:
271                        xE0 = (u8.prefix3 &~ temp36);
272                        xED = (u8.prefix3 & temp20);
273                        xF0 = (u8.prefix4 &~ temp36);
274                        temp72 = (temp4 &~ temp13);
275                        xF4 = (u8.prefix4 & temp72);
276                        u8.xA0_xBF = (u8.suffix & basis_bits.bit_2);
277                        u8.x80_x9F = (u8.suffix &~ basis_bits.bit_2);
278                        u8.x90_xBF = (u8.suffix & temp49);
279                        u8.x80_x8F = (u8.suffix &~ temp49);
280                        xEF = (u8.prefix3 & temp23);
281                        temp73 = (u8.suffix & temp7);
282                        u8.xBF = (temp73 & temp23);
283                        u8.xBE = (temp73 & temp15);
284                        u8.xE0_scope = bitutil.Advance(xE0);
285                        u8.xED_scope = bitutil.Advance(xED);
286                        u8.xF0_scope = bitutil.Advance(xF0);
287                        u8.xF4_scope = bitutil.Advance(xF4);
288                        u8.xEF_scope = bitutil.Advance(xEF);
289                        u8.scope32 = bitutil.Advance(u8.prefix3)
290                        u8.scope33 = bitutil.Advance(u8.scope32)
291                        u8.scope42 = bitutil.Advance(u8.prefix4)
292                        u8.scope43 = bitutil.Advance(u8.scope42)
293                        u8.scope44 = bitutil.Advance(u8.scope43)
294
295                        u8lastscope = u8.scope22 | u8.scope33 | u8.scope44
296                        u8anyscope = u8lastscope | u8.scope32 | u8.scope42 | u8.scope43
297               
298                        u8error1 = u8.xE0_scope & u8.x80_x9F
299                        u8error2 = u8.xED_scope & u8.xA0_xBF
300                        u8error3 = u8.xF0_scope & u8.x80_x8F
301                        u8error4 = u8.xF4_scope & u8.x90_xBF
302       
303                        u8.error |= u8error1 | u8error2 | u8error3 | u8error4
304
305                        EF_BF_pending = bitutil.Advance(u8.xEF_scope & u8.xBF)
306
307                        u8.FFFE_FFFF = (EF_BF_pending & (u8.xBE | u8.xBF))
308        u8mismatch = u8anyscope ^ u8.suffix
309        u8.error |= u8mismatch
310       
311def Add_scope_streams(lex, scope1):
312        #scope1.LAngle = bitutil.Advance(lex.LAngle)
313        #scope1.Hyphen = bitutil.Advance(lex.Hyphen)
314        #scope1.QMark = bitutil.Advance(lex.QMark)
315        v = lex.LAngle | lex.Hyphen
316        w = lex.Hyphen | lex.QMark
317        v1 = bitutil.Advance(v)
318        w1 = bitutil.Advance(w)
319        scope1.LAngle = v1 &~ w1
320        scope1.Hyphen = v1 & w1
321        scope1.QMark = w1 &~ v1
322
323def Parse_CtCDPI(ctCDPI_Callouts, lex, scope1, check_streams):
324        ctCDPI_Callouts.CD_end = 0
325        ctCDPI_Callouts.Ct_starts = 0
326        ctCDPI_Callouts.Ct_ends = 0
327        ctCDPI_Callouts.CD_starts = 0
328        ctCDPI_Callouts.CD_ends = 0
329        ctCDPI_Callouts.PI_starts = 0
330        ctCDPI_Callouts.PI_name_starts = 0
331        ctCDPI_Callouts.PI_name_ends = 0
332        ctCDPI_Callouts.PI_ends = 0
333        ctCDPI_Callouts.CtCDPI_mask = 0
334        ctCDPI_Callouts.error = 0
335        CtCDPI_starts = 0
336        Ct_errors = 0
337
338        if lex.RBracket:
339                scope1.RBracket = bitutil.Advance(lex.RBracket)
340                ctCDPI_Callouts.CD_end = bitutil.Advance(scope1.RBracket & lex.RBracket) & lex.RAngle
341        PI_start = scope1.LAngle & lex.QMark
342        CtCD_start = scope1.LAngle & lex.Exclam
343        CtCDPI_start = PI_start | CtCD_start
344
345        DoubleHyphen = scope1.Hyphen & lex.Hyphen
346        PI_end = scope1.QMark & lex.RAngle
347
348
349        #
350        # Initiate the scan
351        CtCDPI_Cursor = bitutil.ScanToFirst(CtCDPI_start)
352        while CtCDPI_Cursor:
353                CtCDPI_starts |= CtCDPI_Cursor
354                PI_Cursor = CtCDPI_Cursor & PI_start
355                CD_Ct_Cursor = bitutil.Advance(CtCDPI_Cursor & ~PI_Cursor)
356                CD_Cursor = CD_Ct_Cursor & lex.LBracket
357                Ct_Cursor = CD_Ct_Cursor & lex.Hyphen
358                ctCDPI_Callouts.PI_starts |= PI_Cursor
359                ctCDPI_Callouts.CD_starts |= CD_Cursor
360                ctCDPI_Callouts.Ct_starts |= Ct_Cursor
361                Ct_Cursor = bitutil.Advance(Ct_Cursor) 
362                Ct_errors |= Ct_Cursor & ~ lex.Hyphen
363                # Advance twice past <!--, so that we don't treat <!---
364                # as being a terminated comment.
365                Ct_Cursor = bitutil.Advance(bitutil.Advance(Ct_Cursor))
366                PI_Cursor = bitutil.Advance(PI_Cursor)
367                ctCDPI_Callouts.PI_name_starts |= PI_Cursor
368                PI_name_end = bitutil.ScanThru(PI_Cursor, lex.NameScan)
369                ctCDPI_Callouts.PI_name_ends |= PI_name_end
370                PI_Cursor = bitutil.ScanTo(PI_name_end, PI_end)
371                CD_Cursor = bitutil.ScanTo(CD_Cursor, ctCDPI_Callouts.CD_end)
372                Ct_Cursor = bitutil.Advance(bitutil.ScanTo(Ct_Cursor, DoubleHyphen))
373                ctCDPI_Callouts.PI_ends |= PI_Cursor
374                ctCDPI_Callouts.CD_ends |= CD_Cursor
375                ctCDPI_Callouts.Ct_ends |= Ct_Cursor
376                CtCDPI_Cursor = PI_Cursor | CD_Cursor | Ct_Cursor
377                CtCDPI_Cursor = bitutil.ScanTo(CtCDPI_Cursor, CtCDPI_start)
378       
379                ctCDPI_Callouts.CtCDPI_mask = bitutil.Advance(ctCDPI_Callouts.CD_ends | ctCDPI_Callouts.Ct_ends | ctCDPI_Callouts.PI_ends) - CtCDPI_starts             
380                #ctCDPI_Callouts.error = Ct_ends & ~lex.RAngle | Ct_starts & ~ lex.Hyphen
381                ctCDPI_Callouts.error = Ct_errors | ctCDPI_Callouts.Ct_ends & ~lex.RAngle
382                ctCDPI_Callouts.error |= bitutil.Advance(ctCDPI_Callouts.PI_name_ends & ~ lex.WS) & ~ PI_end
383                ctCDPI_Callouts.error |= ctCDPI_Callouts.PI_name_starts & ctCDPI_Callouts.PI_name_ends
384                # If any of the Comment, CDATA or PI markups are unterminated, it is an error.
385                ctCDPI_Callouts.error |= ctCDPI_Callouts.CtCDPI_mask &~ EOF_mask
386               
387        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
388
389def Parse_tags(lex, scope1, ctCDPI_Callouts, tag_Callouts):
390
391       
392        # Delimiters for scans.
393        DQuoteDelim = lex.DQuote | lex.LAngle
394        SQuoteDelim = lex.SQuote | lex.LAngle
395        AttListDelim = lex.Slash | lex.RAngle
396       
397        # Start the parallel parsing by inspecting the character
398        # after the opening "<" of a tag.
399        tag_Callouts.LAngleFollow = scope1.LAngle &~ ctCDPI_Callouts.CtCDPI_mask
400        tag_Callouts.ElemName_starts = tag_Callouts.LAngleFollow & ~lex.Slash
401        tag_Callouts.EndTag_marks = tag_Callouts.LAngleFollow & lex.Slash
402       
403        # Start Tag/Empty Element Tag Parsing
404
405        # Advance all cursors by scanning through the tag name.
406        tag_Callouts.ElemName_ends = bitutil.ScanThru(tag_Callouts.ElemName_starts, lex.NameScan)
407        # Must have at least one name character for a legal start tag.
408        # Mark any occurrences of null names as errors.
409        ParseError = tag_Callouts.ElemName_starts & tag_Callouts.ElemName_ends
410       
411        # Initialize the accumulators for attribute name and value positions.
412        tag_Callouts.AttName_starts = 0 
413        tag_Callouts.AttName_ends = 0
414        EqToCheck = 0
415        tag_Callouts.AttVal_starts = 0
416        AttValEnds = 0
417        tag_Callouts.AttVal_ends = 0
418
419        # After the element name, there may or may not be an attlist.
420        AfterWS = bitutil.ScanThru(tag_Callouts.ElemName_ends, lex.WS)
421        AttListEnd = AfterWS & AttListDelim
422        AttNameStart = AfterWS & ~AttListDelim
423        # At least one WS character is required between ElemNames and AttNames.
424        ParseError |= tag_Callouts.ElemName_ends & AttNameStart
425
426        #
427        # The following loop iterates through attributes within a start tag.
428        # Because all start tags are processed in parallel, the number of
429        # iterations is the maximum number of attributes found in any one
430        # start tag, plus one.
431        while AttNameStart:
432                tag_Callouts.AttName_starts |= AttNameStart
433                AttNameFollow = bitutil.ScanThru(AttNameStart, lex.NameScan)
434                tag_Callouts.AttName_ends |= AttNameFollow
435                # Scan through WS to the expected '=' delimiter.
436                EqExpected = bitutil.ScanThru(AttNameFollow, lex.WS)
437                EqToCheck |= EqExpected
438                AttValPos = bitutil.ScanThru(EqExpected, EqExpected | lex.WS)
439                tag_Callouts.AttVal_starts |= AttValPos
440                DQuoteAttVal = AttValPos & lex.DQuote
441                SQuoteAttVal = AttValPos & lex.SQuote
442                DQuoteAttEnd = bitutil.ScanTo(DQuoteAttVal, DQuoteDelim &~ DQuoteAttVal)
443                SQuoteAttEnd = bitutil.ScanTo(SQuoteAttVal, SQuoteDelim &~ SQuoteAttVal)
444                AttValEnd = DQuoteAttEnd | SQuoteAttEnd
445                AttValEnds |= AttValEnd
446                AttValFollow = bitutil.Advance(AttValEnd)
447                tag_Callouts.AttVal_ends |= AttValFollow
448                AfterWS = bitutil.ScanThru(AttValFollow, lex.WS)
449                AttListEnd |= AfterWS & AttListDelim
450                AttNameStart = AfterWS & ~AttListDelim
451
452        # No more attribute values to process when AttNameStart == 0.
453        STagEnds = AttListEnd & lex.RAngle
454        # Mark any "/" characters found as the ends of empty element tags.
455        tag_Callouts.EmptyTag_marks = bitutil.Advance(AttListEnd & lex.Slash)
456       
457        # Check for errors.
458        ParseError |= tag_Callouts.AttVal_ends & tag_Callouts.AttName_starts # No intervening WS.
459        ParseError |= tag_Callouts.AttName_starts & tag_Callouts.AttName_ends # Null AttName
460        ParseError |= EqToCheck & ~lex.Equals # = not found where expected.
461        ParseError |= tag_Callouts.AttVal_starts & ~ (lex.DQuote | lex.SQuote)
462        ParseError |= AttValEnds & ~ (lex.DQuote | lex.SQuote)
463        ParseError |= tag_Callouts.EmptyTag_marks & ~lex.RAngle
464
465        # End Tag Parsing
466        EndTagEnds = bitutil.ScanThru(bitutil.ScanThru(tag_Callouts.EndTag_marks, tag_Callouts.EndTag_marks | lex.NameScan), lex.WS)
467        ParseError |= EndTagEnds & ~lex.RAngle
468        tag_Callouts.error = ParseError
469               
470        # Attribute value spans
471        tag_Callouts.AttVal_spans = tag_Callouts.AttVal_ends - tag_Callouts.AttVal_starts
472                       
473def Parse_refs(lex, scope1, ctCDPI_Callouts, ref_Callouts):
474        ref_Callouts.GenRef_starts = 0
475        ref_Callouts.GenRef_ends = 0
476        ref_Callouts.DecRef_starts = 0
477        ref_Callouts.DecRef_ends = 0
478        ref_Callouts.HexRef_starts = 0
479        ref_Callouts.HexRef_ends = 0
480        ref_Callouts.error = 0
481
482        Ref1 = lex.RefStart &~ ctCDPI_Callouts.CtCDPI_mask
483        # All remaining "&" must be reference start characters; parse them.
484        if Ref1:
485                scope1.RefStart = bitutil.Advance(Ref1)
486                NumRef2 = scope1.RefStart & lex.Hash
487                ref_Callouts.GenRef_starts = scope1.RefStart &~ lex.Hash
488                NumRef3 = bitutil.Advance(NumRef2)
489                HexRef3 = NumRef3 & lex.x
490                ref_Callouts.DecRef_starts = NumRef3 &~ lex.x
491                ref_Callouts.HexRef_starts = bitutil.Advance(HexRef3) 
492                ref_Callouts.GenRef_ends = bitutil.ScanThru(ref_Callouts.GenRef_starts, lex.NameScan)
493                ref_Callouts.DecRef_ends = bitutil.ScanThru(ref_Callouts.DecRef_starts, lex.Digit)
494                ref_Callouts.HexRef_ends = bitutil.ScanThru(ref_Callouts.HexRef_starts, lex.Hex)
495                # Error checks
496                # At least one digit required for DecRef, one hex digit for HexRef.
497                ref_error1 = ref_Callouts.DecRef_starts &~ lex.Digit
498                ref_error2 = ref_Callouts.HexRef_starts &~ lex.Hex
499                # Semicolon terminator required (also covers unterminated at EOF).
500                ref_ends = ref_Callouts.GenRef_ends | ref_Callouts.DecRef_ends | ref_Callouts.HexRef_ends
501                ref_error3 = ref_ends &~ lex.Semicolon
502                ref_Callouts.error = ref_error1 | ref_error2 | ref_error3
503
504def Validate_xml_names(ctCDPI_Callouts, ref_Callouts, tag_Callouts, lex, u8, xml_names, check_streams):
505        PI_names = ctCDPI_Callouts.PI_name_ends - ctCDPI_Callouts.PI_name_starts
506        GenRefs = ref_Callouts.GenRef_ends - ref_Callouts.GenRef_starts
507        ElemNames = tag_Callouts.ElemName_ends - tag_Callouts.ElemName_starts
508        AttNames = tag_Callouts.AttName_ends - tag_Callouts.AttName_starts
509        qname_stream =  ElemNames | AttNames
510        ncname_stream = PI_names | GenRefs
511        name_stream = qname_stream | ncname_stream
512        name_start = name_stream &~ bitutil.Advance(name_stream)
513        name_cursor = name_stream & ~bitutil.Advance(name_stream)
514        void_prefix_err = name_cursor & lex.Colon
515        namespace_sep = bitutil.ScanThru(name_cursor, lex.NameScan &~ lex.Colon) & lex.Colon
516        local_part_start = bitutil.Advance(namespace_sep)
517        local_part_err = local_part_start &~ lex.NameScan
518        colon2_err = bitutil.ScanThru(local_part_start, lex.NameScan &~ lex.Colon) & lex.Colon
519        ncname_err = ncname_stream & lex.Colon
520        xml_names.namespace_error = void_prefix_err | local_part_err | colon2_err | ncname_err
521                       
522        check_streams.non_ascii_name_starts = name_start &~lex.ASCII_name_start
523        check_streams.non_ascii_names = (name_stream &~ name_start) & ~lex.ASCII_name_char & ~u8.suffix
524   
525def Do_check_streams(ctCDPI_Callouts, tag_Callouts, lex, u8, scope1, ref_Callouts, xml_names, check_streams):
526    # Ensure that no occurrence of ]]> occurs outside of markup.
527    CD_end_error = ctCDPI_Callouts.CD_end & ~(ctCDPI_Callouts.CtCDPI_mask | tag_Callouts.AttVal_spans)
528           
529    # Consolidate and check for errors
530    check_streams.error_mask = lex.error & EOF_mask | u8.error | u8.FFFE_FFFF | ctCDPI_Callouts.error | tag_Callouts.error | CD_end_error | ref_Callouts.error | xml_names.namespace_error
531
532    check_streams.tag_marks = tag_Callouts.EmptyTag_marks | tag_Callouts.LAngleFollow | tag_Callouts.AttName_starts
533    check_streams.name_follows = tag_Callouts.ElemName_ends | tag_Callouts.AttName_ends
534    check_streams.att_refs = tag_Callouts.AttVal_spans & scope1.RefStart
535
536def Form_Length_Group_Bitstreams(tag_Callouts):
537
538    remaining_starts = tag_Callouts.ElemName_starts
539    remaining_ends = tag_Callouts.ElemName_ends
540    temp = tag_Callouts.ElemName_starts
541
542    # Group symbols of length 1
543    temp = bitutil.Advance(temp)
544    tag_Callouts.ElemName_ends_1 = temp & remaining_ends
545    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_1
546
547    # Group symbols of length 2
548    temp = bitutil.Advance(temp)
549    tag_Callouts.ElemName_ends_2 = temp & remaining_ends
550    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_2
551
552    # Group symbols of length 3
553    temp = bitutil.Advance(temp)
554    tag_Callouts.ElemName_ends_3 = temp & remaining_ends
555    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_3
556
557    # Group symbols of length 4
558    temp = bitutil.Advance(temp)
559    tag_Callouts.ElemName_ends_4 = temp & remaining_ends
560    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_4
561
562    # Group symbols of length 5
563    temp = bitutil.Advance(temp)
564    tag_Callouts.ElemName_ends_5 = temp & remaining_ends
565    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_5
566
567    # Group symbols of length 6
568    temp = bitutil.Advance(temp)
569    tag_Callouts.ElemName_ends_6 = temp & remaining_ends
570    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_6
571
572    # Group symbols of length 7
573    temp = bitutil.Advance(temp)
574    tag_Callouts.ElemName_ends_7 = temp & remaining_ends
575    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_7
576
577    # Group symbols of length 8
578    temp = bitutil.Advance(temp)
579    tag_Callouts.ElemName_ends_8 = temp & remaining_ends
580    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_8
581
582    # Group symbols of length 9
583    temp = bitutil.Advance(temp)
584    tag_Callouts.ElemName_ends_9 = temp & remaining_ends
585    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_9
586
587    # Group symbols of length 10
588    temp = bitutil.Advance(temp)
589    tag_Callouts.ElemName_ends_10 = temp & remaining_ends
590    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_10
591
592    # Group symbols of length 11
593    temp = bitutil.Advance(temp)
594    tag_Callouts.ElemName_ends_11 = temp & remaining_ends
595    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_11
596
597    # Group symbols of length 12
598    temp = bitutil.Advance(temp)
599    tag_Callouts.ElemName_ends_12 = temp & remaining_ends
600    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_12
601
602    # Group symbols of length 13
603    temp = bitutil.Advance(temp)
604    tag_Callouts.ElemName_ends_13 = temp & remaining_ends
605    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_13
606
607    # Group symbols of length 14
608    temp = bitutil.Advance(temp)
609    tag_Callouts.ElemName_ends_14 = temp & remaining_ends
610    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_14
611
612    # Group symbols of length 15
613    temp = bitutil.Advance(temp)
614    tag_Callouts.ElemName_ends_15 = temp & remaining_ends
615    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_15
616
617    # Group symbols of length 16
618    temp = bitutil.Advance(temp)
619    tag_Callouts.ElemName_ends_16 = temp & remaining_ends
620    remaining_ends = remaining_ends & ~tag_Callouts.ElemName_ends_16
621
622    # Group symbols of length 17 and longer
623    tag_Callouts.ElemName_ends_17_and_longer = remaining_ends
624
625
626def Compute_Hash_Value_Bitstream(hash_data, basis_bits):
627    hash_data.Hash_value = basis_bits.bit_2 ^ basis_bits.bit_4 ^ basis_bits.bit_6
628    #hash_data.Hash_value = basis_bits.bit_3 ^ basis_bits.bit_5 ^ basis_bits.bit_7
629
630#def main(basis_bits, lex, u8, scope1, ctCDPI_Callouts, masks, check_streams, tag_Callouts, ref_Callouts, xml_names):
631def Main(basis_bits, lex, u8, xml_char, scope1, ctCDPI_Callouts, ref_Callouts, tag_Callouts, masks, xml_names, check_streams, hash_data):
632       
633        # Classify bytes for UTF-8 processing, whitespace and control
634        # processing and XML lexical analysis.
635        # Classify_bytes(basis_bits, lex)
636
637        # Validate UTF-8 multibyte sequences and determine the UTF-8 scope streams
638        # Validate_utf8(basis_bits, u8)
639                               
640        Classify_bytes_Validate_utf8(basis_bits, lex, u8)
641
642        Add_scope_streams(lex, scope1)
643   
644        # Parse all comments, CDATA sections and processing instructions.
645        Parse_CtCDPI(ctCDPI_Callouts, lex, scope1, check_streams)
646               
647        # All remaining '<' must be tag start characters; parse tags.
648        Parse_tags(lex, scope1, ctCDPI_Callouts, tag_Callouts) 
649
650        # All remaining '&' must be reference start characters; parse them.
651        Parse_refs(lex, scope1, ctCDPI_Callouts, ref_Callouts)
652       
653        # Validate XML namespaces and generate bit streams to post validate non-ascii range XML names
654        Validate_xml_names(ctCDPI_Callouts, ref_Callouts, tag_Callouts, lex, u8, xml_names, check_streams)
655   
656   
657        Do_check_streams(ctCDPI_Callouts, tag_Callouts, lex, u8, scope1, ref_Callouts, xml_names, check_streams)
658
659        # These methods are needed to do Paralel Bitstream Based Length Sorting
660        Form_Length_Group_Bitstreams(tag_Callouts)
661
662        Compute_Hash_Value_Bitstream(hash_data, basis_bits);
663               
Note: See TracBrowser for help on using the repository browser.