1#!/usr/bin/python
2
3import os
4import re
5import time
6from pprint import pprint
7
8###
9
10c99URL = 'http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf'
11c99TOC = [('Foreword', 'xi'),
12('Introduction', 'xiv'),
13('1. Scope', '1'),
14('2. Normative references', '2'),
15('3. Terms, definitions, and symbols', '3'),
16('4. Conformance', '7'),
17('5. Environment', '9'),
18('5.1 Conceptual models', '9'),
19('5.1.1 Translation environment', '9'),
20('5.1.2 Execution environments', '11'),
21('5.2 Environmental considerations', '17'),
22('5.2.1 Character sets', '17'),
23('5.2.2 Character display semantics', '19'),
24('5.2.3 Signals and interrupts', '20'),
25('5.2.4 Environmental limits', '20'),
26('6. Language', '29'),
27('6.1 Notation', '29'),
28('6.2 Concepts', '29'),
29('6.2.1 Scopes of identifiers', '29'),
30('6.2.2 Linkages of identifiers', '30'),
31('6.2.3 Name spaces of identifiers', '31'),
32('6.2.4 Storage durations of objects', '32'),
33('6.2.5 Types', '33'),
34('6.2.6 Representations of types', '37'),
35('6.2.7 Compatible type and composite type', '40'),
36('6.3 Conversions', '42'),
37('6.3.1 Arithmetic operands', '42'),
38('6.3.2 Other operands', '46'),
39('6.4 Lexical elements', '49'),
40('6.4.1 Keywords', '50'),
41('6.4.2 Identifiers', '51'),
42('6.4.3 Universal character names', '53'),
43('6.4.4 Constants', '54'),
44('6.4.5 String literals', '62'),
45('6.4.6 Punctuators', '63'),
46('6.4.7 Header names', '64'),
47('6.4.8 Preprocessing numbers', '65'),
48('6.4.9 Comments', '66'),
49('6.5 Expressions', '67'),
50('6.5.1 Primary expressions', '69'),
51('6.5.2 Postfix operators', '69'),
52('6.5.3 Unary operators', '78'),
53('6.5.4 Cast operators', '81'),
54('6.5.5 Multiplicative operators', '82'),
55('6.5.6 Additive operators', '82'),
56('6.5.7 Bitwise shift operators', '84'),
57('6.5.8 Relational operators', '85'),
58('6.5.9 Equality operators', '86'),
59('6.5.10 Bitwise AND operator', '87'),
60('6.5.11 Bitwise exclusive OR operator', '88'),
61('6.5.12 Bitwise inclusive OR operator', '88'),
62('6.5.13 Logical AND operator', '89'),
63('6.5.14 Logical OR operator', '89'),
64('6.5.15 Conditional operator', '90'),
65('6.5.16 Assignment operators', '91'),
66('6.5.17 Comma operator', '94'),
67('6.6 Constant expressions', '95'),
68('6.7 Declarations', '97'),
69('6.7.1 Storage-class specifiers', '98'),
70('6.7.2 Type specifiers', '99'),
71('6.7.3 Type qualifiers', '108'),
72('6.7.4 Function specifiers', '112'),
73('6.7.5 Declarators', '114'),
74('6.7.6 Type names', '122'),
75('6.7.7 Type definitions', '123'),
76('6.7.8 Initialization', '125'),
77('6.8 Statements and blocks', '131'),
78('6.8.1 Labeled statements', '131'),
79('6.8.2 Compound statement', '132'),
80('6.8.3 Expression and null statements', '132'),
81('6.8.4 Selection statements', '133'),
82('6.8.5 Iteration statements', '135'),
83('6.8.6 Jump statements', '136'),
84('6.9 External definitions', '140'),
85('6.9.1 Function definitions', '141'),
86('6.9.2 External object definitions', '143'),
87('6.10 Preprocessing directives', '145'),
88('6.10.1 Conditional inclusion', '147'),
89('6.10.2 Source file inclusion', '149'),
90('6.10.3 Macro replacement', '151'),
91('6.10.4 Line control', '158'),
92('6.10.5 Error directive', '159'),
93('6.10.6 Pragma directive', '159'),
94('6.10.7 Null directive', '160'),
95('6.10.8 Predefined macro names', '160'),
96('6.10.9 Pragma operator', '161'),
97('6.11 Future language directions', '163'),
98('6.11.1 Floating types', '163'),
99('6.11.2 Linkages of identifiers', '163'),
100('6.11.3 External names', '163'),
101('6.11.4 Character escape sequences', '163'),
102('6.11.5 Storage-class specifiers', '163'),
103('6.11.6 Function declarators', '163'),
104('6.11.7 Function definitions', '163'),
105('6.11.8 Pragma directives', '163'),
106('6.11.9 Predefined macro names', '163'),
107('7. Library', '164'),
108('7.1 Introduction', '164'),
109('7.1.1 Definitions of terms', '164'),
110('7.1.2 Standard headers', '165'),
111('7.1.3 Reserved identifiers', '166'),
112('7.1.4 Use of library functions', '166'),
113('7.2 Diagnostics <assert.h>', '169'),
114('7.2.1 Program diagnostics', '169'),
115('7.3 Complex arithmetic <complex.h>', '170'),
116('7.3.1 Introduction', '170'),
117('7.3.2 Conventions', '170'),
118('7.3.3 Branch cuts', '171'),
119('7.3.4 The CX_LIMITED_RANGE pragma', '171'),
120('7.3.5 Trigonometric functions', '172'),
121('7.3.6 Hyperbolic functions', '174'),
122('7.3.7 Exponential and logarithmic functions', '176'),
123('7.3.8 Power and absolute-value functions', '177'),
124('7.3.9 Manipulation functions', '178'),
125('7.4 Character handling <ctype.h>', '181'),
126('7.4.1 Character classification functions', '181'),
127('7.4.2 Character case mapping functions', '184'),
128('7.5 Errors <errno.h>', '186'),
129('7.6 Floating-point environment <fenv.h>', '187'),
130('7.6.1 The FENV_ACCESS pragma', '189'),
131('7.6.2 Floating-point exceptions', '190'),
132('7.6.3 Rounding', '193'),
133('7.6.4 Environment', '194'),
134('7.7 Characteristics of floating types <float.h>', '197'),
135('7.8 Format conversion of integer types <inttypes.h>', '198'),
136('7.8.1 Macros for format specifiers', '198'),
137('7.8.2 Functions for greatest-width integer types', '199'),
138('7.9 Alternative spellings <iso646.h>', '202'),
139('7.10 Sizes of integer types <limits.h>', '203'),
140('7.11 Localization <locale.h>', '204'),
141('7.11.1 Locale control', '205'),
142('7.11.2 Numeric formatting convention inquiry', '206'),
143('7.12 Mathematics <math.h>', '212'),
144('7.12.1 Treatment of error conditions', '214'),
145('7.12.2 The FP_CONTRACT pragma', '215'),
146('7.12.3 Classification macros', '216'),
147('7.12.4 Trigonometric functions', '218'),
148('7.12.5 Hyperbolic functions', '221'),
149('7.12.6 Exponential and logarithmic functions', '223'),
150('7.12.7 Power and absolute-value functions', '228'),
151('7.12.8 Error and gamma functions', '230'),
152('7.12.9 Nearest integer functions', '231'),
153('7.12.10 Remainder functions', '235'),
154('7.12.11 Manipulation functions', '236'),
155('7.12.12 Maximum, minimum, and positive difference functions', '238'),
156('7.12.13 Floating multiply-add', '239'),
157('7.12.14 Comparison macros', '240'),
158('7.13 Nonlocal jumps <setjmp.h>', '243'),
159('7.13.1 Save calling environment', '243'),
160('7.13.2 Restore calling environment', '244'),
161('7.14 Signal handling <signal.h>', '246'),
162('7.14.1 Specify signal handling', '247'),
163('7.14.2 Send signal', '248'),
164('7.15 Variable arguments <stdarg.h>', '249'),
165('7.15.1 Variable argument list access macros', '249'),
166('7.16 Boolean type and values <stdbool.h>', '253'),
167('7.17 Common definitions <stddef.h>', '254'),
168('7.18 Integer types <stdint.h>', '255'),
169('7.18.1 Integer types', '255'),
170('7.18.2 Limits of specified-width integer types', '257'),
171('7.18.3 Limits of other integer types', '259'),
172('7.18.4 Macros for integer constants', '260'),
173('7.19 Input/output <stdio.h>', '262'),
174('7.19.1 Introduction', '262'),
175('7.19.2 Streams', '264'),
176('7.19.3 Files', '266'),
177('7.19.4 Operations on files', '268'),
178('7.19.5 File access functions', '270'),
179('7.19.6 Formatted input/output functions', '274'),
180('7.19.7 Character input/output functions', '296'),
181('7.19.8 Direct input/output functions', '301'),
182('7.19.9 File positioning functions', '302'),
183('7.19.10 Error-handling functions', '304'),
184('7.20 General utilities <stdlib.h>', '306'),
185('7.20.1 Numeric conversion functions', '307'),
186('7.20.2 Pseudo-random sequence generation functions', '312'),
187('7.20.3 Memory management functions', '313'),
188('7.20.4 Communication with the environment', '315'),
189('7.20.5 Searching and sorting utilities', '318'),
190('7.20.6 Integer arithmetic functions', '320'),
191('7.20.7 Multibyte/wide character conversion functions', '321'),
192('7.20.8 Multibyte/wide string conversion functions', '323'),
193('7.21 String handling <string.h>', '325'),
194('7.21.1 String function conventions', '325'),
195('7.21.2 Copying functions', '325'),
196('7.21.3 Concatenation functions', '327'),
197('7.21.4 Comparison functions', '328'),
198('7.21.5 Search functions', '330'),
199('7.21.6 Miscellaneous functions', '333'),
200('7.22 Type-generic math <tgmath.h>', '335'),
201('7.23 Date and time <time.h>', '338'),
202('7.23.1 Components of time', '338'),
203('7.23.2 Time manipulation functions', '339'),
204('7.23.3 Time conversion functions', '341'),
205('7.24 Extended multibyte and wide character utilities <wchar.h>', '348'),
206('7.24.1 Introduction', '348'),
207('7.24.2 Formatted wide character input/output functions', '349'),
208('7.24.3 Wide character input/output functions', '367'),
209('7.24.4 General wide string utilities', '371'),
210('7.24.5 Wide character time conversion functions', '385'),
211('7.24.6 Extended multibyte/wide character conversion utilities', '386'),
212('7.25 Wide character classification and mapping utilities <wctype.h>',
213  '393'),
214('7.25.1 Introduction', '393'),
215('7.25.2 Wide character classification utilities', '394'),
216('7.25.3 Wide character case mapping utilities', '399'),
217('7.26 Future library directions', '401'),
218('7.26.1 Complex arithmetic <complex.h>', '401'),
219('7.26.2 Character handling <ctype.h>', '401'),
220('7.26.3 Errors <errno.h>', '401'),
221('7.26.4 Format conversion of integer types <inttypes.h>', '401'),
222('7.26.5 Localization <locale.h>', '401'),
223('7.26.6 Signal handling <signal.h>', '401'),
224('7.26.7 Boolean type and values <stdbool.h>', '401'),
225('7.26.8 Integer types <stdint.h>', '401'),
226('7.26.9 Input/output <stdio.h>', '402'),
227('7.26.10 General utilities <stdlib.h>', '402'),
228('7.26.11 String handling <string.h>', '402'),
229('<wchar.h>', '402'),
230('<wctype.h>', '402'),
231('Annex A (informative) Language syntax summary', '403'),
232('A.1 Lexical grammar', '403'),
233('A.2 Phrase structure grammar', '409'),
234('A.3 Preprocessing directives', '416'),
235('Annex B (informative) Library summary', '418'),
236('B.1 Diagnostics <assert.h>', '418'),
237('B.2 Complex <complex.h>', '418'),
238('B.3 Character handling <ctype.h>', '420'),
239('B.4 Errors <errno.h>', '420'),
240('B.5 Floating-point environment <fenv.h>', '420'),
241('B.6 Characteristics of floating types <float.h>', '421'),
242('B.7 Format conversion of integer types <inttypes.h>', '421'),
243('B.8 Alternative spellings <iso646.h>', '422'),
244('B.9 Sizes of integer types <limits.h>', '422'),
245('B.10 Localization <locale.h>', '422'),
246('B.11 Mathematics <math.h>', '422'),
247('B.12 Nonlocal jumps <setjmp.h>', '427'),
248('B.13 Signal handling <signal.h>', '427'),
249('B.14 Variable arguments <stdarg.h>', '427'),
250('B.15 Boolean type and values <stdbool.h>', '427'),
251('B.16 Common definitions <stddef.h>', '428'),
252('B.17 Integer types <stdint.h>', '428'),
253('B.18 Input/output <stdio.h>', '428'),
254('B.19 General utilities <stdlib.h>', '430'),
255('B.20 String handling <string.h>', '432'),
256('B.21 Type-generic math <tgmath.h>', '433'),
257('B.22 Date and time <time.h>', '433'),
258('B.23 Extended multibyte/wide character utilities <wchar.h>', '434'),
259('B.24 Wide character classification and mapping utilities <wctype.h>',
260  '436'),
261('Annex C (informative) Sequence points', '438'),
262('Annex D (normative) Universal character names for identifiers', '439'),
263('Annex E (informative) Implementation limits', '441'),
264('Annex F (normative) IEC 60559 floating-point arithmetic', '443'),
265('F.1 Introduction', '443'),
266('F.2 Types', '443'),
267('F.3 Operators and functions', '444'),
268('F.4 Floating to integer conversion', '446'),
269('F.5 Binary-decimal conversion', '446'),
270('F.6 Contracted expressions', '447'),
271('F.7 Floating-point environment', '447'),
272('F.8 Optimization', '450'),
273('F.9 Mathematics <math.h>', '453'),
274('Annex G (informative) IEC 60559-compatible complex arithmetic', '466'),
275('G.1 Introduction', '466'),
276('G.2 Types', '466'),
277('G.3 Conventions', '466'),
278('G.4 Conversions', '467'),
279('G.5 Binary operators', '467'),
280('G.6 Complex arithmetic <complex.h>', '471'),
281('G.7 Type-generic math <tgmath.h>', '479'),
282('Annex H (informative) Language independent arithmetic', '480'),
283('H.1 Introduction', '480'),
284('H.2 Types', '480'),
285('H.3 Notification', '484'),
286('Annex I (informative) Common warnings', '486'),
287('Annex J (informative) Portability issues', '488'),
288('J.1 Unspecified behavior', '488'),
289('J.2 Undefined behavior', '491'),
290('J.3 Implementation-defined behavior', '504'),
291('J.4 Locale-specific behavior', '511'),
292('J.5 Common extensions', '512'),
293('Bibliography', '515'),
294('Index', '517')]
295
296cXXURL = 'http://open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2723.pdf'
297cXXTOC = [('Contents', 'ii'),
298('List of Tables', 'ix'),
299('1 General', '1'),
300('1.1 Scope', '1'),
301('1.2 Normative references', '1'),
302('1.3 Definitions', '2'),
303('1.4 Implementation compliance', '4'),
304('1.5 Structure of this International Standard', '5'),
305('1.6 Syntax notation', '5'),
306('1.7 The C++ memory model', '6'),
307('1.8 The C++ object model', '6'),
308('1.9 Program execution', '7'),
309('1.10 Multi-threaded executions and data races', '10'),
310('1.11 Acknowledgments', '13'),
311('2 Lexical conventions', '15'),
312('2.1 Phases of translation', '15'),
313('2.2 Character sets', '16'),
314('2.3 Trigraph sequences', '17'),
315('2.4 Preprocessing tokens', '17'),
316('2.5 Alternative tokens', '18'),
317('2.6 Tokens', '19'),
318('2.7 Comments', '19'),
319('2.8 Header names', '19'),
320('2.9 Preprocessing numbers', '20'),
321('2.10 Identifiers', '20'),
322('2.11 Keywords', '20'),
323('2.12 Operators and punctuators', '21'),
324('2.13 Literals', '21'),
325('3 Basic concepts', '29'),
326('3.1 Declarations and definitions', '29'),
327('3.2 One definition rule', '31'),
328('3.3 Declarative regions and scopes', '33'),
329('3.4 Name lookup', '38'),
330('3.5 Program and linkage', '51'),
331('3.6 Start and termination', '54'),
332('3.7 Storage duration', '58'),
333('3.8 Object Lifetime', '62'),
334('3.9 Types', '65'),
335('3.10 Lvalues and rvalues', '70'),
336('3.11 Alignment', '72'),
337('4 Standard conversions', '73'),
338('4.1 Lvalue-to-rvalue conversion', '74'),
339('4.2 Array-to-pointer conversion', '74'),
340('4.3 Function-to-pointer conversion', '74'),
341('4.4 Qualification conversions', '74'),
342('4.5 Integral promotions', '75'),
343('4.6 Floating point promotion', '76'),
344('4.7 Integral conversions', '76'),
345('4.8 Floating point conversions', '76'),
346('4.9 Floating-integral conversions', '77'),
347('4.10 Pointer conversions', '77'),
348('4.11 Pointer to member conversions', '77'),
349('4.12 Boolean conversions', '78'),
350('4.13 Integer conversion rank', '78'),
351('5 Expressions', '79'),
352('5.1 Primary expressions', '80'),
353('5.2 Postfix expressions', '85'),
354('5.3 Unary expressions', '96'),
355('5.4 Explicit type conversion (cast notation)', '104'),
356('5.5 Pointer-to-member operators', '105'),
357('5.6 Multiplicative operators', '106'),
358('5.7 Additive operators', '106'),
359('5.8 Shift operators', '107'),
360('5.9 Relational operators', '108'),
361('5.10 Equality operators', '109'),
362('5.11 Bitwise AND operator', '110'),
363('5.12 Bitwise exclusive OR operator', '110'),
364('5.13 Bitwise inclusive OR operator', '110'),
365('5.14 Logical AND operator', '110'),
366('5.15 Logical OR operator', '110'),
367('5.16 Conditional operator', '111'),
368('5.17 Assignment and compound assignment operators', '112'),
369('5.18 Comma operator', '113'),
370('5.19 Constant expressions', '113'),
371('6 Statements', '116'),
372('6.1 Labeled statement', '116'),
373('6.2 Expression statement', '116'),
374('6.3 Compound statement or block', '116'),
375('6.4 Selection statements', '117'),
376('6.5 Iteration statements', '118'),
377('6.6 Jump statements', '121'),
378('6.7 Declaration statement', '122'),
379('6.8 Ambiguity resolution', '123'),
380('7 Declarations', '125'),
381('7.1 Specifiers', '126'),
382('7.2 Enumeration declarations', '140'),
383('7.3 Namespaces', '143'),
384('7.4 The asm declaration', '156'),
385('7.5 Linkage specifications', '156'),
386('8 Declarators', '160'),
387('8.1 Type names', '161'),
388('8.2 Ambiguity resolution', '161'),
389('8.3 Meaning of declarators', '163'),
390('8.4 Function definitions', '175'),
391('8.5 Initializers', '177'),
392('9 Classes', '191'),
393('9.1 Class names', '193'),
394('9.2 Class members', '194'),
395('9.3 Member functions', '197'),
396('9.4 Static members', '200'),
397('9.5 Unions', '202'),
398('9.6 Bit-fields', '203'),
399('9.7 Nested class declarations', '204'),
400('9.8 Local class declarations', '205'),
401('9.9 Nested type names', '206'),
402('10 Derived classes', '207'),
403('10.1 Multiple base classes', '208'),
404('10.2 Member name lookup', '210'),
405('10.3 Virtual functions', '213'),
406('10.4 Abstract classes', '217'),
407('11 Member access control', '219'),
408('11.1 Access specifiers', '221'),
409('11.2 Accessibility of base classes and base class members', '222'),
410('11.3 Access declarations', '224'),
411('11.4 Friends', '225'),
412('11.5 Protected member access', '228'),
413('11.6 Access to virtual functions', '229'),
414('11.7 Multiple access', '230'),
415('11.8 Nested classes', '230'),
416('12 Special member functions', '231'),
417('12.1 Constructors', '231'),
418('12.2 Temporary objects', '233'),
419('12.3 Conversions', '235'),
420('12.4 Destructors', '238'),
421('12.5 Free store', '240'),
422('12.6 Initialization', '242'),
423('12.7 Construction and destruction', '247'),
424('12.8 Copying class objects', '250'),
425('12.9 Inheriting Constructors', '255'),
426('13 Overloading', '259'),
427('13.1 Overloadable declarations', '259'),
428('13.2 Declaration matching', '261'),
429('13.3 Overload resolution', '262'),
430('13.4 Address of overloaded function', '281'),
431('13.5 Overloaded operators', '282'),
432('13.6 Built-in operators', '286'),
433('14 Templates', '290'),
434('14.1 Template parameters', '291'),
435('14.2 Names of template specializations', '294'),
436('14.3 Template arguments', '296'),
437('14.4 Type equivalence', '302'),
438('14.5 Template declarations', '303'),
439('14.6 Name resolution', '318'),
440('14.7 Template instantiation and specialization', '331'),
441('14.8 Function template specializations', '343'),
442('15 Exception handling', '363'),
443('15.1 Throwing an exception', '364'),
444('15.2 Constructors and destructors', '366'),
445('15.3 Handling an exception', '366'),
446('15.4 Exception specifications', '368'),
447('15.5 Special functions', '371'),
448('15.6 Exceptions and access', '372'),
449('16 Preprocessing directives', '373'),
450('16.1 Conditional inclusion', '375'),
451('16.2 Source file inclusion', '376'),
452('16.3 Macro replacement', '377'),
453('16.4 Line control', '382'),
454('16.5 Error directive', '383'),
455('16.6 Pragma directive', '383'),
456('16.7 Null directive', '383'),
457('16.8 Predefined macro names', '383'),
458('16.9 Pragma operator', '384'),
459('17 Library introduction', '386'),
460('17.1 General', '386'),
461('17.2 Overview', '386'),
462('17.3 Definitions', '386'),
463('17.4 Additional definitions', '390'),
464('17.5 Method of description (Informative)', '390'),
465('17.6 Library-wide requirements', '396'),
466('18 Language support library', '407'),
467('18.1 Types', '407'),
468('18.2 Implementation properties', '408'),
469('18.3 Integer types', '417'),
470('18.4 Start and termination', '418'),
471('18.5 Dynamic memory management', '420'),
472('18.6 Type identification', '424'),
473('18.7 Exception handling', '427'),
474('18.8 Initializer lists', '432'),
475('18.9 Other runtime support', '434'),
476('19 Diagnostics library', '435'),
477('19.1 Exception classes', '435'),
478('19.2 Assertions', '439'),
479('19.3 Error numbers', '440'),
480('19.4 System error support', '440'),
481('20 General utilities library', '452'),
482('20.1 Requirements', '452'),
483('20.2 Utility components', '457'),
484('20.3 Compile-time rational arithmetic', '463'),
485('20.4 Tuples', '465'),
486('20.5 Metaprogramming and type traits', '473'),
487('20.6 Function objects', '486'),
488('20.7 Memory', '509'),
489('20.8 Time utilities', '548'),
490('20.9 Date and time functions', '562'),
491('21 Strings library', '563'),
492('21.1 Character traits', '563'),
493('21.2 String classes', '569'),
494('21.3 Class template basic_string', '572'),
495('21.4 Numeric Conversions', '599'),
496('21.5 Null-terminated sequence utilities', '600'),
497('22 Localization library', '604'),
498('22.1 Locales', '604'),
499('22.2 Standard locale categories', '617'),
500('22.3 Standard code conversion facets', '657'),
501('22.4 C Library Locales', '659'),
502('23 Containers library', '660'),
503('23.1 Container requirements', '660'),
504('23.2 Sequence containers', '681'),
505('23.3 Associative containers', '719'),
506('23.4 Unordered associative containers', '744'),
507('24 Iterators library', '759'),
508('24.1 Iterator requirements', '759'),
509('24.2 Header <iterator> synopsis', '764'),
510('24.3 Iterator primitives', '767'),
511('24.4 Predefined iterators', '770'),
512('24.5 Stream iterators', '784'),
513('25 Algorithms library', '792'),
514('25.1 Non-modifying sequence operations', '802'),
515('25.2 Mutating sequence operations', '806'),
516('25.3 Sorting and related operations', '815'),
517('25.4 C library algorithms', '829'),
518('26 Numerics library', '831'),
519('26.1 Numeric type requirements', '831'),
520('26.2 The floating-point environment', '832'),
521('26.3 Complex numbers', '833'),
522('26.4 Random number generation', '842'),
523('26.5 Numeric arrays', '884'),
524('26.6 Generalized numeric operations', '904'),
525('26.7 C Library', '907'),
526('27 Input/output library', '912'),
527('27.1 Iostreams requirements', '912'),
528('27.2 Forward declarations', '912'),
529('27.3 Standard iostream objects', '915'),
530('27.4 Iostreams base classes', '916'),
531('27.5 Stream buffers', '934'),
532('27.6 Formatting and manipulators', '944'),
533('27.7 String-based streams', '972'),
534('27.8 File-based streams', '984'),
535('28 Regular expressions library', '1000'),
536('28.1 Definitions', '1000'),
537('28.2 Requirements', '1000'),
538('28.3 Regular expressions summary', '1002'),
539('28.4 Header <regex> synopsis', '1003'),
540('28.5 Namespace std::regex_constants', '1009'),
541('28.6 Class regex_error', '1012'),
542('28.7 Class template regex_traits', '1012'),
543('28.8 Class template basic_regex', '1015'),
544('28.9 Class template sub_match', '1020'),
545('28.10Class template match_results', '1025'),
546('28.11Regular expression algorithms', '1029'),
547('28.12Regular expression Iterators', '1033'),
548('28.13Modified ECMAScript regular expression grammar', '1039'),
549('29 Atomic operations library', '1042'),
550('29.1 Order and Consistency', '1044'),
551('29.2 Lock-free Property', '1046'),
552('29.3 Atomic Types', '1046'),
553('29.4 Operations on Atomic Types', '1051'),
554('29.5 Flag Type and Operations', '1054'),
555('30 Thread support library', '1057'),
556('30.1 Requirements', '1057'),
557('30.2 Threads', '1058'),
558('30.3 Mutual exclusion', '1063'),
559('30.4 Condition variables', '1077'),
560('A Grammar summary', '1085'),
561('A.1 Keywords', '1085'),
562('A.2 Lexical conventions', '1085'),
563('A.3 Basic concepts', '1089'),
564('A.4 Expressions', '1090'),
565('A.5 Statements', '1093'),
566('A.6 Declarations', '1094'),
567('A.7 Declarators', '1097'),
568('A.8 Classes', '1098'),
569('A.9 Derived classes', '1099'),
570('A.10 Special member functions', '1099'),
571('A.11 Overloading', '1100'),
572('A.12 Templates', '1100'),
573('A.13 Exception handling', '1101'),
574('A.14 Preprocessing directives', '1101'),
575('B Implementation quantities', '1103'),
576('C Compatibility', '1105'),
577('C.1 C++ and ISO C', '1105'),
578('C.2 Standard C library', '1114'),
579('D Compatibility features', '1119'),
580('D.1 Increment operator with bool operand', '1119'),
581('D.2 static keyword', '1119'),
582('D.3 Access declarations', '1119'),
583('D.4 Implicit conversion from const strings', '1119'),
584('D.5 C standard library headers', '1119'),
585('D.6 Old iostreams members', '1120'),
586('D.7 char* streams', '1121'),
587('D.8 Binders', '1130'),
588('D.9 auto_ptr', '1132'),
589('E Universal-character-names', '1135'),
590('F Cross references', '1137'),
591('Index', '1153')]
592
593kDocuments = {
594    'C99' : (c99URL, c99TOC, 12),
595    'C++' : (cXXURL, cXXTOC, 12),
596}
597
598def findClosestTOCEntry(data, target):
599    offset = data[2]
600    best = None
601    for (name,page) in data[1]:
602        if ' ' in name:
603            section,name = name.split(' ',1)
604            if section == 'Annex':
605                section,name = name.split(' ',1)
606                section = 'Annex '+section
607        else:
608            section = None
609        try:
610            page = int(page) + offset
611        except:
612            page = 1
613        try:
614            spec = SpecIndex.fromstring(section)
615        except:
616            spec = None
617
618        # Meh, could be better...
619        if spec is not None:
620            dist = spec - target
621            if best is None or dist < best[0]:
622                best = (dist, (section, name, page))
623    return best[1]
624
625# What a hack. Slow to boot.
626doxyLineRefRE = re.compile(r"<a name=\"l([0-9]+)\"></a>")
627def findClosestLineReference(clangRoot, doxyName, target):
628    try:
629        f = open(os.path.join(clangRoot, 'docs', 'doxygen', 'html', doxyName))
630    except:
631        return None
632
633    best = None
634    for m in doxyLineRefRE.finditer(f.read()):
635        line = int(m.group(1), 10)
636        dist = abs(line - target)
637        if best is None or dist < best[0]:
638            best = (dist,'l'+m.group(1))
639    f.close()
640    if best is not None:
641        return best[1]
642    return None
643
644###
645
646nameAndSpecRefRE = re.compile(r"(C99|C90|C\+\+|H\&S) (([0-9]+)(\.[0-9]+)*(p[0-9]+)?)")
647loneSpecRefRE = re.compile(r" (([0-9]+)(\.[0-9]+){2,100}(p[0-9]+)?)")
648def scanFile(path, filename):
649    try:
650        f = open(path)
651    except IOError:
652        print >>sys.stderr,'WARNING: Unable to open:',path
653        return
654
655    for i,ln in enumerate(f):
656        ignore = set()
657        for m in nameAndSpecRefRE.finditer(ln):
658            section = m.group(2)
659            name = m.group(1)
660            if section.endswith('.'):
661                section = section[:-1]
662            yield RefItem(name, section, filename, path, i+1)
663            ignore.add(section)
664        for m in loneSpecRefRE.finditer(ln):
665            section = m.group(1)
666            if section.endswith('.'):
667                section = section[:-1]
668            if section not in ignore:
669                yield RefItem(None, section, filename, path, i+1)
670
671###
672
673class SpecIndex:
674    @staticmethod
675    def fromstring(str):
676        secs = str.split('.')
677        paragraph = None
678        if 'p' in secs[-1]:
679            secs[-1],p = secs[-1].split('p',1)
680            paragraph = int(p)
681        indices = map(int, secs)
682        return SpecIndex(indices, paragraph)
683
684    def __init__(self, indices, paragraph=None):
685        assert len(indices)>0
686        self.indices = tuple(indices)
687        self.paragraph = paragraph
688
689    def __str__(self):
690        s =  '.'.join(map(str,self.indices))
691        if self.paragraph is not None:
692            s += '.p%d'%(self.paragraph,)
693        return s
694
695    def __repr__(self):
696        return 'SpecIndex(%s, %s)'%(self.indices, self.paragraph)
697
698    def __cmp__(self, b):
699        return cmp((self.indices,self.paragraph),
700                   (b.indices,b.paragraph))
701
702    def __hash__(self):
703        return hash((self.indices,self.paragraph))
704
705    def __sub__(self, indices):
706        def sub(a,b):
707            a = a or 0
708            b = b or 0
709            return abs(a-b)
710        return map(sub,self.indices,indices)
711
712class RefItem:
713    def __init__(self, name, section, filename, path, line):
714        self.name = name
715        self.section = SpecIndex.fromstring(section)
716        self.filename = filename
717        self.path = path
718        self.line = line
719
720    def __str__(self):
721        if self.name is not None:
722            return '%s %s'%(self.name, self.section)
723        else:
724            return '--- %s'%(self.section,)
725
726    def __repr__(self):
727        return 'RefItem(%s, %r, "%s", "%s", %d)'%(self.name,
728                                              self.section,
729                                              self.filename,
730                                              self.path,
731                                              self.line)
732
733    def __cmp__(self, b):
734        return cmp((self.name,self.section,self.filename,self.path,self.line),
735                   (b.name,b.section,self.filename,self.path,self.line))
736
737    def __hash__(self):
738        return hash((self.name,self.section,self.filename,self.path,self.line))
739
740###
741
742def sorted(l):
743    l = list(l)
744    l.sort()
745    return l
746
747def getRevision(path):
748    import subprocess
749    p = subprocess.Popen(['svn', 'info', path],
750                         stdin=open('/dev/null','r'),
751                         stdout=subprocess.PIPE)
752    for ln in p.stdout.read(1024).split('\n'):
753        if ln.startswith('Revision:'):
754            return ln.split(':',1)[1].strip()
755    return None
756
757def buildRefTree(references):
758    root = (None, {}, [])
759
760    def getNode(keys):
761        if not keys:
762            return root
763        key,parent = keys[-1],getNode(keys[:-1])
764        node = parent[1].get(key)
765        if node is None:
766            parent[1][key] = node = (key, {}, [])
767        return node
768
769    for ref in references:
770        n = getNode((ref.name,) + ref.section.indices)
771        n[2].append(ref)
772
773    def flatten((key, children, data)):
774        children = sorted(map(flatten,children.values()))
775        return (key, children, sorted(data))
776
777    return flatten(root)
778
779def preorder(node,parents=(),first=True):
780    (key,children,data) = node
781    if first:
782        yield parents+(node,)
783    for c in children:
784        for item in preorder(c, parents+(node,)):
785            yield item
786
787def main():
788    global options
789    from optparse import OptionParser
790    parser = OptionParser("usage: %prog [options] CLANG_ROOT <output-dir>")
791
792    (options, args) = parser.parse_args()
793
794    if len(args) != 2:
795        parser.error("incorrect number of arguments")
796
797    references = []
798    root,outputDir = args
799    for (dirpath, dirnames, filenames) in os.walk(root):
800        for filename in filenames:
801            name,ext = os.path.splitext(filename)
802            if ext in ('.c', '.cpp', '.h', '.def'):
803                fullpath = os.path.join(dirpath, filename)
804                references.extend(list(scanFile(fullpath, filename)))
805
806    refTree = buildRefTree(references)
807
808    specs = {}
809    for ref in references:
810        spec = specs[ref.name] = specs.get(ref.name,{})
811        items = spec[ref.section] = spec.get(ref.section,[])
812        items.append(ref)
813
814    print 'Found %d references.'%(len(references),)
815
816    referencesPath = os.path.join(outputDir,'references.html')
817    print 'Writing: %s'%(referencesPath,)
818    f = open(referencesPath,'w')
819    print >>f, '<html><head><title>clang: Specification References</title></head>'
820    print >>f, '<body>'
821    print >>f, '\t<h2>Specification References</h2>'
822    for i,node in enumerate(refTree[1]):
823        specName = node[0] or 'Unknown'
824        print >>f, '<a href="#spec%d">%s</a><br>'%(i,specName)
825    for i,node in enumerate(refTree[1]):
826        specName = node[0] or 'Unknown'
827        print >>f, '<hr>'
828        print >>f, '<a name="spec%d">'%(i,)
829        print >>f, '<h3>Document: %s</h3>'%(specName or 'Unknown',)
830        print >>f, '<table border="1" cellspacing="2" width="80%">'
831        print >>f, '<tr><th width="20%">Name</th><th>References</th></tr>'
832        docData = kDocuments.get(specName)
833        for path in preorder(node,first=False):
834            if not path[-1][2]:
835                continue
836            components = '.'.join([str(p[0]) for p in path[1:]])
837            print >>f, '\t<tr>'
838            tocEntry = None
839            if docData is not None:
840                tocEntry = findClosestTOCEntry(docData, [p[0] for p in path[1:]])
841            if tocEntry is not None:
842                section,name,page = tocEntry
843                # If section is exact print the TOC name
844                if page is not None:
845                    linkStr = '<a href="%s#page=%d">%s</a> (pg.%d)'%(docData[0],page,components,page)
846                else:
847                    linkStr = components
848                if section == components:
849                    print >>f, '\t\t<td valign=top>%s<br>%s</td>'%(linkStr,name)
850                else:
851                    print >>f, '\t\t<td valign=top>%s</td>'%(linkStr,)
852            else:
853                print >>f, '\t\t<td valign=top>%s</td>'%(components,)
854            print >>f, '\t\t<td valign=top>'
855            for item in path[-1][2]:
856                # XXX total hack
857                relativePath = item.path[len(root):]
858                if relativePath.startswith('/'):
859                    relativePath = relativePath[1:]
860                # XXX this is broken, how does doxygen mangle w/ multiple
861                # refs? Can we just read its map?
862                filename = os.path.basename(relativePath)
863                doxyName = '%s-source.html'%(filename.replace('.','_8'),)
864                # Grrr, why can't doxygen write line number references.
865                lineReference = findClosestLineReference(root,doxyName,item.line)
866                if lineReference is not None:
867                    linkStr = 'http://clang.llvm.org/doxygen/%s#%s'%(doxyName,lineReference)
868                else:
869                    linkStr = 'http://clang.llvm.org/doxygen/%s'%(doxyName,)
870                if item.section.paragraph is not None:
871                    paraText = '&nbsp;(p%d)'%(item.section.paragraph,)
872                else:
873                    paraText = ''
874                print >>f,'<a href="%s">%s:%d</a>%s<br>'%(linkStr,relativePath,item.line,paraText)
875            print >>f, '\t\t</td>'
876            print >>f, '\t</tr>'
877        print >>f, '</table>'
878    print >>f, '<hr>'
879    print >>f, 'Generated: %s<br>'%(time.strftime('%Y-%m-%d %H:%M'),)
880    print >>f, 'SVN Revision: %s'%(getRevision(root),)
881    print >>f, '</body>'
882    f.close()
883
884if __name__=='__main__':
885   main()
886