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
296kDocuments = {
297    'C99' : (c99URL, c99TOC, 12)
298}
299
300def findClosestTOCEntry(data, target):
301    offset = data[2]
302    best = None
303    for (name,page) in data[1]:
304        if ' ' in name:
305            section,name = name.split(' ',1)
306            if section == 'Annex':
307                section,name = name.split(' ',1)
308                section = 'Annex '+section
309        else:
310            section = None
311        try:
312            page = int(page) + offset
313        except:
314            page = 1
315        try:
316            spec = SpecIndex.fromstring(section)
317        except:
318            spec = None
319
320        # Meh, could be better...
321        if spec is not None:
322            dist = spec - target
323            if best is None or dist < best[0]:
324                best = (dist, (section, name, page))
325    return best[1]
326
327# What a hack. Slow to boot.
328doxyLineRefRE = re.compile(r"<a name=\"l([0-9]+)\"></a>")
329def findClosestLineReference(clangRoot, doxyName, target):
330    try:
331        f = open(os.path.join(clangRoot, 'docs', 'doxygen', 'html', doxyName))
332    except:
333        return None
334
335    best = None
336    for m in doxyLineRefRE.finditer(f.read()):
337        line = int(m.group(1), 10)
338        dist = abs(line - target)
339        if best is None or dist < best[0]:
340            best = (dist,'l'+m.group(1))
341    f.close()
342    if best is not None:
343        return best[1]
344    return None
345
346###
347
348nameAndSpecRefRE = re.compile(r"(C99|C90|C\+\+|H\&S) (([0-9]+)(\.[0-9]+)*(p[0-9]+)?)")
349loneSpecRefRE = re.compile(r" (([0-9]+)(\.[0-9]+){2,100}(p[0-9]+)?)")
350def scanFile(path, filename):
351    try:
352        f = open(path)
353    except IOError:
354        print >>sys.stderr,'WARNING: Unable to open:',path
355        return
356
357    for i,ln in enumerate(f):
358        ignore = set()
359        for m in nameAndSpecRefRE.finditer(ln):
360            section = m.group(2)
361            name = m.group(1)
362            if section.endswith('.'):
363                section = section[:-1]
364            yield RefItem(name, section, filename, path, i+1)
365            ignore.add(section)
366        for m in loneSpecRefRE.finditer(ln):
367            section = m.group(1)
368            if section.endswith('.'):
369                section = section[:-1]
370            if section not in ignore:
371                yield RefItem(None, section, filename, path, i+1)
372
373###
374
375class SpecIndex:
376    @staticmethod
377    def fromstring(str):
378        secs = str.split('.')
379        paragraph = None
380        if 'p' in secs[-1]:
381            secs[-1],p = secs[-1].split('p',1)
382            paragraph = int(p)
383        indices = map(int, secs)
384        return SpecIndex(indices, paragraph)
385
386    def __init__(self, indices, paragraph=None):
387        assert len(indices)>0
388        self.indices = tuple(indices)
389        self.paragraph = paragraph
390
391    def __str__(self):
392        s =  '.'.join(map(str,self.indices))
393        if self.paragraph is not None:
394            s += '.p%d'%(self.paragraph,)
395        return s
396
397    def __repr__(self):
398        return 'SpecIndex(%s, %s)'%(self.indices, self.paragraph)
399
400    def __cmp__(self, b):
401        return cmp((self.indices,self.paragraph),
402                   (b.indices,b.paragraph))
403
404    def __hash__(self):
405        return hash((self.indices,self.paragraph))
406
407    def __sub__(self, indices):
408        def sub(a,b):
409            a = a or 0
410            b = b or 0
411            return abs(a-b)
412        return map(sub,self.indices,indices)
413
414class RefItem:
415    def __init__(self, name, section, filename, path, line):
416        self.name = name
417        self.section = SpecIndex.fromstring(section)
418        self.filename = filename
419        self.path = path
420        self.line = line
421
422    def __str__(self):
423        if self.name is not None:
424            return '%s %s'%(self.name, self.section)
425        else:
426            return '--- %s'%(self.section,)
427
428    def __repr__(self):
429        return 'RefItem(%s, %r, "%s", "%s", %d)'%(self.name,
430                                              self.section,
431                                              self.filename,
432                                              self.path,
433                                              self.line)
434
435    def __cmp__(self, b):
436        return cmp((self.name,self.section,self.filename,self.path,self.line),
437                   (b.name,b.section,self.filename,self.path,self.line))
438
439    def __hash__(self):
440        return hash((self.name,self.section,self.filename,self.path,self.line))
441
442###
443
444def sorted(l):
445    l = list(l)
446    l.sort()
447    return l
448
449def getRevision(path):
450    import subprocess
451    p = subprocess.Popen(['svn', 'info', path],
452                         stdin=open('/dev/null','r'),
453                         stdout=subprocess.PIPE)
454    for ln in p.stdout.read(1024).split('\n'):
455        if ln.startswith('Revision:'):
456            return ln.split(':',1)[1].strip()
457    return None
458
459def buildRefTree(references):
460    root = (None, {}, [])
461
462    def getNode(keys):
463        if not keys:
464            return root
465        key,parent = keys[-1],getNode(keys[:-1])
466        node = parent[1].get(key)
467        if node is None:
468            parent[1][key] = node = (key, {}, [])
469        return node
470
471    for ref in references:
472        n = getNode((ref.name,) + ref.section.indices)
473        n[2].append(ref)
474
475    def flatten((key, children, data)):
476        children = sorted(map(flatten,children.values()))
477        return (key, children, sorted(data))
478
479    return flatten(root)
480
481def preorder(node,parents=(),first=True):
482    (key,children,data) = node
483    if first:
484        yield parents+(node,)
485    for c in children:
486        for item in preorder(c, parents+(node,)):
487            yield item
488
489def main():
490    global options
491    from optparse import OptionParser
492    parser = OptionParser("usage: %prog [options] CLANG_ROOT <output-dir>")
493
494    (options, args) = parser.parse_args()
495
496    if len(args) != 2:
497        parser.error("incorrect number of arguments")
498
499    references = []
500    root,outputDir = args
501    for (dirpath, dirnames, filenames) in os.walk(root):
502        for filename in filenames:
503            name,ext = os.path.splitext(filename)
504            if ext in ('.c', '.cpp', '.h', '.def'):
505                fullpath = os.path.join(dirpath, filename)
506                references.extend(list(scanFile(fullpath, filename)))
507
508    refTree = buildRefTree(references)
509
510    specs = {}
511    for ref in references:
512        spec = specs[ref.name] = specs.get(ref.name,{})
513        items = spec[ref.section] = spec.get(ref.section,[])
514        items.append(ref)
515
516    print 'Found %d references.'%(len(references),)
517
518    referencesPath = os.path.join(outputDir,'references.html')
519    print 'Writing: %s'%(referencesPath,)
520    f = open(referencesPath,'w')
521    print >>f, '<html><head><title>clang: Specification References</title></head>'
522    print >>f, '<body>'
523    print >>f, '\t<h2>Specification References</h2>'
524    for i,node in enumerate(refTree[1]):
525        specName = node[0] or 'Unknown'
526        print >>f, '<a href="#spec%d">%s</a><br>'%(i,specName)
527    for i,node in enumerate(refTree[1]):
528        specName = node[0] or 'Unknown'
529        print >>f, '<hr>'
530        print >>f, '<a name="spec%d">'%(i,)
531        print >>f, '<h3>Document: %s</h3>'%(specName or 'Unknown',)
532        print >>f, '<table border="1" cellspacing="2" width="80%">'
533        print >>f, '<tr><th width="20%">Name</th><th>References</th></tr>'
534        docData = kDocuments.get(specName)
535        for path in preorder(node,first=False):
536            if not path[-1][2]:
537                continue
538            components = '.'.join([str(p[0]) for p in path[1:]])
539            print >>f, '\t<tr>'
540            tocEntry = None
541            if docData is not None:
542                tocEntry = findClosestTOCEntry(docData, [p[0] for p in path[1:]])
543            if tocEntry is not None:
544                section,name,page = tocEntry
545                # If section is exact print the TOC name
546                if page is not None:
547                    linkStr = '<a href="%s#page=%d">%s</a> (pg.%d)'%(docData[0],page,components,page)
548                else:
549                    linkStr = components
550                if section == components:
551                    print >>f, '\t\t<td valign=top>%s<br>%s</td>'%(linkStr,name)
552                else:
553                    print >>f, '\t\t<td valign=top>%s</td>'%(linkStr,)
554            else:
555                print >>f, '\t\t<td valign=top>%s</td>'%(components,)
556            print >>f, '\t\t<td valign=top>'
557            for item in path[-1][2]:
558                # XXX total hack
559                relativePath = item.path[len(root):]
560                if relativePath.startswith('/'):
561                    relativePath = relativePath[1:]
562                # XXX this is broken, how does doxygen mangle w/ multiple
563                # refs? Can we just read its map?
564                filename = os.path.basename(relativePath)
565                doxyName = '%s-source.html'%(filename.replace('.','_8'),)
566                # Grrr, why can't doxygen write line number references.
567                lineReference = findClosestLineReference(root,doxyName,item.line)
568                if lineReference is not None:
569                    linkStr = 'http://clang.llvm.org/doxygen/%s#%s'%(doxyName,lineReference)
570                else:
571                    linkStr = 'http://clang.llvm.org/doxygen/%s'%(doxyName,)
572                if item.section.paragraph is not None:
573                    paraText = '&nbsp;(p%d)'%(item.section.paragraph,)
574                else:
575                    paraText = ''
576                print >>f,'<a href="%s">%s:%d</a>%s<br>'%(linkStr,relativePath,item.line,paraText)
577            print >>f, '\t\t</td>'
578            print >>f, '\t</tr>'
579        print >>f, '</table>'
580    print >>f, '<hr>'
581    print >>f, 'Generated: %s<br>'%(time.strftime('%Y-%m-%d %H:%M'),)
582    print >>f, 'SVN Revision: %s'%(getRevision(root),)
583    print >>f, '</body>'
584    f.close()
585
586if __name__=='__main__':
587   main()
588