xref: /llvm-project/clang/docs/tools/dump_format_style.py (revision 9e8fff26f374766b76aad9531ee5e6860f67e46c) 
1#!/usr/bin/env python3
2# A tool to parse the FormatStyle struct from Format.h and update the
3# documentation in ../ClangFormatStyleOptions.rst automatically.
4# Run from the directory in which this file is located to update the docs.
5
6import inspect
7import os
8import re
9from typing import Set
10
11CLANG_DIR = os.path.join(os.path.dirname(__file__), '../..')
12FORMAT_STYLE_FILE = os.path.join(CLANG_DIR, 'include/clang/Format/Format.h')
13INCLUDE_STYLE_FILE = os.path.join(CLANG_DIR, 'include/clang/Tooling/Inclusions/IncludeStyle.h')
14DOC_FILE = os.path.join(CLANG_DIR, 'docs/ClangFormatStyleOptions.rst')
15
16PLURALS_FILE = os.path.join(os.path.dirname(__file__), 'plurals.txt')
17
18plurals: Set[str] = set()
19with open(PLURALS_FILE, 'a+') as f:
20  f.seek(0)
21  plurals = set(f.read().splitlines())
22
23def substitute(text, tag, contents):
24  replacement = '\n.. START_%s\n\n%s\n\n.. END_%s\n' % (tag, contents, tag)
25  pattern = r'\n\.\. START_%s\n.*\n\.\. END_%s\n' % (tag, tag)
26  return re.sub(pattern, '%s', text, flags=re.S) % replacement
27
28def register_plural(singular: str, plural: str):
29  if plural not in plurals:
30    if not hasattr(register_plural, "generated_new_plural"):
31      print('Plural generation: you can use '
32      f'`git checkout -- {os.path.relpath(PLURALS_FILE)}` '
33      'to reemit warnings or `git add` to include new plurals\n')
34    register_plural.generated_new_plural = True
35
36    plurals.add(plural)
37    with open(PLURALS_FILE, 'a') as f:
38      f.write(plural + '\n')
39    cf = inspect.currentframe()
40    lineno = ''
41    if cf and cf.f_back:
42      lineno = ':' + str(cf.f_back.f_lineno)
43    print(f'{__file__}{lineno} check if plural of {singular} is {plural}', file=os.sys.stderr)
44  return plural
45
46def pluralize(word: str):
47  lword = word.lower()
48  if len(lword) >= 2 and lword[-1] == 'y' and lword[-2] not in 'aeiou':
49    return register_plural(word, word[:-1] + 'ies')
50  elif lword.endswith(('s', 'sh', 'ch', 'x', 'z')):
51    return register_plural(word, word[:-1] + 'es')
52  elif lword.endswith('fe'):
53    return register_plural(word, word[:-2] + 'ves')
54  elif lword.endswith('f') and not lword.endswith('ff'):
55    return register_plural(word, word[:-1] + 'ves')
56  else:
57    return register_plural(word, word + 's')
58
59
60def to_yaml_type(typestr: str):
61  if typestr == 'bool':
62    return 'Boolean'
63  elif typestr == 'int':
64    return 'Integer'
65  elif typestr == 'unsigned':
66    return 'Unsigned'
67  elif typestr == 'std::string':
68    return 'String'
69
70  subtype, napplied = re.subn(r'^std::vector<(.*)>$', r'\1', typestr)
71  if napplied == 1:
72    return 'List of ' + pluralize(to_yaml_type(subtype))
73
74  return typestr
75
76def doxygen2rst(text):
77  text = re.sub(r'<tt>\s*(.*?)\s*<\/tt>', r'``\1``', text)
78  text = re.sub(r'\\c ([^ ,;\.]+)', r'``\1``', text)
79  text = re.sub(r'\\\w+ ', '', text)
80  return text
81
82def indent(text, columns, indent_first_line=True):
83  indent = ' ' * columns
84  s = re.sub(r'\n([^\n])', '\n' + indent + '\\1', text, flags=re.S)
85  if not indent_first_line or s.startswith('\n'):
86    return s
87  return indent + s
88
89class Option(object):
90  def __init__(self, name, type, comment):
91    self.name = name
92    self.type = type
93    self.comment = comment.strip()
94    self.enum = None
95    self.nested_struct = None
96
97  def __str__(self):
98    s = '**%s** (``%s``)\n%s' % (self.name, to_yaml_type(self.type),
99                                 doxygen2rst(indent(self.comment, 2)))
100    if self.enum and self.enum.values:
101      s += indent('\n\nPossible values:\n\n%s\n' % self.enum, 2)
102    if self.nested_struct:
103      s += indent('\n\nNested configuration flags:\n\n%s\n' %self.nested_struct,
104                  2)
105    return s
106
107class NestedStruct(object):
108  def __init__(self, name, comment):
109    self.name = name
110    self.comment = comment.strip()
111    self.values = []
112
113  def __str__(self):
114    return '\n'.join(map(str, self.values))
115
116class NestedField(object):
117  def __init__(self, name, comment):
118    self.name = name
119    self.comment = comment.strip()
120
121  def __str__(self):
122    return '\n* ``%s`` %s' % (
123        self.name,
124        doxygen2rst(indent(self.comment, 2, indent_first_line=False)))
125
126class Enum(object):
127  def __init__(self, name, comment):
128    self.name = name
129    self.comment = comment.strip()
130    self.values = []
131
132  def __str__(self):
133    return '\n'.join(map(str, self.values))
134
135class NestedEnum(object):
136  def __init__(self, name, enumtype, comment, values):
137    self.name = name
138    self.comment = comment
139    self.values = values
140    self.type = enumtype
141
142  def __str__(self):
143    s = '\n* ``%s %s``\n%s' % (to_yaml_type(self.type), self.name,
144                                 doxygen2rst(indent(self.comment, 2)))
145    s += indent('\nPossible values:\n\n', 2)
146    s += indent('\n'.join(map(str, self.values)),2)
147    return s;
148
149class EnumValue(object):
150  def __init__(self, name, comment, config):
151    self.name = name
152    self.comment = comment
153    self.config = config
154
155  def __str__(self):
156    return '* ``%s`` (in configuration: ``%s``)\n%s' % (
157        self.name,
158        re.sub('.*_', '', self.config),
159        doxygen2rst(indent(self.comment, 2)))
160
161def clean_comment_line(line):
162  match = re.match(r'^/// (?P<indent> +)?\\code(\{.(?P<lang>\w+)\})?$', line)
163  if match:
164    indent = match.group('indent')
165    if not indent:
166      indent = ''
167    lang = match.group('lang')
168    if not lang:
169      lang = 'c++'
170    return '\n%s.. code-block:: %s\n\n' % (indent, lang)
171
172  endcode_match = re.match(r'^/// +\\endcode$', line)
173  if endcode_match:
174    return ''
175
176  match = re.match(r'^/// \\warning$', line)
177  if match:
178    return '\n.. warning:: \n\n'
179
180  endwarning_match = re.match(r'^/// +\\endwarning$', line)
181  if endwarning_match:
182    return ''
183  return line[4:] + '\n'
184
185def read_options(header):
186  class State(object):
187    BeforeStruct, Finished, InStruct, InNestedStruct, InNestedFieldComment, \
188    InFieldComment, InEnum, InEnumMemberComment = range(8)
189  state = State.BeforeStruct
190
191  options = []
192  enums = {}
193  nested_structs = {}
194  comment = ''
195  enum = None
196  nested_struct = None
197
198  for line in header:
199    line = line.strip()
200    if state == State.BeforeStruct:
201      if line == 'struct FormatStyle {' or line == 'struct IncludeStyle {':
202        state = State.InStruct
203    elif state == State.InStruct:
204      if line.startswith('///'):
205        state = State.InFieldComment
206        comment = clean_comment_line(line)
207      elif line == '};':
208        state = State.Finished
209        break
210    elif state == State.InFieldComment:
211      if line.startswith('///'):
212        comment += clean_comment_line(line)
213      elif line.startswith('enum'):
214        state = State.InEnum
215        name = re.sub(r'enum\s+(\w+)\s*(:((\s*\w+)+)\s*)?\{', '\\1', line)
216        enum = Enum(name, comment)
217      elif line.startswith('struct'):
218        state = State.InNestedStruct
219        name = re.sub(r'struct\s+(\w+)\s*\{', '\\1', line)
220        nested_struct = NestedStruct(name, comment)
221      elif line.endswith(';'):
222        state = State.InStruct
223        field_type, field_name = re.match(r'([<>:\w(,\s)]+)\s+(\w+);',
224                                          line).groups()
225        option = Option(str(field_name), str(field_type), comment)
226        options.append(option)
227      else:
228        raise Exception('Invalid format, expected comment, field or enum')
229    elif state == State.InNestedStruct:
230      if line.startswith('///'):
231        state = State.InNestedFieldComment
232        comment = clean_comment_line(line)
233      elif line == '};':
234        state = State.InStruct
235        nested_structs[nested_struct.name] = nested_struct
236    elif state == State.InNestedFieldComment:
237      if line.startswith('///'):
238        comment += clean_comment_line(line)
239      else:
240        state = State.InNestedStruct
241        field_type, field_name = re.match(r'([<>:\w(,\s)]+)\s+(\w+);',line).groups()
242        if field_type in enums:
243            nested_struct.values.append(NestedEnum(field_name,field_type,comment,enums[field_type].values))
244        else:
245            nested_struct.values.append(NestedField(field_type + " " + field_name, comment))
246
247    elif state == State.InEnum:
248      if line.startswith('///'):
249        state = State.InEnumMemberComment
250        comment = clean_comment_line(line)
251      elif line == '};':
252        state = State.InStruct
253        enums[enum.name] = enum
254      else:
255        # Enum member without documentation. Must be documented where the enum
256        # is used.
257        pass
258    elif state == State.InEnumMemberComment:
259      if line.startswith('///'):
260        comment += clean_comment_line(line)
261      else:
262        state = State.InEnum
263        val = line.replace(',', '')
264        pos = val.find(" // ")
265        if (pos != -1):
266            config = val[pos+4:]
267            val = val[:pos]
268        else:
269            config = val;
270        enum.values.append(EnumValue(val, comment,config))
271  if state != State.Finished:
272    raise Exception('Not finished by the end of file')
273
274  for option in options:
275    if not option.type in ['bool', 'unsigned', 'int', 'std::string',
276                           'std::vector<std::string>',
277                           'std::vector<IncludeCategory>',
278                           'std::vector<RawStringFormat>']:
279      if option.type in enums:
280        option.enum = enums[option.type]
281      elif option.type in nested_structs:
282        option.nested_struct = nested_structs[option.type]
283      else:
284        raise Exception('Unknown type: %s' % option.type)
285  return options
286
287options = read_options(open(FORMAT_STYLE_FILE))
288options += read_options(open(INCLUDE_STYLE_FILE))
289
290options = sorted(options, key=lambda x: x.name)
291options_text = '\n\n'.join(map(str, options))
292
293contents = open(DOC_FILE).read()
294
295contents = substitute(contents, 'FORMAT_STYLE_OPTIONS', options_text)
296
297with open(DOC_FILE, 'wb') as output:
298  output.write(contents.encode())
299