1#!/usr/bin/env python
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 collections
7import os
8import re
9
10CLANG_DIR = os.path.join(os.path.dirname(__file__), '../..')
11FORMAT_STYLE_FILE = os.path.join(CLANG_DIR, 'include/clang/Format/Format.h')
12INCLUDE_STYLE_FILE = os.path.join(CLANG_DIR, 'include/clang/Tooling/Inclusions/IncludeStyle.h')
13DOC_FILE = os.path.join(CLANG_DIR, 'docs/ClangFormatStyleOptions.rst')
14
15
16def substitute(text, tag, contents):
17  replacement = '\n.. START_%s\n\n%s\n\n.. END_%s\n' % (tag, contents, tag)
18  pattern = r'\n\.\. START_%s\n.*\n\.\. END_%s\n' % (tag, tag)
19  return re.sub(pattern, '%s', text, flags=re.S) % replacement
20
21def doxygen2rst(text):
22  text = re.sub(r'<tt>\s*(.*?)\s*<\/tt>', r'``\1``', text)
23  text = re.sub(r'\\c ([^ ,;\.]+)', r'``\1``', text)
24  text = re.sub(r'\\\w+ ', '', text)
25  return text
26
27def indent(text, columns, indent_first_line=True):
28  indent = ' ' * columns
29  s = re.sub(r'\n([^\n])', '\n' + indent + '\\1', text, flags=re.S)
30  if not indent_first_line or s.startswith('\n'):
31    return s
32  return indent + s
33
34class Option(object):
35  def __init__(self, name, type, comment):
36    self.name = name
37    self.type = type
38    self.comment = comment.strip()
39    self.enum = None
40    self.nested_struct = None
41
42  def __str__(self):
43    s = '**%s** (``%s``)\n%s' % (self.name, self.type,
44                                 doxygen2rst(indent(self.comment, 2)))
45    if self.enum and self.enum.values:
46      s += indent('\n\nPossible values:\n\n%s\n' % self.enum, 2)
47    if self.nested_struct:
48      s += indent('\n\nNested configuration flags:\n\n%s\n' %self.nested_struct,
49                  2)
50    return s
51
52class NestedStruct(object):
53  def __init__(self, name, comment):
54    self.name = name
55    self.comment = comment.strip()
56    self.values = []
57
58  def __str__(self):
59    return '\n'.join(map(str, self.values))
60
61class NestedField(object):
62  def __init__(self, name, comment):
63    self.name = name
64    self.comment = comment.strip()
65
66  def __str__(self):
67    return '\n* ``%s`` %s' % (
68        self.name,
69        doxygen2rst(indent(self.comment, 2, indent_first_line=False)))
70
71class Enum(object):
72  def __init__(self, name, comment):
73    self.name = name
74    self.comment = comment.strip()
75    self.values = []
76
77  def __str__(self):
78    return '\n'.join(map(str, self.values))
79
80class NestedEnum(object):
81  def __init__(self, name, enumtype, comment, values):
82    self.name = name
83    self.comment = comment
84    self.values = values
85    self.type = enumtype
86
87  def __str__(self):
88    s = '\n* ``%s %s``\n%s' % (self.type, self.name,
89                                 doxygen2rst(indent(self.comment, 2)))
90    s += indent('\nPossible values:\n\n', 2)
91    s += indent('\n'.join(map(str, self.values)),2)
92    return s;
93
94class EnumValue(object):
95  def __init__(self, name, comment, config):
96    self.name = name
97    self.comment = comment
98    self.config = config
99
100  def __str__(self):
101    return '* ``%s`` (in configuration: ``%s``)\n%s' % (
102        self.name,
103        re.sub('.*_', '', self.config),
104        doxygen2rst(indent(self.comment, 2)))
105
106def clean_comment_line(line):
107  match = re.match(r'^/// (?P<indent> +)?\\code(\{.(?P<lang>\w+)\})?$', line)
108  if match:
109    indent = match.group('indent')
110    if not indent:
111      indent = ''
112    lang = match.group('lang')
113    if not lang:
114      lang = 'c++'
115    return '\n%s.. code-block:: %s\n\n' % (indent, lang)
116
117  endcode_match = re.match(r'^/// +\\endcode$', line)
118  if endcode_match:
119    return ''
120  return line[4:] + '\n'
121
122def read_options(header):
123  class State(object):
124    BeforeStruct, Finished, InStruct, InNestedStruct, InNestedFieldComent, \
125    InFieldComment, InEnum, InEnumMemberComment = range(8)
126  state = State.BeforeStruct
127
128  options = []
129  enums = {}
130  nested_structs = {}
131  comment = ''
132  enum = None
133  nested_struct = None
134
135  for line in header:
136    line = line.strip()
137    if state == State.BeforeStruct:
138      if line == 'struct FormatStyle {' or line == 'struct IncludeStyle {':
139        state = State.InStruct
140    elif state == State.InStruct:
141      if line.startswith('///'):
142        state = State.InFieldComment
143        comment = clean_comment_line(line)
144      elif line == '};':
145        state = State.Finished
146        break
147    elif state == State.InFieldComment:
148      if line.startswith('///'):
149        comment += clean_comment_line(line)
150      elif line.startswith('enum'):
151        state = State.InEnum
152        name = re.sub(r'enum\s+(\w+)\s*(:((\s*\w+)+)\s*)?\{', '\\1', line)
153        enum = Enum(name, comment)
154      elif line.startswith('struct'):
155        state = State.InNestedStruct
156        name = re.sub(r'struct\s+(\w+)\s*\{', '\\1', line)
157        nested_struct = NestedStruct(name, comment)
158      elif line.endswith(';'):
159        state = State.InStruct
160        field_type, field_name = re.match(r'([<>:\w(,\s)]+)\s+(\w+);',
161                                          line).groups()
162        option = Option(str(field_name), str(field_type), comment)
163        options.append(option)
164      else:
165        raise Exception('Invalid format, expected comment, field or enum')
166    elif state == State.InNestedStruct:
167      if line.startswith('///'):
168        state = State.InNestedFieldComent
169        comment = clean_comment_line(line)
170      elif line == '};':
171        state = State.InStruct
172        nested_structs[nested_struct.name] = nested_struct
173    elif state == State.InNestedFieldComent:
174      if line.startswith('///'):
175        comment += clean_comment_line(line)
176      else:
177        state = State.InNestedStruct
178        field_type, field_name = re.match(r'([<>:\w(,\s)]+)\s+(\w+);',line).groups()
179        if field_type in enums:
180            nested_struct.values.append(NestedEnum(field_name,field_type,comment,enums[field_type].values))
181        else:
182            nested_struct.values.append(NestedField(field_type + " " + field_name, comment))
183
184    elif state == State.InEnum:
185      if line.startswith('///'):
186        state = State.InEnumMemberComment
187        comment = clean_comment_line(line)
188      elif line == '};':
189        state = State.InStruct
190        enums[enum.name] = enum
191      else:
192        # Enum member without documentation. Must be documented where the enum
193        # is used.
194        pass
195    elif state == State.InEnumMemberComment:
196      if line.startswith('///'):
197        comment += clean_comment_line(line)
198      else:
199        state = State.InEnum
200        val = line.replace(',', '')
201        pos = val.find(" // ")
202        if (pos != -1):
203            config = val[pos+4:]
204            val = val[:pos]
205        else:
206            config = val;
207        enum.values.append(EnumValue(val, comment,config))
208  if state != State.Finished:
209    raise Exception('Not finished by the end of file')
210
211  for option in options:
212    if not option.type in ['bool', 'unsigned', 'int', 'std::string',
213                           'std::vector<std::string>',
214                           'std::vector<IncludeCategory>',
215                           'std::vector<RawStringFormat>']:
216      if option.type in enums:
217        option.enum = enums[option.type]
218      elif option.type in nested_structs:
219        option.nested_struct = nested_structs[option.type]
220      else:
221        raise Exception('Unknown type: %s' % option.type)
222  return options
223
224options = read_options(open(FORMAT_STYLE_FILE))
225options += read_options(open(INCLUDE_STYLE_FILE))
226
227options = sorted(options, key=lambda x: x.name)
228options_text = '\n\n'.join(map(str, options))
229
230contents = open(DOC_FILE).read()
231
232contents = substitute(contents, 'FORMAT_STYLE_OPTIONS', options_text)
233
234with open(DOC_FILE, 'wb') as output:
235  output.write(contents.encode())
236