Documentation/sphinx/kernel_include.py

   1 #!/usr/bin/env python3
   2 # -*- coding: utf-8; mode: python -*-
   3 # pylint: disable=R0903, C0330, R0914, R0912, E0401
   4
   5 u"""
   6     kernel-include
   7     ~~~~~~~~~~~~~~
   8
   9     Implementation of the ``kernel-include`` reST-directive.
  10
  11     :copyright:  Copyright (C) 2016  Markus Heiser
  12     :license:    GPL Version 2, June 1991 see linux/COPYING for details.
  13
  14     The ``kernel-include`` reST-directive is a replacement for the ``include``
  15     directive. The ``kernel-include`` directive expand environment variables in
  16     the path name and allows to include files from arbitrary locations.
  17
  18     .. hint::
  19
  20       Including files from arbitrary locations (e.g. from ``/etc``) is a
  21       security risk for builders. This is why the ``include`` directive from
  22       docutils *prohibit* pathnames pointing to locations *above* the filesystem
  23       tree where the reST document with the include directive is placed.
  24
  25     Substrings of the form $name or ${name} are replaced by the value of
  26     environment variable name. Malformed variable names and references to
  27     non-existing variables are left unchanged.
  28 """
  29
  30 # ==============================================================================
  31 # imports
  32 # ==============================================================================
  33
  34 import os.path
  35
  36 from docutils import io, nodes, statemachine
  37 from docutils.utils.error_reporting import SafeString, ErrorString
  38 from docutils.parsers.rst import directives
  39 from docutils.parsers.rst.directives.body import CodeBlock, NumberLines
  40 from docutils.parsers.rst.directives.misc import Include
  41
  42 __version__  = '1.0'
  43
  44 # ==============================================================================
  45 def setup(app):
  46 # ==============================================================================
  47
  48     app.add_directive("kernel-include", KernelInclude)
  49     return dict(
  50         version = __version__,
  51         parallel_read_safe = True,
  52         parallel_write_safe = True
  53     )
  54
  55 # ==============================================================================
  56 class KernelInclude(Include):
  57 # ==============================================================================
  58
  59     u"""KernelInclude (``kernel-include``) directive"""
  60
  61     def run(self):
  62         path = os.path.realpath(
  63             os.path.expandvars(self.arguments[0]))
  64
  65         # to get a bit security back, prohibit /etc:
  66         if path.startswith(os.sep + "etc"):
  67             raise self.severe(
  68                 'Problems with "%s" directive, prohibited path: %s'
  69                 % (self.name, path))
  70
  71         self.arguments[0] = path
  72
  73         #return super(KernelInclude, self).run() # won't work, see HINTs in _run()
  74         return self._run()
  75
  76     def _run(self):
  77         """Include a file as part of the content of this reST file."""
  78
  79         # HINT: I had to copy&paste the whole Include.run method. I'am not happy
  80         # with this, but due to security reasons, the Include.run method does
  81         # not allow absolute or relative pathnames pointing to locations *above*
  82         # the filesystem tree where the reST document is placed.
  83
  84         if not self.state.document.settings.file_insertion_enabled:
  85             raise self.warning('"%s" directive disabled.' % self.name)
  86         source = self.state_machine.input_lines.source(
  87             self.lineno - self.state_machine.input_offset - 1)
  88         source_dir = os.path.dirname(os.path.abspath(source))
  89         path = directives.path(self.arguments[0])
  90         if path.startswith('<') and path.endswith('>'):
  91             path = os.path.join(self.standard_include_path, path[1:-1])
  92         path = os.path.normpath(os.path.join(source_dir, path))
  93
  94         # HINT: this is the only line I had to change / commented out:
  95         #path = utils.relative_path(None, path)
  96
  97         path = nodes.reprunicode(path)
  98         encoding = self.options.get(
  99             'encoding', self.state.document.settings.input_encoding)
 100         e_handler=self.state.document.settings.input_encoding_error_handler
 101         tab_width = self.options.get(
 102             'tab-width', self.state.document.settings.tab_width)
 103         try:
 104             self.state.document.settings.record_dependencies.add(path)
 105             include_file = io.FileInput(source_path=path,
 106                                         encoding=encoding,
 107                                         error_handler=e_handler)
 108         except UnicodeEncodeError as error:
 109             raise self.severe('Problems with "%s" directive path:\n'
 110                               'Cannot encode input file path "%s" '
 111                               '(wrong locale?).' %
 112                               (self.name, SafeString(path)))
 113         except IOError as error:
 114             raise self.severe('Problems with "%s" directive path:\n%s.' %
 115                       (self.name, ErrorString(error)))
 116         startline = self.options.get('start-line', None)
 117         endline = self.options.get('end-line', None)
 118         try:
 119             if startline or (endline is not None):
 120                 lines = include_file.readlines()
 121                 rawtext = ''.join(lines[startline:endline])
 122             else:
 123                 rawtext = include_file.read()
 124         except UnicodeError as error:
 125             raise self.severe('Problem with "%s" directive:\n%s' %
 126                               (self.name, ErrorString(error)))
 127         # start-after/end-before: no restrictions on newlines in match-text,
 128         # and no restrictions on matching inside lines vs. line boundaries
 129         after_text = self.options.get('start-after', None)
 130         if after_text:
 131             # skip content in rawtext before *and incl.* a matching text
 132             after_index = rawtext.find(after_text)
 133             if after_index < 0:
 134                 raise self.severe('Problem with "start-after" option of "%s" '
 135                                   'directive:\nText not found.' % self.name)
 136             rawtext = rawtext[after_index + len(after_text):]
 137         before_text = self.options.get('end-before', None)
 138         if before_text:
 139             # skip content in rawtext after *and incl.* a matching text
 140             before_index = rawtext.find(before_text)
 141             if before_index < 0:
 142                 raise self.severe('Problem with "end-before" option of "%s" '
 143                                   'directive:\nText not found.' % self.name)
 144             rawtext = rawtext[:before_index]
 145
 146         include_lines = statemachine.string2lines(rawtext, tab_width,
 147                                                   convert_whitespace=True)
 148         if 'literal' in self.options:
 149             # Convert tabs to spaces, if `tab_width` is positive.
 150             if tab_width >= 0:
 151                 text = rawtext.expandtabs(tab_width)
 152             else:
 153                 text = rawtext
 154             literal_block = nodes.literal_block(rawtext, source=path,
 155                                     classes=self.options.get('class', []))
 156             literal_block.line = 1
 157             self.add_name(literal_block)
 158             if 'number-lines' in self.options:
 159                 try:
 160                     startline = int(self.options['number-lines'] or 1)
 161                 except ValueError:
 162                     raise self.error(':number-lines: with non-integer '
 163                                      'start value')
 164                 endline = startline + len(include_lines)
 165                 if text.endswith('\n'):
 166                     text = text[:-1]
 167                 tokens = NumberLines([([], text)], startline, endline)
 168                 for classes, value in tokens:
 169                     if classes:
 170                         literal_block += nodes.inline(value, value,
 171                                                       classes=classes)
 172                     else:
 173                         literal_block += nodes.Text(value, value)
 174             else:
 175                 literal_block += nodes.Text(text, text)
 176             return [literal_block]
 177         if 'code' in self.options:
 178             self.options['source'] = path
 179             codeblock = CodeBlock(self.name,
 180                                   [self.options.pop('code')], # arguments
 181                                   self.options,
 182                                   include_lines, # content
 183                                   self.lineno,
 184                                   self.content_offset,
 185                                   self.block_text,
 186                                   self.state,
 187                                   self.state_machine)
 188             return codeblock.run()
 189         self.state_machine.insert_input(include_lines, path)
 190         return []