scripts/create_php_tags.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3 #
   4 # Author:  Enrico Tröger
   5 # License: GPL v2 or later
   6 #
   7 # This script downloads the PHP tag definitions in JSON format from
   8 # http://doc.php.net/downloads/json/php_manual_en.json.
   9 # From those defintions all function tags are extracted and written
  10 # to ../data/php.tags (relative to the script's location, not $CWD).
  11
  12 from json import loads
  13 from os.path import dirname, join
  14 from urllib2 import urlopen
  15 import re
  16
  17
  18 UPSTREAM_TAG_DEFINITION = 'http://doc.php.net/downloads/json/php_manual_en.json'
  19 PROTOTYPE_RE = r'^(?P<return_type>.*) {tag_name}(?P<arg_list>\(.*\))$'
  20
  21 # (from tagmanager/src/tm_tag.c:90)
  22 TA_NAME = 200
  23 TA_TYPE = 204
  24 TA_ARGLIST = 205
  25 TA_SCOPE = 206
  26 TA_VARTYPE = 207
  27 # TMTagType (tagmanager/src/tm_tag.h:49)
  28 TYPE_CLASS = 1
  29 TYPE_FUNCTION = 16
  30 TYPE_MEMBER = 64
  31 TYPE_METHOD = 128
  32 TYPE_VARIABLE = 16384
  33
  34
  35 #----------------------------------------------------------------------
  36 def normalize_name(name):
  37     """ Replace namespace separator with class separators, as Geany only
  38         understands the latter """
  39     return name.replace('\\', '::')
  40
  41
  42 #----------------------------------------------------------------------
  43 def split_scope(name):
  44     """ Splits the scope from the member, and returns (scope, member).
  45         Returned scope is None if the name is not a member """
  46     name = normalize_name(name)
  47     sep_pos = name.rfind('::')
  48     if sep_pos < 0:
  49         return None, name
  50     else:
  51         return name[:sep_pos], name[sep_pos+2:]
  52
  53
  54 #----------------------------------------------------------------------
  55 def parse_and_create_php_tags_file():
  56     # download upstream definition
  57     response = urlopen(UPSTREAM_TAG_DEFINITION)
  58     try:
  59         html = response.read()
  60     finally:
  61         response.close()
  62
  63     # parse JSON
  64     definitions = loads(html)
  65
  66     # generate tags
  67     tag_list = list()
  68     for tag_name, tag_definition in definitions.items():
  69         prototype_re = PROTOTYPE_RE.format(tag_name=re.escape(tag_name))
  70         match = re.match(prototype_re, tag_definition['prototype'])
  71         if match:
  72             return_type = normalize_name(match.group('return_type'))
  73             arg_list = match.group('arg_list')
  74
  75             scope, tag_name = split_scope(tag_name)
  76             if tag_name[0] == '$':
  77                 tag_type = TYPE_MEMBER if scope is not None else TYPE_VARIABLE
  78             else:
  79                 tag_type = TYPE_METHOD if scope is not None else TYPE_FUNCTION
  80             tag_list.append((tag_name, tag_type, return_type, arg_list, scope))
  81             # Also create a class tag when encountering a __construct()
  82             if tag_name == '__construct' and scope is not None:
  83                 scope, tag_name = split_scope(scope)
  84                 tag_list.append((tag_name, TYPE_CLASS, None, arg_list, scope))
  85
  86     # write tags
  87     script_dir = dirname(__file__)
  88     tags_file_path = join(script_dir, '..', 'data', 'php.tags')
  89     with open(tags_file_path, 'w') as tags_file:
  90         tags_file.write('# format=tagmanager\n')
  91         for tag_name, tag_type, return_type, arg_list, scope in sorted(tag_list):
  92             tag_line = '{}'.format(tag_name)
  93             for attr, type in [(tag_type, TA_TYPE),
  94                                (arg_list, TA_ARGLIST),
  95                                (return_type, TA_VARTYPE),
  96                                (scope, TA_SCOPE)]:
  97                 if attr is not None:
  98                     tag_line += '{type:c}{attr}'.format(type=type, attr=attr)
  99
 100             tags_file.write(tag_line + '\n')
 101     print(u'Created: {} with {} tags'.format(tags_file_path, len(tag_list)))
 102
 103
 104 if __name__ == '__main__':
 105     parse_and_create_php_tags_file()