scripts/create_php_tags.py

   1 #!/usr/bin/env python3
   2 #
   3 # Author:  Enrico Tröger
   4 # License: GPL v2 or later
   5 #
   6 # This script downloads the PHP tag definitions in JSON format from
   7 # http://doc.php.net/downloads/json/php_manual_en.json.
   8 # From those defintions all function tags are extracted and written
   9 # to ../data/tags/std.php.tags (relative to the script's location, not $CWD).
  10
  11 import re
  12 import sys
  13 from json import loads
  14 from os.path import dirname, join
  15 from urllib.request import urlopen
  16
  17 from create_tags_helper import format_tag, write_ctags_file
  18
  19 UPSTREAM_TAG_DEFINITION = 'http://doc.php.net/downloads/json/php_manual_en.json'
  20 PROTOTYPE_RE = r'^(?P<return_type>.*) {tag_name}(?P<arg_list>\(.*\))$'
  21
  22 # (from src/tagmanager/tm_tag.c:90)
  23 TA_NAME = 200
  24 TA_TYPE = 204
  25 TA_ARGLIST = 205
  26 TA_SCOPE = 206
  27 TA_VARTYPE = 207
  28
  29 # PHP kinds
  30 KIND_CLASS = 'class'
  31 KIND_FUNCTION = 'function'
  32 KIND_VARIABLE = 'variable'
  33
  34
  35 def normalize_name(name):
  36     """ Replace namespace separator with class separators, as Geany only
  37         understands the latter """
  38     return name.replace('\\', '::')
  39
  40
  41 def split_scope(name):
  42     """ Splits the scope from the member, and returns (scope, member).
  43         Returned scope is None if the name is not a member """
  44     name = normalize_name(name)
  45     sep_pos = name.rfind('::')
  46     if sep_pos < 0:
  47         return None, name
  48     else:
  49         return name[:sep_pos], name[sep_pos + 2:]
  50
  51
  52 def parse_and_create_php_tags_file():
  53     # download upstream definition
  54     response = urlopen(UPSTREAM_TAG_DEFINITION)
  55     try:
  56         html = response.read()
  57     finally:
  58         response.close()
  59
  60     # parse JSON
  61     definitions = loads(html)
  62
  63     # generate tags
  64     tag_list = []
  65     for tag_name, tag_definition in definitions.items():
  66         prototype_re = PROTOTYPE_RE.format(tag_name=re.escape(tag_name))
  67         match = re.match(prototype_re, tag_definition['prototype'])
  68         if match:
  69             return_type = normalize_name(match.group('return_type'))
  70             arg_list = match.group('arg_list')
  71
  72             scope, tag_name = split_scope(tag_name)
  73             if tag_name[0] == '$':
  74                 kind = KIND_VARIABLE
  75             else:
  76                 kind = KIND_FUNCTION
  77
  78             tag_list.append(format_tag(tag_name, kind, arg_list, scope, ('unknown', return_type)))
  79             # Also create a class tag when encountering a __construct()
  80             if tag_name == '__construct' and scope is not None:
  81                 scope, tag_name = split_scope(scope)
  82                 tag_list.append(format_tag(tag_name, KIND_CLASS, arg_list, scope or ''))
  83
  84     # write tags
  85     script_dir = dirname(__file__)
  86     filename = join(script_dir, '..', 'data', 'tags', 'std.php.tags')
  87     write_ctags_file(filename, tag_list, sys.argv[0])
  88     print(f'Created: {filename} with {len(tag_list)} tags')
  89
  90
  91 if __name__ == '__main__':
  92     parse_and_create_php_tags_file()