src/core/wp_utils.py
author ymh <ymh.work@gmail.com>
Sun, 16 Jun 2013 01:03:40 +0200
changeset 4 ae066d797a66
parent 3 221af1052ec4
child 33 61c3ffd94f11
permissions -rw-r--r--
add link to joconde search. Add option to import
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
from .models import Term, TermLabel, TERM_URL_STATUS_DICT
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
from django.conf import settings
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
from django.utils.http import urlquote
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
from wikitools import api, wiki
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
import logging
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
logger = logging.getLogger(__name__)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
def normalize_term(term):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
    if len(term) == 0:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
        return term
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
    term = term.strip()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
    term = term.replace("_", " ")
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
    term = " ".join(term.split())
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
    term = term[0].upper() + term[1:]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
    return term
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
def switch_case_group(term):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
    seg_group = term.split()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
    uc_group = []
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
    lc_group = []
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
    for seg in seg_group:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
        is_all_upper = all(c.isupper() or not c.isalpha() for c in seg) 
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
        if is_all_upper and not lc_group:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
            uc_group.append(seg)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
        elif not is_all_upper and uc_group:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
            lc_group.append(seg)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
        else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    31
            return term
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
            
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
    if uc_group and lc_group and len(uc_group)+len(lc_group) == len(seg_group):        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
        return " ".join(lc_group + [normalize_term(t.lower()) for t in uc_group])
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
    elif uc_group and not lc_group and len(uc_group) == len(seg_group):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
        return " ".join([normalize_term(t.lower()) for t in uc_group])
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
        return term
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    40
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
def urlize_for_wikipedia(label):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
    return urlquote(label.replace(" ", "_"))
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    44
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
def __is_homonymie(page_dict):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    46
    for cat in page_dict.get(u"categories", []):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    47
        if u'Catégorie:Homonymie' in cat.get(u"title", u"") or u'Category:Disambiguation pages' in cat.get(u"title", u""):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
            return True
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
    return False
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
def query_wikipedia_title(site, label=None, pageid=None):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
    params = {'action':'query', 'prop':'info|categories|langlinks', 'inprop':'url', 'lllimit':'500', 'cllimit':'500'}
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
    if label:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
        params['titles'] = label
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
        params['pageids'] = pageid
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
    response = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
    def return_null_result():
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    64
        return { 'new_label': None, 'alternative_label': None, 'status': TERM_URL_STATUS_DICT["null_result"], 'wikipedia_url': None, 'pageid': None, 'alternative_wikipedia_url': None, 'alternative_pageid': None, 'dbpedia_uri': None, 'revision_id': None, 'response': response }
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    65
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
    try:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
        wpquery = api.APIRequest(site, params) #@UndefinedVariable
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
        response = wpquery.query()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
    except:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
        logger.exception("Exception when querying wikipedia")
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
        return return_null_result()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
    original_response = response
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
    query_dict = response['query']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
    # get page if multiple pages or none -> return Tag.null_result
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
    pages = query_dict.get("pages", {})
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
    if len(pages) > 1 or len(pages) == 0:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
        return return_null_result()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
    page = pages.values()[0]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
    if u"invalid" in page or u"missing" in page:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
        return return_null_result()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
    url = page.get(u'fullurl', None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
    pageid = page.get(u'pageid', None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
    new_label = page[u'title']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
    alternative_label = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
    alternative_url = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
    alternative_pageid = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
    if __is_homonymie(page):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
        status = TERM_URL_STATUS_DICT["homonyme"]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
    elif u"redirect" in page:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
        status = TERM_URL_STATUS_DICT["redirection"]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
        status = TERM_URL_STATUS_DICT["match"]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   100
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   101
    if status == TERM_URL_STATUS_DICT["redirection"]:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   102
        params['redirects'] = True
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   103
        try:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   104
            wpquery = api.APIRequest(site, params) #@UndefinedVariable    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
            response = wpquery.query()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
        except:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
            logger.exception("Exception when querying wikipedia for redirects")
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
            return return_null_result()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
        query_dict = response['query']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
        pages = query_dict.get("pages", {})
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
        #we know that we have at least one answer        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
        if len(pages) > 1 or len(pages) == 0:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
            return return_null_result()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
        page = pages.values()[0]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   115
        alternative_label = page.get('title', None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   116
        alternative_url = page.get('fullurl', None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   117
        alternative_pageid = page.get('pageid',None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
    revision_id = page.get('lastrevid', None)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
    if status == TERM_URL_STATUS_DICT['match'] or status == TERM_URL_STATUS_DICT['redirection']:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
        dbpedia_uri = settings.DBPEDIA_URI_TEMPLATE % (urlize_for_wikipedia(new_label))
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
        dbpedia_uri = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   127
            
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   128
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   129
    return { 'new_label': new_label, 'alternative_label': alternative_label, 'status': status, 'wikipedia_url': url, 'pageid': pageid, 'alternative_wikipedia_url': alternative_url, 'alternative_pageid': alternative_pageid, 'dbpedia_uri': dbpedia_uri, 'revision_id': revision_id, 'response': original_response }
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   130
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   131
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   132
4
ae066d797a66 add link to joconde search. Add option to import
ymh <ymh.work@gmail.com>
parents: 3
diff changeset
   133
def get_or_create_term(term_label, term_uri, term_lang, thesaurus, wp_label_transform=(lambda l:l), skip_wp_query=False):
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   134
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   135
    term_label_normalized = normalize_term(term_label)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   136
    # We get the wikipedia references for the tag_label
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   137
    # We get or create the tag object
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   138
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   139
    term = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   140
    if term_uri is not None and Term.objects.filter(uri=term_uri).exists():
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   141
        term = Term.objects.get(uri=term_uri)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   142
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   143
        for t in Term.objects.filter(normalized_label__iexact=term_label_normalized, lang=term_lang, thesaurus=thesaurus):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   144
            if term is None or t.url_status != TERM_URL_STATUS_DICT['null_result']:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   145
                term = t
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   146
                if term.url_status != TERM_URL_STATUS_DICT['null_result']:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   147
                    break
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   148
 
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   149
    if term is None:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   150
        term = Term(label=term_label, thesaurus=thesaurus, normalized_label=term_label_normalized, uri=term_uri, lang = term_lang)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   151
        created = True
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   152
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   153
        created = False
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   154
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   155
    site = wiki.Wiki(settings.WIKIPEDIA_API_URL) #@UndefinedVariable
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   156
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   157
    if created:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   158
        
3
221af1052ec4 Correct process wikipedia
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
   159
        wikipedia_revision_id = process_term(site, term, label=wp_label_transform(term_label_normalized))
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   160
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   161
        term_label_obj = TermLabel(label=term_label, term=term, lang=term_lang)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   162
        term_label_obj.save()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   163
        
4
ae066d797a66 add link to joconde search. Add option to import
ymh <ymh.work@gmail.com>
parents: 3
diff changeset
   164
    elif term.wikipedia_pageid and not skip_wp_query:
0
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   165
        wp_res = query_wikipedia_title(site, pageid=term.wikipedia_pageid)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   166
        wikipedia_revision_id = wp_res['revision_id']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   167
        term.wikipedia_revision_id = wikipedia_revision_id
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   168
        term.save()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   169
    else:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   170
        wikipedia_revision_id = None
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   171
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   172
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   173
    return term, wikipedia_revision_id, created
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   174
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   175
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   176
def process_term(site, term, label=None, verbosity=0):
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   177
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   178
    if site == None:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   179
        site = wiki.Wiki(settings.WIKIPEDIA_API_URL)  # @UndefinedVariable
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   180
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   181
    if not label:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   182
        label = term.label
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   183
    wp_res = query_wikipedia_title(site, label=label)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   184
    new_label = wp_res['new_label']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   185
    alternative_label= wp_res['alternative_label']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   186
    status =  wp_res['status']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   187
    url = wp_res['wikipedia_url']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   188
    alternative_url = wp_res['alternative_wikipedia_url']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   189
    pageid = wp_res['pageid']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   190
    alternative_pageid = wp_res['alternative_pageid']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   191
    response = wp_res['response']
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   192
    dbpedia_uri =  wp_res["dbpedia_uri"]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   193
    revision_id = wp_res["revision_id"]
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   194
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   195
    if verbosity >= 2 :
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   196
        print "response from query to %s with parameters %s :" % (site.apibase, repr(new_label))
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   197
        print repr(response)
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   198
    
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   199
    if new_label is not None:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   200
        term.wp_label = new_label
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   201
    if status is not None:
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   202
        term.url_status = status
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   203
    term.wikipedia_url = url
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   204
    term.wikipedia_pageid = pageid
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   205
    term.dbpedia_uri = dbpedia_uri
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   206
    term.alternative_label = alternative_label
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   207
    term.alternative_wikipedia_url = alternative_url
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   208
    term.alternative_wikipedia_pageid = alternative_pageid
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   209
    term.wikipedia_revision_id=revision_id
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   210
        
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   211
    term.save()
4095911a7830 Jocondelab first commit before design
ymh <ymh.work@gmail.com>
parents:
diff changeset
   212
    
3
221af1052ec4 Correct process wikipedia
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
   213
    return revision_id
221af1052ec4 Correct process wikipedia
ymh <ymh.work@gmail.com>
parents: 0
diff changeset
   214