projects.mako.cc - wikipedia-api-cdsw/blob - mwclient/page.py

   1 import client
   2 import errors
   3 import listing
   4 import compatibility
   5 from page_nowriteapi import OldPage
   6
   7 import urllib
   8 import urlparse
   9 import time
  10
  11
  12 class Page(object):
  13
  14     def __init__(self, site, name, info=None, extra_properties={}):
  15         if type(name) is type(self):
  16             return self.__dict__.update(name.__dict__)
  17         self.site = site
  18         self.name = name
  19         self.section = None
  20
  21         if not info:
  22             if extra_properties:
  23                 prop = 'info|' + '|'.join(extra_properties.iterkeys())
  24                 extra_props = []
  25                 [extra_props.extend(extra_prop) for extra_prop in extra_properties.itervalues()]
  26             else:
  27                 prop = 'info'
  28                 extra_props = ()
  29
  30             if type(name) is int:
  31                 info = self.site.api('query', prop=prop, pageids=name,
  32                                      inprop='protection', *extra_props)
  33             else:
  34                 info = self.site.api('query', prop=prop, titles=name,
  35                                      inprop='protection', *extra_props)
  36             info = info['query']['pages'].itervalues().next()
  37         self._info = info
  38
  39         self.namespace = info.get('ns', 0)
  40         self.name = info.get('title', u'')
  41         if self.namespace:
  42             self.page_title = self.strip_namespace(self.name)
  43         else:
  44             self.page_title = self.name
  45
  46         self.touched = client.parse_timestamp(info.get('touched', '0000-00-00T00:00:00Z'))
  47         self.revision = info.get('lastrevid', 0)
  48         self.exists = 'missing' not in info
  49         self.length = info.get('length')
  50         self.protection = dict([(i['type'], (i['level'], i['expiry'])) for i in info.get('protection', ()) if i])
  51         self.redirect = 'redirect' in info
  52
  53         self.last_rev_time = None
  54         self.edit_time = None
  55
  56     def redirects_to(self):
  57         """ Returns the redirect target page, or None if the page is not a redirect page."""
  58         info = self.site.api('query', prop='pageprops', titles=self.name, redirects='')['query']
  59         if 'redirects' in info:
  60             for page in info['redirects']:
  61                 if page['from'] == self.name:
  62                     return Page(self.site, page['to'])
  63             return None
  64         else:
  65             return None
  66
  67     def resolve_redirect(self):
  68         """ Returns the redirect target page, or the current page if it's not a redirect page."""
  69         target_page = self.redirects_to()
  70         if target_page is None:
  71             return self
  72         else:
  73             return target_page
  74
  75     def __repr__(self):
  76         return "<Page object '%s' for %s>" % (self.name.encode('utf-8'), self.site)
  77
  78     def __unicode__(self):
  79         return self.name
  80
  81     @staticmethod
  82     def strip_namespace(title):
  83         if title[0] == ':':
  84             title = title[1:]
  85         return title[title.find(':') + 1:]
  86
  87     @staticmethod
  88     def normalize_title(title):
  89         # TODO: Make site dependent
  90         title = title.strip()
  91         if title[0] == ':':
  92             title = title[1:]
  93         title = title[0].upper() + title[1:]
  94         title = title.replace(' ', '_')
  95         return title
  96
  97     def can(self, action):
  98         level = self.protection.get(action, (action, ))[0]
  99         if level == 'sysop':
 100             level = compatibility.protectright(self.site.version)
 101
 102         return level in self.site.rights
 103
 104     def get_token(self, type, force=False):
 105         self.site.require(1, 11)
 106
 107         if type not in self.site.tokens:
 108             self.site.tokens[type] = '0'
 109         if self.site.tokens.get(type, '0') == '0' or force:
 110             info = self.site.api('query', titles=self.name,
 111                                  prop='info', intoken=type)
 112             for i in info['query']['pages'].itervalues():
 113                 if i['title'] == self.name:
 114                     self.site.tokens[type] = i['%stoken' % type]
 115         return self.site.tokens[type]
 116
 117     def get_expanded(self):
 118         self.site.require(1, 12)
 119
 120         revs = self.revisions(prop='content', limit=1, expandtemplates=True)
 121         try:
 122             return revs.next()['*']
 123         except StopIteration:
 124             return u''
 125
 126     def edit(self, section=None, readonly=False):
 127         """Returns wikitext for a specified section or for the whole page.
 128
 129         Retrieves the latest edit.
 130
 131         """
 132         if not self.can('read'):
 133             raise errors.InsufficientPermission(self)
 134         if not self.exists:
 135             return u''
 136
 137         revs = self.revisions(prop='content|timestamp', limit=1, section=section)
 138         try:
 139             rev = revs.next()
 140             self.text = rev['*']
 141             self.section = section
 142             self.last_rev_time = rev['timestamp']
 143         except StopIteration:
 144             self.text = u''
 145             self.section = None
 146             self.edit_time = None
 147         self.edit_time = time.gmtime()
 148         return self.text
 149
 150     def save(self, text=u'', summary=u'', minor=False, bot=True, section=None, **kwargs):
 151         """Save text of page."""
 152         if not self.site.logged_in and self.site.force_login:
 153             # Should we really check for this?
 154             raise errors.LoginError(self.site)
 155         if self.site.blocked:
 156             raise errors.UserBlocked(self.site.blocked)
 157         if not self.can('edit'):
 158             raise errors.ProtectedPageError(self)
 159
 160         if not text:
 161             text = self.text
 162         if not section:
 163             section = self.section
 164
 165         if not self.site.writeapi:
 166             return OldPage.save(self, text=text, summary=summary, minor=False)
 167
 168         data = {}
 169         if minor:
 170             data['minor'] = '1'
 171         if not minor:
 172             data['notminor'] = '1'
 173         if self.last_rev_time:
 174             data['basetimestamp'] = time.strftime('%Y%m%d%H%M%S', self.last_rev_time)
 175         if self.edit_time:
 176             data['starttimestamp'] = time.strftime('%Y%m%d%H%M%S', self.edit_time)
 177         if bot:
 178             data['bot'] = '1'
 179         if section:
 180             data['section'] = section
 181
 182         data.update(kwargs)
 183
 184         def do_edit():
 185             result = self.site.api('edit', title=self.name, text=text,
 186                                    summary=summary, token=self.get_token('edit'),
 187                                    **data)
 188             if result['edit'].get('result').lower() == 'failure':
 189                 raise errors.EditError(self, result['edit'])
 190             return result
 191         try:
 192             result = do_edit()
 193         except errors.APIError, e:
 194             if e.code == 'badtoken':
 195                 # Retry, but only once to avoid an infinite loop
 196                 self.get_token('edit', force=True)
 197                 try:
 198                     result = do_edit()
 199                 except errors.APIError, e:
 200                     self.handle_edit_error(e, summary)
 201             else:
 202                 self.handle_edit_error(e, summary)
 203
 204         if result['edit'] == 'Success':
 205             self.last_rev_time = client.parse_timestamp(result['newtimestamp'])
 206         return result['edit']
 207
 208     def handle_edit_error(self, e, summary):
 209         if e.code == 'editconflict':
 210             raise errors.EditError(self, summary, e.info)
 211         elif e.code in ('protectedtitle', 'cantcreate', 'cantcreate-anon', 'noimageredirect-anon',
 212                         'noimageredirect', 'noedit-anon', 'noedit'):
 213             raise errors.ProtectedPageError(self, e.code, e.info)
 214         else:
 215             raise
 216
 217     def get_expanded(self):
 218         self.site.require(1, 12)
 219
 220         revs = self.revisions(prop='content', limit=1, expandtemplates=True)
 221         try:
 222             return revs.next()['*']
 223         except StopIteration:
 224             return u''
 225
 226     def move(self, new_title, reason='', move_talk=True, no_redirect=False):
 227         """Move (rename) page to new_title.
 228
 229         If user account is an administrator, specify no_direct as True to not
 230         leave a redirect.
 231
 232         If user does not have permission to move page, an InsufficientPermission
 233         exception is raised.
 234
 235         """
 236         if not self.can('move'):
 237             raise errors.InsufficientPermission(self)
 238
 239         if not self.site.writeapi:
 240             return OldPage.move(self, new_title=new_title,
 241                                 reason=reason, move_talk=move_talk)
 242
 243         data = {}
 244         if move_talk:
 245             data['movetalk'] = '1'
 246         if no_redirect:
 247             data['noredirect'] = '1'
 248         result = self.site.api('move', ('from', self.name), to=new_title,
 249                                token=self.get_token('move'), reason=reason, **data)
 250         return result['move']
 251
 252     def delete(self, reason='', watch=False, unwatch=False, oldimage=False):
 253         """Delete page.
 254
 255         If user does not have permission to delete page, an InsufficientPermission
 256         exception is raised.
 257
 258         """
 259         if not self.can('delete'):
 260             raise errors.InsufficientPermission(self)
 261
 262         if not self.site.writeapi:
 263             return OldPage.delete(self, reason=reason)
 264
 265         data = {}
 266         if watch:
 267             data['watch'] = '1'
 268         if unwatch:
 269             data['unwatch'] = '1'
 270         if oldimage:
 271             data['oldimage'] = oldimage
 272         result = self.site.api('delete', title=self.name,
 273                                token=self.get_token('delete'),
 274                                reason=reason, **data)
 275         return result['delete']
 276
 277     def purge(self):
 278         """Purge server-side cache of page. This will re-render templates and other
 279         dynamic content.
 280
 281         """
 282         self.site.raw_index('purge', title=self.name)
 283
 284     # def watch: requires 1.14
 285
 286     # Properties
 287     def backlinks(self, namespace=None, filterredir='all', redirect=False, limit=None, generator=True):
 288         self.site.require(1, 9)
 289         # Fix title for < 1.11 !!
 290         prefix = listing.List.get_prefix('bl', generator)
 291         kwargs = dict(listing.List.generate_kwargs(prefix,
 292                                                    namespace=namespace, filterredir=filterredir))
 293         if redirect:
 294             kwargs['%sredirect' % prefix] = '1'
 295         kwargs[compatibility.title(prefix, self.site.require(1, 11, raise_error=False))] = self.name
 296
 297         return listing.List.get_list(generator)(self.site, 'backlinks', 'bl', limit=limit, return_values='title', **kwargs)
 298
 299     def categories(self, generator=True):
 300         self.site.require(1, 11)
 301         if generator:
 302             return listing.PagePropertyGenerator(self, 'categories', 'cl')
 303         else:
 304             # TODO: return sortkey if wanted
 305             return listing.PageProperty(self, 'categories', 'cl', return_values='title')
 306
 307     def embeddedin(self, namespace=None, filterredir='all', redirect=False, limit=None, generator=True):
 308         self.site.require(1, 9)
 309         # Fix title for < 1.11 !!
 310         prefix = listing.List.get_prefix('ei', generator)
 311         kwargs = dict(listing.List.generate_kwargs(prefix,
 312                                                    namespace=namespace, filterredir=filterredir))
 313         if redirect:
 314             kwargs['%sredirect' % prefix] = '1'
 315         kwargs[compatibility.title(prefix, self.site.require(1, 11, raise_error=False))] = self.name
 316
 317         return listing.List.get_list(generator)(self.site, 'embeddedin', 'ei', limit=limit, return_values='title', **kwargs)
 318
 319     def extlinks(self):
 320         self.site.require(1, 11)
 321         return listing.PageProperty(self, 'extlinks', 'el', return_values='*')
 322
 323     def images(self, generator=True):
 324         self.site.require(1, 9)
 325         if generator:
 326             return listing.PagePropertyGenerator(self, 'images', '')
 327         else:
 328             return listing.PageProperty(self, 'images', '', return_values='title')
 329
 330     def iwlinks(self):
 331         self.site.require(1, 9)  # guessing...
 332         return listing.PageProperty(self, 'iwlinks', 'iw', return_values=('prefix', '*'))
 333
 334     def langlinks(self, **kwargs):
 335         self.site.require(1, 9)
 336         return listing.PageProperty(self, 'langlinks', 'll', return_values=('lang', '*'), **kwargs)
 337
 338     def links(self, namespace=None, generator=True, redirects=False):
 339         self.site.require(1, 9)
 340         kwargs = dict(listing.List.generate_kwargs('pl', namespace=namespace))
 341         if redirects:
 342             kwargs['redirects'] = '1'
 343         if generator:
 344             return listing.PagePropertyGenerator(self, 'links', 'pl', **kwargs)
 345         else:
 346             return listing.PageProperty(self, 'links', 'pl', return_values='title', **kwargs)
 347
 348     def revisions(self, startid=None, endid=None, start=None, end=None,
 349                   dir='older', user=None, excludeuser=None, limit=50,
 350                   prop='ids|timestamp|flags|comment|user', expandtemplates=False, section=None):
 351         self.site.require(1, 8)
 352         kwargs = dict(listing.List.generate_kwargs('rv', startid=startid, endid=endid,
 353                                                    start=start, end=end, user=user, excludeuser=excludeuser))
 354         kwargs['rvdir'] = dir
 355         kwargs['rvprop'] = prop
 356         if expandtemplates:
 357             kwargs['rvexpandtemplates'] = '1'
 358         if section:
 359             kwargs['rvsection'] = section
 360
 361         return listing.RevisionsIterator(self, 'revisions', 'rv', limit=limit, **kwargs)
 362
 363     def templates(self, namespace=None, generator=True):
 364         self.site.require(1, 8)
 365         kwargs = dict(listing.List.generate_kwargs('tl', namespace=namespace))
 366         if generator:
 367             return listing.PagePropertyGenerator(self, 'templates', 'tl')
 368         else:
 369             return listing.PageProperty(self, 'templates', 'tl', return_values='title')
 370
 371
 372 class Image(Page):
 373
 374     def __init__(self, site, name, info=None):
 375         site.require(1, 11)
 376         Page.__init__(self, site, name, info,
 377                       extra_properties={'imageinfo': (('iiprop',
 378                                                        compatibility.iiprop(site.version)), )})
 379         self.imagerepository = self._info.get('imagerepository', '')
 380         self.imageinfo = self._info.get('imageinfo', ({}, ))[0]
 381
 382     def imagehistory(self):
 383         return listing.PageProperty(self, 'imageinfo', 'ii',
 384                                     iiprop=compatibility.iiprop(self.site.version))
 385
 386     def imageusage(self, namespace=None, filterredir='all', redirect=False,
 387                    limit=None, generator=True):
 388         self.site.require(1, 11)
 389         # TODO: Fix for versions < 1.11
 390         prefix = listing.List.get_prefix('iu', generator)
 391         kwargs = dict(listing.List.generate_kwargs(prefix, title=self.name,
 392                                                    namespace=namespace, filterredir=filterredir))
 393         if redirect:
 394             kwargs['%sredirect' % prefix] = '1'
 395         return listing.List.get_list(generator)(self.site, 'imageusage', 'iu',
 396                                                 limit=limit, return_values='title', **kwargs)
 397
 398     def duplicatefiles(self, limit=None):
 399         self.require(1, 14)
 400         return listing.PageProperty(self, 'duplicatefiles', 'df',
 401                                     dflimit=limit)
 402
 403     def download(self):
 404         url = self.imageinfo['url']
 405         if not url.startswith('http://'):
 406             url = 'http://' + self.site.host + url
 407         url = urlparse.urlparse(url)
 408         # TODO: query string
 409         return self.site.connection.get(url[1], url[2])
 410
 411     def __repr__(self):
 412         return "<Image object '%s' for %s>" % (self.name.encode('utf-8'), self.site)