Archives/wikiteam
2
0
Fork 0
mirror of https://github.com/WikiTeam/wikiteam synced 2024-11-10 13:10:27 +00:00
Code Issues Projects Releases Wiki Activity

Raise and catch PageMissingError when revisions API result is incomplete

Browse Source 
https://github.com/WikiTeam/wikiteam/issues/317
This commit is contained in:
Federico Leva 2018-05-22 10:16:52 +03:00
parent 7a655f0074
commit d76b4b4e01
1 changed files with 33 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
dumpgenerator.py
View File

@ -871,7 +871,15 @@ def getXMLRevisions(config={}, session=None, allpages=False):
for result in results: for result in results:
pages = result['query']['pages'] pages = result['query']['pages']
for page in pages: for page in pages:
yield makeXmlFromPage(pages[page]) try:
xml = makeXmlFromPage(pages[page])
except PageMissingError:
logerror(
config=config,
text=u'Error: empty revision from API. Could not export page: %s' % (title.decode('utf-8'))
)
continue
yield xml
except wikitools.api.APIError: except wikitools.api.APIError:
print "This wikitools version seems not to work for us. Exiting." print "This wikitools version seems not to work for us. Exiting."
@ -879,6 +887,7 @@ def getXMLRevisions(config={}, session=None, allpages=False):
def makeXmlFromPage(page): def makeXmlFromPage(page):
""" Output an XML document as a string from a page as in the API JSON """ """ Output an XML document as a string from a page as in the API JSON """
try:
p = E.page( p = E.page(
E.title(page['title']), E.title(page['title']),
E.ns(to_unicode(page['ns'])), E.ns(to_unicode(page['ns'])),
@ -896,11 +905,13 @@ def makeXmlFromPage(page):
E.text(rev['*'], space="preserve", bytes=to_unicode(rev['size'])), E.text(rev['*'], space="preserve", bytes=to_unicode(rev['size'])),
) )
if 'contentmodel' in rev: if 'contentmodel' in rev:
revision.append(E.model(rev['contentmodel']) revision.append(E.model(rev['contentmodel']))
# The sha1 may not have been backfilled on older wikis or lack for other reasons (Wikia). # The sha1 may not have been backfilled on older wikis or lack for other reasons (Wikia).
if 'sha1' in rev: if 'sha1' in rev:
revision.append(E.sha1(rev['sha1'])) revision.append(E.sha1(rev['sha1']))
p.append(revision) p.append(revision)
except KeyError:
raise PageMissingError(page['title'], '')
return etree.tostring(p, pretty_print=True) return etree.tostring(p, pretty_print=True)
def readTitles(config={}, start=None): def readTitles(config={}, start=None):