def url_title_dicts(version_number=None):
if version_number:
version = topic_models.TopicVersion.get_by_number(version_number)
else:
version = None
return map(lambda url: {
"title": url.title,
"key": str(url.key()),
"ka_url": url.url,
"id": url.key().id()
}, Url.get_all_live(version=version))
def url_title_dicts(version_number=None):
if version_number:
version = topic_models.TopicVersion.get_by_number(version_number)
else:
version = None
return map(
lambda url: {
"title": url.title,
"key": str(url.key()),
"ka_url": url.url,
"id": url.key().id()
}, Url.get_all_live(version=version))
def importIntoVersion(version):
logging.info("comparing to version number %i" % version.number)
topic = Topic.get_by_id("art-history", version)
if not topic:
parent = Topic.get_by_id("humanities---other", version)
if not parent:
raise Exception("Could not find the Humanities & Other topic to put art history into")
topic = Topic.insert(title="Art History",
parent=parent,
id="art-history",
standalone_title="Art History",
description="Spontaneous conversations about works of art where the speakers are not afraid to disagree with each other or art history orthodoxy. Videos are made by Dr. Beth Harris and Dr. Steven Zucker along with other contributors.")
urls = topic.get_urls(include_descendants=True)
href_to_key_dict = dict((url.url, url.key()) for url in urls)
videos = topic.get_videos(include_descendants=True)
video_dict = dict((v.youtube_id, v) for v in videos)
content = getSmartHistoryContent()
if content is None:
raise Exception("Aborting import, could not read from smarthistory")
subtopics = topic.get_child_topics()
subtopic_dict = dict((t.title, t) for t in subtopics)
subtopic_child_keys = {}
new_subtopic_keys = []
child_keys = []
i = 0
for link in content:
href = link["href"]
title = link["title"]
parent_title = link["parent"]
content = link["content"]
youtube_id = link["youtube_id"] if "youtube_id" in link else None
extra_properties = {"original_url": href}
if parent_title not in subtopic_dict:
subtopic = Topic.insert(title=parent_title,
parent=topic,
standalone_title="Art History: %s"
% parent_title,
description="")
subtopic_dict[parent_title] = subtopic
else:
subtopic = subtopic_dict[parent_title]
if subtopic.key() not in new_subtopic_keys:
new_subtopic_keys.append(subtopic.key())
if parent_title not in subtopic_child_keys:
subtopic_child_keys[parent_title] = []
if youtube_id:
if youtube_id not in video_dict:
# make sure it didn't get imported before, but never put
# into a topic
query = video_models.Video.all()
video = query.filter("youtube_id =", youtube_id).get()
if video is None:
logging.info("adding youtube video %i %s %s %s to %s" %
(i, youtube_id, href, title, parent_title))
video_data = youtube_get_video_data_dict(youtube_id)
# use the title from the webpage not from the youtube
# page
video = None
if video_data:
video_data["title"] = title
video_data["extra_properties"] = extra_properties
video = topic_models.VersionContentChange.add_new_content(
video_models.Video,
version,
video_data)
else:
logging.error(("Could not import youtube_id %s " +
"for %s %s") % (youtube_id, href, title))
raise Exception(("Could not import youtube_id %s " +
" for %s %s") % (youtube_id, href,
title))
else:
video = video_dict[youtube_id]
if video.extra_properties != extra_properties:
logging.info(("changing extra properties of %i %s %s " +
"from %s to %s") % (i, href, title,
video.extra_properties, extra_properties))
video.extra_properties = extra_properties
video.put()
if video:
subtopic_child_keys[parent_title].append(video.key())
elif href not in href_to_key_dict:
logging.info("adding %i %s %s to %s" %
(i, href, title, parent_title))
topic_models.VersionContentChange.add_new_content(
Url,
version,
{"title": title,
"url": href
},
["title", "url"])
url = Url(url=href,
title=title,
id=id)
url.put()
subtopic_child_keys[parent_title].append(url.key())
else:
subtopic_child_keys[parent_title].append(href_to_key_dict[href])
i += 1
logging.info("updating child_keys")
change = False
for parent_title, child_keys in subtopic_child_keys.iteritems():
subtopic = subtopic_dict[parent_title]
if subtopic.child_keys != subtopic_child_keys[parent_title]:
change = True
subtopic.update(child_keys=subtopic_child_keys[parent_title])
if topic.child_keys != new_subtopic_keys:
change = True
topic.update(child_keys=new_subtopic_keys)
if change:
logging.info("finished updating version number %i" % version.number)
else:
logging.info("nothing changed")
return change
def importIntoVersion(version):
logging.info("comparing to version number %i" % version.number)
topic = Topic.get_by_id("art-history", version)
if not topic:
parent = Topic.get_by_id("humanities---other", version)
if not parent:
raise Exception("Could not find the Humanities & Other topic to put art history into")
topic = Topic.insert(title="Art History",
parent=parent,
id="art-history",
standalone_title="Art History",
description="Spontaneous conversations about works of art where the speakers are not afraid to disagree with each other or art history orthodoxy. Videos are made by Dr. Beth Harris and Dr. Steven Zucker along with other contributors.")
urls = topic.get_urls(include_descendants=True)
href_to_key_dict = dict((url.url, url.key()) for url in urls)
videos = topic.get_videos(include_descendants=True)
video_dict = dict((v.youtube_id, v) for v in videos)
content = getSmartHistoryContent()
if content is None:
raise Exception("Aborting import, could not read from smarthistory")
subtopics = topic.get_child_topics()
subtopic_dict = dict((t.title, t) for t in subtopics)
subtopic_child_keys = {}
new_subtopic_keys = []
child_keys = []
i = 0
for link in content:
href = link["href"]
title = link["title"]
parent_title = link["parent"]
content = link["content"]
youtube_id = link["youtube_id"] if "youtube_id" in link else None
extra_properties = {"original_url": href}
if parent_title not in subtopic_dict:
subtopic = Topic.insert(title=parent_title,
parent=topic,
standalone_title="Art History: %s"
% parent_title,
description="")
subtopic_dict[parent_title] = subtopic
else:
subtopic = subtopic_dict[parent_title]
if subtopic.key() not in new_subtopic_keys:
new_subtopic_keys.append(subtopic.key())
if parent_title not in subtopic_child_keys:
subtopic_child_keys[parent_title] = []
if youtube_id:
if youtube_id not in video_dict:
# make sure it didn't get imported before, but never put
# into a topic
query = video_models.Video.all()
video = query.filter("youtube_id =", youtube_id).get()
if video is None:
logging.info("adding youtube video %i %s %s %s to %s" %
(i, youtube_id, href, title, parent_title))
video_data = youtube_get_video_data_dict(youtube_id)
# use the title from the webpage not from the youtube
# page
video = None
if video_data:
video_data["title"] = title
video_data["extra_properties"] = extra_properties
video = topic_models.VersionContentChange.add_new_content(
video_models.Video,
version,
video_data)
else:
logging.error(("Could not import youtube_id %s " +
"for %s %s") % (youtube_id, href, title))
raise Exception(("Could not import youtube_id %s " +
" for %s %s") % (youtube_id, href,
title))
else:
video = video_dict[youtube_id]
if video.extra_properties != extra_properties:
logging.info(("changing extra properties of %i %s %s " +
"from %s to %s") % (i, href, title,
video.extra_properties, extra_properties))
video.extra_properties = extra_properties
video.put()
if video:
subtopic_child_keys[parent_title].append(video.key())
elif href not in href_to_key_dict:
logging.info("adding %i %s %s to %s" %
(i, href, title, parent_title))
topic_models.VersionContentChange.add_new_content(
Url,
version,
{"title": title,
"url": href
},
["title", "url"])
url = Url(url=href,
title=title,
id=id)
url.put()
subtopic_child_keys[parent_title].append(url.key())
else:
subtopic_child_keys[parent_title].append(href_to_key_dict[href])
i += 1
logging.info("updating child_keys")
change = False
for parent_title, child_keys in subtopic_child_keys.iteritems():
subtopic = subtopic_dict[parent_title]
if subtopic.child_keys != subtopic_child_keys[parent_title]:
change = True
subtopic.update(child_keys=subtopic_child_keys[parent_title])
if topic.child_keys != new_subtopic_keys:
change = True
topic.update(child_keys=new_subtopic_keys)
if change:
logging.info("finished updating version number %i" % version.number)
else:
logging.info("nothing changed")
return change
