Skip to content

E
edx-platform
Commit 4131f867 by Christina Roberts
Options

Merge pull request #1694 from MITx/fix/cdodge/disable-metadata-inheritence-computation-on-import 

add an array of courses to not refresh the metadata cache on writes. Thi...
parents e4437b7a 5dbb153a
Showing with 98 additions and 76 deletions
common/lib/xmodule/xmodule/modulestore/mongo.py
......@@ -246,6 +246,7 @@ class MongoModuleStore(ModuleStoreBase):
self.fs_root = path(fs_root)
self.error_tracker = error_tracker
self.render_template = render_template
self.ignore_write_events_on_courses = []
def get_metadata_inheritance_tree(self, location):
'''
......@@ -330,6 +331,11 @@ class MongoModuleStore(ModuleStoreBase):
return tree
def refresh_cached_metadata_inheritance_tree(self, location):
pseudo_course_id = '/'.join([location.org, location.course])
if pseudo_course_id not in self.ignore_write_events_on_courses:
self.get_cached_metadata_inheritance_tree(location, force_refresh = True)
def clear_cached_metadata_inheritance_tree(self, location):
key_name = '{0}/{1}'.format(location.org, location.course)
if self.metadata_inheritance_cache is not None:
......@@ -520,7 +526,7 @@ class MongoModuleStore(ModuleStoreBase):
raise DuplicateItemError(location)
# recompute (and update) the metadata inheritance tree which is cached
self.get_cached_metadata_inheritance_tree(Location(location), force_refresh = True)
self.refresh_cached_metadata_inheritance_tree(Location(location))
def get_course_for_item(self, location, depth=0):
'''
......@@ -590,7 +596,7 @@ class MongoModuleStore(ModuleStoreBase):
self._update_single_item(location, {'definition.children': children})
# recompute (and update) the metadata inheritance tree which is cached
self.get_cached_metadata_inheritance_tree(Location(location), force_refresh = True)
self.refresh_cached_metadata_inheritance_tree(Location(location))
def update_metadata(self, location, metadata):
"""
......@@ -616,7 +622,7 @@ class MongoModuleStore(ModuleStoreBase):
self._update_single_item(location, {'metadata': metadata})
# recompute (and update) the metadata inheritance tree which is cached
self.get_cached_metadata_inheritance_tree(loc, force_refresh = True)
self.refresh_cached_metadata_inheritance_tree(loc)
def delete_item(self, location):
"""
......@@ -639,8 +645,7 @@ class MongoModuleStore(ModuleStoreBase):
# from overriding our default value set in the init method.
safe=self.collection.safe)
# recompute (and update) the metadata inheritance tree which is cached
self.get_cached_metadata_inheritance_tree(Location(location), force_refresh = True)
self.refresh_cached_metadata_inheritance_tree(Location(location))
def get_parent_locations(self, location, course_id):
'''Find all locations that are the parents of this location in this
......
common/lib/xmodule/xmodule/modulestore/xml_importer.py
......@@ -201,100 +201,117 @@ def import_from_xml(store, data_dir, course_dirs=None,
course_items = []
for course_id in module_store.modules.keys():
course_data_path = None
course_location = None
if target_location_namespace is not None:
pseudo_course_id = '/'.join([target_location_namespace.org, target_location_namespace.course])
else:
course_id_components = course_id.split('/')
pseudo_course_id = '/'.join([course_id_components[0], course_id_components[1]])
if verbose:
log.debug("Scanning {0} for course module...".format(course_id))
try:
# turn off all write signalling while importing as this is a high volume operation
if pseudo_course_id not in store.ignore_write_events_on_courses:
store.ignore_write_events_on_courses.append(pseudo_course_id)
# Quick scan to get course module as we need some info from there. Also we need to make sure that the
# course module is committed first into the store
for module in module_store.modules[course_id].itervalues():
if module.category == 'course':
course_data_path = path(data_dir) / module.data_dir
course_location = module.location
course_data_path = None
course_location = None
module = remap_namespace(module, target_location_namespace)
if verbose:
log.debug("Scanning {0} for course module...".format(course_id))
# cdodge: more hacks (what else). Seems like we have a problem when importing a course (like 6.002) which
# does not have any tabs defined in the policy file. The import goes fine and then displays fine in LMS,
# but if someone tries to add a new tab in the CMS, then the LMS barfs because it expects that -
# if there is *any* tabs - then there at least needs to be some predefined ones
if module.tabs is None or len(module.tabs) == 0:
module.tabs = [{"type": "courseware"},
{"type": "course_info", "name": "Course Info"},
{"type": "discussion", "name": "Discussion"},
{"type": "wiki", "name": "Wiki"}] # note, add 'progress' when we can support it on Edge
# Quick scan to get course module as we need some info from there. Also we need to make sure that the
# course module is committed first into the store
for module in module_store.modules[course_id].itervalues():
if module.category == 'course':
course_data_path = path(data_dir) / module.data_dir
course_location = module.location
module = remap_namespace(module, target_location_namespace)
if hasattr(module, 'data'):
store.update_item(module.location, module.data)
store.update_children(module.location, module.children)
store.update_metadata(module.location, dict(own_metadata(module)))
# cdodge: more hacks (what else). Seems like we have a problem when importing a course (like 6.002) which
# does not have any tabs defined in the policy file. The import goes fine and then displays fine in LMS,
# but if someone tries to add a new tab in the CMS, then the LMS barfs because it expects that -
# if there is *any* tabs - then there at least needs to be some predefined ones
if module.tabs is None or len(module.tabs) == 0:
module.tabs = [{"type": "courseware"},
{"type": "course_info", "name": "Course Info"},
{"type": "discussion", "name": "Discussion"},
{"type": "wiki", "name": "Wiki"}] # note, add 'progress' when we can support it on Edge
# a bit of a hack, but typically the "course image" which is shown on marketing pages is hard coded to /images/course_image.jpg
# so let's make sure we import in case there are no other references to it in the modules
verify_content_links(module, course_data_path, static_content_store, '/static/images/course_image.jpg')
course_items.append(module)
if hasattr(module, 'data'):
store.update_item(module.location, module.data)
store.update_children(module.location, module.children)
store.update_metadata(module.location, dict(own_metadata(module)))
# a bit of a hack, but typically the "course image" which is shown on marketing pages is hard coded to /images/course_image.jpg
# so let's make sure we import in case there are no other references to it in the modules
verify_content_links(module, course_data_path, static_content_store, '/static/images/course_image.jpg')
# then import all the static content
if static_content_store is not None:
_namespace_rename = target_location_namespace if target_location_namespace is not None else course_location
course_items.append(module)
# first pass to find everything in /static/
import_static_content(module_store.modules[course_id], course_location, course_data_path, static_content_store,
_namespace_rename, subpath='static', verbose=verbose)
# finally loop through all the modules
for module in module_store.modules[course_id].itervalues():
# then import all the static content
if static_content_store is not None:
_namespace_rename = target_location_namespace if target_location_namespace is not None else course_location
if module.category == 'course':
# we've already saved the course module up at the top of the loop
# so just skip over it in the inner loop
continue
# first pass to find everything in /static/
import_static_content(module_store.modules[course_id], course_location, course_data_path, static_content_store,
_namespace_rename, subpath='static', verbose=verbose)
# remap module to the new namespace
if target_location_namespace is not None:
module = remap_namespace(module, target_location_namespace)
# finally loop through all the modules
for module in module_store.modules[course_id].itervalues():
if verbose:
log.debug('importing module location {0}'.format(module.location))
if module.category == 'course':
# we've already saved the course module up at the top of the loop
# so just skip over it in the inner loop
continue
if hasattr(module, 'data'):
module_data = module.data
# remap module to the new namespace
if target_location_namespace is not None:
module = remap_namespace(module, target_location_namespace)
# cdodge: now go through any link references to '/static/' and make sure we've imported
# it as a StaticContent asset
try:
remap_dict = {}
if verbose:
log.debug('importing module location {0}'.format(module.location))
if hasattr(module, 'data'):
module_data = module.data
# use the rewrite_links as a utility means to enumerate through all links
# in the module data. We use that to load that reference into our asset store
# IMPORTANT: There appears to be a bug in lxml.rewrite_link which makes us not be able to
# do the rewrites natively in that code.
# For example, what I'm seeing is <img src='foo.jpg' /> -> <img src='bar.jpg'>
# Note the dropped element closing tag. This causes the LMS to fail when rendering modules - that's
# no good, so we have to do this kludge
if isinstance(module_data, str) or isinstance(module_data, unicode): # some module 'data' fields are non strings which blows up the link traversal code
lxml_rewrite_links(module_data, lambda link: verify_content_links(module, course_data_path,
static_content_store, link, remap_dict))
# cdodge: now go through any link references to '/static/' and make sure we've imported
# it as a StaticContent asset
try:
remap_dict = {}
for key in remap_dict.keys():
module_data = module_data.replace(key, remap_dict[key])
# use the rewrite_links as a utility means to enumerate through all links
# in the module data. We use that to load that reference into our asset store
# IMPORTANT: There appears to be a bug in lxml.rewrite_link which makes us not be able to
# do the rewrites natively in that code.
# For example, what I'm seeing is <img src='foo.jpg' /> -> <img src='bar.jpg'>
# Note the dropped element closing tag. This causes the LMS to fail when rendering modules - that's
# no good, so we have to do this kludge
if isinstance(module_data, str) or isinstance(module_data, unicode): # some module 'data' fields are non strings which blows up the link traversal code
lxml_rewrite_links(module_data, lambda link: verify_content_links(module, course_data_path,
static_content_store, link, remap_dict))
except Exception, e:
logging.exception("failed to rewrite links on {0}. Continuing...".format(module.location))
for key in remap_dict.keys():
module_data = module_data.replace(key, remap_dict[key])
store.update_item(module.location, module_data)
except Exception, e:
logging.exception("failed to rewrite links on {0}. Continuing...".format(module.location))
if hasattr(module, 'children') and module.children != []:
store.update_children(module.location, module.children)
store.update_item(module.location, module_data)
# NOTE: It's important to use own_metadata here to avoid writing
# inherited metadata everywhere.
store.update_metadata(module.location, dict(own_metadata(module)))
if hasattr(module, 'children') and module.children != []:
store.update_children(module.location, module.children)
# NOTE: It's important to use own_metadata here to avoid writing
# inherited metadata everywhere.
store.update_metadata(module.location, dict(own_metadata(module)))
finally:
# turn back on all write signalling
if pseudo_course_id in store.ignore_write_events_on_courses:
store.ignore_write_events_on_courses.remove(pseudo_course_id)
store.refresh_cached_metadata_inheritance_tree(target_location_namespace if
target_location_namespace is not None else course_location)
return module_store, course_items
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment