def export_olx(self, user_id, course_key_string, language):
"""
Export a course or library to an OLX .tar.gz archive and prepare it for download.
"""
set_code_owner_attribute_from_module(__name__)
courselike_key = CourseKey.from_string(course_key_string)
try:
user = User.objects.get(pk=user_id)
except User.DoesNotExist:
with translation_language(language):
self.status.fail(UserErrors.UNKNOWN_USER_ID.format(user_id))
return
if not has_course_author_access(user, courselike_key):
with translation_language(language):
self.status.fail(UserErrors.PERMISSION_DENIED)
return
if isinstance(courselike_key, LibraryLocator):
courselike_module = modulestore().get_library(courselike_key)
else:
courselike_module = modulestore().get_course(courselike_key)
try:
self.status.set_state('Exporting')
tarball = create_export_tarball(courselike_module, courselike_key, {}, self.status)
artifact = UserTaskArtifact(status=self.status, name='Output')
artifact.file.save(name=os.path.basename(tarball.name), content=File(tarball))
artifact.save()
# catch all exceptions so we can record useful error messages
except Exception as exception: # pylint: disable=broad-except
LOGGER.exception('Error exporting course %s', courselike_key, exc_info=True)
if self.status.state != UserTaskStatus.FAILED:
self.status.fail({'raw_error_msg': str(exception)})
return
def verify_root_name_exists(course_dir, root_name):
"""Verify root xml file exists."""
def get_all_files(directory):
"""
For each file in the directory, yield a 2-tuple of (file-name,
directory-path)
"""
for directory_path, _dirnames, filenames in os.walk(directory):
for filename in filenames:
yield (filename, directory_path)
def get_dir_for_filename(directory, filename):
"""
Returns the directory path for the first file found in the directory
with the given name. If there is no file in the directory with
the specified name, return None.
"""
for name, directory_path in get_all_files(directory):
if name == filename:
return directory_path
return None
dirpath = get_dir_for_filename(course_dir, root_name)
if not dirpath:
message = UserErrors.FILE_MISSING.format(root_name)
with translation_language(language):
self.status.fail(message)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return
return dirpath
def validate_user():
"""Validate if the user exists otherwise log error. """
try:
return User.objects.get(pk=user_id)
except User.DoesNotExist as exc:
with translation_language(language):
self.status.fail(UserErrors.USER_PERMISSION_DENIED)
LOGGER.error(f'{log_prefix}: Unknown User: {user_id}')
monitor_import_failure(courselike_key, current_step, exception=exc)
return
def user_has_access(user):
"""Return True if user has studio write access to the given course."""
has_access = has_course_author_access(user, courselike_key)
if not has_access:
message = f'User permission denied: {user.username}'
with translation_language(language):
self.status.fail(UserErrors.COURSE_PERMISSION_DENIED)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key, current_step, message=message)
return has_access
def file_exists_in_storage():
"""Verify archive path exists in storage."""
archive_path_exists = course_import_export_storage.exists(archive_path)
if not archive_path_exists:
message = f'Uploaded file {archive_path} not found'
with translation_language(language):
self.status.fail(UserErrors.FILE_NOT_FOUND)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key, current_step, message=message)
return archive_path_exists
def file_is_supported():
"""Check if it is a supported file."""
file_is_valid = archive_name.endswith('.tar.gz')
if not file_is_valid:
message = f'Unsupported file {archive_name}'
with translation_language(language):
self.status.fail(UserErrors.INVALID_FILE_TYPE)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key, current_step, message=message)
return file_is_valid
def file_exists_in_storage():
archive_path_exists = course_import_export_storage.exists(archive_path)
if not archive_path_exists:
message = f'Uploaded file {archive_path} not found'
with translation_language(language):
self.status.fail(_('Tar file not found'))
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return archive_path_exists
def import_olx(self, user_id, course_key_string, archive_path, archive_name,
language):
"""
Import a course or library from a provided OLX .tar.gz archive.
"""
set_code_owner_attribute_from_module(__name__)
courselike_key = CourseKey.from_string(course_key_string)
try:
user = User.objects.get(pk=user_id)
except User.DoesNotExist:
with translation_language(language):
self.status.fail(_(u'Unknown User ID: {0}').format(user_id))
return
if not has_course_author_access(user, courselike_key):
with translation_language(language):
self.status.fail(_(u'Permission denied'))
return
is_library = isinstance(courselike_key, LibraryLocator)
is_course = not is_library
if is_library:
root_name = LIBRARY_ROOT
courselike_module = modulestore().get_library(courselike_key)
import_func = import_library_from_xml
else:
root_name = COURSE_ROOT
courselike_module = modulestore().get_course(courselike_key)
import_func = import_course_from_xml
# Locate the uploaded OLX archive (and download it from S3 if necessary)
# Do everything in a try-except block to make sure everything is properly cleaned up.
data_root = path(settings.GITHUB_REPO_ROOT)
subdir = base64.urlsafe_b64encode(
repr(courselike_key).encode('utf-8')).decode('utf-8')
course_dir = data_root / subdir
try:
self.status.set_state(u'Unpacking')
if not archive_name.endswith(u'.tar.gz'):
with translation_language(language):
self.status.fail(
_(u'We only support uploading a .tar.gz file.'))
return
temp_filepath = course_dir / get_valid_filename(archive_name)
if not course_dir.isdir():
os.mkdir(course_dir)
LOGGER.debug(u'importing course to {0}'.format(temp_filepath))
# Copy the OLX archive from where it was uploaded to (S3, Swift, file system, etc.)
if not course_import_export_storage.exists(archive_path):
LOGGER.info(u'Course import %s: Uploaded file %s not found',
courselike_key, archive_path)
with translation_language(language):
self.status.fail(_(u'Tar file not found'))
return
with course_import_export_storage.open(archive_path, 'rb') as source:
with open(temp_filepath, 'wb') as destination:
def read_chunk():
"""
Read and return a sequence of bytes from the source file.
"""
return source.read(FILE_READ_CHUNK)
for chunk in iter(read_chunk, b''):
destination.write(chunk)
LOGGER.info(u'Course import %s: Download from storage complete',
courselike_key)
# Delete from source location
course_import_export_storage.delete(archive_path)
# If the course has an entrance exam then remove it and its corresponding milestone.
# current course state before import.
if is_course:
if courselike_module.entrance_exam_enabled:
fake_request = RequestFactory().get(u'/')
fake_request.user = user
from .views.entrance_exam import remove_entrance_exam_milestone_reference
# TODO: Is this really ok? Seems dangerous for a live course
remove_entrance_exam_milestone_reference(
fake_request, courselike_key)
LOGGER.info(
u'entrance exam milestone content reference for course %s has been removed',
courselike_module.id)
# Send errors to client with stage at which error occurred.
except Exception as exception: # pylint: disable=broad-except
if course_dir.isdir():
shutil.rmtree(course_dir)
LOGGER.info(u'Course import %s: Temp data cleared', courselike_key)
LOGGER.exception(u'Error importing course %s',
courselike_key,
exc_info=True)
self.status.fail(text_type(exception))
return
# try-finally block for proper clean up after receiving file.
try:
tar_file = tarfile.open(temp_filepath)
try:
safetar_extractall(tar_file, (course_dir + u'/'))
except SuspiciousOperation as exc:
LOGGER.info(u'Course import %s: Unsafe tar file - %s',
courselike_key, exc.args[0])
with translation_language(language):
self.status.fail(_(u'Unsafe tar file. Aborting import.'))
return
finally:
tar_file.close()
LOGGER.info(u'Course import %s: Uploaded file extracted',
courselike_key)
self.status.set_state(u'Verifying')
self.status.increment_completed_steps()
# find the 'course.xml' file
def get_all_files(directory):
"""
For each file in the directory, yield a 2-tuple of (file-name,
directory-path)
"""
for directory_path, _dirnames, filenames in os.walk(directory):
for filename in filenames:
yield (filename, directory_path)
def get_dir_for_filename(directory, filename):
"""
Returns the directory path for the first file found in the directory
with the given name. If there is no file in the directory with
the specified name, return None.
"""
for name, directory_path in get_all_files(directory):
if name == filename:
return directory_path
return None
dirpath = get_dir_for_filename(course_dir, root_name)
if not dirpath:
with translation_language(language):
self.status.fail(
_(u'Could not find the {0} file in the package.').format(
root_name))
return
dirpath = os.path.relpath(dirpath, data_root)
LOGGER.debug(u'found %s at %s', root_name, dirpath)
LOGGER.info(u'Course import %s: Extracted file verified',
courselike_key)
self.status.set_state(u'Updating')
self.status.increment_completed_steps()
courselike_items = import_func(modulestore(),
user.id,
settings.GITHUB_REPO_ROOT, [dirpath],
load_error_modules=False,
static_content_store=contentstore(),
target_id=courselike_key)
new_location = courselike_items[0].location
LOGGER.debug(u'new course at %s', new_location)
LOGGER.info(u'Course import %s: Course import successful',
courselike_key)
except Exception as exception: # pylint: disable=broad-except
LOGGER.exception(u'error importing course', exc_info=True)
self.status.fail(text_type(exception))
finally:
if course_dir.isdir():
shutil.rmtree(course_dir)
LOGGER.info(u'Course import %s: Temp data cleared', courselike_key)
if self.status.state == u'Updating' and is_course:
# Reload the course so we have the latest state
course = modulestore().get_course(courselike_key)
if course.entrance_exam_enabled:
entrance_exam_chapter = modulestore().get_items(
course.id,
qualifiers={u'category': u'chapter'},
settings={u'is_entrance_exam': True})[0]
metadata = {
u'entrance_exam_id':
text_type(entrance_exam_chapter.location)
}
CourseMetadata.update_from_dict(metadata, course, user)
from .views.entrance_exam import add_entrance_exam_milestone
add_entrance_exam_milestone(course.id, entrance_exam_chapter)
LOGGER.info(u'Course %s Entrance exam imported', course.id)
def import_olx(self, user_id, course_key_string, archive_path, archive_name,
language):
"""
Import a course or library from a provided OLX .tar.gz archive.
"""
current_step = 'Unpacking'
courselike_key = CourseKey.from_string(course_key_string)
set_code_owner_attribute_from_module(__name__)
set_custom_attributes_for_course_key(courselike_key)
log_prefix = f'Course import {courselike_key}'
self.status.set_state(current_step)
data_root = path(settings.GITHUB_REPO_ROOT)
subdir = base64.urlsafe_b64encode(
repr(courselike_key).encode('utf-8')).decode('utf-8')
course_dir = data_root / subdir
def validate_user():
"""Validate if the user exists otherwise log error. """
try:
return User.objects.get(pk=user_id)
except User.DoesNotExist as exc:
with translation_language(language):
self.status.fail(UserErrors.USER_PERMISSION_DENIED)
LOGGER.error(f'{log_prefix}: Unknown User: {user_id}')
monitor_import_failure(courselike_key, current_step, exception=exc)
return
def user_has_access(user):
"""Return True if user has studio write access to the given course."""
has_access = has_course_author_access(user, courselike_key)
if not has_access:
message = f'User permission denied: {user.username}'
with translation_language(language):
self.status.fail(UserErrors.COURSE_PERMISSION_DENIED)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return has_access
def file_is_supported():
"""Check if it is a supported file."""
file_is_valid = archive_name.endswith('.tar.gz')
if not file_is_valid:
message = f'Unsupported file {archive_name}'
with translation_language(language):
self.status.fail(UserErrors.INVALID_FILE_TYPE)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return file_is_valid
def file_exists_in_storage():
"""Verify archive path exists in storage."""
archive_path_exists = course_import_export_storage.exists(archive_path)
if not archive_path_exists:
message = f'Uploaded file {archive_path} not found'
with translation_language(language):
self.status.fail(UserErrors.FILE_NOT_FOUND)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return archive_path_exists
def verify_root_name_exists(course_dir, root_name):
"""Verify root xml file exists."""
def get_all_files(directory):
"""
For each file in the directory, yield a 2-tuple of (file-name,
directory-path)
"""
for directory_path, _dirnames, filenames in os.walk(directory):
for filename in filenames:
yield (filename, directory_path)
def get_dir_for_filename(directory, filename):
"""
Returns the directory path for the first file found in the directory
with the given name. If there is no file in the directory with
the specified name, return None.
"""
for name, directory_path in get_all_files(directory):
if name == filename:
return directory_path
return None
dirpath = get_dir_for_filename(course_dir, root_name)
if not dirpath:
message = UserErrors.FILE_MISSING.format(root_name)
with translation_language(language):
self.status.fail(message)
LOGGER.error(f'{log_prefix}: {message}')
monitor_import_failure(courselike_key,
current_step,
message=message)
return
return dirpath
user = validate_user()
if not user:
return
if not user_has_access(user):
return
if not file_is_supported():
return
is_library = isinstance(courselike_key, LibraryLocator)
is_course = not is_library
if is_library:
root_name = LIBRARY_ROOT
courselike_module = modulestore().get_library(courselike_key)
import_func = import_library_from_xml
else:
root_name = COURSE_ROOT
courselike_module = modulestore().get_course(courselike_key)
import_func = import_course_from_xml
# Locate the uploaded OLX archive (and download it from S3 if necessary)
# Do everything in a try-except block to make sure everything is properly cleaned up.
try:
LOGGER.info(f'{log_prefix}: unpacking step started')
temp_filepath = course_dir / get_valid_filename(archive_name)
if not course_dir.isdir():
os.mkdir(course_dir)
LOGGER.info(f'{log_prefix}: importing course to {temp_filepath}')
# Copy the OLX archive from where it was uploaded to (S3, Swift, file system, etc.)
if not file_exists_in_storage():
return
with course_import_export_storage.open(archive_path, 'rb') as source:
with open(temp_filepath, 'wb') as destination:
def read_chunk():
"""
Read and return a sequence of bytes from the source file.
"""
return source.read(FILE_READ_CHUNK)
for chunk in iter(read_chunk, b''):
destination.write(chunk)
LOGGER.info(f'{log_prefix}: Download from storage complete')
# Delete from source location
course_import_export_storage.delete(archive_path)
# If the course has an entrance exam then remove it and its corresponding milestone.
# current course state before import.
if is_course:
if courselike_module.entrance_exam_enabled:
fake_request = RequestFactory().get('/')
fake_request.user = user
from .views.entrance_exam import remove_entrance_exam_milestone_reference
# TODO: Is this really ok? Seems dangerous for a live course
remove_entrance_exam_milestone_reference(
fake_request, courselike_key)
LOGGER.info(
f'{log_prefix}: entrance exam milestone content reference has been removed'
)
# Send errors to client with stage at which error occurred.
except Exception as exception: # pylint: disable=broad-except
if course_dir.isdir():
shutil.rmtree(course_dir)
LOGGER.info(f'{log_prefix}: Temp data cleared')
self.status.fail(UserErrors.UNKNOWN_ERROR_IN_UNPACKING)
LOGGER.exception(f'{log_prefix}: Unknown error while unpacking',
exc_info=True)
monitor_import_failure(courselike_key,
current_step,
exception=exception)
return
# try-finally block for proper clean up after receiving file.
try:
tar_file = tarfile.open(temp_filepath)
try:
safetar_extractall(tar_file, (course_dir + '/'))
except SuspiciousOperation as exc:
with translation_language(language):
self.status.fail(UserErrors.UNSAFE_TAR_FILE)
LOGGER.error(f'{log_prefix}: Unsafe tar file')
monitor_import_failure(courselike_key, current_step, exception=exc)
return
finally:
tar_file.close()
current_step = 'Verifying'
self.status.set_state(current_step)
self.status.increment_completed_steps()
LOGGER.info(
f'{log_prefix}: Uploaded file extracted. Verification step started'
)
dirpath = verify_root_name_exists(course_dir, root_name)
if not dirpath:
return
if not validate_course_olx(courselike_key, dirpath, self.status):
return
dirpath = os.path.relpath(dirpath, data_root)
current_step = 'Updating'
self.status.set_state(current_step)
self.status.increment_completed_steps()
LOGGER.info(
f'{log_prefix}: Extracted file verified. Updating course started')
courselike_items = import_func(
modulestore(),
user.id,
settings.GITHUB_REPO_ROOT,
[dirpath],
load_error_modules=False,
static_content_store=contentstore(),
target_id=courselike_key,
verbose=True,
)
new_location = courselike_items[0].location
LOGGER.debug('new course at %s', new_location)
LOGGER.info(f'{log_prefix}: Course import successful')
set_custom_attribute('course_import_completed', True)
except (CourseImportException, InvalidProctoringProvider,
DuplicateCourseError) as known_exe:
handle_course_import_exception(courselike_key, known_exe, self.status)
except Exception as exception: # pylint: disable=broad-except
handle_course_import_exception(courselike_key,
exception,
self.status,
known=False)
finally:
if course_dir.isdir():
shutil.rmtree(course_dir)
LOGGER.info(f'{log_prefix}: Temp data cleared')
if self.status.state == 'Updating' and is_course:
# Reload the course so we have the latest state
course = modulestore().get_course(courselike_key)
if course.entrance_exam_enabled:
entrance_exam_chapter = modulestore().get_items(
course.id,
qualifiers={'category': 'chapter'},
settings={'is_entrance_exam': True})[0]
metadata = {
'entrance_exam_id': str(entrance_exam_chapter.location)
}
CourseMetadata.update_from_dict(metadata, course, user)
from .views.entrance_exam import add_entrance_exam_milestone
add_entrance_exam_milestone(course.id, entrance_exam_chapter)
LOGGER.info(
f'Course import {course.id}: Entrance exam imported')