def burst_pdf(**kwargs):
"""
A task that creates a sub-folder next to a PDF file and extracts all
pages from the PDF as PNG files into the sub-folder.
"""
from flask_app import app
from filesystem_manager import get_abs_path, get_burst_path, get_file_data
from filesystem_manager import delete_dir, make_dirs, path_exists
from filesystem_sync import delete_folder
from imagemagick import imagemagick_burst_pdf
from models import Folder
from util import get_file_extension
(src,) = _extract_parameters(["src"], **kwargs)
burst_folder_rel = get_burst_path(src)
# Ensure src is a PDF
if get_file_extension(src) not in app.config["PDF_FILE_TYPES"]:
app.log.warn("Cannot burst non-PDF file: " + src)
return
# See if the burst folder already exists (in the database and on disk)
db_folder = app.data_engine.get_folder(folder_path=burst_folder_rel)
if db_folder is not None and db_folder.status == Folder.STATUS_ACTIVE:
# Wipe the folder, old images, data, and uncache the old images
delete_folder(db_folder, None, app.data_engine, None, app.log)
deleted_ids = app.data_engine.list_image_ids(db_folder)
for image_id in deleted_ids:
app.image_engine._uncache_image_id(image_id)
# See if the burst folder already exists (just on disk)
if path_exists(burst_folder_rel, require_directory=True):
# Wipe the folder and old images
delete_dir(burst_folder_rel, recursive=True)
# Create the burst folder and burst
pdf_data = get_file_data(src)
if pdf_data is not None:
make_dirs(burst_folder_rel)
burst_folder_abs = get_abs_path(burst_folder_rel)
if not imagemagick_burst_pdf(pdf_data, burst_folder_abs, app.config["PDF_BURST_DPI"]):
app.log.warn("Failed to burst PDF: " + src)
else:
app.log.warn("Cannot burst PDF, file not found: " + src)
def auto_sync_existing_file(rel_path, data_manager, task_manager,
anon_history=True, burst_pdf='auto', _db_session=None):
"""
Returns the database record for an image file that is known to exist,
creating a new record or un-deleting an old record if required,
and always returning a value.
This method creates anonymous image history entries when anon_history is
True. If the current user should be recorded against an action, the caller
should set anon_history to False and manually add a history record.
The bursting of PDF files is also initiated here. By default, a PDF file
will be burst if no burst folder already exists. Setting burst_pdf to False
disables this, or setting burst_pdf to True will force it to be burst again.
Raises a DoesNotExistError if the image path is in fact invalid.
Raises a SecurityError if the image path is outside of IMAGES_BASE_DIR.
Raises a DBError if the database record cannot be created.
"""
db_own = (_db_session is None)
db_session = _db_session or data_manager.db_get_session()
db_error = False
try:
# Get (or create) db record for the file
on_create = on_image_db_create_anon_history if anon_history \
else on_image_db_create
db_image = data_manager.get_or_create_image(
rel_path, on_create, _db_session=db_session
)
if not db_image:
# Not expected
raise DBError('Failed to add image to database: ' + rel_path)
# Burst PDF if we need to
# TODO This would be better in on_image_db_create if we can get a task_manager without
# importing the one from flask_app. Needs to be compatible with the task server.
if burst_pdf and app.config['PDF_BURST_TO_PNG']:
can_burst = get_file_extension(rel_path) in app.config['PDF_FILE_TYPES']
if can_burst:
if burst_pdf == 'auto':
burst_pdf = not path_exists(
get_burst_path(rel_path),
require_directory=True
)
if burst_pdf:
burst_pdf_file(rel_path, task_manager)
if db_image.status == Image.STATUS_ACTIVE:
# The normal case
return db_image
else:
# We need to undelete the database record
db_image.status = Image.STATUS_ACTIVE
if anon_history:
on_image_db_create_anon_history(db_image)
else:
on_image_db_create(db_image)
# Check whether the file's folder needs to be undeleted too
if db_image.folder.status == Folder.STATUS_DELETED:
auto_sync_existing_folder(
db_image.folder.path,
data_manager,
_db_session=db_session
)
return db_image
except:
db_error = True
raise
finally:
if db_own:
try:
if db_error:
db_session.rollback()
else:
db_session.commit()
finally:
db_session.close()
