def test_get_nb_page(file_path):
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
with pytest.raises(UnsupportedMimeType):
manager.get_page_nb(
file_path=file_path,
file_ext='.bin'
)
def test_get_nb_page() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
nb_page = manager.get_page_nb(file_path=PDF_FILE_PATH)
assert nb_page == 2
nb_page = manager.get_page_nb(file_path=PDF_FILE_PATH__ENCRYPTED)
assert nb_page == 2
nb_page = manager.get_page_nb(file_path=PDF_FILE_PATH__A4)
assert nb_page == 2
def test_page_number__extension_forced():
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
page_number = manager.get_page_nb(
file_path=IMAGE_FILE_PATH_NO_EXTENSION,
file_ext=".txt"
)
assert page_number == 1
def pdf_data_to_thumbnails_by_preview_generator(pdf_data,
page=None,
width_max=256,
height_max=256):
"""A more robust preview generator than imagemagick (wand).
:param page: an int for one page or a list of ints for multiple pages
:return: dict map from page number to encoded image of that page.
"""
# Installation:
# - pip install preview-generator
# - pakges to install: perl-image-exiftool, inskcape, scribus
# Testcase:
# - http://arxiv.org/abs/1612.01033v2
# - where preview_generator succeed but wand failed.
if not isinstance(page, (tuple, list)):
page_list = [page]
else:
page_list = page
if page_list is None:
page_list = list(range(num_pages))
from preview_generator.manager import PreviewManager
cache_dir = tempfile.mkdtemp(prefix='preview-cache-')
try:
# save pdf
fd, pdf_path = tempfile.mkstemp(dir=cache_dir)
os.close(fd)
with open(pdf_path, 'wb') as f:
f.write(pdf_data)
manager = PreviewManager(cache_dir, create_folder=True)
num_pages = manager.get_page_nb(pdf_path)
rst = {}
for page in page_list:
if not (0 <= page < num_pages):
continue
preview_path = manager.get_jpeg_preview(pdf_path,
width=width_max,
height=height_max,
page=page)
with open(preview_path, 'rb') as f:
rst[page] = f.read()
finally:
shutil.rmtree(cache_dir)
return rst
def pdf_data_to_thumbnails_by_preview_generator(pdf_data: bytes,
pages: List[int],
width_max: int,
height_max: int):
"""
Convert pdf data to images with preview generator, which is sometimes more robust.
:param page: an int for one page or a list of ints for multiple pages
:return: dict map from page number to encoded image of that page.
"""
# Installation:
# - pip install preview-generator
# - pakges to install: perl-image-exiftool, inskcape, scribus
# Testcase:
# - http://arxiv.org/abs/1612.01033v2
# - where preview_generator succeed but wand failed.
from preview_generator.manager import PreviewManager
cache_dir = tempfile.mkdtemp(prefix="preview-cache-")
try:
# save pdf
fd, pdf_path = tempfile.mkstemp(dir=cache_dir)
os.close(fd)
with open(pdf_path, "wb") as f:
f.write(pdf_data)
manager = PreviewManager(cache_dir, create_folder=True)
num_pages = manager.get_page_nb(pdf_path)
rst = {}
for page in pages:
if not (0 <= page < num_pages):
continue
preview_path = manager.get_jpeg_preview(pdf_path,
width=width_max,
height=height_max,
page=page)
with open(preview_path, "rb") as f:
data = f.read()
if len(data) == 0:
raise ValueError("preview_generator gives zero-sized image")
rst[page] = data
finally:
shutil.rmtree(cache_dir)
return rst
def get_page_nb():
jsonRequest = request.json
if 'path' not in jsonRequest:
raise ValueError('path required')
else:
source_document = request.json['path']
print('source_document: ', source_document)
cache_path = '/var/preview-cache'
manager = PreviewManager('/var/preview-cache/', create_folder=True)
get_page_nb = manager.get_page_nb(source_document)
response = app.response_class(
response=json.dumps({"pages": get_page_nb}),
status=200,
mimetype='application/json'
)
return response
def test_get_nb_page() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
nb_page = manager.get_page_nb(
file_path=IMAGE_FILE_PATH
) # FIXME must add parameter force=True/False in the API
assert nb_page == 1
def get_one(self, file_id, revision_id=None):
file_id = int(file_id)
cache_path = CFG.get_instance().PREVIEW_CACHE_DIR
preview_manager = PreviewManager(cache_path, create_folder=True)
user = tmpl_context.current_user
workspace = tmpl_context.workspace
current_user_content = Context(CTX.CURRENT_USER,
current_user=user).toDict(user)
current_user_content.roles.sort(key=lambda role: role.workspace.name)
content_api = ContentApi(user, show_archived=True, show_deleted=True)
if revision_id:
file = content_api.get_one_from_revision(file_id, self._item_type,
workspace, revision_id)
else:
file = content_api.get_one(file_id, self._item_type, workspace)
revision_id = file.revision_id
file_path = content_api.get_one_revision_filepath(revision_id)
nb_page = 0
enable_pdf_buttons = False # type: bool
preview_urls = []
try:
nb_page = preview_manager.get_page_nb(file_path=file_path)
for page in range(int(nb_page)):
url_str = '/previews/{}/pages/{}?revision_id={}'
url = url_str.format(file_id, page, revision_id)
preview_urls.append(url)
enable_pdf_buttons = \
preview_manager.has_pdf_preview(file_path=file_path)
except PreviewGeneratorException as e:
# INFO - A.P - Silently intercepts preview exception
# As preview generation isn't mandatory, just register it
logger.debug(self,
'Preview Generator Exception: {}'.format(e.__str__))
except Exception as e:
# INFO - D.A - 2017-08-11 - Make Tracim robust to pg exceptions
# Preview generator may potentially raise any type of exception
# so we prevent user interface crashes by catching all exceptions
logger.error(
self,
'Preview Generator Generic Exception: {}'.format(e.__str__))
pdf_available = 'true' if enable_pdf_buttons else 'false' # type: str
fake_api_breadcrumb = self.get_breadcrumb(file_id)
fake_api_content = DictLikeClass(breadcrumb=fake_api_breadcrumb,
current_user=current_user_content)
fake_api = Context(CTX.FOLDER, current_user=user)\
.toDict(fake_api_content)
dictified_file = Context(self._get_one_context,
current_user=user).toDict(file, 'file')
result = DictLikeClass(result=dictified_file,
fake_api=fake_api,
nb_page=nb_page,
url=preview_urls,
pdf_available=pdf_available)
return result
def test_page_number__no_extension() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
page_number = manager.get_page_nb(file_path=HTML_FILE_PATH_NO_EXTENSION)
assert page_number == 7
def test_page_number() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
page_number = manager.get_page_nb(file_path=HTML_FILE_PATH)
assert page_number == 7
def test_get_nb_page_no_extension() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
nb_page = manager.get_page_nb(file_path=ODT_FILE_PATH_NO_EXTENSION, file_ext=ODT_FILE_EXT)
assert nb_page == 2
def test_get_nb_page() -> None:
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
nb_page = manager.get_page_nb(file_path=ODP_FILE_PATH)
assert nb_page == 5
def test_page_number():
manager = PreviewManager(cache_folder_path=CACHE_DIR, create_folder=True)
page_number = manager.get_page_nb(
file_path=IMAGE_FILE_PATH,
)
assert page_number == 1
