def txt_url(self):
result = PagePath(document_ep=self.document.path,
page_num=self.number,
page_count=self.page_count)
return result.txt_url()
def test_txt_url(self):
"""
Without any arguments
page_ep.url() returns page_ep.txt_url()
"""
doc_ep = DocumentPath(user_id=1, document_id=3, file_name="x.pdf")
page_ep = PagePath(document_ep=doc_ep,
page_num=1,
step=Step(1),
page_count=3)
self.assertEqual(page_ep.url(), page_ep.txt_url())
def apply_automates(document_id, page_num):
logger.debug("apply_automates: Begin.")
try:
document = Document.objects.get(id=document_id)
except Document.DoesNotExist:
logger.error(f"Provided document_id={document_id}, does not exists")
return
# use text files from the original version of the document
doc_path = DocumentPath.copy_from(
document.path,
version=0
)
page_count = get_pagecount(
default_storage.abspath(doc_path.url())
)
page_path = PagePath(
document_path=doc_path,
page_num=page_num,
page_count=page_count,
step=Step(),
)
user = document.user
text_path = default_storage.abspath(page_path.txt_url())
text = ""
with open(text_path, "r") as f:
text = f.read()
automates = Automate.objects.filter(user=user)
# are there automates for the user?
if automates.count() == 0:
logger.debug(
f"No automates for user {user}. Quit."
)
return
# check all automates for given user (the owner of the document)
matched = []
for automate in automates:
if automate.is_a_match(text):
logger.debug(f"Automate {automate} matched document={document}")
plugin_klass = get_plugin_by_module_name(
automate.plugin_name
)
plugin = plugin_klass() if plugin_klass else None
automate.apply(
document=document,
page_num=page_num,
hocr=text,
# Notice () - plugin passed is instance of the class
plugin=plugin
)
matched.append(automate)
else:
logger.debug(
f"No match for automate={automate}"
f" doc_id={document_id} page_num={page_num}"
)
message = ""
message = _(
"%(count)s of %(total)s Automate(s) matched. ") % {
'count': len(matched),
'total': automates.count()
}
if len(matched) > 0:
message += _("List of matched Automates: %(matched_automates)s") % {
'matched_automates': matched
}
automates_matching.send(
sender="papermerge.core.automate",
user_id=document.user.id,
document_id=document_id,
level=logging.INFO,
message=message,
page_num=page_num,
text=text
)