def test_paragraphs():
from docx import Document
doc = Document(test_dir / 'test1.docx')
summary, reference_csv_details, title = create_upload_variables(doc)
summary_detail = summary['Paragraphs']
assert summary_detail[
'ok'], "Paragraphs has issues but should have no issues"
for item in summary_detail['details']:
assert item[
'style_ok'], "Paragraphs has style issues but should have no style issues"
def test_author():
from docx import Document
doc = Document(test_dir / 'test1.docx')
summary, reference_csv_details, title = create_upload_variables(doc)
summary_detail = summary['Authors']
assert summary_detail['ok'], "Author has issues but should have no issues"
assert len(summary_detail['details']), "Author is more than one paragraph"
details = summary_detail['details'][0]
assert details[
'style_ok'], "Author has style issues but should have no style issues"
assert details[
'title_style_ok'], "Author has strict style issues but should have no strict style issues"
def test_no_abstract():
from docx import Document
doc = Document(test_dir / 'correct_a4_margins.docx')
summary, reference_csv_details, title = None, None, None
try:
summary, reference_csv_details, title = create_upload_variables(doc)
assert False, 'AbstractNotFoundError not raised'
except AbstractNotFoundError as err:
assert str(
err
) == "Abstract header not found", 'AbstractNotFoundError found but not correct message'
# these should be empty because Error raised before values returned
assert summary is None, 'summary not empty'
assert reference_csv_details is None, 'reference_csv_details not empty'
def upload_from_spms(paper_name, parse_type):
base_path = 'C:\\\\Users\Hawke\\\\PycharmProjects\\\\jacow_files\\\\IPAC19 Test Files\\\\'
try:
spms_summary = ''
full_path = base_path+paper_name+'.'+parse_type
conference_path = 'C:\\\\Users\\\\Hawke\\\\PycharmProjects\\\\References_ipac19.csv'
if parse_type == 'docx':
doc = Document(full_path)
# check whether tracking on (will raise an error if it is)
check_tracking_on(doc)
# get variables
summary, authors, title = create_upload_variables(doc)
spms_summary, reference_csv_details = create_spms_variables(paper_name, authors, title, conference_path)
elif parse_type == 'tex':
doc = TexSoup(open(full_path, encoding="utf8"))
summary, authors, title = create_upload_variables_latex(doc)
spms_summary, reference_csv_details = create_spms_variables(paper_name, authors, title, conference_path)
print(spms_summary)
return spms_summary
except (PackageNotFoundError, ValueError):
message = f"Failed to open document. Is it a valid {parse_type} document?"
except TrackingOnError as err:
message = err
except OSError:
message = f"It seems the file {paper_name} is corrupted"
except PaperNotFoundError:
message = f"It seems the file {paper_name} has no corresponding entry in the SPMS ({conference_path}) " \
f"references list. Is your filename the same as your Paper name?"
except AbstractNotFoundError as err:
message = err
except Exception:
app.logger.exception("Failed to process document")
message = f"Failed to process document: {paper_name}"
print(message)
return message
def upload_common(documents, args):
admin = 'DEV_DEBUG' in os.environ and os.environ['DEV_DEBUG'] == 'True'
conferences = [
conference.short_name for conference in Conference.query.filter_by(
is_active=True).order_by(Conference.display_order.asc()).all()
]
if request.method == "POST" and documents.name in request.files:
try:
filename = documents.save(request.files[documents.name])
paper_name = os.path.splitext(filename)[0]
except UploadNotAllowed:
return render_template(
"upload.html",
error=
f"Wrong file extension. Please upload {args['extension']} files only",
admin=admin,
args=args)
full_path = documents.path(filename)
# set a default
conference_id = False # next(iter(conferences))
conference_path = ''
if 'conference_id' in request.form and request.form[
"conference_id"] in conferences:
conference_id = request.form["conference_id"]
conference_path = get_conference_path(conference_id)
try:
if args['description'] == 'Word':
doc = Document(full_path)
parse_type = 'docx'
metadata = doc.core_properties
# check whether tracking on
result = check_tracking_on(doc)
# get variables to pass to template
summary, authors, title = create_upload_variables(doc)
if conference_id:
spms_summary, reference_csv_details = \
create_spms_variables(paper_name, authors, title, conference_path, conference_id)
if spms_summary:
summary.update(spms_summary)
elif args['description'] == 'Latex':
doc = TexSoup(open(full_path, encoding="utf8"))
summary, authors, title = create_upload_variables_latex(doc)
metadata = []
if conference_id:
spms_summary, reference_csv_details = \
create_spms_variables(paper_name, authors, title, conference_path, conference_id)
if spms_summary:
summary.update(spms_summary)
save_log(filename, conference_id, 'OK', locals())
return render_template("upload.html", processed=True, **locals())
except (PackageNotFoundError, ValueError):
save_log(filename, conference_id, 'PackageNotFoundError', locals())
return render_template(
"upload.html",
filename=filename,
conferences=conferences,
error=
f"Failed to open document {filename}. Is it a valid {args['description']} document?",
admin=admin,
args=args)
except TrackingOnError as err:
save_log(filename, conference_id, 'TrackingOnError', locals())
return render_template("upload.html",
filename=filename,
conferences=conferences,
error=err,
admin=admin,
args=args)
except OSError:
save_log(filename, conference_id, 'OSError', locals())
return render_template(
"upload.html",
filename=filename,
conferences=conferences,
error=f"It seems the file {filename} is corrupted",
admin=admin,
args=args)
except PaperNotFoundError:
save_log(filename, conference_id, 'PaperNotFoundError', locals())
return render_template(
"upload.html",
processed=True,
**locals(),
error=
f"It seems the file {filename} has no corresponding entry in the SPMS ({conference_id}) references list. "
f"Is your filename the same as your Paper name?")
except AbstractNotFoundError as err:
save_log(filename, conference_id, 'AbstractNotFoundError',
locals())
return render_template("upload.html",
filename=filename,
conferences=conferences,
error=err,
admin=admin,
args=args)
except Exception:
# TODO work out why there is an OK log followed by an Exception one.
save_log(filename, conference_id, 'Exception', locals())
if app.debug:
raise
else:
app.logger.exception("Failed to process document")
return render_template(
"upload.html",
error=f"Failed to process document: {filename}",
conferences=conferences,
args=args)
finally:
os.remove(full_path)
return render_template("upload.html",
admin=admin,
args=args,
conferences=conferences)
