def extract_metadata(report):
report_path = path_for(report, report['file_type'])
file_type_lower = report['file_type'].lower()
if file_type_lower == "pdf":
real_report_path = os.path.abspath(
os.path.expandvars(os.path.join(utils.data_dir(), report_path)))
if utils.check_pdf_decryption(real_report_path):
real_decrypted_path = real_report_path[:-4] + ".decrypted.pdf"
decrypted_path = report_path[:-4] + ".decrypted.pdf"
if os.path.isfile(real_decrypted_path) or utils.decrypt_pdf(
real_report_path, real_decrypted_path):
metadata = utils.metadata_from_pdf(decrypted_path)
else:
metadata = None
else:
metadata = utils.metadata_from_pdf(report_path)
if metadata:
report['pdf'] = metadata
return metadata
elif file_type_lower == "doc":
metadata = utils.metadata_from_doc(report_path)
if metadata:
report['doc'] = metadata
return metadata
elif file_type_lower == "docx":
metadata = utils.metadata_from_docx(report_path)
if metadata:
report['docx'] = metadata
return metadata
elif file_type_lower in FILE_EXTENSIONS_HTML:
return None
else:
logging.warn("Unknown file type, don't know how to extract metadata!")
return None
def extract_metadata(report):
report_path = path_for(report, report['file_type'])
file_type_lower = report['file_type'].lower()
if file_type_lower == "pdf":
real_report_path = os.path.abspath(os.path.expandvars(os.path.join(utils.data_dir(), report_path)))
if utils.check_pdf_decryption(real_report_path):
real_decrypted_path = real_report_path[:-4] + ".decrypted.pdf"
decrypted_path = report_path[:-4] + ".decrypted.pdf"
if os.path.isfile(real_decrypted_path) or utils.decrypt_pdf(real_report_path, real_decrypted_path):
metadata = utils.metadata_from_pdf(decrypted_path)
else:
metadata = None
else:
metadata = utils.metadata_from_pdf(report_path)
if metadata:
report['pdf'] = metadata
return metadata
elif file_type_lower == "doc":
metadata = utils.metadata_from_doc(report_path)
if metadata:
report['doc'] = metadata
return metadata
elif file_type_lower == "docx":
metadata = utils.metadata_from_docx(report_path)
if metadata:
report['docx'] = metadata
return metadata
elif file_type_lower in FILE_EXTENSIONS_HTML:
return None
else:
logging.warn("Unknown file type, don't know how to extract metadata!")
return None
def extract_metadata(report):
report_path = path_for(report, report["file_type"])
real_report_path = os.path.join(utils.data_dir(), report_path)
file_type_lower = report["file_type"].lower()
if file_type_lower == "pdf":
metadata = utils.metadata_from_pdf(report_path)
if metadata:
report["pdf"] = metadata
return metadata
elif file_type_lower == "htm" or file_type_lower == "html":
return None
else:
logging.warn("Unknown file type, don't know how to extract metadata!")
return None
def extract_metadata(report):
report_path = path_for(report, report['file_type'])
real_report_path = os.path.join(utils.data_dir(), report_path)
file_type_lower = report['file_type'].lower()
if file_type_lower == "pdf":
metadata = utils.metadata_from_pdf(report_path)
if metadata:
report['pdf'] = metadata
return metadata
elif file_type_lower in FILE_EXTENSIONS_HTML:
return None
else:
logging.warn("Unknown file type, don't know how to extract metadata!")
return None
def extract_metadata(report):
report_path = path_for(report, report['file_type'])
file_type_lower = report['file_type'].lower()
if file_type_lower == "pdf":
metadata = utils.metadata_from_pdf(report_path)
if metadata:
report['pdf'] = metadata
return metadata
elif file_type_lower == "doc":
metadata = utils.metadata_from_doc(report_path)
if metadata:
report['doc'] = metadata
return metadata
elif file_type_lower in FILE_EXTENSIONS_HTML:
return None
else:
logging.warn("Unknown file type, don't know how to extract metadata!")
return None