From 83371cfb1f71a09218aabc1adaf0df51521d8169 Mon Sep 17 00:00:00 2001
From: Varsha U N <varshamaddur2006@gmail.com>
Date: Tue, 14 Jan 2025 18:49:17 +0530
Subject: [PATCH 1/2] Support extracting Libre office documents
Signed-off-by: Varsha U N <varshamaddur2006@gmail.com>
---
src/extractcode/extract.py | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/src/extractcode/extract.py b/src/extractcode/extract.py
index 8c547f2..1a30ab3 100644
--- a/src/extractcode/extract.py
+++ b/src/extractcode/extract.py
@@ -230,6 +230,14 @@ def extract_files(
logger.debug('extract:walk:recurse:extraction event: %(xevent)r' % locals())
yield xevent
+def extract_libre_office_document(location, target):
+ """ Extract Libre Office documents (e.g., .ods files) as ZIP archives. """
+ if not zipfile.is_zipfile(location):
+ return
+ with zipfile.ZipFile(location, 'r') as zip_ref:
+ zip_ref.extractall(target)
+ print(f"Extracted Libre Office document from {location} to {target}")
+
def extract_file(
location,
From fd532ff2130fd5d047b9e37dcf0467bcb733b9aa Mon Sep 17 00:00:00 2001
From: Varsha U N <varshamaddur2006@gmail.com>
Date: Sun, 19 Jan 2025 08:39:11 +0530
Subject: [PATCH 2/2] Allow extraction of libre-office-docs
Signed-off-by: Varsha U N <varshamaddur2006@gmail.com>
---
src/extractcode/extract.py | 16 +++++++++-------
1 file changed, 9 insertions(+), 7 deletions(-)
diff --git a/src/extractcode/extract.py b/src/extractcode/extract.py
index 1a30ab3..5dfcf6a 100644
--- a/src/extractcode/extract.py
+++ b/src/extractcode/extract.py
@@ -9,6 +9,7 @@
import logging
import traceback
+import zipfile
from collections import namedtuple
from functools import partial
@@ -229,14 +230,15 @@ def extract_files(
if TRACE:
logger.debug('extract:walk:recurse:extraction event: %(xevent)r' % locals())
yield xevent
-
+
def extract_libre_office_document(location, target):
- """ Extract Libre Office documents (e.g., .ods files) as ZIP archives. """
- if not zipfile.is_zipfile(location):
- return
- with zipfile.ZipFile(location, 'r') as zip_ref:
- zip_ref.extractall(target)
- print(f"Extracted Libre Office document from {location} to {target}")
+ """Extract Libre Office documents (e.g., .ods files) as ZIP archives."""
+ try:
+ with zipfile.ZipFile(location, 'r') as zip_ref:
+ zip_ref.extractall(target)
+ logger.info(f"Extracted Libre Office document from {location} to {target}")
+ except zipfile.BadZipFile:
+ raise ValueError(f"File at {location} is not a valid ZIP archive.")
def extract_file(