seanpedrickcase
commited on
Commit
•
230fcc3
1
Parent(s):
bc4bdbd
Works correctly with images again
Browse files- tools/file_conversion.py +7 -1
- tools/file_redaction.py +8 -1
tools/file_conversion.py
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
from pdf2image import convert_from_path, pdfinfo_from_path
|
2 |
-
from tools.helper_functions import get_file_path_end, output_folder
|
3 |
from PIL import Image
|
4 |
import os
|
5 |
from gradio import Progress
|
@@ -159,6 +159,12 @@ def prepare_image_or_text_pdf(
|
|
159 |
|
160 |
print("file_path:", file_path)
|
161 |
|
|
|
|
|
|
|
|
|
|
|
|
|
162 |
#if file_path:
|
163 |
# file_path_without_ext = get_file_path_end(file_path)
|
164 |
if not file_path:
|
|
|
1 |
from pdf2image import convert_from_path, pdfinfo_from_path
|
2 |
+
from tools.helper_functions import get_file_path_end, output_folder, detect_file_type
|
3 |
from PIL import Image
|
4 |
import os
|
5 |
from gradio import Progress
|
|
|
159 |
|
160 |
print("file_path:", file_path)
|
161 |
|
162 |
+
file_extension = os.path.splitext(file_path)[1].lower()
|
163 |
+
|
164 |
+
# Check if the file is an image type
|
165 |
+
if file_extension in ['.jpg', '.jpeg', '.png']:
|
166 |
+
in_redact_method = "Image analysis"
|
167 |
+
|
168 |
#if file_path:
|
169 |
# file_path_without_ext = get_file_path_end(file_path)
|
170 |
if not file_path:
|
tools/file_redaction.py
CHANGED
@@ -190,7 +190,12 @@ def redact_image_pdf(file_path:str, image_paths:List[str], language:str, chosen_
|
|
190 |
|
191 |
image_paths = process_file(file_path)
|
192 |
|
193 |
-
|
|
|
|
|
|
|
|
|
|
|
194 |
|
195 |
#print("Image paths:", image_paths)
|
196 |
number_of_pages = len(image_paths[0])
|
@@ -216,6 +221,8 @@ def redact_image_pdf(file_path:str, image_paths:List[str], language:str, chosen_
|
|
216 |
|
217 |
#for i in progress.tqdm(range(0,number_of_pages), total=number_of_pages, unit="pages", desc="Redacting pages"):
|
218 |
|
|
|
|
|
219 |
for n in range(0, number_of_pages):
|
220 |
|
221 |
try:
|
|
|
190 |
|
191 |
image_paths = process_file(file_path)
|
192 |
|
193 |
+
print("image_paths:", image_paths)
|
194 |
+
|
195 |
+
|
196 |
+
if not isinstance(image_paths, list):
|
197 |
+
print("Converting image_paths to list")
|
198 |
+
image_paths = [image_paths]
|
199 |
|
200 |
#print("Image paths:", image_paths)
|
201 |
number_of_pages = len(image_paths[0])
|
|
|
221 |
|
222 |
#for i in progress.tqdm(range(0,number_of_pages), total=number_of_pages, unit="pages", desc="Redacting pages"):
|
223 |
|
224 |
+
images = []
|
225 |
+
|
226 |
for n in range(0, number_of_pages):
|
227 |
|
228 |
try:
|