prthm11 commited on
Commit
0d00d37
·
verified ·
1 Parent(s): ad0a9ec

Update app_main.py

Browse files
Files changed (1) hide show
  1. app_main.py +5 -0
app_main.py CHANGED
@@ -238,6 +238,9 @@ def extract_images_from_pdf(pdf_path, output_json_path):
238
  except Exception as e:
239
  raise RuntimeError(f"❌ Error in extract_images_from_pdf: {str(e)}")
240
 
 
 
 
241
  def similarity_matching(input_json_path: str) -> str:
242
  import uuid, shutil, tempfile
243
  from langchain_experimental.open_clip.open_clip import OpenCLIPEmbeddings
@@ -318,6 +321,8 @@ def similarity_matching(input_json_path: str) -> str:
318
  # ============================== #
319
  # COMPUTE SIMILARITIES #
320
  # ============================== #
 
 
321
  similarity = np.matmul(np.array(sprite_features), np.array(img_features).T)
322
  most_similar_indices = np.argmax(similarity, axis=1)
323
 
 
238
  except Exception as e:
239
  raise RuntimeError(f"❌ Error in extract_images_from_pdf: {str(e)}")
240
 
241
+ os.environ["TRANSFORMERS_CACHE"] = "/app/cache"
242
+ os.environ["HF_HOME"] = "/app/cache"
243
+
244
  def similarity_matching(input_json_path: str) -> str:
245
  import uuid, shutil, tempfile
246
  from langchain_experimental.open_clip.open_clip import OpenCLIPEmbeddings
 
321
  # ============================== #
322
  # COMPUTE SIMILARITIES #
323
  # ============================== #
324
+ if not sprite_features or not img_features:
325
+ raise ValueError("❌ Embedding generation failed: One of the image feature lists is empty.")
326
  similarity = np.matmul(np.array(sprite_features), np.array(img_features).T)
327
  most_similar_indices = np.argmax(similarity, axis=1)
328