Update app.py

app.py

@@ -30,13 +30,31 @@ print("Version ---- 2")
 app = FastAPI()
 
 
-app = FastAPI()
-
 pretrained_path = "aubmindlab/bert-base-arabertv2"  # must match training
 tokenizer = AutoTokenizer.from_pretrained(pretrained_path)
 encoder = AutoModel.from_pretrained(pretrained_path).eval()
 
 
+def download_file_from_hf(repo_id, filename):
+    target_dir = os.path.expanduser("~/.sinatools/Wj27012000.tar")
+    os.makedirs(target_dir, exist_ok=True)
+
+    file_path = hf_hub_download(
+        repo_id=repo_id,
+        filename=filename,
+        local_dir=target_dir,
+        local_dir_use_symlinks=False
+    )
+
+    return file_path
+
+download_file_from_hf("SinaLab/Nested-v1","args.json")
+download_file_from_hf("SinaLab/Nested-v1","tag_vocab.pkl")
+
+snapshot_download(repo_id="SinaLab/Nested", allow_patterns="checkpoints/")
+
+
+
 checkpoint_path = snapshot_download(repo_id="SinaLab/Nested", allow_patterns="checkpoints/")
 
 args_path = hf_hub_download(
@@ -188,6 +206,7 @@ def NER(sentence, mode):
             return json_short
 
 BASE_DIR = os.path.expanduser("~/.sinatools")
+NER_DIR = os.path.join(BASE_DIR, "Wj27012000.tar")
 
 # Paths expected by sinatools
 RELATION_MODEL_DIR = os.path.join(BASE_DIR, "relation_model")
@@ -205,6 +224,21 @@ if not os.path.exists(RELATION_MODEL_DIR) or not os.listdir(RELATION_MODEL_DIR):
     )
 
 
+if not os.path.exists(NER_DIR):
+    os.makedirs(NER_DIR, exist_ok=True)
+
+    nested_repo_path = snapshot_download(
+        repo_id="SinaLab/Nested"
+    )
+
+    # Copy tag_vocab.pkl to expected location
+    src_vocab = os.path.join(nested_repo_path, "Nested", "utils", "tag_vocab.pkl")
+    dst_vocab = os.path.join(NER_DIR, "tag_vocab.pkl")
+
+    if os.path.exists(src_vocab):
+        shutil.copy(src_vocab, dst_vocab)
+
+
 from sinatools.relations.relation_extractor import relation_extraction
 from sinatools.relations.event_relation_extractor import event_argument_relation_extraction