Update app.py

app.py

@@ -21,7 +21,6 @@ REL_LABELS = ["has_symptom", "has_medicine", "no_relation"]
 REL_ID2LABEL = {i: l for i, l in enumerate(REL_LABELS)}
 REL_LABEL2ID = {l: i for i, l in enumerate(REL_LABELS)}
 
-
 def load_models():
     ner_model = AutoModelForTokenClassification.from_pretrained(
         "DanielNRU/pollen-ner",
@@ -41,14 +40,11 @@ def load_models():
     re_model.eval()
     return ner_model, ner_tokenizer, re_model, re_tokenizer
 
-
 ner_model, ner_tokenizer, re_model, re_tokenizer = load_models()
 
-
 def split_sentences(text):
     return [s.strip() for s in re.split(r"[.!?]", text) if s.strip()]
 
-
 def predict_entities(text):
     inputs = ner_tokenizer(text, return_tensors="pt", return_offsets_mapping=True, truncation=True, max_length=512)
     offset_mapping = inputs.pop("offset_mapping")[0]
@@ -71,22 +67,28 @@ def predict_entities(text):
         else:
             if current:
                 entities.append(current)
-                current = None
+            current = None
     if current:
         entities.append(current)
     return entities
 
-
 def insert_entity_markers(text, ent1, ent2):
     if ent1["start"] < ent2["start"]:
         first, second = ent1, ent2
     else:
         first, second = ent2, ent1
-    t = (text[:second["start"]] + f"[{second['label']}]" + text[second["start"]:second["end"]] + f"[/{second['label']}]" + text[second["end"]:])
-    t = (t[:first["start"]] + f"[{first['label']}]" + t[first["start"]:first["end"]] + f"[/{first['label']}]" + t[first["end"]:])
+    t = (text[:second["start"]]
+         + f"[{second['label']}]"
+         + text[second["start"]:second["end"]]
+         + f"[/{second['label']}]"
+         + text[second["end"]:])
+    t = (t[:first["start"]]
+         + f"[{first['label']}]"
+         + t[first["start"]:first["end"]]
+         + f"[/{first['label']}]"
+         + t[first["end"]:])
     return t
 
-
 def predict_relations(text, entities):
     relations = []
     for sent in split_sentences(text):
@@ -108,7 +110,6 @@ def predict_relations(text, entities):
                         relations.append({"head": ent1, "tail": ent2, "relation": rel})
     return relations
 
-
 def analyze_text(text):
     if not text or not text.strip():
         return [], "", "", "", ""
@@ -127,7 +128,6 @@ def analyze_text(text):
     symptoms = [f"{r['head']['text']} {r['tail']['text']}" for r in relations if r["relation"] == "has_symptom" and r["tail"]["label"] == "SYMPTOM"]
     return highlights, ", ".join(toponyms), ", ".join(medicines), ", ".join(symptoms), ", ".join(allergens)
 
-
 EXAMPLES = [
     ["В Московской области у меня началась аллергия на пыльцу березы, потекли глаза, нос, принимаю Зиртек и Назонекс."],
     ["У ребенка в Новокузнецке чешутся глаза, уши и течет нос, врач прописал Кромогексал, Назонекс в нос."],
@@ -150,5 +150,4 @@ with gr.Blocks() as demo:
             out_allergens = gr.Textbox(label="Аллергены")
     run_btn.click(fn=analyze_text, inputs=input_text, outputs=[out_highlight, out_toponyms, out_medicines, out_symptoms, out_allergens])
 
-# ✅ ОБЯЗАТЕЛЬНО для Gradio 5.x и 6.x на Spaces
-demo.launch(ssr_mode=False)
+demo.launch()