Update searchschool.py

searchschool.py

@@ -3,6 +3,7 @@ import os
 import pandas as pd
 from huggingface_hub import hf_hub_download
 from rapidfuzz import process, fuzz
+from web_search import tavily_search_codes
 
 # ====================================================
 # CONFIG: columns, states, HF dataset
@@ -36,6 +37,130 @@ except Exception:
     normalize_with_patterns_dynamic = None
 
 
+def on_search_web(
+    school_name: str,
+    state_name: str,
+    district: str = None,
+    block: str = None
+):
+    """
+    1. Performs Tavily search → returns list of valid UDISE codes.
+    2. Looks up these UDISE codes in our HF Schools dataset using
+       get_school_rows_by_udise().
+    3. Converts results into the standard DataFrame your Gradio app expects.
+
+    Returns:
+        pandas.DataFrame with columns:
+        School_Name, State, District, Block, UDISE_Code, Score
+    """
+
+    # Step 1: Tavily → list of UDISE codes
+    udise_list = tavily_search_codes(
+        school_name=school_name,
+        state_name=state_name,
+        district=district,
+        api_key=None,               # use HuggingFace secret instead
+        enforce_state_prefix=True
+    )
+
+    if not udise_list:
+        # Always return an empty DF with correct schema
+        return pd.DataFrame(
+            columns=["School_Name", "State", "District", "Block", "UDISE_Code"]
+        )
+
+    # Step 2: HF dataset lookup
+    rows = get_school_rows_by_udise(state_name, udise_list, try_global=True)
+
+    # Step 3: Convert list → DataFrame
+    df = pd.DataFrame(rows)
+
+    # Make sure all expected columns exist
+    expected = ["School_Name", "State", "District", "Block", "UDISE_Code"]
+    for col in expected:
+        if col not in df.columns:
+            df[col] = None   # keep schema consistent
+
+    # Reorder to canonical format
+    df = df[expected]
+
+    # Score is not applicable for web search → keep None
+    return df
+
+
+def get_school_rows_by_udise(state_name: str, udise_codes: list[str], try_global: bool = True) -> list:
+    """
+    Very simplified UDISE → school rows lookup.
+    Returns list of dicts:
+        School_Name, State, District, Block, UDISE_Code
+    """
+
+    if not udise_codes:
+        return []
+
+    udise_codes = list({str(u) for u in udise_codes})  # unique + cast to str
+
+    results = []
+
+    # --- Normalize state key ---
+    state_key = None
+    if state_name:
+        upper = state_name.strip().upper()
+        for k in STATE_HF_FILES.keys():
+            if k.upper() == upper:
+                state_key = k
+                break
+
+    # --- Helper: read CSV safely ---
+    def load_csv(filename):
+        try:
+            path = hf_hub_download(
+                repo_id=HF_SCHOOLS_DATASET,
+                repo_type="dataset",
+                filename=filename
+            )
+            return pd.read_csv(path, dtype=str).fillna("")
+        except Exception:
+            return pd.DataFrame()
+
+    # --- Helper: extract rows for given DF ---
+    def extract_rows(df, state_label):
+        if df.empty or MASTER_UDISE_COL not in df.columns:
+            return []
+        matched = df[df[MASTER_UDISE_COL].isin(udise_codes)]
+        if matched.empty:
+            return []
+        rows = []
+        for _, r in matched.iterrows():
+            rows.append({
+                "School_Name": r.get(MASTER_SCHOOL_COL, ""),
+                "State": r.get(MASTER_STATE_COL, state_label),
+                "District": r.get(MASTER_DISTRICT_COL, ""),
+                "Block": r.get(MASTER_BLOCK_COL, ""),
+                "UDISE_Code": r.get(MASTER_UDISE_COL, "")
+            })
+        return rows
+
+    # --- 1) Try requested state first ---
+    if state_key:
+        fname = STATE_HF_FILES[state_key]
+        df_state = load_csv(fname)
+        rows = extract_rows(df_state, state_label=state_key)
+        if rows:
+            return rows
+
+    # --- 2) Try all states (global fallback) ---
+    if try_global:
+        for sk, fname in STATE_HF_FILES.items():
+            df = load_csv(fname)
+            rows = extract_rows(df, state_label=sk)
+            if rows:
+                results.extend(rows)
+
+    return results
+
+
+
 def load_master_for_state(state_key: str | None):
     """
     Load the master CSV for a state from Hugging Face Hub (dataset repo),