Update app.py

app.py

@@ -1,4 +1,4 @@
-from flask import Flask, request, render_template, send_file, redirect, url_for
+from flask import Flask, request, render_template, send_file, redirect, url_for, jsonify
 import os
 import re
 import uuid
@@ -42,12 +42,12 @@ def clear_uploads_folder():
 clear_uploads_folder()
 print("✅ Uploads folder cleared.")
 
-# runtime cache keyed by search‑id → (paragraphs, embeddings, faiss‑index)
+# runtime cache keyed by search-id → (paragraphs, embeddings, faiss-index)
 index_data = {}
 
 # ── helpers ────────────────────────────────────────────────────────────────────
 def get_paths(sid: str):
-    """Return per‑search folders & files, creating them if needed."""
+    """Return per-search folders & files, creating them if needed."""
     up_folder   = os.path.join(BASE_UPLOADS, sid)
     res_folder  = os.path.join(BASE_RESULTS, sid)
     os.makedirs(up_folder,  exist_ok=True)
@@ -70,7 +70,7 @@ def extract_text(file_path):
     return ""
 
 def rebuild_merged_and_index(sid: str):
-    """Re‑embed everything for *this* search id."""
+    """Re-embed everything for *this* search id."""
     up_folder, _, merged_file, _ = get_paths(sid)
 
     merged_text = ""
@@ -107,7 +107,7 @@ def index():
     up_folder, _, _, _ = get_paths(sid)
     paragraphs, embeddings, index_faiss = index_data.get(sid, ([], None, None))
 
-    uploaded_filenames = sorted(os.listdir(up_folder))  # ✅ added this line
+    uploaded_filenames = sorted(os.listdir(up_folder))
 
     results = []
     query = ""
@@ -127,12 +127,14 @@ def index():
                 q_embed = q_embed[np.newaxis, :]
             faiss.normalize_L2(q_embed)
             D, I = index_faiss.search(q_embed, k=min(k, len(paragraphs)))
-            results = [paragraphs[i] for i in I[0]]
+
+            # Keep both the text and the FAISS paragraph index so the UI can fetch context.
+            results = [{"idx": int(i), "text": paragraphs[i]} for i in I[0]]
 
             _, res_folder, _, result_file = get_paths(sid)
             with open(result_file, "w", encoding='utf-8') as f:
-                for para in results:
-                    f.write(para + "\n\n")
+                for r in results:
+                    f.write(r["text"] + "\n\n")
 
     return render_template(
         "index.html",
@@ -140,10 +142,9 @@ def index():
         query=query,
         topk=k,
         sid=sid,
-        uploaded_filenames=uploaded_filenames  # ✅ pass to template
+        uploaded_filenames=uploaded_filenames
     )
 
-
 @app.route("/upload", methods=["POST"])
 def upload_file():
     sid = request.args.get("sid")
@@ -181,7 +182,6 @@ def download_merged():
         return ("Nothing to download", 404)
     return send_file(merged_file, as_attachment=True)
 
-
 @app.route("/reset")
 def reset():
     sid = request.args.get("sid")
@@ -201,12 +201,48 @@ def reset():
 def ping():
     return "pong", 200
 
+@app.route("/api/context")
+def api_context():
+    """
+    Return an excerpt of the merged paragraphs centered on the requested paragraph index.
+    Query params: sid, idx (int), window (int, optional, default 3) – number of surrounding paras on each side.
+    """
+    sid = request.args.get("sid")
+    try:
+        idx = int(request.args.get("idx", "-1"))
+    except (TypeError, ValueError):
+        return jsonify(error="Bad idx"), 400
+
+    try:
+        window = int(request.args.get("window", "3"))
+    except (TypeError, ValueError):
+        window = 3
+
+    if not sid or idx < 0:
+        return jsonify(error="Missing sid or idx"), 400
+
+    paragraphs, _, _ = index_data.get(sid, (None, None, None))
+    if paragraphs is None:
+        return jsonify(error="No index for this sid. Upload files first."), 404
+    if idx >= len(paragraphs):
+        return jsonify(error="idx out of range"), 400
+
+    start = max(0, idx - window)
+    end = min(len(paragraphs), idx + window + 1)
+    context_paras = paragraphs[start:end]
+    center_local = idx - start  # where the highlighted paragraph sits in that slice
+
+    return jsonify(
+        paras=context_paras,
+        center=center_local,
+        start=start,
+        end=end,
+        total=len(paragraphs)
+    )
+
 #if __name__ == "__main__":
 #    from waitress import serve
-#    # Use threads to approximate “workers” on Windows (Waitress is single‑process).
 #    serve(app, host="0.0.0.0", port=9001, threads=4)
 
-
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=7860)
-