From ac0eb1e5f62b284950555cb3044784d3851c1b5d Mon Sep 17 00:00:00 2001
From: jon <janvanwan@gmail.com>
Date: Thu, 23 Jul 2026 13:48:32 -0500
Subject: [PATCH] feat(rag): PDF/docx ingest + chat citations

- Upload endpoint (base64 JSON, no multipart dep): extracts text from
  pdf/docx/txt/md via pypdf + python-docx, then runs the existing
  chunk/embed pipeline. Documents page uploads files straight through.
- Citations: the chat stream emits an SSE `sources` event listing the
  documents that fed the answer; the UI shows them as chips under the reply.
- Deps: pypdf, python-docx (both pure-Python, Windows-safe).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 interface/web/src/Chatbot.jsx   | 22 ++++++++++++++++
 interface/web/src/Documents.jsx | 40 +++++++++++++++++++++-------
 requirements-base.txt           |  4 +++
 requirements-windows.txt        |  4 +++
 synapse/main.py                 | 46 +++++++++++++++++++++++++++++++++
 tests/test_documents.py         | 16 ++++++++++++
 6 files changed, 123 insertions(+), 9 deletions(-)

diff --git a/interface/web/src/Chatbot.jsx b/interface/web/src/Chatbot.jsx
index ceb4070..b332753 100644
--- a/interface/web/src/Chatbot.jsx
+++ b/interface/web/src/Chatbot.jsx
@@ -215,6 +215,18 @@ export function Chatbot({ conversationId, setConversationId, onConversationChang
               pendingEventType = null;
               continue;
             }
+            if (pendingEventType === "sources") {
+              try {
+                const src = JSON.parse(payload).sources;
+                setMessages(prev => {
+                  const updated = [...prev];
+                  updated[assistantIndex] = { ...updated[assistantIndex], sources: src };
+                  return updated;
+                });
+              } catch { /* ignore */ }
+              pendingEventType = null;
+              continue;
+            }
             if (pendingEventType === "done") {
               // Answer is complete; re-enable input while the backend finishes
               // slow post-processing (title, memory) on the still-open stream.
@@ -526,6 +538,16 @@ export function Chatbot({ conversationId, setConversationId, onConversationChang
                       : <span style={{ color: "#aaa" }}>Thinking...</span>
                   }
                 </div>
+                {msg.sources && msg.sources.length > 0 && (
+                  <div style={{ marginTop: "0.35rem", display: "flex", flexWrap: "wrap", gap: "0.35rem", alignItems: "center" }}>
+                    <span style={{ fontSize: "0.7rem", color: "#777" }}>Sources:</span>
+                    {msg.sources.map((s, i) => (
+                      <span key={i} style={{ fontSize: "0.72rem", color: "#8ab4ff", background: "#1a2433", border: "1px solid #2a3a52", borderRadius: "6px", padding: "0.1rem 0.45rem" }}>
+                        📄 {s}
+                      </span>
+                    ))}
+                  </div>
+                )}
                 {msg.content && editingIdx !== idx && (
                   <div style={{ display: "flex", gap: "0.25rem", marginTop: "0.25rem" }}>
                     <button
diff --git a/interface/web/src/Documents.jsx b/interface/web/src/Documents.jsx
index 91092ed..5100a2b 100644
--- a/interface/web/src/Documents.jsx
+++ b/interface/web/src/Documents.jsx
@@ -27,16 +27,37 @@ export function Documents() {
 
   useEffect(() => { load(); }, []);
 
+  // Files (pdf/docx/txt/md) upload straight to the server, which extracts the
+  // text. Base64 in JSON — no multipart dependency.
   const onFile = (e) => {
     const file = e.target.files?.[0];
-    if (!file) return;
-    const reader = new FileReader();
-    reader.onload = () => {
-      setContent(String(reader.result || ""));
-      if (!title.trim()) setTitle(file.name.replace(/\.[^.]+$/, ""));
-    };
-    reader.readAsText(file);
     e.target.value = ""; // allow re-selecting the same file
+    if (!file) return;
+    setBusy(true);
+    setMessage(`Reading ${file.name}…`);
+    const reader = new FileReader();
+    reader.onload = async () => {
+      const b64 = String(reader.result || "").split(",")[1];
+      try {
+        const r = await fetch(`${API_BASE}/documents/upload`, {
+          method: "POST",
+          headers: { "Content-Type": "application/json" },
+          body: JSON.stringify({ filename: file.name, data: b64 }),
+        });
+        if (r.ok) {
+          const d = await r.json();
+          setMessage(`Indexed "${d.title}" (${d.chunks} chunk${d.chunks === 1 ? "" : "s"}).`);
+          load();
+        } else {
+          setMessage((await r.json().catch(() => ({}))).detail || "Failed to index file.");
+        }
+      } catch (err) {
+        setMessage(`Error: ${err.message}`);
+      } finally {
+        setBusy(false);
+      }
+    };
+    reader.readAsDataURL(file);
   };
 
   const addDoc = async () => {
@@ -88,10 +109,11 @@ export function Documents() {
       <div style={{ display: "flex", flexDirection: "column", gap: "0.7rem", marginBottom: "1.5rem" }}>
         <input type="text" placeholder="Title" value={title}
           onChange={e => setTitle(e.target.value)} style={input} />
-        <textarea rows={8} placeholder="Paste text here, or load a .txt/.md file below"
+        <textarea rows={8} placeholder="Paste text here, or upload a .pdf/.docx/.txt/.md file below"
           value={content} onChange={e => setContent(e.target.value)} style={input} />
         <div style={{ display: "flex", gap: "0.7rem", alignItems: "center" }}>
-          <input type="file" accept=".txt,.md,.markdown,text/*" onChange={onFile}
+          <input type="file" accept=".pdf,.docx,.txt,.md,.markdown,text/*" onChange={onFile} disabled={busy}
+            title="Upload a file — text extracted server-side"
             style={{ color: "#aaa", flex: 1 }} />
           <button onClick={addDoc} disabled={busy}
             style={{ padding: "0.9rem 1.5rem", background: busy ? "#555" : "#007acc", color: "#fff", border: "none", borderRadius: "8px", cursor: busy ? "default" : "pointer" }}>
diff --git a/requirements-base.txt b/requirements-base.txt
index bac09ff..d01ff2c 100644
--- a/requirements-base.txt
+++ b/requirements-base.txt
@@ -37,6 +37,10 @@ rich
 python-dotenv
 PyYAML
 
+# Document ingest (RAG): pure-Python text extraction, no native deps
+pypdf
+python-docx
+
 # Documentation Support
 markdown-it-py
 MarkupSafe
diff --git a/requirements-windows.txt b/requirements-windows.txt
index fc54690..6417f53 100755
--- a/requirements-windows.txt
+++ b/requirements-windows.txt
@@ -9,6 +9,10 @@ PyYAML
 python-dotenv
 psutil
 
+# Document ingest (RAG): pure-Python text extraction, no native deps
+pypdf
+python-docx
+
 # Native desktop window for the UI (WebView2 on Windows; pulls pythonnet).
 # Used by bin/nexus_window.py, launched from launch_nexus.ps1.
 pywebview
diff --git a/synapse/main.py b/synapse/main.py
index 60747b1..bc35657 100644
--- a/synapse/main.py
+++ b/synapse/main.py
@@ -287,10 +287,12 @@ async def chat_stream_endpoint(payload: Dict[str, Any]):
 
         # Retrieve relevant uploaded documents (RAG) and inject the top chunks.
         doc_hits = await store.search_documents(message, get_ollama_manager().embed, limit=3)
+        doc_titles: list = []
         if doc_hits:
             doc_block = "\n\n".join(f"[{d['title']}]\n{d['text']}" for d in doc_hits)
             separator = "\n\n---\nRelevant documents (cite as source material):\n\n"
             system_prompt = (system_prompt + separator + doc_block) if system_prompt else doc_block
+            doc_titles = list(dict.fromkeys(d["title"] for d in doc_hits))  # unique, order-preserving
 
         # Fetch web search results for time-sensitive queries
         search_results = ""
@@ -373,6 +375,10 @@ async def chat_stream_endpoint(payload: Dict[str, Any]):
             meta: dict = {}
             final_model = model
 
+            # Tell the client which documents fed this answer (RAG citations).
+            if doc_titles:
+                yield f"event: sources\ndata: {_json.dumps({'sources': doc_titles})}\n\n"
+
             # ── Phase 1: stream primary model response ────────────────────
             try:
                 async for chunk in stream_chat_response(
@@ -953,6 +959,46 @@ async def add_document(payload: Dict[str, Any] = Body(...)):
     return result
 
 
+def _extract_text(filename: str, data: bytes) -> str:
+    """Pull plain text from an uploaded file by extension. PDF/DOCX use
+    pure-Python parsers; anything else is decoded as UTF-8."""
+    import io
+    name = (filename or "").lower()
+    if name.endswith(".pdf"):
+        from pypdf import PdfReader
+        reader = PdfReader(io.BytesIO(data))
+        return "\n\n".join((page.extract_text() or "") for page in reader.pages)
+    if name.endswith(".docx"):
+        import docx
+        doc = docx.Document(io.BytesIO(data))
+        return "\n\n".join(p.text for p in doc.paragraphs if p.text.strip())
+    return data.decode("utf-8", errors="replace")
+
+
+@app.post("/documents/upload")
+async def upload_document(payload: Dict[str, Any] = Body(...)):
+    """Ingest a file (pdf/docx/txt/md) sent as base64. Extracts text, then runs
+    the same chunk/embed pipeline as a pasted document."""
+    import base64
+    import os as _os
+    filename = (payload.get("filename") or "").strip()
+    b64 = payload.get("data") or ""
+    if not filename or not b64:
+        raise HTTPException(status_code=400, detail="filename and data are required")
+    try:
+        raw = base64.b64decode(b64)
+    except Exception:
+        raise HTTPException(status_code=400, detail="data must be base64")
+    try:
+        text = _extract_text(filename, raw).strip()
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"could not read {filename}: {e}")
+    if not text:
+        raise HTTPException(status_code=400, detail="no extractable text in file")
+    title = _os.path.splitext(_os.path.basename(filename))[0] or filename
+    return await store.add_document(title, text, get_ollama_manager().embed)
+
+
 @app.get("/documents/{doc_id}")
 async def get_document(doc_id: str):
     chunks = store.get_document(doc_id)
diff --git a/tests/test_documents.py b/tests/test_documents.py
index a1c3a95..101071c 100644
--- a/tests/test_documents.py
+++ b/tests/test_documents.py
@@ -55,3 +55,19 @@ def test_add_list_search_delete_roundtrip():
 def test_search_empty_query_returns_nothing():
     s = _store()
     assert asyncio.run(s.search_documents("", _fake_embed)) == []
+
+
+def test_extract_text_by_type():
+    from synapse.main import _extract_text
+    # plain text / markdown -> UTF-8 decode
+    assert _extract_text("notes.md", b"# Title\n\nbody") == "# Title\n\nbody"
+    assert _extract_text("x.txt", "café".encode("utf-8")) == "café"
+    # a real (tiny) PDF built with pypdf -> text extracted back out
+    from pypdf import PdfWriter, PdfReader
+    import io
+    w = PdfWriter()
+    w.add_blank_page(width=200, height=200)
+    buf = io.BytesIO(); w.write(buf)
+    out = _extract_text("blank.pdf", buf.getvalue())
+    assert isinstance(out, str)  # blank page -> "" or whitespace, never raises
+    assert PdfReader(io.BytesIO(buf.getvalue())).pages  # sanity: it was a valid PDF