deploy 2025-08-09 16:48:55

.dockerignore

@@ -1 +1,5 @@
 node_modules
+.git
+__pycache__
+*.pyc
+.env

.gitattributes

@@ -40,3 +40,6 @@ figures/RAG_pipeline.png filter=lfs diff=lfs merge=lfs -text
 figures/demo.gif filter=lfs diff=lfs merge=lfs -text
 figures/demo.mp4 filter=lfs diff=lfs merge=lfs -text
 figures/grafana.png filter=lfs diff=lfs merge=lfs -text
+downloaded_pdfs/AmazingFacts.pdf filter=lfs diff=lfs merge=lfs -text
+downloaded_pdfs/contrat_2c969eb69791039801980498c3333638_Liasse_DocumentASigner_RESIDENCE[[:space:]]COUREILLES_T1-20250713-182255.929_11.pdf.pdf filter=lfs diff=lfs merge=lfs -text
+downloaded_pdfs/hop-dong-thue-nha-tro_2810165040_2011153218_0804150519.pdf filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.10-slim
+FROM python:3.10-slim AS base
 
 ENV PYTHONDONTWRITEBYTECODE=1 \
     PYTHONUNBUFFERED=1 \
@@ -7,14 +7,21 @@ ENV PYTHONDONTWRITEBYTECODE=1 \
 
 WORKDIR /app
 
-# Rest of the code
-COPY . .
+# Install build dependencies only if needed (optional)
+# RUN apt-get update && apt-get install -y --no-install-recommends build-essential && rm -rf /var/lib/apt/lists/*
+
+# Copy only requirements first (cacheable layer)
+COPY requirements.txt .
 
-# Install backend dependencies
-RUN pip install --no-cache-dir -r requirements.txt
+# Install backend dependencies (cached if requirements.txt hasn't changed)
+RUN --mount=type=cache,target=/root/.cache \
+    pip install -r requirements.txt
+
+# Now copy the rest of the application
+COPY . .
 
 # Expose backend and frontend ports
 EXPOSE 3012 8080
 
-# Start both backend (FastAPI) and serve frontend (using Python's http.server)
+# Start both backend and frontend
 CMD ["sh", "-c", "uvicorn backend.main:app --host 0.0.0.0 --port 3012 & python3 -m http.server 8080 --directory frontend"]

README.md

@@ -1,12 +1,3 @@
----
-title: Basic Docker SDK Space
-emoji: 🐳
-colorFrom: purple
-colorTo: gray
-sdk: docker
-app_port: 8000
----
-
 # Legal Contract Analyzer
 A MLOps project of an AI-powered RAG Chatbot for understanding and querying legal documents. Together with CI/CD, monitoring and visualization.
 
@@ -35,6 +26,8 @@ With the power of RAG, the answers are now more precise, the LLM experiences les
 
 [Watch the full demo video](https://youtu.be/kvJwAMWmvj0)
 
+Demo on Render: https://legalcontractanalyzer.onrender.com/
+
 ### Monitoring and Visualization with Prometheus + Grafana:
 
 ![Prometheus](./figures/prometheus.png)
@@ -124,19 +117,20 @@ docker compose build .
 docker compose up -d
 ```
 
-Access the web app frontend at: http://localhost:8080
+3. Access the web app frontend at: http://localhost:8080
 
-Acces the monitoring Grafana at: http://localhost:3000 
+Start by uploading a file then ask a question like: "Thời hạn hợp đồng kéo dài trong bao lâu?"
+
+4. Acces the monitoring Grafana at: http://localhost:3000 
 
 *Note*: Username / password for Grafana should be admin / admin, go to Dashboards and select the panel to view metrics
 
-Other ports if you're interested:
+5. Other ports if you're interested:
 
 Backend: http://localhost:3012
 
 Prometheus: http://localhost:9090
 
-
 ## API Endpoints
 
 Available endpoints for backend:
@@ -148,7 +142,7 @@ Available endpoints for backend:
 |POST /ingest             | Embeds and stores chunks
 |POST /query              | Retrieves top-K chunks for a query
 |POST /rag_chat           | Full chat with RAG streaming
-GET  /api_key            | Exposes env vars (for dev)
+|GET  /api_key            | Exposes env vars (for dev)
 
 ## Monitoring
 

backend/chroma_vector_db/chroma.sqlite3

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7503eebf53b9f9e4e18950d5a04cedfe613e13bcd2215857e79a643241795d31
-size 540672
+oid sha256:bedaaed932573c2da9f3c8ad609773f192cb3e027c9aabc8927f9567f35b288b
+size 3514368

backend/query.py

@@ -1,43 +1,67 @@
-import os
+# backend/query.py
+import numpy as np
 import chromadb
 from openai import OpenAI
 from dotenv import load_dotenv
 from backend.config import CHROMA_DB_PATH
-# ─── ENVIRONMENT ──────────────────────────────────────────────────────────────
+import os
+
 load_dotenv()
-API_KEY = os.getenv("OPENAI_API_KEY", "TRANMINHDUONGDEPTRAI")
-BASE_URL = "https://glowing-workable-arachnid.ngrok-free.app/v1"  # or ngrok URL
+API_KEY = os.getenv("OPENAI_API_KEY")
+BASE_URL = os.getenv("FASTCHAT_URL", "https://glowing-workable-arachnid.ngrok-free.app/v1")
 openai_client = OpenAI(api_key=API_KEY, base_url=BASE_URL)
 
-# ─── CHROMA SETUP ─────────────────────────────────────────────────────────────
 chroma_client = chromadb.PersistentClient(path=CHROMA_DB_PATH)
 collection = chroma_client.get_or_create_collection("legal_docs")
 
-# ─── EMBEDDING FUNCTION ───────────────────────────────────────────────────────
-def embed_query(query_text):
+def embed_texts(texts):
     resp = openai_client.embeddings.create(
         model="Qwen3-0.6B",
-        input=[query_text]
+        input=texts
     )
-    return resp.data[0].embedding
-
-# ─── TOP-K RETRIEVAL ──────────────────────────────────────────────────────────
-def query_top_k(query_text, k=5):
-    query_emb = embed_query(query_text)
-    results = collection.query(
-        query_embeddings=[query_emb],
-        n_results=k
-    )
-    # results['documents'] is a list of lists (one per query)
-    # results['distances'] is a list of lists (one per query)
-    # We'll return a list of (chunk, distance) tuples
-    docs = results['documents'][0] if results['documents'] else []
-    dists = results['distances'][0] if results['distances'] else []
-    return list(zip(docs, dists))
+    # ensure order
+    return [item.embedding for item in sorted(resp.data, key=lambda d: d.index)]
+
+def normalize(vec):
+    arr = np.array(vec, dtype=np.float32)
+    return arr / (np.linalg.norm(arr) + 1e-10)
+
+def query_top_k(query_text, k=10, rerank_top_n=5):
+    # 1) embed
+    q_emb = embed_texts([query_text])[0]
+    q_norm = normalize(q_emb)
+
+    # 2) dense retrieval (get more candidates)
+    results = collection.query(query_embeddings=[q_emb], n_results=k)
+    docs = results.get('documents', [[]])[0]
+    dists = results.get('distances', [[]])[0]
+
+    # Note: Chroma distances are lower = better. We'll compute cosine from stored embeddings if available.
+    # If you stored embeddings in collection, pull them (some Chroma versions allow include=['embeddings'])
+    # Here we fallback to converting distance -> similarity (if the metric is cosine)
+    sims = []
+    for idx, doc in enumerate(docs):
+        # try to get the stored embedding if available:
+        try:
+            emb = results['embeddings'][0][idx]
+            sim = float(np.dot(q_norm, normalize(emb)))
+        except Exception:
+            # fallback: invert distance (only approximate)
+            dist = dists[idx] if idx < len(dists) else 1.0
+            sim = 1.0 - float(dist)
+        sims.append((doc, sim))
+
+    # sort by similarity desc
+    sims.sort(key=lambda x: x[1], reverse=True)
+
+    # optional: rerank top candidates with a cross-encoder here
+
+    return sims[:rerank_top_n]  # return top rerank_top_n with similarity
+
 
 # Example usage:
 if __name__ == "__main__":
-    q = "Sẽ ra sao nếu một trong hai bên muốn chấm dứt hợp đồng trước thời hạn?"
+    q = "An interesting fact about the humming bird"
     top_chunks = query_top_k(q, k=3)
     for chunk, dist in top_chunks:
         print(f"Score: {dist:.4f}\n{chunk}\n{'-'*40}")

downloaded_pdfs/AmazingFacts.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97d6ce2512ad2c8df2ef58709e2ad00186acc7eb3dad9da3554eb92f13eac531
+size 127901

downloaded_pdfs/contrat_2c969eb69791039801980498c3333638_Liasse_DocumentASigner_RESIDENCE COUREILLES_T1-20250713-182255.929_11.pdf.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c75cb43e380be568c00346f4b56dfab7447de39a35c77bae63c25a881ba7b7aa
+size 993576

downloaded_pdfs/hop-dong-thue-nha-tro_2810165040_2011153218_0804150519.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:797d41945ea3b81f01163d86215cb7f7a0277c71f085429bc6820aa18e2f4a50
+size 490064

frontend/script.js

@@ -254,14 +254,30 @@ document.addEventListener("DOMContentLoaded", function () {
         embeddedFiles[file.name] = true;
         uploadedCount++;
         renderFileList();
-        updateStatus(`Uploaded ${file.name} to Supabase!`, "success");
+        updateStatus("Uploaded " + file.name + " to Supabase!", "success");
       } catch (err) {
-        updateStatus(`Error uploading ${file.name}: ${err.message || err}`, "error");
+        updateStatus("Error uploading " + file.name + ": " + (err.message || err), "error");
       }
     }
+    // After uploading, call backend to retrieve and ingest documents
     if (uploadedCount > 0) {
-      updateStatus(`Uploaded ${uploadedCount} file(s) to Supabase!`, "success");
-      embeddingStatus.textContent = "Uploaded";
+      disableChat()
+      updateStatus("Calling backend to retrieve documents...", "processing");
+      try {
+        const retrieveRes = await fetch("http://localhost:3012/retrieve_documents", { method: "POST" });
+        if (!retrieveRes.ok) throw new Error("Failed to retrieve documents");
+        updateStatus("Retrieving documents complete. Now ingesting...", "processing");
+        console.log("Retrieving documents complete. Now ingesting...")
+        const ingestRes = await fetch("http://localhost:3012/ingest", { method: "POST" });
+        if (!ingestRes.ok) throw new Error("Failed to ingest documents");
+        updateStatus("Uploaded and embedded " + uploadedCount + " file(s) successfully!", "success");
+        console.log("Successfully embedded the documents!")
+        embeddingStatus.textContent = "Uploaded";
+      } catch (err) {
+        updateStatus("Error embedding documents: " + (err.message || err), "error");
+        embeddingStatus.textContent = "Error";
+      }
+      enableChat()
     } else {
       updateStatus("No new files uploaded.", "warning");
       embeddingStatus.textContent = "Ready";