Upload zero

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+data/AC_33_7-1.pdf filter=lfs diff=lfs merge=lfs -text
+data/AC_33-2C.pdf filter=lfs diff=lfs merge=lfs -text
+data/CFR-2024-title14-vol1-part33.pdf filter=lfs diff=lfs merge=lfs -text
+data/CFR-2024-title14-vol1-part43.pdf filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,14 +1,14 @@
 ---
-title: Faa Agentic Ai
-emoji: 🦀
-colorFrom: purple
-colorTo: yellow
+title: FAA Engine Compliance Report Generator
+emoji: ✈️
+colorFrom: gray
+colorTo: blue
 sdk: gradio
-sdk_version: 5.32.0
+sdk_version: "4.25.0"
 app_file: app.py
 pinned: false
-license: mit
-short_description: This app uses semantic search over FAA PDF documents and the
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# ✈️ FAA Engine Anomaly Compliance Report Generator
+
+This app uses semantic search over FAA PDF documents and the Mistral-7B model (via Hugging Face Inference API) to generate FAA-compliant reports for engine anomalies.

app.py

@@ -0,0 +1,29 @@
+import streamlit as st
+from pathlib import Path
+from refactored_mistral_demo_pdfs import main
+
+# Load FAISS index once per session
+if "faiss_index" not in st.session_state:
+    from vector_search import load_and_index_pdfs
+    st.session_state["faiss_index"], _, st.session_state["chunks"] = load_and_index_pdfs("data")
+
+st.set_page_config(page_title="FAA Report Generator", layout="centered")
+st.title("✈️ FAA Engine Anomaly Report Generator")
+
+anomaly_input = st.text_input("Enter engine anomaly (e.g., 'Oil temp exceeds 110°C')")
+
+if st.button("Generate Report"):
+    if not anomaly_input:
+        st.warning("Please enter an anomaly.")
+    else:
+        st.info("Generating report... please wait ⏳")
+        main(anomaly_input, st.session_state.faiss_index, st.session_state.chunks)
+
+        safe_name = "".join([c if c.isalnum() or c in (' ', '-') else '_' for c in anomaly_input])[:50]
+        report_name = f"{safe_name} Report.md"
+        if Path(report_name).exists():
+            st.success(f"✅ Report generated: {report_name}")
+            with open(report_name, "r") as file:
+                st.markdown(file.read())
+        else:
+            st.error("❌ Report generation failed. Please try again.")

data/AC_33-2C.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2537d3f282c27c4e742422eac09ecb57889505f81da4c5401c55d541fc8e1fd0
+size 809940

data/AC_33_7-1.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f98c78050680fe9d492ccb25acf37b2c6f68475610a92e60f7dd0050260df4c4
+size 110485

data/CFR-2024-title14-vol1-part33.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eac212f346a2f2208cc35040b8a2c9808fbe628151b92bb25f05321fc7109104
+size 1078402

data/CFR-2024-title14-vol1-part43.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6a7376b6ffb3782fdcc71247ee1be711985a6c4c123678169d778939d79adde
+size 307028

refactored_mistral_demo_pdfs.py

@@ -0,0 +1,70 @@
+import os
+import re
+import requests
+from pathlib import Path
+
+def call_mistral(prompt):
+    url = "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct-v0.3"
+    token = os.environ.get("HF_TOKEN", "")
+
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Content-Type": "application/json"
+    }
+
+    data = {
+        "inputs": prompt,
+        "parameters": {
+            "temperature": 0.5,
+            "max_new_tokens": 512
+        }
+    }
+
+    response = requests.post(url, headers=headers, json=data)
+
+    if response.status_code != 200:
+        raise Exception("Hugging Face API error:", response.text)
+
+    result = response.json()
+    if isinstance(result, list) and "generated_text" in result[0]:
+        return result[0]["generated_text"]
+    else:
+        raise Exception("Unexpected API output:", result)
+
+def write_report(anomaly_description: str, plan_info: str) -> str:
+    prompt = f"""You are a compliance report assistant. Your task is to generate an FAA-compliant maintenance report.
+
+Anomaly: {anomaly_description}
+Regulatory Guidance: {plan_info}
+
+Requirements:
+- Include an FAA regulation reference (e.g., CFR 43.13)
+- Recommend actionable steps
+- Output format: Markdown
+"""
+    return call_mistral(prompt)
+
+def validate_report(report: str) -> str:
+    return "Pass" if "CFR" in report and "action" in report.lower() else "Fail"
+
+def clean_report(report: str) -> str:
+    match = re.search(r"(#|FAA Report|##)", report)
+    return report[match.start():].strip() if match else report.strip()
+
+def main(user_input, faiss_index, chunks):
+    from vector_search import query_guidance
+
+    plan_info = query_guidance(user_input, faiss_index, None, chunks)
+    plan_text = "\n\n".join(plan_info)
+
+    report = write_report(user_input, plan_text)
+    report = clean_report(report)
+
+    validation_result = validate_report(report)
+
+    if validation_result == "Pass":
+        safe_name = re.sub(r"[^\w\- ]", "_", user_input)[:50]
+        report_name = f"{safe_name} Report.md"
+        Path(report_name).write_text(report)
+    else:
+        print("❌ Validation failed.")

requirements.txt

@@ -0,0 +1,6 @@
+streamlit
+huggingface_hub
+requests
+pypdf
+faiss-cpu
+sentence-transformers

vector_search.py

@@ -0,0 +1,77 @@
+import os
+import json
+from pathlib import Path
+from typing import List, Tuple
+
+import faiss
+import numpy as np
+from pypdf import PdfReader
+from sentence_transformers import SentenceTransformer
+
+
+# Paths
+DATA_DIR = Path("data")
+INDEX_FILE = DATA_DIR / "faa_index.faiss"
+CHUNKS_FILE = DATA_DIR / "faa_chunks.json"
+
+# Model (load once)
+MODEL = SentenceTransformer("all-MiniLM-L6-v2")
+
+
+def extract_text_from_pdf(pdf_path: str) -> str:
+    reader = PdfReader(pdf_path)
+    return "\n".join([page.extract_text() or "" for page in reader.pages])
+
+
+def chunk_text(text: str, chunk_size: int = 500) -> List[str]:
+    words = text.split()
+    return [' '.join(words[i:i + chunk_size]) for i in range(0, len(words), chunk_size)]
+
+
+def embed_chunks(chunks: List[str]) -> np.ndarray:
+    return MODEL.encode(chunks, show_progress_bar=True)
+
+
+def save_faiss_index(index: faiss.IndexFlatL2, embeddings: np.ndarray, chunks: List[str]):
+    faiss.write_index(index, str(INDEX_FILE))
+    with open(CHUNKS_FILE, "w", encoding="utf-8") as f:
+        json.dump(chunks, f)
+    print("💾 Saved FAISS index and chunk metadata.")
+
+
+def load_faiss_index() -> Tuple[faiss.IndexFlatL2, np.ndarray, List[str]]:
+    index = faiss.read_index(str(INDEX_FILE))
+    with open(CHUNKS_FILE, "r", encoding="utf-8") as f:
+        chunks = json.load(f)
+    print("🔁 Loaded FAISS index and chunks.")
+    return index, None, chunks  # `None` because we don't reuse original embeddings
+
+
+def build_faiss_index(chunks: List[str]) -> Tuple[faiss.IndexFlatL2, np.ndarray, List[str]]:
+    embeddings = embed_chunks(chunks)
+    index = faiss.IndexFlatL2(embeddings.shape[1])
+    index.add(embeddings)
+    return index, embeddings, chunks
+
+
+def load_and_index_pdfs(pdf_folder: str = "data") -> Tuple[faiss.IndexFlatL2, np.ndarray, List[str]]:
+    if INDEX_FILE.exists() and CHUNKS_FILE.exists():
+        return load_faiss_index()
+
+    all_chunks = []
+    pdf_folder = Path(pdf_folder)
+    for pdf_path in pdf_folder.glob("*.pdf"):
+        print(f"📄 Processing {pdf_path.name}")
+        raw_text = extract_text_from_pdf(str(pdf_path))
+        chunks = chunk_text(raw_text)
+        all_chunks.extend(chunks)
+
+    index, embeddings, chunks = build_faiss_index(all_chunks)
+    save_faiss_index(index, embeddings, chunks)
+    return index, embeddings, chunks
+
+
+def query_guidance(query: str, index: faiss.IndexFlatL2, _, chunks: List[str], top_k: int = 3) -> List[str]:
+    query_vec = MODEL.encode([query])
+    distances, indices = index.search(query_vec, top_k)
+    return [chunks[i] for i in indices[0]]