Update app.py

app.py

@@ -1,23 +1,12 @@
-# ======================================================
-# app.py — FastAPI + Gradio Hybrid (Court / Research Modes)
-# ======================================================
-
 import gradio as gr
-import time
-import hashlib
-import io
-import base64
-from datetime import datetime, timedelta
+import time, hashlib, io, zipfile, os, tempfile
+import xml.etree.ElementTree as ET
+from datetime import datetime
 from urllib.parse import quote_plus
-import requests
-from fastapi import FastAPI
-from fastapi.middleware.cors import CORSMiddleware
 
 from reportlab.platypus import SimpleDocTemplate, Paragraph, PageBreak
 from reportlab.lib.styles import getSampleStyleSheet
-
-from citations import bluebook_exhibit
-from foia_requests import generate_foia_request_text
+from reportlab.lib.pagesizes import LETTER
 
 # ======================================================
 # HARD FEATURE FLAGS (GOVERNANCE ENFORCED)
@@ -25,345 +14,255 @@ from foia_requests import generate_foia_request_text
 
 ENABLE_FAISS_PHASE_4 = False
 ENABLE_AI = True
-ENABLE_PDF_EXTRACTION = True
 
 # ======================================================
-# OPTIONAL PDF SUPPORT
+# FIPS MODE
 # ======================================================
 
-PDF_TEXT_AVAILABLE = False
-PDF_THUMBNAIL_AVAILABLE = False
+FIPS_140_MODE = False
 
-try:
-    from pdfminer.high_level import extract_text
-    PDF_TEXT_AVAILABLE = True
-except Exception:
-    pass
+# ======================================================
+# SESSION STATE
+# ======================================================
 
-try:
-    from pdf2image import convert_from_bytes
-    PDF_THUMBNAIL_AVAILABLE = True
-except Exception:
-    pass
+LAST_RESULTS = []
 
 # ======================================================
-# FASTAPI CORE
+# CRYPTOGRAPHIC CORE
 # ======================================================
 
-api = FastAPI(
-    title="Federal FOIA Intelligence Search",
-    version="1.0.0",
-    description="Public FOIA Reading Room Intelligence Tool"
-)
+def sha256_text(t: str):
+    return hashlib.sha256(t.encode()).hexdigest()
 
-api.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],
-    allow_methods=["*"],
-    allow_headers=["*"],
-)
+def provenance_headers(payload: str):
+    return {
+        "Tool-Version": "1.7.0",
+        "Generated-UTC": datetime.utcnow().isoformat(),
+        "Content-SHA256": sha256_text(payload),
+        "Public-Source-Only": "true",
+        "AI-Assisted": "formatting-only",
+        "FIPS-140-Mode": str(FIPS_140_MODE).lower(),
+    }
+
+def render_provenance_block(text: str):
+    return "\n".join(f"{k}: {v}" for k, v in provenance_headers(text).items())
 
 # ======================================================
-# SESSION STATE
+# ECF NUMBER (LOCAL / PRE-FILING ONLY)
 # ======================================================
 
-LAST_RESULTS = []
-SELECTED_INDEX = None
-APP_MODE = "research"  # research | court | newsroom
+def generate_ecf_filing_number():
+    return f"ECF-PREFILE-{datetime.utcnow().strftime('%Y%m%d-%H%M%S')}"
 
 # ======================================================
-# HELPERS
+# DISTRICT DEFINITIONS
 # ======================================================
 
-def citation_hash(r):
-    return hashlib.sha256(
-        f"{r['agency']}|{r['resolved_url']}|{r['timestamp']}".encode()
-    ).hexdigest()[:16]
+DISTRICT_SCHEMAS = {
+    "Generic": {},
+    "D.D.C.": {},
+    "S.D.N.Y.": {},
+    "N.D. Cal.": {},
+}
 
-def sha256_text(text):
-    return hashlib.sha256(text.encode()).hexdigest()
+# ======================================================
+# COVER SHEET PDF
+# ======================================================
 
-def fre_callout():
-    return (
-        "Federal Rules of Evidence (Educational Reference):\n"
-        "• Rule 901 – Authentication\n"
-        "• Rule 803(8) – Public Records Exception\n"
-        "• Rule 1005 – Copies of Public Records\n"
-        "Not legal advice."
-    )
+def generate_cover_sheet_pdf(district, ecf_no):
+    buf = io.BytesIO()
+    styles = getSampleStyleSheet()
 
-def ai_disclosure():
-    return (
-        "\n\n---\n"
-        "AI DISCLOSURE\n"
-        "• User-initiated only\n"
-        "• Public FOIA documents only\n"
-        "• No legal advice\n"
-        "• Verify against cited exhibit\n"
+    doc = SimpleDocTemplate(buf, pagesize=LETTER)
+
+    body = (
+        f"<b>CM/ECF PRE-FILING COVER SHEET</b><br/><br/>"
+        f"<b>District:</b> {district}<br/>"
+        f"<b>Reference No.:</b> {ecf_no}<br/><br/>"
+        "This submission is a <b>pre-filing informational bundle</b> generated "
+        "from publicly available FOIA electronic reading rooms.<br/><br/>"
+        "No document in this bundle is filed, certified, or authenticated "
+        "by any court, clerk, or agency."
     )
 
+    doc.build([
+        Paragraph(body, styles["Normal"]),
+        PageBreak(),
+        Paragraph(render_provenance_block(body).replace("\n", "<br/>"), styles["Code"])
+    ])
+
+    buf.seek(0)
+    return buf
+
 # ======================================================
-# FOIA EXHAUSTION TIMELINE
+# EXHIBIT LIST
 # ======================================================
 
-def build_foia_exhaustion_timeline(request_date):
-    base = datetime.fromisoformat(request_date)
-    return [
-        ("FOIA Request Filed", base),
-        ("20-Day Statutory Response Due", base + timedelta(days=20)),
-        ("Administrative Appeal Window", base + timedelta(days=20)),
-        ("Constructive Exhaustion Eligible", base + timedelta(days=30)),
-        ("Judicial Review Eligible", base + timedelta(days=90)),
-    ]
+def generate_proposed_exhibit_list():
+    lines = ["PROPOSED EXHIBIT LIST\n"]
+    for i, r in enumerate(LAST_RESULTS, 1):
+        lines.append(
+            f"Exhibit {i:03d}: {r['agency']} FOIA Reading Room "
+            f"({r['url']})"
+        )
+    return "\n".join(lines)
 
 # ======================================================
-# PDF RESOLUTION
+# CLERK VERIFICATION CHECKLIST
 # ======================================================
 
-def resolve_pdf_url(url):
-    try:
-        r = requests.get(
-            url,
-            timeout=15,
-            allow_redirects=True,
-            headers={"User-Agent": "FOIA-Research-Tool"}
+def clerk_verification_checklist():
+    return (
+        "CLERK VERIFICATION CHECKLIST\n\n"
+        "☐ Confirm exhibit URLs resolve to issuing agency domains\n"
+        "☐ Confirm SHA-256 hash matches downloaded agency document\n"
+        "☐ Confirm document is publicly released\n"
+        "☐ Note: Tool does NOT certify authenticity\n"
+        "☐ Note: No sealed or restricted material included\n\n"
+        "Relevant Rules:\n"
+        "• FRE 902(5)\n"
+        "• FRE 803(8)\n"
+        "• FRE 1005\n"
+    )
+
+# ======================================================
+# PDF GENERATION (WITH AI / ETHICS FOOTER)
+# ======================================================
+
+def generate_pdf(title, body, exhibit_no, ecf_no):
+    buf = io.BytesIO()
+    styles = getSampleStyleSheet()
+
+    def footer(canvas, doc):
+        canvas.setFont("Helvetica", 8)
+        canvas.drawString(
+            40, 20,
+            "AI-assisted formatting only; no substantive analysis or factual assertions."
+        )
+        canvas.drawRightString(
+            580, 20,
+            f"{ecf_no} — Exhibit {exhibit_no}"
         )
-        ct = r.headers.get("content-type", "").lower()
-        is_pdf = r.url.lower().endswith(".pdf") or "application/pdf" in ct
-        return is_pdf, r.url
-    except Exception:
-        return False, url
-
-def generate_pdf_thumbnails(url, max_pages=3):
-    if not PDF_THUMBNAIL_AVAILABLE:
-        return []
-    try:
-        r = requests.get(url, timeout=15)
-        images = convert_from_bytes(r.content, first_page=1, last_page=max_pages)
-        thumbs = []
-        for img in images:
-            buf = io.BytesIO()
-            img.save(buf, format="PNG")
-            thumbs.append(base64.b64encode(buf.getvalue()).decode())
-        return thumbs
-    except Exception:
-        return []
+
+    doc = SimpleDocTemplate(
+        buf,
+        pagesize=LETTER,
+        onFirstPage=footer,
+        onLaterPages=footer
+    )
+
+    doc.build([
+        Paragraph(f"<b>{title}</b>", styles["Title"]),
+        Paragraph(body.replace("\n", "<br/>"), styles["Normal"]),
+        PageBreak(),
+        Paragraph(render_provenance_block(body).replace("\n", "<br/>"), styles["Code"]),
+    ])
+
+    buf.seek(0)
+    return buf
 
 # ======================================================
-# FOIA ADAPTERS (LINK-OUT ONLY)
+# FOIA ADAPTERS
 # ======================================================
 
 class FOIAAdapter:
-    agency = "UNKNOWN"
-    search_url = ""
-
-    def search(self, query):
-        start = time.time()
-        url = self.search_url.format(q=quote_plus(query))
-        latency = round((time.time() - start) * 1000, 1)
+    agency = ""
+    url = ""
+    def search(self, q):
         return [{
             "agency": self.agency,
-            "title": f"{self.agency} FOIA Reading Room Result",
-            "url": url,
-            "timestamp": datetime.utcnow().isoformat(),
-            "latency_ms": latency,
-            "sealed": False,
-            "redacted": False
+            "title": f"{self.agency} FOIA Reading Room",
+            "url": self.url.format(q=quote_plus(q)),
         }]
 
 class CIA(FOIAAdapter):
     agency = "CIA"
-    search_url = "https://www.cia.gov/readingroom/search/site/{q}"
+    url = "https://www.cia.gov/readingroom/search/site/{q}"
 
 class FBI(FOIAAdapter):
     agency = "FBI"
-    search_url = "https://vault.fbi.gov/search?SearchableText={q}"
-
-class DOJ(FOIAAdapter):
-    agency = "DOJ"
-    search_url = "https://www.justice.gov/foia/library?search={q}"
-
-class DHS(FOIAAdapter):
-    agency = "DHS"
-    search_url = "https://www.dhs.gov/foia-library/search?search={q}"
-
-class STATE(FOIAAdapter):
-    agency = "State Department"
-    search_url = "https://foia.state.gov/Search/Search.aspx?q={q}"
-
-class NSA(FOIAAdapter):
-    agency = "NSA"
-    search_url = "https://www.nsa.gov/resources/everyone/foia/reading-room/?q={q}"
-
-ALL_ADAPTERS = {
-    "CIA": CIA(),
-    "FBI": FBI(),
-    "DOJ": DOJ(),
-    "DHS": DHS(),
-    "State": STATE(),
-    "NSA": NSA()
-}
+    url = "https://vault.fbi.gov/search?SearchableText={q}"
+
+ALL_ADAPTERS = {"CIA": CIA(), "FBI": FBI()}
 
 # ======================================================
 # SEARCH
 # ======================================================
 
 def run_search(query, agencies):
-    global LAST_RESULTS, SELECTED_INDEX
-    SELECTED_INDEX = None
+    global LAST_RESULTS
     LAST_RESULTS = []
     rows = []
-
-    for name in agencies:
-        adapter = ALL_ADAPTERS[name]
-        for r in adapter.search(query):
-            r["resolved_pdf"], r["resolved_url"] = resolve_pdf_url(r["url"])
-            r["hash"] = citation_hash(r)
-            r["thumbnails"] = (
-                generate_pdf_thumbnails(r["resolved_url"])
-                if r["resolved_pdf"] else []
-            )
+    for a in agencies:
+        for r in ALL_ADAPTERS[a].search(query):
+            r["hash"] = sha256_text(r["url"])
             LAST_RESULTS.append(r)
-            rows.append([
-                r["agency"],
-                r["title"],
-                r["resolved_url"],
-                r["hash"],
-                f"{r['latency_ms']} ms"
-            ])
-
-    return rows, render_cards(), "Ready"
+            rows.append([r["agency"], r["title"], r["url"], r["hash"]])
+    return rows
 
 # ======================================================
-# CARD RENDER
+# COURT BUNDLE
 # ======================================================
 
-def render_cards():
-    cards = []
-    for idx, r in enumerate(LAST_RESULTS):
-        thumbs = "".join(
-            f'<img src="data:image/png;base64,{t}" style="width:32%;margin:4px;border-radius:6px" />'
-            for t in r["thumbnails"]
-        )
-
-        preview = thumbs or f'<a href="{r["resolved_url"]}" target="_blank">Open Source</a>'
+def generate_court_bundle(district):
+    ecf_no = generate_ecf_filing_number()
 
-        cards.append(f"""
-        <div class="card">
-            <div class="card-header">
-                <b>{r['agency']}</b>
-                <span class="badge">PUBLIC</span>
-            </div>
-            <div>{preview}</div>
-            <small>Hash: {r['hash']}</small>
-        </div>
-        """)
+    with tempfile.TemporaryDirectory() as td:
+        zpath = os.path.join(td, "court_bundle.zip")
 
-    return "".join(cards)
-
-# ======================================================
-# AI ASK
-# ======================================================
-
-def ask_ai(opt_in, pdf_opt_in, question):
-    if not opt_in:
-        return "Explicit AI opt-in required."
-    if SELECTED_INDEX is None:
-        return "Select a document first."
-
-    r = LAST_RESULTS[SELECTED_INDEX]
-    if not r["resolved_pdf"]:
-        return "AI available only for public PDFs."
-
-    context = ""
-    pin = "n.p."
-
-    if pdf_opt_in and PDF_TEXT_AVAILABLE:
-        try:
-            raw = extract_text(io.BytesIO(
-                requests.get(r["resolved_url"], timeout=15).content
-            ))
-            context = raw[:4000]
-            pin = "p. 1"
-        except Exception:
-            pass
-
-    analysis = (
-        f"{bluebook_exhibit(r, SELECTED_INDEX + 1, pin=pin)}\n\n"
-        f"{fre_callout()}\n\n"
-        f"Question:\n{question}\n\nContext:\n{context}"
-    )
+        with zipfile.ZipFile(zpath, "w") as z:
+            # Cover sheet
+            cover = generate_cover_sheet_pdf(district, ecf_no)
+            z.writestr("00_Cover_Sheet.pdf", cover.read())
 
-    final = analysis + ai_disclosure()
-    return final + f"\n\nIntegrity Hash: {sha256_text(final)}"
+            # Exhibits
+            for i, r in enumerate(LAST_RESULTS, 1):
+                pdf = generate_pdf(
+                    "Judicial Appendix",
+                    f"{r['agency']} FOIA Reading Room\n{r['url']}",
+                    f"{i:03d}",
+                    ecf_no,
+                )
+                z.writestr(f"Exhibit_{i:03d}.pdf", pdf.read())
+                z.writestr(
+                    f"Exhibit_{i:03d}.sha256",
+                    sha256_text(r["url"])
+                )
 
-# ======================================================
-# MODE SWITCH
-# ======================================================
+            # Support docs
+            z.writestr("proposed_exhibit_list.txt", generate_proposed_exhibit_list())
+            z.writestr("clerk_verification_checklist.txt", clerk_verification_checklist())
 
-def set_mode(mode):
-    global APP_MODE
-    APP_MODE = mode
-    return f"Mode: {mode.upper()}"
+        return open(zpath, "rb")
 
 # ======================================================
 # UI
 # ======================================================
 
-CSS = """
-.card {border:1px solid #ddd;border-radius:14px;padding:14px;margin:14px 0;background:#fafafa;}
-.badge {background:#eef;padding:4px 10px;border-radius:999px;font-size:0.75em;}
-.gradio-container .tab-nav {position:sticky;top:0;z-index:1000;background:white}
-@media (max-width: 640px){
-  .tab-nav button span {display:none}
-}
-"""
-
-with gr.Blocks(css=CSS, title="FOIA Intelligence Search") as gradio_app:
+with gr.Blocks(title="Federal FOIA Intelligence Search") as app:
     gr.Markdown("## Federal FOIA Intelligence Search")
 
-    with gr.Row():
-        mode = gr.Radio(
-            ["research", "court", "newsroom"],
-            value="research",
-            label="Operating Mode"
-        )
-        mode_status = gr.Textbox(interactive=False)
-
-    mode.change(set_mode, mode, mode_status)
-
     with gr.Tab("Search"):
-        agencies = gr.CheckboxGroup(list(ALL_ADAPTERS.keys()), value=list(ALL_ADAPTERS.keys()))
+        agencies = gr.CheckboxGroup(
+            list(ALL_ADAPTERS.keys()),
+            value=list(ALL_ADAPTERS.keys())
+        )
         query = gr.Textbox()
-        table = gr.Dataframe(headers=["Agency","Title","URL","Hash","Latency"])
-        cards = gr.HTML()
-        gr.Button("Search").click(run_search, [query, agencies], [table, cards, mode_status])
-
-    with gr.Tab("AI"):
-        ai_opt = gr.Checkbox(label="Enable AI")
-        pdf_opt = gr.Checkbox(label="Allow PDF Extraction")
-        q = gr.Textbox(lines=4)
-        a = gr.Textbox(lines=18)
-        gr.Button("Ask").click(ask_ai, [ai_opt, pdf_opt, q], a)
-
-    with gr.Tab("Timeline"):
-        d = gr.Textbox(value=datetime.utcnow().date().isoformat())
-        t = gr.Textbox(lines=8)
-        gr.Button("Build").click(
-            lambda d: "\n".join(f"{k}: {v.date()}" for k,v in build_foia_exhaustion_timeline(d)),
-            d, t
+        table = gr.Dataframe(
+            headers=["Agency", "Title", "URL", "SHA-256"]
         )
+        gr.Button("Search").click(run_search, [query, agencies], table)
 
-# ======================================================
-# MOUNT GRADIO INTO FASTAPI
-# ======================================================
-
-api = gr.mount_gradio_app(api, gradio_app, path="/")
+    with gr.Tab("Court Bundle"):
+        district = gr.Dropdown(
+            list(DISTRICT_SCHEMAS.keys()),
+            value="Generic"
+        )
+        gr.File(label="Download Court Bundle").upload(
+            lambda d=district: generate_court_bundle(d)
+        )
 
-# ======================================================
-# ENTRYPOINT
-# ======================================================
+    with gr.Tab("Clerk Checklist"):
+        gr.Textbox(value=clerk_verification_checklist(), lines=16)
 
-if __name__ == "__main__":
-    import uvicorn
-    uvicorn.run(api, host="0.0.0.0", port=7860)
+app.launch()