src/pageparse/web.py

from __future__ import annotations

import hashlib
import json
import time
import uuid
from pathlib import Path

from fastapi import FastAPI, File, Request, UploadFile
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import HTMLResponse
from fastapi.staticfiles import StaticFiles

from pageparse.config import settings
from pageparse.store import Store
from pageparse.telemetry import get_telemetry

app = FastAPI(title="PageParse")

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

HERE = Path(__file__).resolve().parent.parent.parent
web_static = HERE / "web" / "static"
web_templates = HERE / "web" / "templates"

if not web_static.exists():
    cwd_static = Path.cwd() / "web" / "static"
    if cwd_static.exists():
        web_static = cwd_static
if not web_templates.exists():
    cwd_templates = Path.cwd() / "web" / "templates"
    if cwd_templates.exists():
        web_templates = cwd_templates

if web_static.exists():
    app.mount("/static", StaticFiles(directory=str(web_static)), name="static")

_store_instance: Store | None = None


def _get_store() -> Store:
    global _store_instance
    if _store_instance is None:
        _store_instance = Store()
        _store_instance.init_db()
    return _store_instance


def _get_ocr():
    from pageparse.ocr.handwriting import HandwritingOCR
    return HandwritingOCR()


def _get_printed_ocr():
    from pageparse.ocr.printed import PrintedOCR
    return PrintedOCR()


def _get_extractor():
    from pageparse.extract import Extractor
    return Extractor()


# Structured logging
import structlog
logger = structlog.get_logger()

# Request tracing
TRACING_ENABLED = settings.tracing_enabled

# Prometheus metrics
if settings.prometheus_enabled:
    try:
        from prometheus_client import Counter, Histogram, generate_latest, REGISTRY, CONTENT_TYPE_LATEST
        from starlette.responses import Response

        HTTP_REQUESTS = Counter("pageparse_http_requests_total", "Total HTTP requests", ["method", "endpoint", "status"])
        HTTP_REQUEST_DURATION = Histogram("pageparse_http_request_duration_seconds", "HTTP request duration", ["method", "endpoint"])
        PROCESSED_SOURCES = Counter("pageparse_processed_sources_total", "Total processed sources", ["source_type", "status"])
        OCR_INFERENCES = Counter("pageparse_ocr_inferences_total", "Total OCR inferences", ["ocr_type"])
    except ImportError:
        settings.prometheus_enabled = False


@app.middleware("http")
async def add_tracing_and_metrics(request: Request, call_next):
    request_id = str(uuid.uuid4())[:8]
    request.state.request_id = request_id
    start_time = time.time()

    response = await call_next(request)

    duration = time.time() - start_time
    status_code = response.status_code

    if settings.prometheus_enabled:
        try:
            HTTP_REQUESTS.labels(method=request.method, endpoint=request.url.path, status=status_code).inc()
            HTTP_REQUEST_DURATION.labels(method=request.method, endpoint=request.url.path).observe(duration)
        except Exception:
            pass

    logger.info(
        "request",
        request_id=request_id,
        method=request.method,
        path=request.url.path,
        status=status_code,
        duration_ms=round(duration * 1000),
    )

    response.headers["X-Request-ID"] = request_id
    return response


if settings.prometheus_enabled:
    @app.get("/metrics")
    async def metrics():
        return Response(content=generate_latest(REGISTRY), media_type=CONTENT_TYPE_LATEST)


@app.get("/", response_class=HTMLResponse)
async def index() -> str:
    static_index = web_static / "index.html"
    if static_index.exists():
        html = static_index.read_text(encoding="utf-8")
        btn = (
            '<a href="/file-explorer" style="position:fixed;bottom:24px;right:24px;z-index:9999;'
            'display:inline-flex;align-items:center;gap:8px;padding:12px 20px;'
            'background:linear-gradient(135deg,#238636,#2ea043);color:#fff;'
            'border-radius:12px;text-decoration:none;font-size:14px;font-weight:600;'
            'box-shadow:0 4px 16px rgba(35,134,54,0.4);'
            'transition:transform .2s,box-shadow .2s;"'
            'onmouseover="this.style.transform=\'scale(1.05)\';this.style.boxShadow=\'0 6px 24px rgba(35,134,54,0.6)\'"'
            'onmouseout="this.style.transform=\'scale(1)\';this.style.boxShadow=\'0 4px 16px rgba(35,134,54,0.4)\'">'
            '&#x1F4CA; DSA Visualizer</a>'
        )
        close_tag = '</body>'
        if close_tag in html:
            return html.replace(close_tag, btn + close_tag)
        return html + btn
    index_path = web_templates / "index.html"
    if index_path.exists():
        return index_path.read_text(encoding="utf-8")
    return "<html><body><h1>PageParse</h1></body></html>"


@app.post("/upload")
async def upload(
    file: UploadFile = File(...), language: str = "English", schema_type: str = "auto"
) -> dict:
    import tempfile

    from pageparse import pipelines

    contents = await file.read()
    filename = file.filename or "upload"

    content_hash = hashlib.sha256(contents).hexdigest()

    store = _get_store()
    if settings.diff_sync_enabled:
        existing = store.get_source_by_hash(content_hash)
        if existing:
            records = store.get_records(source_id=existing["id"])
            return {
                "source_id": existing["id"],
                "page_id": existing["id"],
                "message": "Duplicate content skipped (already ingested)",
                "duplicate": True,
                "raw_text": existing.get("raw_text", ""),
                "image_path": existing.get("image_path", ""),
                "cleaned_image_path": existing.get("cleaned_image_path", ""),
                "result": {
                    "source_file": existing.get("filename", ""),
                    "source_type": existing.get("source_type", ""),
                    "captured_date": existing.get("captured_date", ""),
                    "title": existing.get("title", ""),
                    "summary": existing.get("summary", ""),
                    "records": [
                        {
                            "type": r.get("type", "task"),
                            "content": r.get("content", ""),
                            "due_date": r.get("due_date"),
                            "priority": r.get("priority"),
                            "category": r.get("category"),
                            "speaker": r.get("speaker"),
                            "timestamp": r.get("timestamp"),
                            "status": r.get("status"),
                            "confidence": r.get("confidence", 0.9),
                        }
                        for r in records
                    ],
                },
            }

    temp_dir = Path(tempfile.gettempdir())
    temp_path = temp_dir / filename
    temp_path.write_bytes(contents)

    uploads_dir = web_static / "uploads"
    uploads_dir.mkdir(exist_ok=True, parents=True)
    timestamp = int(time.time())
    original_filename = f"original_{timestamp}_{filename}"
    cleaned_filename = f"cleaned_{timestamp}_{filename}"
    original_path = uploads_dir / original_filename
    cleaned_path = uploads_dir / cleaned_filename
    original_path.write_bytes(contents)

    filename_lower = filename.lower()
    ext = Path(filename_lower).suffix

    image_exts = {".jpg", ".jpeg", ".png", ".bmp", ".tiff", ".tif", ".webp"}
    audio_exts = {".mp3", ".wav", ".m4a", ".ogg", ".flac"}
    video_exts = {".mp4", ".mkv", ".mov", ".avi"}
    doc_exts = {".pdf", ".docx", ".txt"}
    sheet_exts = {".csv", ".xlsx", ".xls"}

    source_type = "document"
    cleaned_image_url = None
    barcodes = []
    tables = []

    if ext in image_exts:
        source_type = "image"
        raw_text, image_url, cleaned_image_url, barcodes, tables = pipelines.process_image(temp_path, language)
        clean_temp = temp_dir / f"cleaned_{file.filename}"
        if clean_temp.exists():
            cleaned_path.write_bytes(clean_temp.read_bytes())
        if settings.prometheus_enabled:
            try:
                OCR_INFERENCES.labels(ocr_type="image").inc()
            except Exception:
                pass
    elif ext in audio_exts:
        source_type = "audio"
        raw_text, image_url = pipelines.process_audio(temp_path, language)
        if settings.prometheus_enabled:
            try:
                OCR_INFERENCES.labels(ocr_type="audio").inc()
            except Exception:
                pass
    elif ext in video_exts:
        source_type = "video"
        raw_text, image_url = pipelines.process_video(temp_path, language)
    elif ext in doc_exts:
        source_type = "document"
        raw_text, image_url = pipelines.process_document(temp_path, language)
    elif ext in sheet_exts:
        source_type = "spreadsheet"
        raw_text, image_url = pipelines.process_spreadsheet(temp_path, language)
    else:
        source_type = "document"
        raw_text, image_url = pipelines.process_document(temp_path, language)

    if settings.prometheus_enabled:
        try:
            PROCESSED_SOURCES.labels(source_type=source_type, status="processed").inc()
        except Exception:
            pass

    english_raw_text = raw_text
    has_non_ascii = not all(ord(c) < 128 for c in raw_text)

    if language != "English" and has_non_ascii:
        from pageparse.translation import translate_text
        try:
            english_raw_text = translate_text(raw_text, "English", source_lang=language)
        except Exception as e:
            print(f"Failed to translate native text to English for extraction: {e}")

    extractor = _get_extractor()
    if schema_type == "auto":
        schema_type = extractor.detect_schema_type(english_raw_text)
    result = extractor.extract(english_raw_text, file.filename, schema_type)
    result.source_type = source_type
    result.barcodes = barcodes
    result.tables = tables

    if language != "English":
        from pageparse.translation import translate_text
        for record in result.records:
            if record.content:
                record.content = translate_text(record.content, language, source_lang="English")
        if not has_non_ascii:
            raw_text = translate_text(raw_text, language)

    final_image_url = f"/static/uploads/{original_filename}"
    final_cleaned_image_url = (
        f"/static/uploads/{cleaned_filename}" if cleaned_image_url or (ext in image_exts) else None
    )

    source_id = store.save(
        result,
        raw_text,
        image_path=final_image_url,
        cleaned_image_path=final_cleaned_image_url,
        content_hash=content_hash,
    )

    return {
        "source_id": source_id,
        "page_id": source_id,
        "result": result.model_dump(mode="json"),
        "raw_text": raw_text,
        "image_path": final_image_url,
        "cleaned_image_path": final_cleaned_image_url,
        "barcodes": barcodes,
        "tables": tables,
        "detected_schema": schema_type if schema_type != "todo" else None,
    }


@app.post("/upload/batch")
async def upload_batch(files: list[UploadFile] = File(...), language: str = "English", schema_type: str = "auto") -> list[dict]:
    results = []
    for file in files:
        try:
            single = await upload(file, language, schema_type)
            results.append(single)
        except Exception as e:
            results.append({"filename": file.filename, "error": str(e)})
    return results


@app.get("/telemetry")
async def telemetry() -> dict:
    return get_telemetry()


@app.get("/stats")
async def stats() -> dict:
    store = _get_store()
    return store.get_stats()


@app.get("/airgap")
async def get_airgap() -> dict:
    return {"airgap": settings.airgap}


@app.post("/airgap")
async def set_airgap(request: Request) -> dict:
    data = await request.json()
    val = data.get("airgap", False)
    settings.airgap = val
    return {"airgap": settings.airgap}


@app.post("/translate")
async def translate(request: Request) -> dict:
    from pageparse.translation import translate_text
    data = await request.json()
    text = data.get("text", "")
    target_lang = data.get("target_lang", "English")
    source_lang = data.get("source_lang", "auto")
    translated_text = translate_text(text, target_lang, source_lang)
    return {"translated_text": translated_text}


@app.get("/sources")
@app.get("/pages")
async def get_sources() -> list[dict]:
    return _get_store().list_sources()


@app.get("/records")
@app.get("/tasks")
async def get_records(priority: str | None = None, type: str | None = None) -> list[dict]:
    return _get_store().get_records(priority=priority, type=type)


@app.patch("/records/{record_id}")
@app.patch("/tasks/{record_id}")
async def patch_record(record_id: int, request: Request) -> dict:
    updates = await request.json()
    success = _get_store().update_record(record_id, updates)
    return {"success": success}


@app.delete("/records/{record_id}")
@app.delete("/tasks/{record_id}")
async def delete_record(record_id: int) -> dict:
    success = _get_store().delete_record(record_id)
    return {"success": success}


@app.post("/sources/{source_id}/summarize")
@app.post("/pages/{source_id}/summarize")
async def summarize_source(source_id: int) -> dict:
    store = _get_store()
    sources = store.list_sources()
    source = next((s for s in sources if s["id"] == source_id), None)
    if not source:
        return {"error": "Source not found"}

    raw_text = source.get("raw_text", "")
    if not raw_text.strip() or raw_text in ("[UNCLEAR]", "[inaudible]"):
        return {"summary": "No text available to summarize."}

    summary = _llm_summarize(raw_text)
    if not summary.strip():
        summary = f"Summary of {source.get('filename')}: Contains extracted records and raw text."

    store.update_source_summary(source_id, summary)
    return {"summary": summary}


@app.post("/tts")
async def text_to_speech(request: Request) -> dict:
    data = await request.json()
    text = data.get("text", "")
    if not text.strip():
        return {"error": "No text provided"}
    try:
        import tempfile
        import pyttsx3
        engine = pyttsx3.init()
        engine.setProperty("rate", 150)
        engine.setProperty("volume", 0.9)
        tts_dir = web_static / "tts"
        tts_dir.mkdir(exist_ok=True)
        tts_path = tts_dir / f"tts_{int(time.time())}.wav"
        engine.save_to_file(text, str(tts_path))
        engine.runAndWait()
        return {"tts_url": f"/static/tts/{tts_path.name}"}
    except ImportError:
        return {"error": "TTS not available (pyttsx3 not installed)"}
    except Exception as e:
        return {"error": f"TTS failed: {e}"}


@app.post("/webhook")
async def set_webhook(request: Request) -> dict:
    data = await request.json()
    url = data.get("url", "")
    secret = data.get("secret", "")
    settings.webhook_url = url
    settings.webhook_secret = secret
    return {"webhook_url": url, "configured": bool(url)}


@app.get("/webhook")
async def get_webhook() -> dict:
    return {
        "webhook_url": settings.webhook_url or "",
        "configured": bool(settings.webhook_url),
    }


@app.post("/analyze-code")
async def analyze_code(request: Request) -> dict:
    from fastapi.responses import JSONResponse

    data = await request.json()
    code = data.get("code", "").strip()
    language_hint = data.get("language", "auto")

    if not code:
        return JSONResponse(status_code=400, content={"error": "No code provided"})

    # Check Ollama connectivity first
    ollama_ok = _check_ollama_connection()
    if not ollama_ok:
        return JSONResponse(
            status_code=503,
            content={
                "error": f"SLM (Ollama) not reachable at {settings.ollama_url}. Make sure Ollama is running and the model '{settings.slm_model}' is available.",
                "language": language_hint if language_hint != "auto" else "unknown",
                "explanation": "",
                "summary": "",
            },
        )

    lang = language_hint if language_hint != "auto" else _detect_language(code)

    prompt = (
        "You are a code analysis assistant. Analyze the following code and provide:\n"
        "1. EXPLANATION: A clear explanation of what this code does (2-3 sentences)\n"
        "2. SUMMARY: A one-line summary of the code's purpose\n\n"
        f"Language: {lang}\n\n"
        f"Code:\n```{lang}\n{code}\n```\n\n"
        "Format your response as:\n"
        "EXPLANATION:\n<explanation>\n\nSUMMARY:\n<summary>"
    )
    response = _llm_complete(prompt, system="You are a precise code analyst.", max_tokens=512)

    if not response:
        return JSONResponse(
            status_code=500,
            content={
                "error": f"SLM returned empty response. Check that Ollama model '{settings.slm_model}' is downloaded and running at {settings.ollama_url}.",
                "language": lang,
                "explanation": "",
                "summary": "",
            },
        )

    explanation = ""
    summary = ""

    if "EXPLANATION:" in response and "SUMMARY:" in response:
        parts = response.split("SUMMARY:", 1)
        summary = parts[1].strip()
        expl_part = parts[0].replace("EXPLANATION:", "").strip()
        explanation = expl_part
    else:
        explanation = response
        summary = response.split(".")[0] + "."

    return {
        "language": lang,
        "explanation": explanation,
        "summary": summary,
    }


def _check_ollama_connection() -> bool:
    import urllib.request
    try:
        req = urllib.request.Request(f"{settings.ollama_url}/api/tags")
        with urllib.request.urlopen(req, timeout=5) as resp:
            return resp.status == 200
    except Exception:
        return False


def _detect_language(code: str) -> str:
    import re
    code_stripped = code.strip()
    if not code_stripped:
        return "unknown"

    # Python
    if re.search(r'^\s*(import |from |def |class |print\(|if __name__)', code_stripped, re.MULTILINE):
        return "python"
    # JavaScript/TypeScript
    if re.search(r'^\s*(import |export |function |const |let |var |console\.log|document\.)', code_stripped, re.MULTILINE):
        if re.search(r':\s*(string|number|boolean|void|any)\s*[=;),]', code_stripped):
            return "typescript"
        return "javascript"
    # Java
    if re.search(r'^\s*(public |private |protected |class |import java\.)', code_stripped, re.MULTILINE):
        return "java"
    # C/C++
    if re.search(r'^\s*(#include|int main|void main|#define)', code_stripped, re.MULTILINE):
        return "c" if re.search(r'printf\(|scanf\(', code_stripped) else "cpp"
    # Rust
    if re.search(r'^\s*(fn |let mut|use std|impl |pub )', code_stripped, re.MULTILINE):
        return "rust"
    # Go
    if re.search(r'^\s*(package main|func |import \()', code_stripped, re.MULTILINE):
        return "go"
    # HTML
    if re.search(r'^\s*<!DOCTYPE|<html|<div|<body|<head', code_stripped, re.MULTILINE):
        return "html"
    # CSS
    if re.search(r'^\s*[a-zA-Z-]+\s*\{', code_stripped, re.MULTILINE):
        return "css"
    # SQL
    if re.search(r'^\s*(SELECT|INSERT|UPDATE|DELETE|CREATE TABLE|ALTER|DROP)\s', code_stripped, re.MULTILINE):
        return "sql"
    # Ruby
    if re.search(r'^\s*(def |class |require |module |end\s*$)', code_stripped, re.MULTILINE):
        return "ruby"
    # Shell/Bash
    if re.search(r'^\s*(#!/bin/|#!|\. )', code_stripped, re.MULTILINE) or re.search(r'\b(if\s+\[|fi\s*$|then\s*$|esac\s*$)', code_stripped, re.MULTILINE):
        return "bash"

    return language_hint if language_hint != "auto" else "unknown"


def _llm_complete(prompt: str, system: str = "", temperature: float = 0.3, max_tokens: int = 256) -> str:
    import urllib.request

    payload = {
        "model": settings.slm_model,
        "prompt": f"{system}\n\n{prompt}" if system else prompt,
        "stream": False,
        "options": {"temperature": temperature, "num_predict": max_tokens},
    }
    url = f"{settings.ollama_url}/api/generate"
    try:
        data = json.dumps(payload).encode("utf-8")
        req = urllib.request.Request(url, data=data, headers={"Content-Type": "application/json"})
        with urllib.request.urlopen(req, timeout=90) as response:
            res = json.loads(response.read().decode("utf-8"))
            return res.get("response", "").strip()
    except Exception as e:
        print(f"Ollama request failed: {e}")
        return ""


def _llm_summarize(text: str) -> str:
    prompt = (
        "Summarize the following raw extracted text concisely in 2-3 sentences. "
        "Focus on the main topics and action items.\n\n"
        f"Text:\n{text}\n\n"
        "Summary:"
    )
    return _llm_complete(prompt, system="You are a concise summarizer.")


@app.post("/chat")
async def chat(request: Request) -> dict:
    data = await request.json()
    prompt = data.get("prompt", "")
    source_id = data.get("source_id", None)

    store = _get_store()
    records = []

    try:
        if source_id:
            records = store.get_records(source_id=int(source_id))
        else:
            from pageparse.search import SemanticSearch
            searcher = SemanticSearch()
            records = searcher.search(prompt, top_k=7)
    except Exception:
        all_recs = store.get_records(source_id=int(source_id) if source_id else None)
        keywords = prompt.lower().split()
        matched = []
        for r in all_recs:
            if any(
                k in r["content"].lower() or (r["category"] and k in r["category"].lower())
                for k in keywords
            ):
                matched.append(r)
        records = matched[:7] if matched else all_recs[:10]

    context = "\n".join(
        [
            f"- [{r['type']}] {r['content']} (Due: {r['due_date']}, Priority: {r['priority']}, Category: {r['category']})"
            for r in records
        ]
    )

    llm_prompt = (
        "You are PageParse AI, a local notes and task assistant. "
        "Use the following user records context to answer the user's question. "
        "Answer concisely and clearly in English. If the information is not in the context, "
        "use your general knowledge but state clearly if it's not present in the user's records.\n\n"
        f"Context:\n{context}\n\n"
        f"User Question: {prompt}\n\n"
        "Response:"
    )

    response_text = _llm_complete(llm_prompt, system="You are a helpful task assistant.", max_tokens=256)
    if not response_text:
        ollama_ok = _check_ollama_connection()
        if not ollama_ok:
            response_text = (
                "The AI assistant requires Ollama to be running locally.\n\n"
                "To enable AI features:\n"
                "1. Install Ollama from https://ollama.com\n"
                "2. Pull a model: `ollama pull llama3.2:1b`\n"
                "3. Start Ollama and refresh this page\n\n"
                "Meanwhile, you can still upload files or explore DSA templates."
            )
        elif context:
            response_text = f"Here is what I found in your uploaded files:\n\n{context}"
        else:
            response_text = (
                "I don't have enough context from your files to answer that.\n\n"
                "Try:\n"
                "- Upload a file first and ask about it\n"
                "- Ask about the DSA visualizer templates\n"
                "- Check Ollama is running with a compatible model"
            )

    return {"response": response_text}


@app.get("/ollama/status")
async def ollama_status() -> dict:
    import urllib.request

    url = f"{settings.ollama_url}/api/tags"
    connected = False
    models: list[str] = []
    model_available = False
    version = ""
    try:
        req = urllib.request.Request(url)
        with urllib.request.urlopen(req, timeout=5) as response:
            data = json.loads(response.read().decode("utf-8"))
            models = [m["name"] for m in data.get("models", [])]
            connected = True
            model_available = settings.slm_model in models
    except Exception:
        pass

    # Also try getting version
    try:
        ver_req = urllib.request.Request(f"{settings.ollama_url}/api/version")
        with urllib.request.urlopen(ver_req, timeout=3) as response:
            ver_data = json.loads(response.read().decode("utf-8"))
            version = ver_data.get("version", "")
    except Exception:
        pass

    return {
        "connected": connected,
        "url": settings.ollama_url,
        "model": settings.slm_model,
        "models": models,
        "model_available": model_available,
        "version": version,
    }


@app.post("/ollama/configure")
async def ollama_configure(request: Request) -> dict:
    data = await request.json()
    url = data.get("url", "").strip()
    model = data.get("model", "").strip()
    if url:
        settings.ollama_url = url.rstrip("/")
    if model:
        settings.slm_model = model
    return {"url": settings.ollama_url, "model": settings.slm_model}


CODE_EXTENSIONS = {
    ".py", ".js", ".ts", ".jsx", ".tsx", ".html", ".css", ".scss", ".less",
    ".c", ".cpp", ".cc", ".cxx", ".h", ".hpp", ".java", ".kt", ".scala",
    ".rs", ".go", ".rb", ".php", ".swift", ".m", ".mm",
    ".sql", ".r", ".m", ".sh", ".bash", ".zsh", ".ps1",
    ".yaml", ".yml", ".json", ".xml", ".toml", ".ini", ".cfg",
    ".md", ".rst", ".tex", ".txt",
    ".vue", ".svelte", ".lua", ".pl", ".pm", ".hs", ".ex", ".exs",
}
IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".bmp", ".tiff", ".tif", ".webp", ".gif", ".svg"}
AUDIO_EXTS = {".mp3", ".wav", ".m4a", ".ogg", ".flac", ".aac", ".wma"}
VIDEO_EXTS = {".mp4", ".mkv", ".mov", ".avi", ".webm", ".flv", ".wmv"}
DOC_EXTS = {".pdf", ".docx", ".txt"}
SHEET_EXTS = {".csv", ".xlsx", ".xls"}


@app.post("/analyze-file")
async def analyze_file(file: UploadFile = File(...)) -> dict:
    import tempfile

    contents = await file.read()
    filename = file.filename or "upload"
    ext = Path(filename).suffix.lower()
    size_kb = len(contents) / 1024

    result = {
        "filename": filename,
        "extension": ext,
        "size_kb": round(size_kb, 2),
        "file_type": "unknown",
        "content_preview": "",
        "explanation": "",
        "summary": "",
        "step_by_step": [],
    }

    if ext in CODE_EXTENSIONS:
        result["file_type"] = "code"
        try:
            code_text = contents.decode("utf-8")
        except UnicodeDecodeError:
            try:
                code_text = contents.decode("latin-1")
            except Exception:
                code_text = "[Binary or non-text content]"

        preview_lines = code_text.split("\n")[:50]
        result["content_preview"] = "\n".join(preview_lines)
        result["total_lines"] = code_text.count("\n") + 1

        lang = ext.lstrip(".")
        lang_map = {
            "py": "python", "js": "javascript", "ts": "typescript",
            "html": "html", "css": "css", "c": "c", "cpp": "cpp",
            "java": "java", "rs": "rust", "go": "go", "rb": "ruby",
            "kt": "kotlin", "swift": "swift", "sh": "bash", "bash": "bash",
            "php": "php", "sql": "sql", "r": "r", "lua": "lua",
            "vue": "vue", "svelte": "svelte", "scala": "scala",
        }
        language = lang_map.get(lang, lang)

        from pageparse import dsa_visualizer
        analysis = dsa_visualizer.analyze_code(code_text, language)
        result["explanation"] = analysis.get("explanation", "")
        result["summary"] = analysis.get("summary", "")
        result["dsa_type"] = analysis.get("dsa_type", "general")
        result["time_complexity"] = analysis.get("time_complexity", "")
        result["space_complexity"] = analysis.get("space_complexity", "")

        step_by_step = dsa_visualizer.explain_code_step_by_step(code_text, language)
        raw_steps = step_by_step.get("explanation", "").split("\n")
        result["step_by_step"] = [s.strip() for s in raw_steps if s.strip()]

        if not result["explanation"]:
            lines = code_text.split("\n")
            result["explanation"] = f"This {language} file has {result['total_lines']} lines. "
            result["step_by_step"] = [f"Line {i+1}: {lines[i][:100]}" for i in range(min(20, len(lines)))]

    elif ext in IMAGE_EXTS:
        result["file_type"] = "image"
        import base64
        b64 = base64.b64encode(contents).decode("utf-8")
        mime_key = {"jpg": "jpeg", "jpeg": "jpeg"}.get(ext.lstrip("."), ext.lstrip("."))
        result["image_data_url"] = f"data:image/{mime_key};base64,{b64}"
        result["mime"] = f"image/{mime_key}"

        temp_dir = Path(tempfile.gettempdir())
        temp_path = temp_dir / filename
        temp_path.write_bytes(contents)
        try:
            from PIL import Image as PILImage
            pil_img = PILImage.open(temp_path)
            img_w, img_h = pil_img.size
            img_mode = pil_img.mode
            img_format = pil_img.format or ext.lstrip(".").upper()
            meta_desc = f"Image dimensions: {img_w}x{img_h} pixels, Format: {img_format}, Color mode: {img_mode}, File size: {size_kb:.1f} KB"
            result["image_metadata"] = {"width": img_w, "height": img_h, "format": img_format, "mode": img_mode}
            result["content_preview"] = f"[{img_format} Image: {img_w}x{img_h}, {size_kb:.1f} KB]"

            from pageparse import pipelines
            raw_text, image_url, cleaned_url, barcodes, tables = pipelines.process_image(temp_path)
            text = raw_text.strip()
            if text and text not in ("[UNCLEAR]", "[inaudible]", ""):
                result["explanation"] = text
                result["barcodes"] = barcodes
                result["tables"] = tables
                result["step_by_step"] = text.split("\n")[:30]
                sm = _llm_summarize(text)
                result["summary"] = sm if sm else f"Extracted {len(text)} characters of text from this image."
            else:
                result["explanation"] = f"This is a {img_format} image ({img_w}x{img_h} pixels, {img_mode} mode, {size_kb:.1f} KB). No text was detected in the image."
                result["summary"] = f"{img_format} image — {img_w}x{img_h}px, {size_kb:.1f}KB"
        except Exception as e:
            result["explanation"] = f"Image file: {filename} ({ext}, {size_kb:.1f} KB). Details: {e}"
            result["summary"] = f"Image: {filename}"
            result["content_preview"] = f"[Image: {filename}, {size_kb:.1f} KB]"

    elif ext in AUDIO_EXTS:
        result["file_type"] = "audio"
        temp_dir = Path(tempfile.gettempdir())
        temp_path = temp_dir / filename
        temp_path.write_bytes(contents)
        audio_info = f"Audio file: {filename} ({ext}, {size_kb:.1f} KB)"
        try:
            from pageparse import pipelines
            transcript, audio_url = pipelines.process_audio(temp_path)
            text = transcript.strip()
            if text and text not in ("[UNCLEAR]", "[inaudible]", ""):
                result["content_preview"] = text[:2000]
                result["explanation"] = text
                result["step_by_step"] = [s.strip() + "." for s in text.replace("?", ".").replace("!", ".").split(".") if len(s.strip()) > 20][:30]
                sm = _llm_summarize(text)
                result["summary"] = sm if sm else f"Transcribed {len(text)} characters from audio."
            else:
                result["content_preview"] = audio_info
                result["explanation"] = f"{audio_info}. The audio was processed but no speech could be transcribed (may be silence, music, or an unsupported format)."
                result["summary"] = f"Audio — no speech detected"
        except Exception as e:
            result["content_preview"] = audio_info
            result["explanation"] = f"{audio_info}. Could not transcribe: {e}"
            result["summary"] = f"Audio: {filename}"

    elif ext in VIDEO_EXTS:
        result["file_type"] = "video"
        temp_dir = Path(tempfile.gettempdir())
        temp_path = temp_dir / filename
        temp_path.write_bytes(contents)
        video_info = f"Video file: {filename} ({ext}, {size_kb:.1f} KB)"
        try:
            import cv2
            cap = cv2.VideoCapture(str(temp_path))
            v_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
            v_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
            v_fps = cap.get(cv2.CAP_PROP_FPS)
            v_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
            v_duration = v_frames / v_fps if v_fps > 0 else 0
            v_duration_str = f"{int(v_duration // 60)}m {int(v_duration % 60)}s" if v_duration > 0 else "unknown"
            cap.release()
            video_info = f"Video: {v_width}x{v_height}, {v_fps:.1f} fps, {v_duration_str}, {size_kb:.1f} KB"
            result["video_metadata"] = {"width": v_width, "height": v_height, "fps": round(v_fps, 1), "frames": v_frames, "duration_sec": round(v_duration, 1)}

            from pageparse import pipelines
            combined_text, video_url = pipelines.process_video(temp_path)
            text = combined_text.strip()
            if text and text not in ("[UNCLEAR]", "[inaudible]", ""):
                result["content_preview"] = text[:2000]
                result["explanation"] = text
                result["step_by_step"] = text.split("\n\n")[:30]
                sm = _llm_summarize(text)
                result["summary"] = sm if sm else f"Extracted {len(text)} characters from video."
            else:
                result["content_preview"] = video_info
                result["explanation"] = f"This video ({v_width}x{v_height}, {v_fps:.1f} fps, {v_duration_str}, {size_kb:.1f} KB). No text or speech could be extracted."
                result["summary"] = f"Video — {v_width}x{v_height}, {v_duration_str}"
        except Exception as e:
            result["content_preview"] = video_info
            result["explanation"] = f"{video_info}. Could not process: {e}"
            result["summary"] = f"Video: {filename}"

    elif ext in DOC_EXTS:
        result["file_type"] = "document"
        try:
            text_content = contents.decode("utf-8")
        except Exception:
            text_content = ""
        result["content_preview"] = text_content[:2000] if text_content else f"[Document: {filename}]"
        result["explanation"] = f"This is a document file ({ext}). "
        result["summary"] = f"Document: {filename}"
        if text_content:
            result["step_by_step"] = text_content.split("\n")[:30]

    elif ext in SHEET_EXTS:
        result["file_type"] = "spreadsheet"
        result["content_preview"] = f"[Spreadsheet file: {filename}, {size_kb:.1f} KB]"
        result["explanation"] = f"This is a spreadsheet file ({ext}). Upload to /upload for data extraction."
        result["summary"] = f"Spreadsheet: {filename}"

    else:
        result["file_type"] = "binary" if ext else "unknown"
        result["content_preview"] = f"[{size_kb:.1f} KB binary file]"
        result["explanation"] = f"File type '{ext}' is not recognized. Try uploading via the standard /upload endpoint."
        result["summary"] = f"Unknown file type: {filename}"

    return result


@app.post("/dsa/analyze-code")
async def dsa_analyze_code(request: Request) -> dict:
    from fastapi.responses import JSONResponse

    data = await request.json()
    code = data.get("code", "").strip()
    language = data.get("language", "auto")

    if not code:
        return JSONResponse(status_code=400, content={"error": "No code provided"})

    from pageparse import dsa_visualizer
    result = dsa_visualizer.analyze_code(code, language)
    return result


@app.post("/dsa/explain-code")
async def dsa_explain_code(request: Request) -> dict:
    from fastapi.responses import JSONResponse

    data = await request.json()
    code = data.get("code", "").strip()
    language = data.get("language", "auto")

    if not code:
        return JSONResponse(status_code=400, content={"error": "No code provided"})

    from pageparse import dsa_visualizer
    result = dsa_visualizer.explain_code_step_by_step(code, language)
    return result


@app.post("/dsa/visualize")
async def dsa_visualize(request: Request) -> dict:
    from fastapi.responses import JSONResponse

    data = await request.json()
    code = data.get("code", "").strip()
    language = data.get("language", "auto")
    input_data = data.get("input_data", None)

    if not code:
        return JSONResponse(status_code=400, content={"error": "No code provided"})

    from pageparse import dsa_visualizer
    result = dsa_visualizer.analyze_code(code, language)

    if input_data and isinstance(input_data, list):
        dsa_type = result.get("dsa_type", "")
        if dsa_type in ("bubble_sort", "selection_sort", "insertion_sort"):
            result["steps"] = dsa_visualizer.generate_sorting_steps(input_data, dsa_type)
        elif dsa_type == "binary_tree":
            result["steps"] = dsa_visualizer.generate_bst_steps(input_data)
        elif dsa_type == "linked_list":
            result["steps"] = dsa_visualizer.generate_linked_list_steps(input_data)

    return result


@app.get("/dsa/templates")
async def dsa_templates() -> dict:
    from pageparse import dsa_visualizer
    return {"templates": dsa_visualizer.list_templates()}


@app.post("/dsa/visualize-template")
async def dsa_visualize_template(request: Request) -> dict:
    from fastapi.responses import JSONResponse

    data = await request.json()
    template_name = data.get("template", "").strip()
    input_data = data.get("input_data", None)

    from pageparse import dsa_visualizer
    template = dsa_visualizer.get_template(template_name)
    if not template:
        return JSONResponse(status_code=404, content={"error": f"Template '{template_name}' not found"})

    result = dsa_visualizer.analyze_code(template["code"], "python")

    if input_data and isinstance(input_data, list):
        dsa_type = result.get("dsa_type", "")
        if dsa_type in ("bubble_sort", "selection_sort", "insertion_sort"):
            result["steps"] = dsa_visualizer.generate_sorting_steps(input_data, dsa_type)
        elif dsa_type == "binary_tree":
            result["steps"] = dsa_visualizer.generate_bst_steps(input_data)
        elif dsa_type == "linked_list":
            result["steps"] = dsa_visualizer.generate_linked_list_steps(input_data)

    return result


@app.get("/file-explorer")
async def file_explorer_page() -> HTMLResponse:
    fe_path = web_templates / "file_explorer.html"
    if fe_path.exists():
        return HTMLResponse(content=fe_path.read_text(encoding="utf-8"))
    return HTMLResponse(content="<html><body><h1>Page not found</h1></body></html>")