app.py

"""
app.py
Step 7: Allow uploading .txt + .pdf + .html files from the Admin panel (then auto-reindex).

(Objective)
- Accept multiple files (txt/pdf/html) in the Admin section.
- Save them into data/seed/ (overwrites by filename).
- Rebuild index and clear the cached searcher.
"""

from pathlib import Path
from typing import List

import gradio as gr

from indexer import build_index
from rag_search import (
    RAGSearcher,
    format_sources,
    extract_links,
    split_form_links,
    links_markdown,
)
from evo_inference import synthesize_with_evo
from utils_lang import SUPPORTED, normalize_lang, L

_searcher = None
DATA_SEED_DIR = Path("data/seed")
ALLOWED_EXTS = {".txt", ".pdf", ".html", ".htm", ".xhtml"}  # (Objective)


def ensure_searcher():
    global _searcher
    if _searcher is None:
        _searcher = RAGSearcher()
    return _searcher


def on_build_index():
    try:
        msg = build_index()
        global _searcher
        _searcher = None
    except Exception as e:
        msg = f"Error while building index: {e}"
    return msg


def _save_files(files: List[gr.File]) -> List[str]:
    """
    (Objective) Save uploaded files with allowed extensions into data/seed/.
    Supports different Gradio file object shapes.
    """
    saved = []
    DATA_SEED_DIR.mkdir(parents=True, exist_ok=True)
    if not files:
        return saved

    for f in files:
        try:
            # Resolve a path from the gradio file object
            path = None
            if isinstance(f, dict) and "name" in f:
                path = Path(f["name"])
            elif hasattr(f, "name"):
                path = Path(getattr(f, "name"))
            elif isinstance(f, str):
                path = Path(f)
            if not path or not path.exists():
                continue

            if path.suffix.lower() not in ALLOWED_EXTS:
                continue

            dest = DATA_SEED_DIR / path.name
            dest.write_bytes(path.read_bytes())
            saved.append(path.name)
        except Exception:
            continue
    return saved


def on_upload_and_reindex(files) -> str:
    saved = _save_files(files or [])
    if not saved:
        return "No valid .txt/.pdf/.html uploaded."
    status = on_build_index()
    return f"Uploaded: {', '.join(saved)}\n{status}"


def on_ask(question: str, top_k: int, lang_code: str, history: list):
    lang = normalize_lang(lang_code)
    if not question or len(question.strip()) < 3:
        return history, L(lang, "intro_err"), f"{L(lang, 'sources')}: (none)", "", ""

    try:
        searcher = ensure_searcher()
        hits = searcher.search(question, k=int(top_k))
        answer = synthesize_with_evo(question, lang, hits)

        srcs = format_sources(hits)
        if srcs.startswith("Sources:"):
            srcs = srcs.replace("Sources:", f"{L(lang, 'sources')}:")

        all_links = extract_links(hits)
        form_links, other_links = split_form_links(all_links)
        links_md = links_markdown(L(lang, "links"), other_links)
        forms_md = links_markdown(L(lang, "forms"), form_links)

    except Exception as e:
        answer = f"Error: {e}\n\n{L(lang, 'tip_build')}"
        srcs = f"{L(lang, 'sources')}: (none)"
        links_md = f"**{L(lang, 'links')}:** (none)"
        forms_md = f"**{L(lang, 'forms')}:** (none)"

    history = history + [(question, answer)]
    return history, answer, srcs, links_md, forms_md


with gr.Blocks(title=L("en", "title")) as demo:
    gr.Markdown(f"# 🇲🇺 **{L('en','title')}** — Step 7")

    with gr.Row():
        build_btn = gr.Button(L("en", "build_btn"))
        status = gr.Markdown(L("en", "status_idle"))
    build_btn.click(fn=on_build_index, outputs=status)

    # Admin: upload txt/pdf/html, then reindex
    with gr.Accordion("Admin: upload TXT / PDF / HTML and auto-reindex", open=False):
        upload = gr.File(
            file_count="multiple",
            file_types=["file"],  # allow any file; we'll filter by extension
            label="Upload .txt / .pdf / .html"
        )
        save_btn = gr.Button("Save & Reindex")
        upload_status = gr.Markdown()
        save_btn.click(fn=on_upload_and_reindex, inputs=upload, outputs=upload_status)

    gr.Markdown("### Language / Langue / Langaz")
    lang = gr.Dropdown(choices=list(SUPPORTED.keys()), value="en", label="EN / FR / MFE")

    gr.Markdown("### Ask a question / Posez une question / Poz enn kestyon")
    with gr.Row():
        q = gr.Textbox(placeholder=L("en", "ask_placeholder"), label=L("en", "ask_label"))
        k = gr.Slider(3, 12, step=1, value=6, label=L("en", "k_label"))
        ask_btn = gr.Button("Ask / Demander / Rode")

    chat = gr.Chatbot(label="Assistant", height=360)
    answer_md = gr.Markdown()
    sources_md = gr.Markdown()
    links_md = gr.Markdown()
    forms_md = gr.Markdown()

    ask_btn.click(
        fn=on_ask,
        inputs=[q, k, lang, chat],
        outputs=[chat, answer_md, sources_md, links_md, forms_md],
    )
    q.submit(
        fn=on_ask,
        inputs=[q, k, lang, chat],
        outputs=[chat, answer_md, sources_md, links_md, forms_md],
    )

demo.launch()