app.py

"""
Streamlit Universal File‑Format Changer
--------------------------------------
A Streamlit app ready for Hugging Face Spaces that **actually converts file
contents when possible**, instead of merely renaming extensions.

* Image ↔ image conversions via **Pillow** (JPEG, PNG, GIF, BMP, TIFF, ICO, WEBP)
* Plain‑text files kept intact but re‑encoded (UTF‑8) when changing among
  text‑like extensions (txt, md, csv, json, xml, html, css, js)
* Disallowed uploads: `.exe`, `.bin`
* Everything is bundled into one ZIP download.

Created 2025‑05‑22 • v2
"""
from __future__ import annotations

# NOTE: Set env vars *before* importing Streamlit ------------------------------
import os, pathlib
os.environ.setdefault("STREAMLIT_HOME", "/tmp/.streamlit")
os.environ.setdefault("HOME", "/tmp")
pathlib.Path(os.environ["STREAMLIT_HOME"]).mkdir(parents=True, exist_ok=True)

import io
import zipfile
from datetime import datetime
from pathlib import Path

import streamlit as st
from PIL import Image  # Pillow for real image conversion

# -----------------------------------------------------------------------------
# Supported extensions ---------------------------------------------------------
# -----------------------------------------------------------------------------
IMAGE_EXTS = {
    ".jpg", ".jpeg", ".png", ".gif", ".bmp", ".tiff", ".ico", ".webp",
}
TEXT_EXTS = {
    ".txt", ".md", ".csv", ".json", ".xml", ".html", ".css", ".js",
}
ARCHIVE_EXTS = {".zip", ".tar", ".gz", ".7z"}
MEDIA_EXTS = {".mp3", ".wav", ".mp4", ".avi", ".mkv", ".mov"}
DOC_EXTS   = {".pdf", ".doc", ".docx"}

ALLOWED_TARGET_EXTS: list[str] = sorted(
    IMAGE_EXTS | TEXT_EXTS | ARCHIVE_EXTS | MEDIA_EXTS | DOC_EXTS
)

DISALLOWED_SOURCE_EXTS = {".exe", ".bin"}

# -----------------------------------------------------------------------------
# Helpers ----------------------------------------------------------------------
# -----------------------------------------------------------------------------

def sidebar_target_extension() -> str:
    st.sidebar.header("Settings")
    query = st.sidebar.text_input("Filter extensions… (optional)")
    filtered = [e for e in ALLOWED_TARGET_EXTS if query.lower() in e]
    if not filtered:
        st.sidebar.error("No extension matches that filter.")
    target_ext = st.sidebar.selectbox(
        "Choose target extension (applied to **all** files)",
        filtered or ALLOWED_TARGET_EXTS,
        index=(filtered or ALLOWED_TARGET_EXTS).index(".png")
        if ".png" in (filtered or ALLOWED_TARGET_EXTS) else 0,
    )
    st.sidebar.markdown(
        "*Images are truly converted. Text files are re‑saved as UTF‑8. "
        "Other combinations fall back to a safe rename.*"
    )
    return target_ext


def uploader():
    return st.file_uploader(
        "Upload any files (multiple allowed)",
        accept_multiple_files=True,
        type=None,  # accept *all* extensions
        help="Drag‑and‑drop or click to browse.",
    )


# -----------------------------------------------------------------------------
# Conversion logic -------------------------------------------------------------
# -----------------------------------------------------------------------------

def convert_image(data: bytes, target_ext: str) -> bytes:
    """Return `bytes` of the image converted to `target_ext`. Raises if Pillow
    cannot save in that format."""
    img = Image.open(io.BytesIO(data))
    buf = io.BytesIO()
    # Map certain extensions to Pillow format names
    pil_fmt = {
        ".jpg": "JPEG", ".jpeg": "JPEG", ".png": "PNG", ".gif": "GIF",
        ".bmp": "BMP", ".tiff": "TIFF", ".ico": "ICO", ".webp": "WEBP",
    }[target_ext]
    img.save(buf, format=pil_fmt)
    buf.seek(0)
    return buf.read()


def convert_text(data: bytes, _target_ext: str) -> bytes:
    """Return data re‑encoded as UTF‑8 (no format change)."""
    text = data.decode("utf‑8", errors="ignore")
    return text.encode("utf‑8")


def convert_file(file: st.runtime.uploaded_file_manager.UploadedFile, target_ext: str) -> tuple[bytes, str]:
    """Try to convert and return (bytes, conversion_note). On failure, return
    original data with a note that only rename happened."""
    orig_ext = Path(file.name).suffix.lower()
    raw = file.read()

    try:
        if orig_ext in IMAGE_EXTS and target_ext in IMAGE_EXTS:
            return convert_image(raw, target_ext), "image converted"
        if orig_ext in TEXT_EXTS and target_ext in TEXT_EXTS:
            return convert_text(raw, target_ext), "text re‑encoded"
    except Exception as err:
        st.warning(f"⚠️  Could not convert **{file.name}**: {err}. Falling back to rename.")

    # Fallback: no conversion
    return raw, "renamed only"


# -----------------------------------------------------------------------------
# Zip packaging ---------------------------------------------------------------
# -----------------------------------------------------------------------------

def package_zip(files: list[st.runtime.uploaded_file_manager.UploadedFile], target_ext: str) -> io.BytesIO:
    buf = io.BytesIO()
    with zipfile.ZipFile(buf, "w", zipfile.ZIP_DEFLATED) as zf:
        for file in files:
            orig_path = Path(file.name)
            if orig_path.suffix.lower() in DISALLOWED_SOURCE_EXTS:
                st.warning(f"⏭️  Skipping disallowed file: **{orig_path.name}**")
                continue
            data, note = convert_file(file, target_ext)
            new_name = orig_path.with_suffix(target_ext).name
            zf.writestr(new_name, data)
            st.success(f"✅ {note} • **{orig_path.name}** → **{new_name}**")
    buf.seek(0)
    return buf


# -----------------------------------------------------------------------------
# Main ------------------------------------------------------------------------
# -----------------------------------------------------------------------------

def main():
    st.set_page_config("Universal Format Changer", page_icon="🔄", layout="centered")
    st.title("🔄 Universal File‑Format Changer")
    st.write("Upload files, pick a target extension, and download a ZIP with the converted files.")

    target_ext = sidebar_target_extension()
    files = uploader()

    if files and st.button("🚀 Convert & Download"):
        zip_buf = package_zip(files, target_ext)
        ts = datetime.utcnow().strftime("%Y%m%dT%H%M%SZ")
        st.download_button(
            "⬇️ Get ZIP", zip_buf, file_name=f"converted_{ts}.zip", mime="application/zip"
        )

    st.caption("© 2025 Universal Changer • Streamlit • Hugging Face Spaces")


if __name__ == "__main__":
    main()