===== scripts/libarchive/build.sh =====

#!/usr/bin/env bash
# Reproducible build of libarchive for the browser (7Z and RAR extraction):
# libarchive + zlib + bzip2 + xz/liblzma compiled with a pinned Emscripten,
# with a UTF-8 locale set at startup (glue.c). Output: src/vendor/libarchive/
# (libarchive.mjs + libarchive.wasm + BUILD.md), committed to the repo.
#
# Usage: scripts/libarchive/build.sh [work-dir]
# Needs: git, curl, cmake, make, python3. Downloads ~1 GB (emsdk) the first time.
set -euo pipefail

EMSDK_VERSION=6.0.13
LIBARCHIVE_VERSION=3.8.10
ZLIB_VERSION=1.3.2
BZIP2_VERSION=1.0.8
XZ_VERSION=5.8.4
# sha256 of each source tarball, recorded when the build was first made (2026-10-22).
SOURCES=(
  "https://libarchive.org/downloads/libarchive-${LIBARCHIVE_VERSION}.tar.xz 56cad765459d5ee25a3b761f90c47d99379d28297fbfb72d908880d6b875fb96"
  "https://zlib.net/zlib-${ZLIB_VERSION}.tar.xz d7a0654783a4da529d1bb793b7ad9c3318020af77667bcae35f95d0e42a792f3"
  "https://sourceware.org/pub/bzip2/bzip2-${BZIP2_VERSION}.tar.gz ab5a03176ee106d3f0fa90e381da478ddae405918153cca248e682cd0c4a2269"
  "https://github.com/tukaani-project/xz/releases/download/v${XZ_VERSION}/xz-${XZ_VERSION}.tar.xz 4ce24038fd4221e0d13bc1a2de7a4db56e90b92b3bf75321f6c14be73f65de4b"
)

HERE="$(cd "$(dirname "$0")" && pwd)"
ROOT="$(cd "$HERE/../.." && pwd)"
WORK="$(mkdir -p "${1:-${TMPDIR:-/tmp}/libarchive-build}" && cd "${1:-${TMPDIR:-/tmp}/libarchive-build}" && pwd)"
OUT="$ROOT/src/vendor/libarchive"
PREFIX="$WORK/prefix"
mkdir -p "$WORK/src" "$PREFIX"

# --- Emscripten ---
if [ ! -d "$WORK/emsdk" ]; then git clone --quiet https://github.com/emscripten-core/emsdk.git "$WORK/emsdk"; fi
"$WORK/emsdk/emsdk" install "$EMSDK_VERSION" >/dev/null
"$WORK/emsdk/emsdk" activate "$EMSDK_VERSION" >/dev/null
# shellcheck disable=SC1091
source "$WORK/emsdk/emsdk_env.sh" >/dev/null 2>&1

# --- Sources (checked against the recorded sha256) ---
cd "$WORK/src"
for line in "${SOURCES[@]}"; do
  url="${line% *}"
  sum="${line##* }"
  file="$(basename "$url")"
  [ -f "$file" ] || curl -sSfLo "$file" "$url"
  echo "$sum  $file" | sha256sum -c --quiet -
  rm -rf "${file%.tar.*}"
  tar -xf "$file"
done

CFLAGS_COMMON="-O3 -fPIC"
export CFLAGS="$CFLAGS_COMMON"

# --- zlib ---
cd "$WORK/src/zlib-$ZLIB_VERSION"
emconfigure ./configure --static --prefix="$PREFIX" >/dev/null
emmake make -j"$(nproc)" libz.a >/dev/null
emmake make install >/dev/null

# --- bzip2 ---
cd "$WORK/src/bzip2-$BZIP2_VERSION"
emmake make -j"$(nproc)" libbz2.a CC=emcc AR=emar RANLIB=emranlib CFLAGS="$CFLAGS_COMMON -D_FILE_OFFSET_BITS=64" >/dev/null
mkdir -p "$PREFIX/lib" "$PREFIX/include"
cp libbz2.a "$PREFIX/lib/" && cp bzlib.h "$PREFIX/include/"

# --- xz (liblzma only) ---
cd "$WORK/src/xz-$XZ_VERSION"
emconfigure ./configure --prefix="$PREFIX" --host=wasm32-unknown-emscripten --disable-shared --enable-static \
  --disable-threads --disable-nls --disable-xz --disable-xzdec --disable-lzmadec --disable-lzmainfo \
  --disable-lzma-links --disable-scripts --disable-doc >/dev/null
emmake make -j"$(nproc)" -C src/liblzma >/dev/null
emmake make -C src/liblzma install >/dev/null

# --- libarchive (readers only: no tools, no crypto, no iconv) ---
cd "$WORK/src/libarchive-$LIBARCHIVE_VERSION"
rm -rf wasm-build && mkdir wasm-build && cd wasm-build
emcmake cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_PREFIX_PATH="$PREFIX" -DCMAKE_FIND_ROOT_PATH="$PREFIX" \
  -DBUILD_SHARED_LIBS=OFF -DENABLE_TEST=OFF -DENABLE_INSTALL=OFF \
  -DENABLE_TAR=OFF -DENABLE_CPIO=OFF -DENABLE_CAT=OFF -DENABLE_UNZIP=OFF \
  -DENABLE_ZLIB=ON -DENABLE_BZip2=ON -DENABLE_LZMA=ON \
  -DENABLE_ZSTD=OFF -DENABLE_LZ4=OFF -DENABLE_LIBB2=OFF -DENABLE_LZO=OFF \
  -DENABLE_OPENSSL=OFF -DENABLE_MBEDTLS=OFF -DENABLE_NETTLE=OFF -DENABLE_CNG=OFF \
  -DENABLE_LIBXML2=OFF -DENABLE_EXPAT=OFF -DENABLE_PCREPOSIX=OFF -DENABLE_PCRE2POSIX=OFF -DENABLE_REGEX=OFF \
  -DENABLE_ICONV=OFF -DENABLE_ACL=OFF -DENABLE_XATTR=OFF -DENABLE_WIN32_XMLLITE=OFF \
  -DZLIB_LIBRARY="$PREFIX/lib/libz.a" -DZLIB_INCLUDE_DIR="$PREFIX/include" \
  -DBZIP2_LIBRARIES="$PREFIX/lib/libbz2.a" -DBZIP2_INCLUDE_DIR="$PREFIX/include" \
  -DLIBLZMA_LIBRARY="$PREFIX/lib/liblzma.a" -DLIBLZMA_INCLUDE_DIR="$PREFIX/include" >/dev/null
emmake make -j"$(nproc)" archive_static >/dev/null

# --- Link: the functions libarchive-wasm's wrapper calls, as an ES module ---
FUNCTIONS=(
  archive_version_number archive_version_string archive_version_details
  archive_read_new_memory archive_read_new archive_read_support_filter_all archive_read_support_format_all
  archive_read_open_memory archive_read_next_entry archive_read_has_encrypted_entries archive_read_data
  archive_read_data_skip archive_read_add_passphrase archive_read_free archive_error_string
  archive_entry_filetype archive_entry_pathname_utf8 archive_entry_symlink_utf8 archive_entry_hardlink_utf8
  archive_entry_size archive_entry_atime archive_entry_birthtime archive_entry_ctime archive_entry_mtime
  archive_entry_is_encrypted malloc free
)
EXPORTS="$(printf '"_%s",' "${FUNCTIONS[@]}")"
mkdir -p "$OUT"
emcc -O3 "$HERE/glue.c" "$HERE/read_memory.c" -I"$WORK/src/libarchive-$LIBARCHIVE_VERSION/libarchive" \
  libarchive/libarchive.a "$PREFIX/lib/libz.a" "$PREFIX/lib/libbz2.a" "$PREFIX/lib/liblzma.a" \
  -o "$OUT/libarchive.mjs" \
  -sMODULARIZE=1 -sEXPORT_ES6=1 -sEXPORT_NAME=createLibarchive -sENVIRONMENT=web,worker \
  -sWASM_BIGINT=0 -sALLOW_MEMORY_GROWTH=1 -sMAXIMUM_MEMORY=2GB -sINCOMING_MODULE_JS_API=locateFile,wasmBinary \
  -sEXPORTED_FUNCTIONS="[${EXPORTS%,}]" -sEXPORTED_RUNTIME_METHODS='["cwrap","HEAP8","HEAPU8"]'

cat > "$OUT/BUILD.md" <<INFO
# libarchive para el navegador (generado; no editar)

Generado por \`scripts/libarchive/build.sh\` el $(date -u +%Y-%m-%d) con Emscripten ${EMSDK_VERSION}.

| Componente | Versión | Licencia |
|---|---|---|
| libarchive | ${LIBARCHIVE_VERSION} | BSD-2-Clause |
| zlib | ${ZLIB_VERSION} | Zlib |
| bzip2 | ${BZIP2_VERSION} | bzip2 (BSD-like) |
| xz / liblzma | ${XZ_VERSION} | 0BSD |

- BLAKE2 (para RAR 5): copia incluida en libarchive (`libb2/bundled`, CC0/Apache-2.0).
- `-sWASM_BIGINT=0`: los enteros de 64 bits llegan a JS como `number` (el envoltorio de libarchive-wasm no acepta BigInt).
- Locale \`C.UTF-8\` al iniciar (\`glue.c\`): nombres no ASCII de 7Z y RAR 4 en UTF-8.
- Sin OpenSSL/Nettle/mbedTLS (no descifra: los cifrados se detectan y se rechazan), sin iconv, zstd, lz4, libxml2 ni expat.
- sha256: libarchive.wasm $(sha256sum "$OUT/libarchive.wasm" | cut -d' ' -f1)
INFO
echo "libarchive: $OUT ready ($(du -h "$OUT/libarchive.wasm" | cut -f1) of WASM)"

===== scripts/libarchive/glue.c =====

// Runs once when the module starts: a UTF-8 locale makes libarchive convert
// entry names (UTF-16 in 7z, OEM/Unicode in RAR 4) to UTF-8 instead of
// replacing every non-ASCII letter, which is what the default "C" locale does.
#include <locale.h>

__attribute__((constructor)) static void fileflect_utf8_locale(void) {
  setlocale(LC_ALL, "C.UTF-8");
}

===== scripts/libarchive/read_memory.c =====

// The two helpers libarchive-wasm's JavaScript wrapper expects on top of
// libarchive's API (same names and contracts as its own build).
#include <stddef.h>
#include "archive.h"
#include "archive_entry.h"

// A reader over `size` bytes at `data` with every filter and format enabled;
// NULL if the data can't be opened.
struct archive *archive_read_new_memory(const void *data, size_t size, const char *passphrase) {
  struct archive *a = archive_read_new();
  if (a == NULL) return NULL;
  if (archive_read_support_filter_all(a) != ARCHIVE_OK || archive_read_support_format_all(a) != ARCHIVE_OK ||
      (passphrase != NULL && archive_read_add_passphrase(a, passphrase) != ARCHIVE_OK) ||
      archive_read_open_memory(a, data, size) != ARCHIVE_OK) {
    archive_read_free(a);
    return NULL;
  }
  return a;
}

// The next entry's header, or NULL at the end of the archive or on an error
// (then archive_error_string tells which).
struct archive_entry *archive_read_next_entry(struct archive *a) {
  struct archive_entry *entry;
  int r = archive_read_next_header(a, &entry);
  return r == ARCHIVE_OK || r == ARCHIVE_WARN ? entry : NULL;
}