#!/bin/bash
set -euo pipefail

# Fix for AltSnap repo encoding issues.
#
# We want to use a .gitattributes with working-tree-encoding=utf-16le-bom, as
# this enables git diffs on UTF-16 files, which is nice. Also it explicitly
# declares the required encoding for these files, also nice. However
# working-tree-encoding always requires that the blobs be stored as UTF-8 in
# the git repository, regardless of what encoding they will ultimately be
# checked out to. In this repo though, most of the blobs were committed as raw
# UTF-16LE — never converted to UTF-8. When the .gitattributes was added later,
# it confuses git. It results in warnings and results in working-tree-encoding
# not function properly. A broken .gitattributes makes it difficult to correct
# the encoding of the blobs to UTF-8, so we have to do it in multiple steps,
# the first of which is to temporarily remove .gitattributes (we will restore it
# later).
#
# This script:
#   1. Removes .gitattributes so git treats files as raw bytes
#   2. Converts the UTF-16LE blobs to UTF-8 (the format git expects)
#   3. Normalizes line endings to DOS (CRLF)
#   4. Restores .gitattributes (with eol=crlf added)
#   5. Lets git apply the encoding rules, producing proper UTF-16LE-BOM
#      working tree files from the now-correct UTF-8 blobs

REPO_URL="https://github.com/RamonUnch/AltSnap"
CLONE_DIR="AltSnap"

# Fresh clone (expect encoding errors — that's the bug we're fixing)
git clone "$REPO_URL" "$CLONE_DIR"
cd "$CLONE_DIR"

# Step 1: Remove .gitattributes and commit
git rm .gitattributes
git commit -m "remove .gitattributes"

# Re-checkout files as raw bytes (without .gitattributes filtering)
git rm --cached -r . > /dev/null
git reset HEAD > /dev/null 2>&1
git checkout -- .

# Step 2: Convert UTF-16 to UTF-8 and normalize line endings
# Using -f utf-16 (not utf-16le) so iconv reads and consumes the BOM.
# ko_KR.ini is already UTF-8 so skip the iconv step for it.
for f in AltSnap.dni Lang/*; do
    if [ "$f" != "Lang/ko_KR.ini" ]; then
        iconv -f utf-16 -t utf-8 "$f" > "$f.tmp"
        mv "$f.tmp" "$f"
    fi
    unix2dos -e "$f"
done

git add .
git commit -m "normalize git-internal encoding to UTF-8 and line endings to DOS"

# Step 3: Restore .gitattributes with an additional specifier of eol=crlf to 
# enforce Windows compatibility and avoid EOL conversion warnings.
cat > .gitattributes << 'EOF'
*.dni diff merge text eol=crlf linguist-language=ini working-tree-encoding=utf-16le-bom
Lang/* diff merge text eol=crlf linguist-language=ini working-tree-encoding=utf-16le-bom
EOF

git add .gitattributes
git commit -m "add back .gitattributes with eol=crlf"

# Step 4: Force git to re-checkout with the new .gitattributes rules
# Clearing the index is necessary — git won't re-apply working-tree-encoding
# if it thinks the checkout is already current.
git rm --cached -r . > /dev/null 2>&1
git reset --hard HEAD

# Step 5: Commit the re-encoded working tree
git add .
git commit -m "allow .gitattributes to take effect"

# Verify
echo ""
echo "=== Verification ==="
unix2dos -idumbteph AltSnap.dni Lang/*

# You can clone my feature branch as a way to validate the effect of the PR, 
# since the PR's content is identical to the feature branch (it's literally symlinked by GitHub:)
git clone -b fix/utf16le-line-endings https://github.com/mike-clark-8192/AltSnap AS-MC # no warnings
git clone https://github.com/RamonUnch/AltSnap AS-RU # gives warnings