#!/usr/bin/env bash
#
# fetch-reference.sh
#
# Clones every repository and downloads every external HTML/text document
# referenced by HTML_RENDERING_ENGINES.md, RESEARCH.md, and X11_COMPOSITING_WMS.md
# into ./reference/repos and ./reference/docs respectively.
#
# - All git clones use --depth 1 --single-branch (shallow, single branch only).
# - All operations are strictly sequential: one git clone at a time, one curl
#   at a time. No backgrounding, no parallelism.
# - Safe to re-run: existing repos are refreshed with a shallow fetch so they
#   stay at depth 1; existing docs are re-downloaded.

set -u

# Resolve repo root (script lives in <repo>/bin/)
SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" &>/dev/null && pwd)"
REPO_ROOT="$(cd -- "${SCRIPT_DIR}/.." &>/dev/null && pwd)"

REPOS_DIR="${REPO_ROOT}/reference/repos"
DOCS_DIR="${REPO_ROOT}/reference/docs"

mkdir -p "${REPOS_DIR}" "${DOCS_DIR}"

# ---------------------------------------------------------------------------
# Repositories to clone
# ---------------------------------------------------------------------------
REPOS=(
    # HTML/CSS rendering engines
    "https://github.com/chromiumembedded/cef.git"
    "https://github.com/WebKit/WebKit.git"
    "https://github.com/VehicularEpic/webkit2-osr.git"
    "https://github.com/WebPlatformForEmbedded/WPEWebKit.git"
    "https://github.com/WebPlatformForEmbedded/libwpe.git"
    "https://github.com/servo/servo.git"
    "https://github.com/paulrouget/servo-embedding-example.git"
    "https://github.com/mikke89/RmlUi.git"
    "https://github.com/NimbusFox/raylib-RmlUi.git"
    "https://github.com/litehtml/litehtml.git"
    "https://github.com/lexbor/lexbor.git"
    "https://github.com/lexborisov/Modest.git"
    "https://github.com/webview/webview.git"

    # Reference compositor / WM implementations from RESEARCH.md
    "https://github.com/obiwac/x-compositing-wm.git"
    "https://gitlab.freedesktop.org/xorg/app/xcompmgr.git"
    "https://github.com/yshui/picom.git"
    "https://gitlab.com/compiz/compiz-core.git"
    "https://github.com/mackstann/tinywm.git"
    "https://github.com/jichu4n/basic_wm.git"
    "https://github.com/jaelpark/chamferwm.git"
    "https://github.com/deurzen/wzrd.git"

    # Desktop-environment compositing WMs
    "https://github.com/KDE/kwin.git"
    "https://gitlab.gnome.org/GNOME/mutter.git"
    "https://gitlab.xfce.org/xfce/xfwm4.git"
    "https://github.com/linuxmint/muffin.git"
    "https://github.com/mate-desktop/marco.git"
    "https://github.com/elementary/gala.git"
    "https://github.com/Enlightenment/enlightenment.git"
    "https://github.com/JeffHoogland/moksha.git"

    # Standalone compositing WMs
    "https://github.com/compiz-reloaded/compiz.git"
    "https://github.com/noodlylight/fusilli.git"
    "https://github.com/ammen99/fire.git"
    "https://github.com/gschwind/page.git"
    "https://github.com/michaelshiel/eink-wm.git"

    # Picom forks & siblings
    "https://github.com/pijulius/picom.git"
    "https://github.com/jonaburg/picom.git"
    "https://github.com/ibhagwan/picom.git"
    "https://github.com/chjj/compton.git"
    "https://github.com/comick/finalcm.git"
)

# ---------------------------------------------------------------------------
# Documents / web pages to curl
# ---------------------------------------------------------------------------
DOCS=(
    # Project websites referenced by HTML_RENDERING_ENGINES.md
    "https://webkitgtk.org/"
    "https://wpewebkit.org/"
    "https://servo.org/"
    "https://mikke89.github.io/RmlUiDoc/"
    "http://www.litehtml.com/"
    "https://lexbor.com/"

    # Specifications & articles referenced by RESEARCH.md
    "https://www.x.org/releases/X11R7.5/doc/compositeproto/compositeproto.txt"
    "https://www.x.org/releases/current/doc/man/man3/Xcomposite.3.xhtml"
    "https://www.x.org/archive/X11R7.5/doc/damageproto/damageproto.txt"
    "https://registry.khronos.org/OpenGL/extensions/EXT/GLX_EXT_texture_from_pixmap.txt"
    "https://specifications.freedesktop.org/wm-spec/1.5/index.html"
    "https://jichu4n.com/posts/how-x-window-managers-work-and-how-to-write-one-part-i/"
    "https://jichu4n.com/posts/how-x-window-managers-work-and-how-to-write-one-part-ii/"
    "https://wingolog.org/archives/2008/07/26/so-you-want-to-build-a-compositor"
    "https://magcius.github.io/xplain/article/composite.html"
    "https://deepwiki.com/yshui/picom"
    "https://download.nvidia.com/XFree86/Linux-x86_64/396.51/README/xcompositeextension.html"
    "https://registry.khronos.org/OpenGL/extensions/SGI/GLX_SGI_swap_control.txt"
    "https://www.x.org/releases/X11R7.7/doc/fixesproto/fixesproto.txt"
)

# ---------------------------------------------------------------------------
# Clone repos — STRICTLY SEQUENTIAL, one at a time, all shallow (depth 1).
# ---------------------------------------------------------------------------
echo "==> Cloning ${#REPOS[@]} repositories into ${REPOS_DIR} (sequential, depth 1)"
repo_ok=0
repo_fail=0
for url in "${REPOS[@]}"; do
    # Derive a directory name: "<owner>__<repo>" to avoid collisions
    # (e.g. yshui/picom vs pijulius/picom).
    stripped="${url%.git}"
    repo_name="${stripped##*/}"
    owner_path="${stripped%/*}"
    owner="${owner_path##*/}"
    dest_name="${owner}__${repo_name}"
    dest="${REPOS_DIR}/${dest_name}"

    if [[ -d "${dest}/.git" ]]; then
        # Re-run path: refresh via a shallow fetch so the clone stays at depth 1.
        echo "--> [${dest_name}] already present, shallow-refreshing"
        if git -C "${dest}" fetch --depth 1 origin \
            && git -C "${dest}" reset --hard FETCH_HEAD; then
            repo_ok=$((repo_ok + 1))
        else
            echo "    !! refresh failed for ${url}"
            repo_fail=$((repo_fail + 1))
        fi
    else
        echo "--> [${dest_name}] cloning ${url}"
        if git clone --depth 1 --single-branch "${url}" "${dest}"; then
            repo_ok=$((repo_ok + 1))
        else
            echo "    !! clone failed for ${url}"
            repo_fail=$((repo_fail + 1))
        fi
    fi
    # Wait a moment so the previous clone has fully released resources before
    # the next one starts. (Belt-and-suspenders for "one at a time".)
    wait
done

# ---------------------------------------------------------------------------
# Download docs — STRICTLY SEQUENTIAL, one curl at a time.
# ---------------------------------------------------------------------------
echo
echo "==> Downloading ${#DOCS[@]} documents into ${DOCS_DIR} (sequential)"
doc_ok=0
doc_fail=0
for url in "${DOCS[@]}"; do
    # Build a filesystem-safe filename from the URL.
    # Strip scheme, replace path separators, and ensure a sensible extension.
    no_scheme="${url#*://}"
    safe="${no_scheme//\//_}"
    safe="${safe//\?/_}"
    safe="${safe//&/_}"
    safe="${safe//=/_}"
    safe="${safe//:/_}"
    # Trim trailing underscores (from URLs ending with "/").
    safe="${safe%_}"
    # If there's no file extension, append .html so browsers/editors know.
    base="${safe##*/}"
    if [[ "${base}" != *.* ]]; then
        safe="${safe}.html"
    fi
    dest="${DOCS_DIR}/${safe}"

    echo "--> [${safe}] ${url}"
    if curl --fail --location --silent --show-error \
            --user-agent "Mozilla/5.0 (fetch-reference.sh)" \
            --output "${dest}" \
            "${url}"; then
        doc_ok=$((doc_ok + 1))
    else
        echo "    !! download failed for ${url}"
        rm -f "${dest}"
        doc_fail=$((doc_fail + 1))
    fi
    # Belt-and-suspenders: make sure the previous curl has fully exited before
    # the next one is launched.
    wait
done

# ---------------------------------------------------------------------------
# Summary
# ---------------------------------------------------------------------------
echo
echo "==> Done."
echo "    repos:  ${repo_ok} ok, ${repo_fail} failed  (in ${REPOS_DIR})"
echo "    docs:   ${doc_ok} ok, ${doc_fail} failed  (in ${DOCS_DIR})"

# Non-zero exit if anything failed, so callers/CI can detect it.
if (( repo_fail > 0 || doc_fail > 0 )); then
    exit 1
fi