cache-bust

#!/usr/bin/env python3
import hashlib
import json
import re
import subprocess
import sys
from pathlib import Path

# Cache busting is fun.

# Note: Firefox only supports inline importmaps (bug 1765745)

ROOT = Path(__file__).parent
INDEX_FILE = ROOT / "index.html"
APP_PATH = ROOT / "app"


def update_importmap():
    # build relative urls with md5 suffixes
    imports = {}
    for filename in sorted(APP_PATH.rglob("*.mjs")):
        name = str(filename.relative_to(APP_PATH)).replace(".mjs", "")
        assert name not in imports
        md5 = hashlib.md5(filename.read_bytes()).hexdigest()
        imports[name] = f"./{filename.relative_to(ROOT)}?{md5}"

    # update inline importmap in index.html
    importmap = json.dumps({"imports": imports}, indent=4, sort_keys=True)
    index = INDEX_FILE.read_text()
    assert '<script type="importmap">' in index
    index = re.sub(
        r'<script type="importmap">[^<]+</script>',
        f'<script type="importmap">{importmap}</script>',
        index,
    )
    INDEX_FILE.write_text(index)

    print(f"{INDEX_FILE.relative_to(ROOT)} importmap updated")


def check_importmap():
    # parse inline importmap from index.html
    rel_index_file = INDEX_FILE.relative_to(ROOT)
    m = re.search(r'<script type="importmap">([^<]+)</script>', INDEX_FILE.read_text())
    if not m:
        print(f"{rel_index_file}: failed to read importmap", file=sys.stderr)
        sys.exit(1)
    try:
        imports = json.loads(m[1])["imports"]
    except (KeyError, ValueError):
        print(f"{rel_index_file}: importmap is malformed", file=sys.stderr)
        sys.exit(1)

    errors = False

    # verify md5
    for ref in sorted(imports.values()):
        filename, map_md5 = ref.split("?", maxsplit=1)
        md5 = hashlib.md5(Path(filename).read_bytes()).hexdigest()
        if md5 != map_md5:
            print(f"{filename[2:]}: invalid md5 in importmap", file=sys.stderr)
            errors = True

    # verify each import statement references a key in the importmap
    for filename in sorted(APP_PATH.rglob("*.mjs")):
        lines = filename.read_text().splitlines()
        for i, line in enumerate(lines):
            m = re.search(r'^import .+ from "([^"]+)";$', line)
            if not m:
                continue
            name = m[1]
            if name not in imports:
                rel_name = filename.relative_to(ROOT)
                print(
                    f"{rel_name}:{i+1}: '{name}' is not a valid importmap key",
                    file=sys.stderr,
                )
                errors = True
    if errors:
        print("run `make format` to correct issues", file=sys.stderr)
        sys.exit(1)
    print("imports ok")


def split_attribute(value):
    if "?" in value:
        name, match_md5 = value.split("?", maxsplit=1)
    else:
        name = value
        match_md5 = ""
    return name, match_md5


def update_attribute(m):
    match_full = m[0]
    match_file, match_md5 = split_attribute(m[1])
    if match_file.endswith((".css", ".mjs")):
        filename = ROOT / match_file
        assert filename.exists()
        md5 = hashlib.md5(filename.read_bytes()).hexdigest()
        if "?" in match_full:
            match_full = match_full.split("?", maxsplit=1)[0]
        match_full = f"{match_full}?{md5}"
    return match_full


def update_attributes():
    # add md5 suffix to css and js attribute references
    html = INDEX_FILE.read_text()
    html = re.sub(r'(?:href|src)="([^"]+)', update_attribute, html, flags=re.A)
    INDEX_FILE.write_text(html)
    print("css and js references updated")


def check_attributes():
    # verify md5 exists and is correct when referencing css and js
    rel_index_file = INDEX_FILE.relative_to(ROOT)
    lines = INDEX_FILE.read_text().splitlines()
    errors = False
    for i, line in enumerate(lines):
        for m in re.finditer(r'(?:href|src)="([^"]+)', line):
            name, match_md5 = split_attribute(m[1])
            if not name.endswith((".css", ".mjs")):
                continue
            filename = ROOT / name
            assert filename.exists(), name
            md5 = hashlib.md5(filename.read_bytes()).hexdigest()
            if md5 != match_md5:
                print(f"{rel_index_file}:{i+1}: invalid md5 for {name}")
                errors = True
    if errors:
        sys.exit(1)
    print("css and js references ok")


def main():
    if len(sys.argv) != 2 or sys.argv[1] not in ("update", "check"):
        print("usage: cache-bust {update|check}", file=sys.stderr)
        print("       update : update cache busting references", file=sys.stderr)
        print("       check  : verify cache busting references", file=sys.stderr)
        sys.exit(1)
    if sys.argv[1] == "update":
        update_importmap()
        update_attributes()
    else:
        check_importmap()
        check_attributes()


if __name__ == "__main__":
    main()