zulip/tools/check-templates

#!/usr/bin/env python3
from lib.template_parser import validate
from lib.html_branches import build_id_dict
from lib.pretty_print import validate_indent_html
import argparse
import sys
import logging

from six.moves import filter

# check for the venv
from lib import sanity_check
sanity_check.check_venv(__file__)

import lister
from typing import cast, Callable, Dict, Iterable, List

EXCLUDED_FILES = [
    ## Test data Files for testing modules in tests
    "tools/tests/test_template_data",
]

def check_our_files(modified_only, all_dups, targets):
    # type: (bool, bool, List[str]) -> None
    by_lang = cast(
        Dict[str, List[str]],
        lister.list_files(
            targets=targets,
            modified_only=args.modified,
            ftypes=['handlebars', 'html'],
            group_by_ftype=True, exclude=EXCLUDED_FILES))

    check_handlebar_templates(by_lang['handlebars'])
    check_html_templates(by_lang['html'], args.all_dups)

def check_html_templates(templates, all_dups):
    # type: (Iterable[str], bool) -> None
    # Our files with .html extensions are usually for Django, but we also
    # have a few static .html files.
    #
    # We also have .html files that we vendored from Casper.
    # The casperjs files use HTML5 (whereas Zulip prefers XHTML), and
    # there are also cases where Casper deliberately uses invalid HTML,
    # so we exclude them from our linter.
    logging.basicConfig(format='%(levelname)s:%(message)s')
    templates = filter(
        lambda fn: ('casperjs' not in fn),
        templates)
    templates = sorted(list(templates))

    template_id_dict = build_id_dict(templates)
    # TODO: Clean up these cases of duplicate ids in the code
    IGNORE_IDS = [
        'api-example-tabs',
        'errors',
        'email',
        'messages',
        'registration',
        'pw_strength',
        'id_password',
        'top_navbar',
        'id_email',
        'id_terms',
        'send_confirm',
        'register',
    ]
    bad_ids_dict = {ids: fns for ids, fns in template_id_dict.items()
                    if (ids not in IGNORE_IDS) and len(fns) > 1}

    if all_dups:
        ignorable_ids_dict = {ids: fns for ids, fns in template_id_dict.items()
                              if ids in IGNORE_IDS and len(fns) > 1}

        for ids, fns in ignorable_ids_dict.items():
            logging.warning("Duplicate ID(s) detected :Id '" + ids +
                            "' present at following files:")
            for fn in fns:
                print(fn)

    for ids, fns in bad_ids_dict.items():
        logging.error("Duplicate ID(s) detected :Id '" + ids +
                      "' present at following files:")
        for fn in fns:
            print(fn)

    if list(bad_ids_dict.keys()):
        print('Exiting--please clean up all duplicates before running this again.')
        sys.exit(1)

    for fn in templates:
        # Many of our Django templates have strange indentation.  The
        # indentation errors are often harmless, even stylistically
        # harmless, but they tend to be in files that might be old
        # and might eventually require more scrutiny for things like
        # localization.  See github #1236.
        bad_files = [
            'templates/corporate/zephyr-mirror.html',
            'templates/zerver/api.html',
            'templates/zerver/api_endpoints.html',
            'templates/zerver/features.html',
            'templates/zerver/home.html',
            'templates/zerver/login.html',
            'templates/zerver/markdown_help.html',
            'zerver/webhooks/taiga/doc.html',
            'zerver/webhooks/zendesk/doc.html',
        ]
        validate(fn=fn, check_indent=(fn not in bad_files))

    # Ignore these files since these have not been cleaned yet :/
    IGNORE_FILES = [
        'static/html/5xx.html',
        'templates/analytics/activity.html',
        'templates/analytics/stats.html',
        'templates/corporate/zephyr-mirror.html',
        'templates/zerver/api.html',
        'templates/zerver/compose.html',
        'templates/zerver/hello.html',
        'templates/zerver/home.html',
        'templates/zerver/index.html',
        'templates/zerver/keyboard_shortcuts.html',
        'templates/zerver/login.html',
        'templates/zerver/markdown_help.html',
        'templates/zerver/navbar.html',
        'zerver/webhooks/stripe/doc.html',
        'zerver/webhooks/wordpress/doc.html',
    ]
    # TODO: Clean these files
    for fn in templates:
        if fn not in IGNORE_FILES:
            if not validate_indent_html(fn):
                sys.exit(1)

def check_handlebar_templates(templates):
    # type: (Iterable[str]) -> None
    # Check all our handlebars templates.
    templates = [fn for fn in templates if fn.endswith('.handlebars')]

    IGNORE_FILES = [
        # TODO: Add some exclude mechanism for the line-wrapping issue here.
        'static/templates/recipient_row.handlebars',
    ]

    for fn in templates:
        if fn in IGNORE_FILES:
            continue
        validate(fn=fn, check_indent=True)

    for fn in templates:
        if fn in IGNORE_FILES:
            continue
        if not validate_indent_html(fn):
            sys.exit(1)

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('-m', '--modified',
                        action='store_true', default=False,
                        help='only check modified files')
    parser.add_argument('--all-dups',
                        action="store_true", default=False,
                        help='Run lint tool to detect duplicate ids on ignored files as well')
    parser.add_argument('targets', nargs=argparse.REMAINDER)
    args = parser.parse_args()
    check_our_files(args.modified, args.all_dups, args.targets)