html-grep

#!/usr/bin/env python3
from lib.html_grep import grep
import argparse
import sys

from typing import List

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

from zulint import lister

USAGE = '''
    This file greps HTML files for keywords in a context-sensitive manner.

    Example:

        $ ./tools/html-grep '#group-pms' '.filters'
        templates/zerver/right_sidebar.html 45

            div.right-sidebar#right-sidebar
                div#group-pm-list
                    ul.filters.scrolling_list#group-pms

    Keyword syntax:

        When supplying keywords, they must correspond to HTML elements,
        classes, or ids.  Typical keywords are:

            li // search for "<li>"
            '.modal' // search for '<foo class="modal">'
            '#intro' // search for '<bar id="intro">'

        Searches are context-sensitive in the sense that the keyword
        may be found among parent tags of a leaf tag, not just the leaf
        tag itself.  (Think of this as being similar to the way a
        browser applies nested CSS styles; it looks at the whole HTML
        tree.)

    Finding files to search:

        The tool currently searches your git repo for files with
        the extension .html or .handlebars., and then it does
        the grep against all those files.

        TODO: allow specific files to be searched.'''

def check_our_files():
    # type: () -> None

    parser = argparse.ArgumentParser(description=USAGE,
                                     formatter_class=argparse.RawTextHelpFormatter)
    parser.add_argument('--modified', '-m',
                        action='store_true', default=False,
                        help='Only check modified files')
    parser.add_argument('--no-filter', '-a',
                        dest='show_all',
                        action='store_true', default=False,
                        help='Show all HTML files (no filtering)')
    parser.add_argument("keyword", nargs="*", help='keyword to be searched')
    options = parser.parse_args()

    if options.show_all:
        keywords = []  # type: List[str]
    else:
        if not options.keyword:
            print('No keywords specified...try --help or --no-filter')
            sys.exit(1)
        else:
            keywords = options.keyword

    files = lister.list_files(
        modified_only=options.modified,
        ftypes=['handlebars', 'html'],
    )

    fns = [fn for fn in files if ('casperjs' not in fn) and ('puppet/zulip' not in fn)]
    grep(fns, set(keywords))

if __name__ == '__main__':
    check_our_files()