check-templates

#!/usr/bin/env python
import os
import sys
import subprocess

class Record:
    pass

def validate(fn, check_indent=True):
    text = open(fn).read()

    state = Record()

    def NoStartTag(end_tag):
        raise Exception('''
            No start tag
            fn: %s
            end tag:
                %s
                line %d, col %d
            ''' % (fn, end_tag, state.line, state.col))

    def start_tag_matcher(s, start_tag):
        start_line = state.line
        start_col = state.col
        state.depth += 1

        old_matcher = state.matcher
        def f(end_tag):
            problem = None
            if start_tag != end_tag:
                problem = 'Mismatched tag.'
            elif check_indent and state.line > start_line + 1 and state.col != start_col:
                problem = 'Bad indentation.'
            if problem:
                raise Exception('''
                    fn: %s
                    %s
                    start:
                        %s
                        line %d, col %d
                    end tag:
                        %s
                        line %d, col %d
                    ''' % (fn, problem, s, start_line, start_col, end_tag, state.line, state.col))
            state.matcher = old_matcher
            state.depth -= 1
        state.matcher = f

    state.depth = 0
    state.i = 0
    state.line = 1
    state.col = 1
    state.matcher = NoStartTag

    def advance(n):
        for _ in range(n):
            state.i += 1
            if state.i >= 0 and text[state.i - 1] == '\n':
                state.line += 1
                state.col = 1
            else:
                state.col += 1

    def looking_at(s):
        return text[state.i:state.i+len(s)] == s

    while state.i < len(text):

        # HTML tags
        if looking_at("<") and not looking_at("</"):
            s = get_html_tag(text, state.i)
            tag = s[1:-1].split()[0]
            ignore = s.startswith('<!--') or s.endswith('/>') or tag in ['meta', '!DOCTYPE']
            if not ignore:
                start_tag_matcher(s, tag)
            advance(len(s))
            continue

        if looking_at("</"):
            s = get_html_tag(text, state.i)
            end_tag = s[2:-1]
            state.matcher(end_tag)
            advance(len(s))
            continue

        # Handlebar tags
        if looking_at("{{#") or looking_at("{{^"):
            s = get_handlebars_tag(text, state.i)
            tag = s[3:-2].split()[0]
            start_tag_matcher(s, tag)
            advance(len(s))
            continue

        if looking_at("{{/"):
            s = get_handlebars_tag(text, state.i)
            end_tag = s[3:-2]
            state.matcher(end_tag)
            advance(len(s))
            continue

        # Django tags
        if looking_at("{% ") and not looking_at("{% end"):
            s = get_django_tag(text, state.i)
            tag = s[3:-2].split()[0]
            if is_django_block_tag(tag):
                start_tag_matcher(s, tag)
            advance(len(s))
            continue

        if looking_at("{% end"):
            s = get_django_tag(text, state.i)
            end_tag = s[6:-3]
            state.matcher(end_tag)
            advance(len(s))
            continue

        advance(1)

    if state.depth != 0:
        return state.matcher("(NO TAG)")

def is_django_block_tag(tag):
    return tag in [
        'autoescape',
        'block',
        'comment',
        'for',
        'if',
        'ifequal',
        'verbatim',
    ]

def get_handlebars_tag(text, i):
    end = i + 2
    while end < len(text) -1 and text[end] != '}':
        end += 1
    if text[end] != '}' or text[end+1] != '}':
        raise Exception('Tag missing }}')
    s = text[i:end+2]
    return s

def get_django_tag(text, i):
    end = i + 2
    while end < len(text) -1 and text[end] != '%':
        end += 1
    if text[end] != '%' or text[end+1] != '}':
        raise Exception('Tag missing %}')
    s = text[i:end+2]
    return s

def get_html_tag(text, i):
    end = i + 1
    while end < len(text) and text[end] != '>':
        end += 1
    if text[end] != '>':
        raise Exception('Tag missing >')
    s = text[i:end+1]
    return s

def check_our_files():
    git_files = map(str.strip, subprocess.check_output(['git', 'ls-files']).split('\n'))

    # Check all our handlebars templates.
    templates = [fn for fn in git_files if fn.endswith('.handlebars')]
    assert len(templates) >= 10 # sanity check that we are actually doing work
    for fn in templates:
        validate(fn)

    # Django templates are pretty messy now, so we do minimal checking.
    templates = sorted([fn for fn in git_files if fn.endswith('.html') and 'templates' in fn])

    def ok(fn):
        if 'api.html' in fn: return False
        if 'base.html' in fn: return False
        return True

    templates = filter(ok, templates)


    assert len(templates) >= 10 # sanity check that we are actually doing work
    for fn in templates:
        validate(fn, check_indent=False)

if __name__ == '__main__':
    check_our_files()