zulip/tools/lib/pretty_print.py

from __future__ import absolute_import
from __future__ import print_function

from typing import Any, Dict, List

from .template_parser import (
    tokenize,
    Token,
)
from six.moves import range

def pretty_print_html(html, num_spaces=4):
    # type: (str, int) -> str
    # We use 1-based indexing for both rows and columns.
    tokens = tokenize(html)
    lines = html.split('\n')

    # We will keep a stack of "start" tags so that we know
    # when HTML ranges end.  Note that some start tags won't
    # be blocks from an indentation standpoint.
    stack = []  # type: List[Dict[str, Any]]

    # Seed our stack with a pseudo entry to make depth calculations
    # easier.
    info = dict(
        block=False,
        depth=-1,
        line=-1,
        token_kind='html_start',
        extra_indent=0)
    stack.append(info)

    # Our main job is to figure out offsets that we use to nudge lines
    # over by.
    offsets = {}  # type: Dict[int, int]

    # Loop through our start/end tokens, and calculate offsets.  As
    # we proceed, we will push/pop info dictionaries on/off a stack.
    for token in tokens:

        if token.kind in ('html_start', 'handlebars_start',
                          'html_singleton', 'django_start'):
            # An HTML start tag should only cause a new indent if we
            # are on a new line.
            if token.tag not in ('extends', 'include', 'else', 'elif'):
                is_block = token.line > stack[-1]['line']

                if is_block:
                    if (token.kind == 'handlebars_start' and
                            stack[-1]['token_kind'] == 'handlebars_start' and
                            not stack[-1]['indenting']):
                        info = stack.pop()
                        info['depth'] = info['depth'] + 1
                        info['indenting'] = True
                        stack.append(info)
                    new_depth = stack[-1]['depth'] + 1
                    extra_indent = stack[-1]['extra_indent']
                    line = lines[token.line - 1]
                    adjustment = len(line)-len(line.lstrip()) + 1
                    offset = (1 + extra_indent + new_depth * num_spaces) - adjustment
                    info = dict(
                        block=True,
                        depth=new_depth,
                        actual_depth=new_depth,
                        line=token.line,
                        token_kind=token.kind,
                        offset=offset,
                        extra_indent=token.col - adjustment + extra_indent,
                        indenting=True
                    )
                    if token.kind == 'handlebars_start':
                        info.update(dict(depth=new_depth - 1, indenting=False))
                else:
                    info = dict(
                        block=False,
                        depth=stack[-1]['depth'],
                        actual_depth=stack[-1]['depth'],
                        line=token.line,
                        token_kind=token.kind,
                        extra_indent=stack[-1]['extra_indent']
                    )
                stack.append(info)
        elif token.kind in ('html_end', 'handlebars_end',
                            'html_singleton_end', 'django_end'):
            info = stack.pop()
            if info['block']:
                # We are at the end of an indentation block.  We
                # assume the whole block was formatted ok before, just
                # possibly at an indentation that we don't like, so we
                # nudge over all lines in the block by the same offset.
                start_line = info['line']
                end_line = token.line
                offsets[start_line] = info['offset']
                offsets[end_line] = info['offset']
                if token.tag != 'pre' and token.kind != 'html_singleton_end' and token.tag != 'script':
                    for line_num in range(start_line + 1, end_line):
                        # Be careful not to override offsets that happened
                        # deeper in the HTML within our block.
                        if line_num not in offsets:
                            line = lines[line_num - 1]
                            new_depth = info['depth'] + 1
                            if (line.lstrip().startswith('{{else}}') or
                                    line.lstrip().startswith('{% else %}') or
                                    line.lstrip().startswith('{% elif')):
                                new_depth = info['actual_depth']
                            extra_indent = info['extra_indent']
                            adjustment = len(line)-len(line.lstrip()) + 1
                            offset = (1 + extra_indent + new_depth * num_spaces) - adjustment
                            offsets[line_num] = offset
                else:
                    for line_num in range(start_line + 1, end_line):
                        if line_num not in offsets:
                            offsets[line_num] = info['offset']

    # Now that we have all of our offsets calculated, we can just
    # join all our lines together, fixing up offsets as needed.
    formatted_lines = []
    for i, line in enumerate(html.split('\n')):
        row = i + 1
        offset = offsets.get(row, 0)
        pretty_line = line
        if line.strip() == '':
            pretty_line = ''
        else:
            if offset > 0:
                pretty_line = (' ' * offset) + pretty_line
            elif offset < 0:
                pretty_line = pretty_line[-1 * offset:]
                assert line.strip() == pretty_line.strip()
        formatted_lines.append(pretty_line)

    return '\n'.join(formatted_lines)