mercurial/minirst.py
author Martin Geisler <mg@aragost.com>
Thu, 22 Apr 2010 10:04:53 +0200
changeset 10972 0a2c6948f5f4
parent 10937 a9d5943d2a30
child 10983 287a5cdf7743
permissions -rw-r--r--
doc, minirst: support hg interpreted text role
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     1
# minirst.py - minimal reStructuredText parser
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     2
#
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
     3
# Copyright 2009, 2010 Matt Mackall <mpm@selenic.com> and others
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     4
#
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     5
# This software may be used and distributed according to the terms of the
10263
25e572394f5c Update license to GPLv2+
Matt Mackall <mpm@selenic.com>
parents: 9741
diff changeset
     6
# GNU General Public License version 2 or any later version.
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     7
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     8
"""simplified reStructuredText parser.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
     9
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    10
This parser knows just enough about reStructuredText to parse the
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    11
Mercurial docstrings.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    12
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    13
It cheats in a major way: nested blocks are not really nested. They
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    14
are just indented blocks that look like they are nested. This relies
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    15
on the user to keep the right indentation for the blocks.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    16
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    17
It only supports a small subset of reStructuredText:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    18
9741
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    19
- sections
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    20
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    21
- paragraphs
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    22
9741
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    23
- literal blocks
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    24
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    25
- definition lists
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    26
9741
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    27
- bullet lists (items must start with '-')
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    28
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    29
- enumerated lists (no autonumbering)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    30
9293
e48a48b754d3 minirst: parse field lists
Martin Geisler <mg@lazybytes.net>
parents: 9292
diff changeset
    31
- field lists (colons cannot be escaped)
e48a48b754d3 minirst: parse field lists
Martin Geisler <mg@lazybytes.net>
parents: 9292
diff changeset
    32
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    33
- option lists (supports only long options without arguments)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    34
9741
245689e7f869 minirst: update module docstring
Martin Geisler <mg@lazybytes.net>
parents: 9739
diff changeset
    35
- inline literals (no other inline markup is not recognized)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    36
"""
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    37
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    38
import re, sys, textwrap
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    39
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    40
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    41
def findblocks(text):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    42
    """Find continuous blocks of lines in text.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    43
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    44
    Returns a list of dictionaries representing the blocks. Each block
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    45
    has an 'indent' field and a 'lines' field.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    46
    """
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    47
    blocks = [[]]
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    48
    lines = text.splitlines()
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    49
    for line in lines:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    50
        if line.strip():
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    51
            blocks[-1].append(line)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    52
        elif blocks[-1]:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    53
            blocks.append([])
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    54
    if not blocks[-1]:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    55
        del blocks[-1]
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    56
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    57
    for i, block in enumerate(blocks):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    58
        indent = min((len(l) - len(l.lstrip())) for l in block)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    59
        blocks[i] = dict(indent=indent, lines=[l[indent:] for l in block])
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    60
    return blocks
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    61
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    62
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    63
def findliteralblocks(blocks):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    64
    """Finds literal blocks and adds a 'type' field to the blocks.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    65
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    66
    Literal blocks are given the type 'literal', all other blocks are
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    67
    given type the 'paragraph'.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    68
    """
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    69
    i = 0
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    70
    while i < len(blocks):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    71
        # Searching for a block that looks like this:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    72
        #
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    73
        # +------------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    74
        # | paragraph                    |
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    75
        # | (ends with "::")             |
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    76
        # +------------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    77
        #    +---------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    78
        #    | indented literal block    |
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    79
        #    +---------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    80
        blocks[i]['type'] = 'paragraph'
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
    81
        if blocks[i]['lines'][-1].endswith('::') and i + 1 < len(blocks):
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    82
            indent = blocks[i]['indent']
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
    83
            adjustment = blocks[i + 1]['indent'] - indent
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    84
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    85
            if blocks[i]['lines'] == ['::']:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    86
                # Expanded form: remove block
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    87
                del blocks[i]
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    88
                i -= 1
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    89
            elif blocks[i]['lines'][-1].endswith(' ::'):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    90
                # Partially minimized form: remove space and both
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    91
                # colons.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    92
                blocks[i]['lines'][-1] = blocks[i]['lines'][-1][:-3]
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    93
            else:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    94
                # Fully minimized form: remove just one colon.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    95
                blocks[i]['lines'][-1] = blocks[i]['lines'][-1][:-1]
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    96
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    97
            # List items are formatted with a hanging indent. We must
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    98
            # correct for this here while we still have the original
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
    99
            # information on the indentation of the subsequent literal
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   100
            # blocks available.
9738
f52c4f7a4732 minirst: prepare for general types of bullet lists
Martin Geisler <mg@lazybytes.net>
parents: 9737
diff changeset
   101
            m = _bulletre.match(blocks[i]['lines'][0])
f52c4f7a4732 minirst: prepare for general types of bullet lists
Martin Geisler <mg@lazybytes.net>
parents: 9737
diff changeset
   102
            if m:
f52c4f7a4732 minirst: prepare for general types of bullet lists
Martin Geisler <mg@lazybytes.net>
parents: 9737
diff changeset
   103
                indent += m.end()
f52c4f7a4732 minirst: prepare for general types of bullet lists
Martin Geisler <mg@lazybytes.net>
parents: 9737
diff changeset
   104
                adjustment -= m.end()
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   105
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   106
            # Mark the following indented blocks.
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   107
            while i + 1 < len(blocks) and blocks[i + 1]['indent'] > indent:
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   108
                blocks[i + 1]['type'] = 'literal'
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   109
                blocks[i + 1]['indent'] -= adjustment
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   110
                i += 1
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   111
        i += 1
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   112
    return blocks
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   113
10447
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   114
_bulletre = re.compile(r'(-|[0-9A-Za-z]+\.|\(?[0-9A-Za-z]+\)|\|) ')
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   115
_optionre = re.compile(r'^(--[a-z-]+)((?:[ =][a-zA-Z][\w-]*)?  +)(.*)$')
10065
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   116
_fieldre = re.compile(r':(?![: ])([^:]*)(?<! ):[ ]+(.*)')
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   117
_definitionre = re.compile(r'[^ ]')
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   118
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   119
def splitparagraphs(blocks):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   120
    """Split paragraphs into lists."""
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   121
    # Tuples with (list type, item regexp, single line items?). Order
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   122
    # matters: definition lists has the least specific regexp and must
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   123
    # come last.
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   124
    listtypes = [('bullet', _bulletre, True),
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   125
                 ('option', _optionre, True),
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   126
                 ('field', _fieldre, True),
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   127
                 ('definition', _definitionre, False)]
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   128
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   129
    def match(lines, i, itemre, singleline):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   130
        """Does itemre match an item at line i?
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   131
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   132
        A list item can be followed by an idented line or another list
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   133
        item (but only if singleline is True).
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   134
        """
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   135
        line1 = lines[i]
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   136
        line2 = i + 1 < len(lines) and lines[i + 1] or ''
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   137
        if not itemre.match(line1):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   138
            return False
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   139
        if singleline:
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   140
            return line2 == '' or line2[0] == ' ' or itemre.match(line2)
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   141
        else:
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   142
            return line2.startswith(' ')
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   143
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   144
    i = 0
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   145
    while i < len(blocks):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   146
        if blocks[i]['type'] == 'paragraph':
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   147
            lines = blocks[i]['lines']
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   148
            for type, itemre, singleline in listtypes:
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   149
                if match(lines, 0, itemre, singleline):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   150
                    items = []
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   151
                    for j, line in enumerate(lines):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   152
                        if match(lines, j, itemre, singleline):
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   153
                            items.append(dict(type=type, lines=[],
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   154
                                              indent=blocks[i]['indent']))
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   155
                        items[-1]['lines'].append(line)
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   156
                    blocks[i:i + 1] = items
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   157
                    break
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   158
        i += 1
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   159
    return blocks
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   160
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   161
10065
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   162
_fieldwidth = 12
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   163
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   164
def updatefieldlists(blocks):
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   165
    """Find key and maximum key width for field lists."""
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   166
    i = 0
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   167
    while i < len(blocks):
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   168
        if blocks[i]['type'] != 'field':
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   169
            i += 1
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   170
            continue
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   171
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   172
        keywidth = 0
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   173
        j = i
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   174
        while j < len(blocks) and blocks[j]['type'] == 'field':
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   175
            m = _fieldre.match(blocks[j]['lines'][0])
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   176
            key, rest = m.groups()
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   177
            blocks[j]['lines'][0] = rest
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   178
            blocks[j]['key'] = key
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   179
            keywidth = max(keywidth, len(key))
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   180
            j += 1
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   181
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   182
        for block in blocks[i:j]:
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   183
            block['keywidth'] = keywidth
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   184
        i = j + 1
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   185
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   186
    return blocks
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   187
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   188
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   189
def prunecontainers(blocks, keep):
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   190
    """Prune unwanted containers.
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   191
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   192
    The blocks must have a 'type' field, i.e., they should have been
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   193
    run through findliteralblocks first.
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   194
    """
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   195
    pruned = []
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   196
    i = 0
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   197
    while i + 1 < len(blocks):
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   198
        # Searching for a block that looks like this:
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   199
        #
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   200
        # +-------+---------------------------+
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   201
        # | ".. container ::" type            |
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   202
        # +---+                               |
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   203
        #     | blocks                        |
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   204
        #     +-------------------------------+
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   205
        if (blocks[i]['type'] == 'paragraph' and
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   206
            blocks[i]['lines'][0].startswith('.. container::')):
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   207
            indent = blocks[i]['indent']
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   208
            adjustment = blocks[i + 1]['indent'] - indent
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   209
            containertype = blocks[i]['lines'][0][15:]
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   210
            prune = containertype not in keep
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   211
            if prune:
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   212
                pruned.append(containertype)
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   213
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   214
            # Always delete "..container:: type" block
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   215
            del blocks[i]
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   216
            j = i
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   217
            while j < len(blocks) and blocks[j]['indent'] > indent:
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   218
                if prune:
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   219
                    del blocks[j]
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   220
                    i -= 1 # adjust outer index
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   221
                else:
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   222
                    blocks[j]['indent'] -= adjustment
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   223
                    j += 1
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   224
        i += 1
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   225
    return blocks, pruned
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   226
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   227
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   228
def findsections(blocks):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   229
    """Finds sections.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   230
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   231
    The blocks must have a 'type' field, i.e., they should have been
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   232
    run through findliteralblocks first.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   233
    """
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   234
    for block in blocks:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   235
        # Searching for a block that looks like this:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   236
        #
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   237
        # +------------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   238
        # | Section title                |
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   239
        # | -------------                |
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   240
        # +------------------------------+
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   241
        if (block['type'] == 'paragraph' and
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   242
            len(block['lines']) == 2 and
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   243
            block['lines'][1] == '-' * len(block['lines'][0])):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   244
            block['type'] = 'section'
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   245
    return blocks
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   246
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   247
9623
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   248
def inlineliterals(blocks):
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   249
    for b in blocks:
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   250
        if b['type'] == 'paragraph':
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   251
            b['lines'] = [l.replace('``', '"') for l in b['lines']]
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   252
    return blocks
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   253
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   254
10972
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   255
_hgrolere = re.compile(r':hg:`([^`]+)`')
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   256
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   257
def hgrole(blocks):
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   258
    for b in blocks:
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   259
        if b['type'] == 'paragraph':
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   260
            b['lines'] = [_hgrolere.sub(r'"hg \1"', l) for l in b['lines']]
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   261
    return blocks
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   262
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   263
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   264
def addmargins(blocks):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   265
    """Adds empty blocks for vertical spacing.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   266
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   267
    This groups bullets, options, and definitions together with no vertical
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   268
    space between them, and adds an empty block between all other blocks.
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   269
    """
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   270
    i = 1
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   271
    while i < len(blocks):
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
   272
        if (blocks[i]['type'] == blocks[i - 1]['type'] and
10936
2853c891ac41 minirst: add margin around definition items
Martin Geisler <mg@lazybytes.net>
parents: 10447
diff changeset
   273
            blocks[i]['type'] in ('bullet', 'option', 'field')):
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   274
            i += 1
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   275
        else:
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   276
            blocks.insert(i, dict(lines=[''], indent=0, type='margin'))
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   277
            i += 2
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   278
    return blocks
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   279
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   280
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   281
def formatblock(block, width):
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   282
    """Format a block according to width."""
9417
4c3fb45123e5 util, minirst: do not crash with COLUMNS=0
Martin Geisler <mg@lazybytes.net>
parents: 9293
diff changeset
   283
    if width <= 0:
4c3fb45123e5 util, minirst: do not crash with COLUMNS=0
Martin Geisler <mg@lazybytes.net>
parents: 9293
diff changeset
   284
        width = 78
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   285
    indent = ' ' * block['indent']
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   286
    if block['type'] == 'margin':
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   287
        return ''
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   288
    if block['type'] == 'literal':
9291
cd5b6a11b607 minirst: indent literal blocks with two spaces
Martin Geisler <mg@lazybytes.net>
parents: 9156
diff changeset
   289
        indent += '  '
cd5b6a11b607 minirst: indent literal blocks with two spaces
Martin Geisler <mg@lazybytes.net>
parents: 9156
diff changeset
   290
        return indent + ('\n' + indent).join(block['lines'])
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   291
    if block['type'] == 'section':
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   292
        return indent + ('\n' + indent).join(block['lines'])
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   293
    if block['type'] == 'definition':
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   294
        term = indent + block['lines'][0]
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   295
        hang = len(block['lines'][-1]) - len(block['lines'][-1].lstrip())
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   296
        defindent = indent + hang * ' '
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   297
        text = ' '.join(map(str.strip, block['lines'][1:]))
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   298
        return "%s\n%s" % (term, textwrap.fill(text, width=width,
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   299
                                               initial_indent=defindent,
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   300
                                               subsequent_indent=defindent))
10937
a9d5943d2a30 minirst: removed unnecessary initindent variable
Martin Geisler <mg@lazybytes.net>
parents: 10936
diff changeset
   301
    subindent = indent
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   302
    if block['type'] == 'bullet':
10447
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   303
        if block['lines'][0].startswith('| '):
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   304
            # Remove bullet for line blocks and add no extra
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   305
            # indention.
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   306
            block['lines'][0] = block['lines'][0][2:]
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   307
        else:
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   308
            m = _bulletre.match(block['lines'][0])
e957cc7cbd14 minirst: support line blocks
Martin Geisler <mg@lazybytes.net>
parents: 10444
diff changeset
   309
            subindent = indent + m.end() * ' '
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   310
    elif block['type'] == 'field':
10065
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   311
        keywidth = block['keywidth']
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   312
        key = block['key']
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   313
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   314
        subindent = indent + _fieldwidth * ' '
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   315
        if len(key) + 2 > _fieldwidth:
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   316
            # key too large, use full line width
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   317
            key = key.ljust(width)
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   318
        elif keywidth + 2 < _fieldwidth:
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   319
            # all keys are small, add only two spaces
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   320
            key = key.ljust(keywidth + 2)
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   321
            subindent = indent + (keywidth + 2) * ' '
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   322
        else:
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   323
            # mixed sizes, use fieldwidth for this one
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   324
            key = key.ljust(_fieldwidth)
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   325
        block['lines'][0] = key + block['lines'][0]
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   326
    elif block['type'] == 'option':
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   327
        m = _optionre.match(block['lines'][0])
10064
6f30c35766d6 minirst: don't test regexps twice
Martin Geisler <mg@lazybytes.net>
parents: 10063
diff changeset
   328
        option, arg, rest = m.groups()
6f30c35766d6 minirst: don't test regexps twice
Martin Geisler <mg@lazybytes.net>
parents: 10063
diff changeset
   329
        subindent = indent + (len(option) + len(arg)) * ' '
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   330
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   331
    text = ' '.join(map(str.strip, block['lines']))
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   332
    return textwrap.fill(text, width=width,
10937
a9d5943d2a30 minirst: removed unnecessary initindent variable
Martin Geisler <mg@lazybytes.net>
parents: 10936
diff changeset
   333
                         initial_indent=indent,
9735
97d0d910fa5d minirst: remove unnecessary "elif:" statements
Martin Geisler <mg@lazybytes.net>
parents: 9623
diff changeset
   334
                         subsequent_indent=subindent)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   335
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   336
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   337
def format(text, width, indent=0, keep=None):
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   338
    """Parse and format the text according to width."""
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   339
    blocks = findblocks(text)
9540
cad36e496640 help: un-indent help topics
Martin Geisler <mg@lazybytes.net>
parents: 9417
diff changeset
   340
    for b in blocks:
cad36e496640 help: un-indent help topics
Martin Geisler <mg@lazybytes.net>
parents: 9417
diff changeset
   341
        b['indent'] += indent
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   342
    blocks = findliteralblocks(blocks)
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   343
    blocks, pruned = prunecontainers(blocks, keep or [])
9623
32727ce029de minirst: convert ``foo`` into "foo" upon display
Martin Geisler <mg@lazybytes.net>
parents: 9540
diff changeset
   344
    blocks = inlineliterals(blocks)
10972
0a2c6948f5f4 doc, minirst: support hg interpreted text role
Martin Geisler <mg@aragost.com>
parents: 10937
diff changeset
   345
    blocks = hgrole(blocks)
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   346
    blocks = splitparagraphs(blocks)
10065
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   347
    blocks = updatefieldlists(blocks)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   348
    blocks = findsections(blocks)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   349
    blocks = addmargins(blocks)
10444
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   350
    text = '\n'.join(formatblock(b, width) for b in blocks)
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   351
    if keep is None:
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   352
        return text
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   353
    else:
e99e0e077bc4 minirst: report pruned container types
Martin Geisler <mg@lazybytes.net>
parents: 10443
diff changeset
   354
        return text, pruned
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   355
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   356
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   357
if __name__ == "__main__":
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   358
    from pprint import pprint
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   359
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   360
    def debug(func, *args):
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   361
        blocks = func(*args)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   362
        print "*** after %s:" % func.__name__
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   363
        pprint(blocks)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   364
        print
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   365
        return blocks
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   366
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   367
    text = open(sys.argv[1]).read()
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   368
    blocks = debug(findblocks, text)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   369
    blocks = debug(findliteralblocks, blocks)
10443
62d484a81dfe minirst: support containers
Martin Geisler <mg@lazybytes.net>
parents: 10282
diff changeset
   370
    blocks = debug(prunecontainers, blocks, sys.argv[2:])
10063
a46478b80ea3 minirst: run inlineliterals too in debug mode
Martin Geisler <mg@lazybytes.net>
parents: 9741
diff changeset
   371
    blocks = debug(inlineliterals, blocks)
9737
5f101af4a921 minirst: combine list parsing in one function
Martin Geisler <mg@lazybytes.net>
parents: 9735
diff changeset
   372
    blocks = debug(splitparagraphs, blocks)
10065
a1ae0ed78d1a minirst: improve layout of field lists
Martin Geisler <mg@lazybytes.net>
parents: 10064
diff changeset
   373
    blocks = debug(updatefieldlists, blocks)
9156
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   374
    blocks = debug(findsections, blocks)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   375
    blocks = debug(addmargins, blocks)
c9c7e8cdac9c minimal reStructuredText parser
Martin Geisler <mg@lazybytes.net>
parents:
diff changeset
   376
    print '\n'.join(formatblock(b, 30) for b in blocks)