Mercurial > repos > bcclaywell > argo_navis

annotate venv/lib/python2.7/site-packages/docutils/parsers/rst/tableparser.py @ 0:d67268158946 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
author bcclaywell
date Mon, 12 Oct 2015 17:43:33 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
1 # $Id: tableparser.py 7320 2012-01-19 22:33:02Z milde $
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
2 # Author: David Goodger <goodger@python.org>
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
3 # Copyright: This module has been placed in the public domain.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
4
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
5 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
6 This module defines table parser classes,which parse plaintext-graphic tables
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
7 and produce a well-formed data structure suitable for building a CALS table.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
8
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
9 :Classes:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
10 - `GridTableParser`: Parse fully-formed tables represented with a grid.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
11 - `SimpleTableParser`: Parse simple tables, delimited by top & bottom
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
12 borders.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
13
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
14 :Exception class: `TableMarkupError`
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
15
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
16 :Function:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
17 `update_dict_of_lists()`: Merge two dictionaries containing list values.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
18 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
19
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
20 __docformat__ = 'reStructuredText'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
21
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
22
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
23 import re
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
24 import sys
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
25 from docutils import DataError
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
26 from docutils.utils import strip_combining_chars
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
27
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
28
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
29 class TableMarkupError(DataError):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
30
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
31 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
32 Raise if there is any problem with table markup.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
33
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
34 The keyword argument `offset` denotes the offset of the problem
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
35 from the table's start line.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
36 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
37
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
38 def __init__(self, *args, **kwargs):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
39 self.offset = kwargs.pop('offset', 0)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
40 DataError.__init__(self, *args)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
41
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
42
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
43 class TableParser:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
44
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
45 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
46 Abstract superclass for the common parts of the syntax-specific parsers.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
47 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
48
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
49 head_body_separator_pat = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
50 """Matches the row separator between head rows and body rows."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
51
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
52 double_width_pad_char = '\x00'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
53 """Padding character for East Asian double-width text."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
54
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
55 def parse(self, block):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
56 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
57 Analyze the text `block` and return a table data structure.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
58
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
59 Given a plaintext-graphic table in `block` (list of lines of text; no
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
60 whitespace padding), parse the table, construct and return the data
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
61 necessary to construct a CALS table or equivalent.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
62
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
63 Raise `TableMarkupError` if there is any problem with the markup.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
64 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
65 self.setup(block)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
66 self.find_head_body_sep()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
67 self.parse_table()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
68 structure = self.structure_from_cells()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
69 return structure
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
70
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
71 def find_head_body_sep(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
72 """Look for a head/body row separator line; store the line index."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
73 for i in range(len(self.block)):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
74 line = self.block[i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
75 if self.head_body_separator_pat.match(line):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
76 if self.head_body_sep:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
77 raise TableMarkupError(
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
78 'Multiple head/body row separators '
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
79 '(table lines %s and %s); only one allowed.'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
80 % (self.head_body_sep+1, i+1), offset=i)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
81 else:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
82 self.head_body_sep = i
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
83 self.block[i] = line.replace('=', '-')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
84 if self.head_body_sep == 0 or self.head_body_sep == (len(self.block)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
85 - 1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
86 raise TableMarkupError('The head/body row separator may not be '
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
87 'the first or last line of the table.',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
88 offset=i)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
89
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
90
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
91 class GridTableParser(TableParser):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
92
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
93 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
94 Parse a grid table using `parse()`.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
95
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
96 Here's an example of a grid table::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
97
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
98 +------------------------+------------+----------+----------+
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
99 | Header row, column 1 | Header 2 | Header 3 | Header 4 |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
100 +========================+============+==========+==========+
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
101 | body row 1, column 1 | column 2 | column 3 | column 4 |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
102 +------------------------+------------+----------+----------+
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
103 | body row 2 | Cells may span columns. |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
104 +------------------------+------------+---------------------+
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
105 | body row 3 | Cells may | - Table cells |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
106 +------------------------+ span rows. | - contain |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
107 | body row 4 | | - body elements. |
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
108 +------------------------+------------+---------------------+
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
109
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
110 Intersections use '+', row separators use '-' (except for one optional
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
111 head/body row separator, which uses '='), and column separators use '|'.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
112
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
113 Passing the above table to the `parse()` method will result in the
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
114 following data structure::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
115
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
116 ([24, 12, 10, 10],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
117 [[(0, 0, 1, ['Header row, column 1']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
118 (0, 0, 1, ['Header 2']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
119 (0, 0, 1, ['Header 3']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
120 (0, 0, 1, ['Header 4'])]],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
121 [[(0, 0, 3, ['body row 1, column 1']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
122 (0, 0, 3, ['column 2']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
123 (0, 0, 3, ['column 3']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
124 (0, 0, 3, ['column 4'])],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
125 [(0, 0, 5, ['body row 2']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
126 (0, 2, 5, ['Cells may span columns.']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
127 None,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
128 None],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
129 [(0, 0, 7, ['body row 3']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
130 (1, 0, 7, ['Cells may', 'span rows.', '']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
131 (1, 1, 7, ['- Table cells', '- contain', '- body elements.']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
132 None],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
133 [(0, 0, 9, ['body row 4']), None, None, None]])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
134
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
135 The first item is a list containing column widths (colspecs). The second
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
136 item is a list of head rows, and the third is a list of body rows. Each
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
137 row contains a list of cells. Each cell is either None (for a cell unused
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
138 because of another cell's span), or a tuple. A cell tuple contains four
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
139 items: the number of extra rows used by the cell in a vertical span
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
140 (morerows); the number of extra columns used by the cell in a horizontal
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
141 span (morecols); the line offset of the first line of the cell contents;
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
142 and the cell contents, a list of lines of text.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
143 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
144
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
145 head_body_separator_pat = re.compile(r'\+=[=+]+=\+ *$')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
146
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
147 def setup(self, block):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
148 self.block = block[:] # make a copy; it may be modified
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
149 self.block.disconnect() # don't propagate changes to parent
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
150 self.bottom = len(block) - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
151 self.right = len(block[0]) - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
152 self.head_body_sep = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
153 self.done = [-1] * len(block[0])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
154 self.cells = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
155 self.rowseps = {0: [0]}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
156 self.colseps = {0: [0]}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
157
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
158 def parse_table(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
159 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
160 Start with a queue of upper-left corners, containing the upper-left
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
161 corner of the table itself. Trace out one rectangular cell, remember
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
162 it, and add its upper-right and lower-left corners to the queue of
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
163 potential upper-left corners of further cells. Process the queue in
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
164 top-to-bottom order, keeping track of how much of each text column has
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
165 been seen.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
166
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
167 We'll end up knowing all the row and column boundaries, cell positions
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
168 and their dimensions.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
169 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
170 corners = [(0, 0)]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
171 while corners:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
172 top, left = corners.pop(0)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
173 if top == self.bottom or left == self.right \
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
174 or top <= self.done[left]:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
175 continue
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
176 result = self.scan_cell(top, left)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
177 if not result:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
178 continue
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
179 bottom, right, rowseps, colseps = result
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
180 update_dict_of_lists(self.rowseps, rowseps)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
181 update_dict_of_lists(self.colseps, colseps)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
182 self.mark_done(top, left, bottom, right)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
183 cellblock = self.block.get_2D_block(top + 1, left + 1,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
184 bottom, right)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
185 cellblock.disconnect() # lines in cell can't sync with parent
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
186 cellblock.replace(self.double_width_pad_char, '')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
187 self.cells.append((top, left, bottom, right, cellblock))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
188 corners.extend([(top, right), (bottom, left)])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
189 corners.sort()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
190 if not self.check_parse_complete():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
191 raise TableMarkupError('Malformed table; parse incomplete.')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
192
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
193 def mark_done(self, top, left, bottom, right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
194 """For keeping track of how much of each text column has been seen."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
195 before = top - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
196 after = bottom - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
197 for col in range(left, right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
198 assert self.done[col] == before
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
199 self.done[col] = after
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
200
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
201 def check_parse_complete(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
202 """Each text column should have been completely seen."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
203 last = self.bottom - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
204 for col in range(self.right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
205 if self.done[col] != last:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
206 return False
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
207 return True
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
208
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
209 def scan_cell(self, top, left):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
210 """Starting at the top-left corner, start tracing out a cell."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
211 assert self.block[top][left] == '+'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
212 result = self.scan_right(top, left)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
213 return result
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
214
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
215 def scan_right(self, top, left):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
216 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
217 Look for the top-right corner of the cell, and make note of all column
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
218 boundaries ('+').
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
219 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
220 colseps = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
221 line = self.block[top]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
222 for i in range(left + 1, self.right + 1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
223 if line[i] == '+':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
224 colseps[i] = [top]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
225 result = self.scan_down(top, left, i)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
226 if result:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
227 bottom, rowseps, newcolseps = result
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
228 update_dict_of_lists(colseps, newcolseps)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
229 return bottom, i, rowseps, colseps
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
230 elif line[i] != '-':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
231 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
232 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
233
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
234 def scan_down(self, top, left, right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
235 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
236 Look for the bottom-right corner of the cell, making note of all row
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
237 boundaries.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
238 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
239 rowseps = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
240 for i in range(top + 1, self.bottom + 1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
241 if self.block[i][right] == '+':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
242 rowseps[i] = [right]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
243 result = self.scan_left(top, left, i, right)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
244 if result:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
245 newrowseps, colseps = result
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
246 update_dict_of_lists(rowseps, newrowseps)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
247 return i, rowseps, colseps
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
248 elif self.block[i][right] != '|':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
249 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
250 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
251
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
252 def scan_left(self, top, left, bottom, right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
253 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
254 Noting column boundaries, look for the bottom-left corner of the cell.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
255 It must line up with the starting point.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
256 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
257 colseps = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
258 line = self.block[bottom]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
259 for i in range(right - 1, left, -1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
260 if line[i] == '+':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
261 colseps[i] = [bottom]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
262 elif line[i] != '-':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
263 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
264 if line[left] != '+':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
265 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
266 result = self.scan_up(top, left, bottom, right)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
267 if result is not None:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
268 rowseps = result
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
269 return rowseps, colseps
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
270 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
271
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
272 def scan_up(self, top, left, bottom, right):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
273 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
274 Noting row boundaries, see if we can return to the starting point.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
275 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
276 rowseps = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
277 for i in range(bottom - 1, top, -1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
278 if self.block[i][left] == '+':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
279 rowseps[i] = [left]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
280 elif self.block[i][left] != '|':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
281 return None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
282 return rowseps
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
283
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
284 def structure_from_cells(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
285 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
286 From the data collected by `scan_cell()`, convert to the final data
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
287 structure.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
288 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
289 rowseps = self.rowseps.keys() # list of row boundaries
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
290 rowseps.sort()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
291 rowindex = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
292 for i in range(len(rowseps)):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
293 rowindex[rowseps[i]] = i # row boundary -> row number mapping
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
294 colseps = self.colseps.keys() # list of column boundaries
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
295 colseps.sort()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
296 colindex = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
297 for i in range(len(colseps)):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
298 colindex[colseps[i]] = i # column boundary -> col number map
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
299 colspecs = [(colseps[i] - colseps[i - 1] - 1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
300 for i in range(1, len(colseps))] # list of column widths
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
301 # prepare an empty table with the correct number of rows & columns
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
302 onerow = [None for i in range(len(colseps) - 1)]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
303 rows = [onerow[:] for i in range(len(rowseps) - 1)]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
304 # keep track of # of cells remaining; should reduce to zero
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
305 remaining = (len(rowseps) - 1) * (len(colseps) - 1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
306 for top, left, bottom, right, block in self.cells:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
307 rownum = rowindex[top]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
308 colnum = colindex[left]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
309 assert rows[rownum][colnum] is None, (
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
310 'Cell (row %s, column %s) already used.'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
311 % (rownum + 1, colnum + 1))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
312 morerows = rowindex[bottom] - rownum - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
313 morecols = colindex[right] - colnum - 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
314 remaining -= (morerows + 1) * (morecols + 1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
315 # write the cell into the table
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
316 rows[rownum][colnum] = (morerows, morecols, top + 1, block)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
317 assert remaining == 0, 'Unused cells remaining.'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
318 if self.head_body_sep: # separate head rows from body rows
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
319 numheadrows = rowindex[self.head_body_sep]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
320 headrows = rows[:numheadrows]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
321 bodyrows = rows[numheadrows:]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
322 else:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
323 headrows = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
324 bodyrows = rows
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
325 return (colspecs, headrows, bodyrows)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
326
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
327
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
328 class SimpleTableParser(TableParser):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
329
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
330 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
331 Parse a simple table using `parse()`.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
332
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
333 Here's an example of a simple table::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
334
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
335 ===== =====
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
336 col 1 col 2
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
337 ===== =====
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
338 1 Second column of row 1.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
339 2 Second column of row 2.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
340 Second line of paragraph.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
341 3 - Second column of row 3.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
342
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
343 - Second item in bullet
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
344 list (row 3, column 2).
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
345 4 is a span
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
346 ------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
347 5
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
348 ===== =====
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
349
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
350 Top and bottom borders use '=', column span underlines use '-', column
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
351 separation is indicated with spaces.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
352
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
353 Passing the above table to the `parse()` method will result in the
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
354 following data structure, whose interpretation is the same as for
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
355 `GridTableParser`::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
356
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
357 ([5, 25],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
358 [[(0, 0, 1, ['col 1']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
359 (0, 0, 1, ['col 2'])]],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
360 [[(0, 0, 3, ['1']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
361 (0, 0, 3, ['Second column of row 1.'])],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
362 [(0, 0, 4, ['2']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
363 (0, 0, 4, ['Second column of row 2.',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
364 'Second line of paragraph.'])],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
365 [(0, 0, 6, ['3']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
366 (0, 0, 6, ['- Second column of row 3.',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
367 '',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
368 '- Second item in bullet',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
369 ' list (row 3, column 2).'])],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
370 [(0, 1, 10, ['4 is a span'])],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
371 [(0, 0, 12, ['5']),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
372 (0, 0, 12, [''])]])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
373 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
374
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
375 head_body_separator_pat = re.compile('=[ =]*$')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
376 span_pat = re.compile('-[ -]*$')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
377
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
378 def setup(self, block):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
379 self.block = block[:] # make a copy; it will be modified
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
380 self.block.disconnect() # don't propagate changes to parent
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
381 # Convert top & bottom borders to column span underlines:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
382 self.block[0] = self.block[0].replace('=', '-')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
383 self.block[-1] = self.block[-1].replace('=', '-')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
384 self.head_body_sep = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
385 self.columns = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
386 self.border_end = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
387 self.table = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
388 self.done = [-1] * len(block[0])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
389 self.rowseps = {0: [0]}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
390 self.colseps = {0: [0]}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
391
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
392 def parse_table(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
393 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
394 First determine the column boundaries from the top border, then
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
395 process rows. Each row may consist of multiple lines; accumulate
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
396 lines until a row is complete. Call `self.parse_row` to finish the
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
397 job.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
398 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
399 # Top border must fully describe all table columns.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
400 self.columns = self.parse_columns(self.block[0], 0)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
401 self.border_end = self.columns[-1][1]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
402 firststart, firstend = self.columns[0]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
403 offset = 1 # skip top border
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
404 start = 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
405 text_found = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
406 while offset < len(self.block):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
407 line = self.block[offset]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
408 if self.span_pat.match(line):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
409 # Column span underline or border; row is complete.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
410 self.parse_row(self.block[start:offset], start,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
411 (line.rstrip(), offset))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
412 start = offset + 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
413 text_found = None
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
414 elif line[firststart:firstend].strip():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
415 # First column not blank, therefore it's a new row.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
416 if text_found and offset != start:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
417 self.parse_row(self.block[start:offset], start)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
418 start = offset
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
419 text_found = 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
420 elif not text_found:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
421 start = offset + 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
422 offset += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
423
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
424 def parse_columns(self, line, offset):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
425 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
426 Given a column span underline, return a list of (begin, end) pairs.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
427 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
428 cols = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
429 end = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
430 while True:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
431 begin = line.find('-', end)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
432 end = line.find(' ', begin)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
433 if begin < 0:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
434 break
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
435 if end < 0:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
436 end = len(line)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
437 cols.append((begin, end))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
438 if self.columns:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
439 if cols[-1][1] != self.border_end:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
440 raise TableMarkupError('Column span incomplete in table '
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
441 'line %s.' % (offset+1),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
442 offset=offset)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
443 # Allow for an unbounded rightmost column:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
444 cols[-1] = (cols[-1][0], self.columns[-1][1])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
445 return cols
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
446
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
447 def init_row(self, colspec, offset):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
448 i = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
449 cells = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
450 for start, end in colspec:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
451 morecols = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
452 try:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
453 assert start == self.columns[i][0]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
454 while end != self.columns[i][1]:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
455 i += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
456 morecols += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
457 except (AssertionError, IndexError):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
458 raise TableMarkupError('Column span alignment problem '
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
459 'in table line %s.' % (offset+2),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
460 offset=offset+1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
461 cells.append([0, morecols, offset, []])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
462 i += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
463 return cells
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
464
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
465 def parse_row(self, lines, start, spanline=None):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
466 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
467 Given the text `lines` of a row, parse it and append to `self.table`.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
468
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
469 The row is parsed according to the current column spec (either
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
470 `spanline` if provided or `self.columns`). For each column, extract
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
471 text from each line, and check for text in column margins. Finally,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
472 adjust for insignificant whitespace.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
473 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
474 if not (lines or spanline):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
475 # No new row, just blank lines.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
476 return
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
477 if spanline:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
478 columns = self.parse_columns(*spanline)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
479 span_offset = spanline[1]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
480 else:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
481 columns = self.columns[:]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
482 span_offset = start
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
483 self.check_columns(lines, start, columns)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
484 row = self.init_row(columns, start)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
485 for i in range(len(columns)):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
486 start, end = columns[i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
487 cellblock = lines.get_2D_block(0, start, len(lines), end)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
488 cellblock.disconnect() # lines in cell can't sync with parent
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
489 cellblock.replace(self.double_width_pad_char, '')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
490 row[i][3] = cellblock
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
491 self.table.append(row)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
492
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
493 def check_columns(self, lines, first_line, columns):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
494 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
495 Check for text in column margins and text overflow in the last column.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
496 Raise TableMarkupError if anything but whitespace is in column margins.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
497 Adjust the end value for the last column if there is text overflow.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
498 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
499 # "Infinite" value for a dummy last column's beginning, used to
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
500 # check for text overflow:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
501 columns.append((sys.maxint, None))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
502 lastcol = len(columns) - 2
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
503 # combining characters do not contribute to the column width
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
504 lines = [strip_combining_chars(line) for line in lines]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
505
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
506 for i in range(len(columns) - 1):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
507 start, end = columns[i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
508 nextstart = columns[i+1][0]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
509 offset = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
510 for line in lines:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
511 if i == lastcol and line[end:].strip():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
512 text = line[start:].rstrip()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
513 new_end = start + len(text)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
514 columns[i] = (start, new_end)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
515 main_start, main_end = self.columns[-1]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
516 if new_end > main_end:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
517 self.columns[-1] = (main_start, new_end)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
518 elif line[end:nextstart].strip():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
519 raise TableMarkupError('Text in column margin '
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
520 'in table line %s.' % (first_line+offset+1),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
521 offset=first_line+offset)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
522 offset += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
523 columns.pop()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
524
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
525 def structure_from_cells(self):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
526 colspecs = [end - start for start, end in self.columns]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
527 first_body_row = 0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
528 if self.head_body_sep:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
529 for i in range(len(self.table)):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
530 if self.table[i][0][2] > self.head_body_sep:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
531 first_body_row = i
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
532 break
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
533 return (colspecs, self.table[:first_body_row],
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
534 self.table[first_body_row:])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
535
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
536
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
537 def update_dict_of_lists(master, newdata):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
538 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
539 Extend the list values of `master` with those from `newdata`.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
540
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
541 Both parameters must be dictionaries containing list values.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
542 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
543 for key, values in newdata.items():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
544 master.setdefault(key, []).extend(values)