|
1
|
1 |
|
import csv |
|
2
|
1 |
|
import io |
|
3
|
1 |
|
import re |
|
4
|
1 |
|
from typing import Any, Dict, List, Optional |
|
5
|
|
|
|
|
6
|
1 |
|
from cleo.io.io import IO |
|
7
|
|
|
|
|
8
|
1 |
|
from sdoc.sdoc2 import in_scope, out_scope |
|
9
|
1 |
|
from sdoc.sdoc2.helper.Enumerable import Enumerable |
|
10
|
1 |
|
from sdoc.sdoc2.node.CaptionNode import CaptionNode |
|
11
|
1 |
|
from sdoc.sdoc2.node.LabelNode import LabelNode |
|
12
|
1 |
|
from sdoc.sdoc2.node.Node import Node |
|
13
|
1 |
|
from sdoc.sdoc2.NodeStore import NodeStore |
|
14
|
|
|
|
|
15
|
|
|
|
|
16
|
1 |
View Code Duplication |
class TableNode(Node): |
|
|
|
|
|
|
17
|
|
|
""" |
|
18
|
|
|
SDoc2 node for table. |
|
19
|
|
|
""" |
|
20
|
|
|
|
|
21
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
22
|
1 |
|
def __init__(self, io_object: IO, options: Dict[str, str]): |
|
23
|
|
|
""" |
|
24
|
|
|
Object constructor. |
|
25
|
|
|
|
|
26
|
|
|
:param OutputStyle io_object: The IO object. |
|
27
|
|
|
:param dict[str,str] options: The options of this table. |
|
28
|
|
|
""" |
|
29
|
|
|
super().__init__(io_object, 'table', options) |
|
30
|
|
|
|
|
31
|
|
|
self.rows: List[List[str]] = [] |
|
32
|
|
|
""" |
|
33
|
|
|
The table rows. |
|
34
|
|
|
""" |
|
35
|
|
|
|
|
36
|
|
|
self.column_headers: List[str] = [] |
|
37
|
|
|
""" |
|
38
|
|
|
The column headers of the table (if any). |
|
39
|
|
|
""" |
|
40
|
|
|
|
|
41
|
|
|
self.alignments: List[Optional[str]] = [] |
|
42
|
|
|
""" |
|
43
|
|
|
The text alignments in the table columns. |
|
44
|
|
|
""" |
|
45
|
|
|
|
|
46
|
|
|
self.caption: Optional[str] = None |
|
47
|
|
|
""" |
|
48
|
|
|
The caption for the table. |
|
49
|
|
|
""" |
|
50
|
|
|
|
|
51
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
52
|
1 |
|
def get_command(self) -> str: |
|
53
|
|
|
""" |
|
54
|
|
|
Returns the command of this node, i.e. table. |
|
55
|
|
|
""" |
|
56
|
|
|
return 'table' |
|
57
|
|
|
|
|
58
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
59
|
1 |
|
def is_block_command(self) -> bool: |
|
60
|
|
|
""" |
|
61
|
|
|
Returns True. |
|
62
|
|
|
""" |
|
63
|
|
|
return True |
|
64
|
|
|
|
|
65
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
66
|
1 |
|
def is_inline_command(self) -> bool: |
|
67
|
|
|
""" |
|
68
|
|
|
Returns False. |
|
69
|
|
|
""" |
|
70
|
|
|
return False |
|
71
|
|
|
|
|
72
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
73
|
1 |
|
def number(self, numbers: Dict[str, Any]) -> None: |
|
74
|
|
|
""" |
|
75
|
|
|
Numbers all numerable nodes such as chapters, sections, figures, and, items. |
|
76
|
|
|
|
|
77
|
|
|
:param dict[str,any] numbers: The current numbers. |
|
78
|
|
|
""" |
|
79
|
|
|
if 'table' not in numbers: |
|
80
|
|
|
numbers['table'] = Enumerable() |
|
81
|
|
|
numbers['table'].generate_numeration(1) |
|
82
|
|
|
numbers['table'].increment_last_level() |
|
83
|
|
|
else: |
|
84
|
|
|
numbers['table'].increment_last_level() |
|
85
|
|
|
|
|
86
|
|
|
self._options['number'] = numbers['table'].get_string() |
|
87
|
|
|
|
|
88
|
|
|
super().number(numbers) |
|
89
|
|
|
|
|
90
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
91
|
1 |
|
def prepare_content_tree(self) -> None: |
|
92
|
|
|
""" |
|
93
|
|
|
Prepares this node for further processing. |
|
94
|
|
|
""" |
|
95
|
|
|
table_nodes = [] |
|
96
|
|
|
|
|
97
|
|
|
for node_id in self.child_nodes: |
|
98
|
|
|
node = in_scope(node_id) |
|
99
|
|
|
|
|
100
|
|
|
if isinstance(node, LabelNode): |
|
101
|
|
|
self.setup_label(node) |
|
102
|
|
|
|
|
103
|
|
|
elif isinstance(node, CaptionNode): |
|
104
|
|
|
self.caption = node.argument |
|
105
|
|
|
|
|
106
|
|
|
else: |
|
107
|
|
|
table_nodes.append(node) |
|
108
|
|
|
|
|
109
|
|
|
node.prepare_content_tree() |
|
110
|
|
|
|
|
111
|
|
|
out_scope(node) |
|
112
|
|
|
|
|
113
|
|
|
self.generate_table(table_nodes) |
|
114
|
|
|
|
|
115
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
116
|
1 |
|
def setup_label(self, node): |
|
117
|
|
|
""" |
|
118
|
|
|
Sets the data of a label to current table. |
|
119
|
|
|
|
|
120
|
|
|
:param sdoc.sdoc2.node.LabelNode.LabelNode node: The label node. |
|
121
|
|
|
""" |
|
122
|
|
|
self._options['id'] = node.argument |
|
123
|
|
|
|
|
124
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
125
|
1 |
|
def generate_table(self, nodes): |
|
126
|
|
|
""" |
|
127
|
|
|
Generates the table node. |
|
128
|
|
|
|
|
129
|
|
|
:param list[sdoc.sdoc2.node.Node.Node] nodes: The list with nodes. |
|
130
|
|
|
""" |
|
131
|
|
|
table_data = TableNode.divide_text_nodes(nodes) |
|
132
|
|
|
split_data = TableNode.split_by_new_lines(table_data) |
|
133
|
|
|
rows = self.generate_output_rows(split_data) |
|
134
|
|
|
|
|
135
|
|
|
if self.has_header(rows): |
|
136
|
|
|
self.column_headers = rows[0] |
|
137
|
|
|
self.rows = rows[2:] |
|
138
|
|
|
self.alignments = self.get_column_alignments(rows[1]) |
|
139
|
|
|
else: |
|
140
|
|
|
self.rows = rows |
|
141
|
|
|
|
|
142
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
143
|
1 |
|
@staticmethod |
|
144
|
1 |
|
def divide_text_nodes(nodes: List[Any]) -> List[Any]: |
|
145
|
|
|
""" |
|
146
|
|
|
Divides text nodes from other type of nodes. |
|
147
|
|
|
|
|
148
|
|
|
:param: list[any] nodes: The list with nodes. |
|
149
|
|
|
|
|
150
|
|
|
:rtype: list[any] |
|
151
|
|
|
""" |
|
152
|
|
|
table_data = [] |
|
153
|
|
|
table_text_repr = '' |
|
154
|
|
|
|
|
155
|
|
|
for node in nodes: |
|
156
|
|
|
if node.get_command() == 'TEXT': |
|
157
|
|
|
table_text_repr += node.argument |
|
158
|
|
|
else: |
|
159
|
|
|
table_data.append(table_text_repr) |
|
160
|
|
|
table_data.append(node) |
|
161
|
|
|
table_text_repr = '' |
|
162
|
|
|
|
|
163
|
|
|
if table_text_repr: |
|
164
|
|
|
table_data.append(table_text_repr) |
|
165
|
|
|
|
|
166
|
|
|
return table_data |
|
167
|
|
|
|
|
168
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
169
|
1 |
|
@staticmethod |
|
170
|
1 |
|
def split_by_new_lines(table_data: List[Any]) -> List[Any]: |
|
171
|
|
|
""" |
|
172
|
|
|
Splits data by newline symbols. |
|
173
|
|
|
|
|
174
|
|
|
:param list[any] table_data: |
|
175
|
|
|
|
|
176
|
|
|
:rtype: list[any] |
|
177
|
|
|
""" |
|
178
|
|
|
split_data = [] |
|
179
|
|
|
|
|
180
|
|
|
for data in table_data: |
|
181
|
|
|
if isinstance(data, str): |
|
182
|
|
|
split_data.append(data.split('\n')) |
|
183
|
|
|
else: |
|
184
|
|
|
split_data.append(data) |
|
185
|
|
|
|
|
186
|
|
|
for data in split_data: |
|
187
|
|
|
if isinstance(data, list): |
|
188
|
|
|
for element in data: |
|
189
|
|
|
if element.isspace() or element == '': |
|
190
|
|
|
data.remove(element) |
|
191
|
|
|
|
|
192
|
|
|
return split_data |
|
193
|
|
|
|
|
194
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
195
|
1 |
|
def generate_output_rows(self, split_data: List[Any]) -> List[List[Any]]: |
|
196
|
|
|
""" |
|
197
|
|
|
Generates the rows for final representation. |
|
198
|
|
|
|
|
199
|
|
|
:param list[any] split_data: The split data. |
|
200
|
|
|
|
|
201
|
|
|
:rtype: list[list[any]] |
|
202
|
|
|
""" |
|
203
|
|
|
separated_data = TableNode.split_by_vertical_separators(split_data) |
|
204
|
|
|
|
|
205
|
|
|
rows = [] |
|
206
|
|
|
|
|
207
|
|
|
for item in separated_data: |
|
208
|
|
|
row = [] |
|
209
|
|
|
|
|
210
|
|
|
for data in item: |
|
211
|
|
|
if data and isinstance(data, str) and not data.isspace(): |
|
212
|
|
|
string = io.StringIO(data) |
|
213
|
|
|
self.parse_vertical_separators(string, row) |
|
214
|
|
|
|
|
215
|
|
|
else: |
|
216
|
|
|
if data: |
|
217
|
|
|
row.append(data) |
|
218
|
|
|
|
|
219
|
|
|
if row: |
|
220
|
|
|
rows.append(row) |
|
221
|
|
|
|
|
222
|
|
|
return rows |
|
223
|
|
|
|
|
224
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
225
|
1 |
|
@staticmethod |
|
226
|
1 |
|
def split_by_vertical_separators(split_data: List[Any]) -> List[List[Any]]: |
|
227
|
|
|
""" |
|
228
|
|
|
Splits data by vertical separators and creates rows with data. |
|
229
|
|
|
|
|
230
|
|
|
:param list[any] split_data: The split data. |
|
231
|
|
|
|
|
232
|
|
|
:rtype: list[list[any]] |
|
233
|
|
|
""" |
|
234
|
|
|
rows = [] |
|
235
|
|
|
row = [] |
|
236
|
|
|
|
|
237
|
|
|
for item in split_data: |
|
238
|
|
|
# If we have a list we pass for each element. |
|
239
|
|
|
# In list we have only text elements. |
|
240
|
|
|
if isinstance(item, list): |
|
241
|
|
|
for element in item: |
|
242
|
|
|
|
|
243
|
|
|
# If element starts with '|' we just add element to row. |
|
244
|
|
|
if element.strip().startswith('|'): |
|
245
|
|
|
row.append(element) |
|
246
|
|
|
|
|
247
|
|
|
# If element ends with '|' we add row to rows list, reset row to empty list |
|
248
|
|
|
# and append there current element. We do it because we know that if string ends with '|', |
|
249
|
|
|
# after it we have non-text element. |
|
250
|
|
|
elif element.strip().endswith('|'): |
|
251
|
|
|
row = TableNode.reset_data(row, rows) |
|
252
|
|
|
row.append(element) |
|
253
|
|
|
|
|
254
|
|
|
# If we have element which not starts and not ends in '|' we do this block. |
|
255
|
|
|
else: |
|
256
|
|
|
# If last added element of row is not string we add row to rows list, reset row to empty. |
|
257
|
|
|
if row and not isinstance(row[-1], str): |
|
258
|
|
|
row = TableNode.reset_data(row, rows) |
|
259
|
|
|
# And just add full element with text like a row to list of rows. |
|
260
|
|
|
rows.append([element]) |
|
261
|
|
|
|
|
262
|
|
|
# Do this block if element is not a list. |
|
263
|
|
|
else: |
|
264
|
|
|
# If last added element not ends with '|' we add row to rows list, and reset row to empty list. |
|
265
|
|
|
if row and not row[-1].strip().endswith('|'): |
|
266
|
|
|
row = TableNode.reset_data(row, rows) |
|
267
|
|
|
# Add item to row. |
|
268
|
|
|
row.append(item) |
|
269
|
|
|
|
|
270
|
|
|
return rows |
|
271
|
|
|
|
|
272
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
273
|
1 |
|
@staticmethod |
|
274
|
1 |
|
def reset_data(row: List[Any], rows: List[List[any]]) -> List[Any]: |
|
275
|
|
|
""" |
|
276
|
|
|
Appends row with data to list of rows, and clear row from any elements. |
|
277
|
|
|
|
|
278
|
|
|
Warning! This method changes original list 'rows'. |
|
279
|
|
|
|
|
280
|
|
|
:param list[any] row: The row with elements |
|
281
|
|
|
:param list[list[any]] rows: The list with many rows. |
|
282
|
|
|
""" |
|
283
|
|
|
rows.append(row) |
|
284
|
|
|
return [] |
|
285
|
|
|
|
|
286
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
287
|
1 |
|
def parse_vertical_separators(self, string: io.StringIO, row: List[any]) -> None: |
|
288
|
|
|
""" |
|
289
|
|
|
Splits row by vertical separator for future output. |
|
290
|
|
|
|
|
291
|
|
|
:param StringIO string: The string which we will separate. |
|
292
|
|
|
:param list[mixed] row: The list with the row in which we append data. |
|
293
|
|
|
""" |
|
294
|
|
|
reader = csv.reader(string, delimiter='|') |
|
295
|
|
|
for line in reader: |
|
296
|
|
|
new_row = self.prune_whitespace(line) |
|
297
|
|
|
|
|
298
|
|
|
for item in new_row: |
|
299
|
|
|
if item: |
|
300
|
|
|
row.append(item) |
|
301
|
|
|
|
|
302
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
303
|
1 |
|
@staticmethod |
|
304
|
1 |
|
def has_header(rows: List[str]) -> bool: |
|
305
|
|
|
""" |
|
306
|
|
|
Returns True if the table has a table header. |
|
307
|
|
|
|
|
308
|
|
|
:param list[str] rows: The second row of the table. |
|
309
|
|
|
""" |
|
310
|
|
|
is_header = True |
|
311
|
|
|
|
|
312
|
|
|
if len(rows) == 1: |
|
313
|
|
|
return False |
|
314
|
|
|
|
|
315
|
|
|
for align in rows[1]: |
|
316
|
|
|
header_part = re.findall(':?---+-*:?', align) |
|
317
|
|
|
if not header_part: |
|
318
|
|
|
is_header = False |
|
319
|
|
|
break |
|
320
|
|
|
|
|
321
|
|
|
return is_header |
|
322
|
|
|
|
|
323
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
324
|
1 |
|
@staticmethod |
|
325
|
1 |
|
def get_column_alignments(row: List[str]) -> List[str]: |
|
326
|
|
|
""" |
|
327
|
|
|
Sets alignments on table columns. |
|
328
|
|
|
|
|
329
|
|
|
:param list[str] row: The row with hyphens for creating column headers. |
|
330
|
|
|
""" |
|
331
|
|
|
alignments = [] |
|
332
|
|
|
for hyphens in row: |
|
333
|
|
|
hyphens = hyphens.strip() |
|
334
|
|
|
if hyphens.startswith(':') and hyphens.endswith(':'): |
|
335
|
|
|
alignments.append('center') |
|
336
|
|
|
elif hyphens.startswith(':'): |
|
337
|
|
|
alignments.append('left') |
|
338
|
|
|
elif hyphens.endswith(':'): |
|
339
|
|
|
alignments.append('right') |
|
340
|
|
|
else: |
|
341
|
|
|
alignments.append('') |
|
342
|
|
|
|
|
343
|
|
|
return alignments |
|
344
|
|
|
|
|
345
|
|
|
# ------------------------------------------------------------------------------------------------------------------ |
|
346
|
1 |
|
@staticmethod |
|
347
|
1 |
|
def prune_whitespace(row: List[str]) -> List[str]: |
|
348
|
|
|
""" |
|
349
|
|
|
Strips whitespaces from the text of an each cell. |
|
350
|
|
|
|
|
351
|
|
|
:param list[str] row: The row with text of an each cell. |
|
352
|
|
|
""" |
|
353
|
|
|
clear_row = [] |
|
354
|
|
|
for item in row: |
|
355
|
|
|
clear_text = item.strip() |
|
356
|
|
|
clear_text = re.sub(r'\s+', ' ', clear_text) |
|
357
|
|
|
clear_row.append(clear_text) |
|
358
|
|
|
|
|
359
|
|
|
return clear_row |
|
360
|
|
|
|
|
361
|
|
|
|
|
362
|
|
|
# ---------------------------------------------------------------------------------------------------------------------- |
|
363
|
|
|
NodeStore.register_block_command('table', TableNode) |
|
364
|
|
|
|