realign-text-table: tool to rebuild plain-text table

- takes a malformed ASCII-drawn table as stdin or file, for example:
    +-----+-----+
    | foo | bar |
    +-----+-----+
    | a long cell | short |
    | x | longer cell ? |
    +--+--+

- outputs a well-formed table:
    +-------------+---------------+
    |     foo     |      bar      |
    +-------------+---------------+
    | a long cell |     short     |
    |      x      | longer cell ? |
    +-------------+---------------+

README.rst

@@ -33,6 +33,7 @@ This repository hosts various small personal tools.
 * r2w_plugins: 2 rest2web plugins
 * radiodump: circular buffer and dump to file
 * random-line: take a random line from stdin
+* realign-text-table: takes a malformed ASCII-drawn table and redraw borders properly
 * redmine2ical: convert Redmine’s timesheet to iCalendar format
 * set-cachedir: basic tool to create CACHEDIR.TAG files (prevent a folder from being backed up)
 * stickimage: display an image always-on-top like a sticky note

realign-text-table/realign-text-table

+#!/usr/bin/env python3
+# license: Do What the Fuck You Want to Public License v2 [http://www.wtfpl.net/]
+
+# - takes a malformed ASCII-drawn table as stdin or file, for example:
+#    +-----+-----+
+#    | foo | bar |
+#    +-----+-----+
+#    | a long cell | short |
+#    | x | longer cell ? |
+#    +--+--+
+#
+# - outputs a well-formed table:
+#    +-------------+---------------+
+#    |     foo     |      bar      |
+#    +-------------+---------------+
+#    | a long cell |     short     |
+#    |      x      | longer cell ? |
+#    +-------------+---------------+
+
+from fileinput import input
+import re
+
+from prettytable import PrettyTable
+
+
+replacements = {
+    '─=': '-',
+    '│': '|',
+    '┘┐└┌├┬┼┴┤': '+',
+}
+box_chars = ''.join(replacements.keys()) + ''.join(replacements.values())
+box_chars_regex = re.compile(r'[%s\s]+' % re.escape(box_chars))
+
+
+table_regex = re.compile(
+    r"""
+    ^
+    \s*(
+        [-─=]+
+        | [+|│┌└├] [+-┬┼┴─\s]* [+|│┤┘┐]
+        | [|│] .* [|│]
+    )\s*
+    $
+    """,
+    re.X
+)
+
+
+def dump_table(rows):
+    if not rows:
+        return
+
+    table = PrettyTable()
+
+    header, rows = rows[0], rows[1:]
+
+    assert not header[0]
+    assert not header[-1]
+    table.field_names = header[1:-1]
+    for row in rows:
+        assert not row[0]
+        assert not row[-1]
+        table.add_row(row[1:-1])
+    print(table)
+
+
+rows = []
+for line in input():
+    if not table_regex.fullmatch(line):
+        dump_table(rows)
+        rows = []
+        print(line, end='')
+        continue
+
+    if box_chars_regex.fullmatch(line):
+        continue
+
+    line = line.rstrip()
+
+    row = [el.strip() for el in re.split('[│|]', line)]
+    rows.append(row)
+dump_table(rows)

realign-text-table/tests.py

+#!/usr/bin/env python3
+# license: Do What the Fuck You Want to Public License v2 [http://www.wtfpl.net/]
+
+from pathlib import Path
+import subprocess
+
+
+tests = """
+| foo | bar |
+| long text | other                                                |
+| x | y |
+
++---+--+
+| foo | bar |
++-------+--+
+| a | b |
++-------+--+
+| c | d |
++-------+--+
+
+┌─────┬─────┐
+│ foo │ bar │
+├─────┼─────┤
+│  a  │  b  │
+│  c  │  d  │
+└─────┴─────┘
+
+--------
+| 1 | 2 | 3 |
+-----------------------
+| 4 | 5 | 6 |
+========
+| 7 | 8 | nine |
+""".strip().split('\n\n')
+
+
+expecteds = """
++-----------+-------+
+|    foo    |  bar  |
++-----------+-------+
+| long text | other |
+|     x     |   y   |
++-----------+-------+
+
++-----+-----+
+| foo | bar |
++-----+-----+
+|  a  |  b  |
+|  c  |  d  |
++-----+-----+
+
++-----+-----+
+| foo | bar |
++-----+-----+
+|  a  |  b  |
+|  c  |  d  |
++-----+-----+
+
++---+---+------+
+| 1 | 2 |  3   |
++---+---+------+
+| 4 | 5 |  6   |
+| 7 | 8 | nine |
++---+---+------+
+""".strip().split('\n\n')
+
+
+def run_one(test, expected):
+    output = subprocess.check_output(
+        [Path(__file__).resolve().with_name('realign-text-table')],
+        input=test, encoding='utf-8'
+    ).rstrip('\n')
+    expected = expected.rstrip('\n')
+    assert output == expected
+
+
+def test_basic():
+    assert len(tests) == len(expecteds)
+    for test, expected in zip(tests, expecteds):
+        run_one(test, expected)
+        run_one(expected, expected)
+
+
+doc_test = """
+a document
+
+    something
+
+this is a table:
+
+| foo | bar |
+| long text | other                                                |
+| x | y |
+
+end
+"""
+
+doc_expected = """
+a document
+
+    something
+
+this is a table:
+
++-----------+-------+
+|    foo    |  bar  |
++-----------+-------+
+| long text | other |
+|     x     |   y   |
++-----------+-------+
+
+end
+"""
+
+
+def test_embedded():
+    run_one(doc_test, doc_expected)