wrap_rst_table

wrap_rst_table.py

#!/usr/bin/env python

import sys
import textwrap

WRAP_WIDTH = 120


def split_line(line):
    stripped = line.lstrip()
    prefix_len = len(line) - len(stripped)

    core = stripped.rstrip()
    suffix_len = len(stripped) - len(core)

    prefix = line[:prefix_len]
    suffix = stripped[len(core) :]

    return prefix, core, suffix


def parse_row_block(block):
    rows = [line.strip()[1:-1].split("|") for line in block]

    num_cols = len(rows[0])
    cols = [[] for _ in range(num_cols)]

    for row in rows:
        for col, cell in zip(cols, row):
            col.append(cell.rstrip())

    return [" ".join(c.strip() for c in col if c.strip()) for col in cols]


def wrap_cells(cells, col_widths):
    wrapped = []

    for col_width, cell in zip(col_widths, cells):
        width = col_width - 2
        lines = textwrap.wrap(cell, width=width, break_long_words=True, break_on_hyphens=False) or [""]
        wrapped.append(lines)

    max_lines = max(len(c) for c in wrapped)

    for col in wrapped:
        col += [""] * (max_lines - len(col))

    return wrapped


def format_row(wrapped_cols, col_widths, prefix="", suffix=""):
    lines = []
    height = len(wrapped_cols[0])

    for i in range(height):
        parts = []
        for col_width, col in zip(col_widths, wrapped_cols):
            content = col[i]
            parts.append(" " + content.ljust(col_width - 1))
        lines.append(prefix + "|" + "|".join(parts) + "|" + suffix)

    return lines


def compute_last_col_width(col_widths, target_width):
    borders = len(col_widths) + 1
    other = sum(col_widths[:-1])
    return max(5, target_width - (other + borders))


def split_fragment_blocks(lines):
    blocks = []
    current = []

    for line in lines:
        stripped_line = line.strip()
        if stripped_line.startswith("+"):
            if current:
                blocks.append(current)
                current = []
            blocks.append(line)  # keep separator
        elif stripped_line.startswith("|"):
            current.append(stripped_line)

    if current:
        blocks.append(current)

    return blocks


def reformat_fragment(lines, wrap_width):
    first_line = next(l for l in lines if "|" in l)
    prefix, core, suffix = split_line(first_line)

    parts = core[1:-1].split("|")
    col_widths = [len(p) for p in parts]
    col_widths[-1] = compute_last_col_width(col_widths, wrap_width)

    blocks = split_fragment_blocks(lines)

    output = []

    for block in blocks:
        if isinstance(block, str):
            output.append(block)
        else:
            cells = parse_row_block(block)
            wrapped = wrap_cells(cells, col_widths)
            output.extend(format_row(wrapped, col_widths, prefix, suffix))

    return output


def reformat_table(table_text, wrap_width):
    lines = table_text.strip("\n").splitlines()
    return "\n".join(reformat_fragment(lines, wrap_width))


def main():
    print(reformat_table(sys.stdin.read(), wrap_width=WRAP_WIDTH))


if __name__ == "__main__":
    main()