Convert <TABLE> to TSV.

tabletotsv.py

# coding: utf-8
from html.parser import HTMLParser
import sys

text = sys.stdin.read()
output = ''

class TableToTSV(HTMLParser):
    def initialize(self):
        self.cols = []
        self.rows = []
        self.nowTd = False

    def handle_starttag(self, tag, attrs):
        if tag == 'td':
            self.nowTd = True

    def handle_endtag(self, tag):
        if tag == 'table':
            global output
            for cols in self.rows:
                output += '\t'.join(cols) + '\n'
        elif tag == 'tr':
            self.rows.append(self.cols)
            self.cols = []
        elif tag == 'td':
            self.nowTd = False

    def handle_data(self, data):
        if self.nowTd:
            self.cols.append(data)

parser = TableToTSV()
parser.initialize()
parser.feed(text)
print(output)