bs4visualizer.py

#! /usr/bin/env python
# -*- coding: utf-8 -*-
__docformat__ = 'reStructuredText'
import sys
from bs4 import BeautifulSoup, NavigableString, Tag

from PyQt5.Qt import (
    QApplication, QMainWindow, QMenuBar, QMenu, QTreeWidget, QTreeWidgetItem)


class MyMain(QMainWindow):
    def __init__(self, parent=None):
        super(MyMain, self).__init__(parent)
        mb = QMenuBar(self)
        mb.setNativeMenuBar(False)
        qm = QMenu('File', self)
        open_act = qm.addAction('Open')
        # qm.addMenu('Foo')
        # mb.addMenu(qm)
        self.setMenuBar(mb)
        tw = QTreeWidget(self)
        tw.setColumnCount(1)
        tw.setHeaderLabel('Elements')
        html_fn = sys.argv[1]
        with open(html_fn) as fh:
            soup = BeautifulSoup(fh.read().decode('utf-8'))

        def make_tree(p, t):
            """
            :type p: QTreeWidgetItem
            :type t: Tag
            """
            for t_ch in t.children:
                #: :type: bs4.Tag
                t_ch = t_ch
                if isinstance(t_ch, NavigableString):
                    ch = repr(unicode(t_ch))
                    t_ch = None
                elif t_ch.hidden:
                    ch = 'Hidden<%r>' % t_ch
                else:
                    if t_ch.namespace:
                        ch = '{%s}%s' % (t_ch.namespace, t_ch.name)
                    else:
                        ch = t_ch.name
                    if t_ch.attrs:
                        ch = '<%s %s>'\
                             % (ch, ' '.join(['%s=%s' % it
                                              for it in t_ch.attrs.items()]))
                ch_twi = QTreeWidgetItem(None, [ch])
                if t_ch is not None:
                    make_tree(ch_twi, t_ch)
                p.addChild(ch_twi)
        twi = QTreeWidgetItem(None, [u'Root'])
        make_tree(twi, soup)
        tw.insertTopLevelItems(0, [twi])
        self.setCentralWidget(tw)


def main(argv):
    app = QApplication(argv)
    mine = MyMain()
    mine.show()
    app.exec_()

if __name__ == '__main__':
    main(sys.argv)