def _html_to_text(html, full_document=True):
"""Convert the given html string into a plain text string. If
full_document is True, only convert data within the body element--ignore
everything else."""
htt = HTMLtoText(full_document)
htt.feed(html)
return str(htt)
def test_two_paragraphs(self):
body = "<p>A paragraph.</p><p>Another paragraph.</p>"
html = self.htmloutline.format(body)
htt = HTMLtoText()
htt.feed(html)
text = str(htt)
self.assertEquals(text, "\nA paragraph.\n\nAnother paragraph.\n")
def test_basic(self):
html = self.htmloutline.format("This is some text.")
htt = HTMLtoText()
htt.feed(html)
text = str(htt)
self.assertEquals(text, "This is some text.")
def test_linebreak(self):
html = self.htmloutline.format("One line.<br/>Two lines.")
htt = HTMLtoText()
htt.feed(html)
text = str(htt)
self.assertEquals(text, "One line.\nTwo lines.")
def test_simple_paragraph(self):
html = self.htmloutline.format("<p>A paragraph.</p>")
htt = HTMLtoText()
htt.feed(html)
text = str(htt)
self.assertEquals(text, "\nA paragraph.\n")