diff --git a/scrapy/trunk/scrapy/tests/test_link_extraction.py b/scrapy/trunk/scrapy/tests/test_link_extraction.py
deleted file mode 100644
index 9a177be55..000000000
--- a/scrapy/trunk/scrapy/tests/test_link_extraction.py
+++ /dev/null
@@ -1,26 +0,0 @@
-import os
-import unittest
-
-from decobot.utils.link_extraction import extract_urls
-
-class LinkExtractionTest(unittest.TestCase):
- def setUp(self):
- self.datadir = os.path.join(os.path.abspath(os.path.dirname(__file__)), "sample_data", "link_extraction")
-
- def test_extract_urls(self):
- text = open(os.path.join(self.datadir, "extract_urls.html")).read()
- links_expected = {'/products/eco-mattress-protector.html': 'Eco Mattress Protector',
- '/products.html': 'Our Products',
- '/products/eco-duvet.html': 'Eco Duvet',
- '/terms-and-conditions.html': 'Terms & Conditions',
- '/returns-policy.html': 'Returns Policy',
- '/products/eco-pillow.html': 'Eco Pillow',
- '/': 'Home',
- '/privacy-policy.html': 'Privacy Policy',
- '/eco-friendly.html': 'Eco-Friendly',
- '/faqs.html': 'FAQs',
- '/contact-us.html': 'Contact Us'}
- self.assertEqual(links_expected, extract_urls(text))
-
-if __name__ == "__main__":
- unittest.main()
diff --git a/scrapy/trunk/scrapy/tests/test_text_extraction.py b/scrapy/trunk/scrapy/tests/test_text_extraction.py
deleted file mode 100644
index 544cdece7..000000000
--- a/scrapy/trunk/scrapy/tests/test_text_extraction.py
+++ /dev/null
@@ -1,111 +0,0 @@
-# -*- coding: utf8 -*-
-import unittest
-from decimal import Decimal
-
-from decobot.utils.text_extraction import clean_markup, unquote_html, read_decimal, read_price
-
-class TextExtractionTest(unittest.TestCase):
- def test_clean_markup(self):
-
- #with html tags
-
- html = u" \xa0 small test
"
-
- self.assertEqual(clean_markup(html),
- u'small test')
- self.assertEqual(clean_markup(html, remove_tags=False, remove_root=False, remove_spaces=False, strip=False),
- u' \xa0 small test
')
- self.assertEqual(clean_markup(html, remove_tags=False, remove_root=True, strip=False),
- u' small test ')
- self.assertEqual(clean_markup(html, remove_tags=False, remove_root=False, remove_spaces=False, strip=True),
- u' \xa0 small test
')
- self.assertEqual(clean_markup(html, remove_tags=False, remove_spaces=True, strip=False),
- u' small test ')
- self.assertEqual(clean_markup(html, remove_tags=True, remove_spaces=False, strip=False),
- u' \xa0 small test ')
- self.assertEqual(clean_markup(html, remove_tags=True, remove_spaces=True, strip=False),
- u' small test ')
- self.assertEqual(clean_markup(html, remove_tags=True, remove_root=True, remove_spaces=True, strip=True),
- u'small test')
-
- #with xml tags
-
- self.assertEqual(clean_markup(u'', xml_doc=True),
- u'')
-
- self.assertEqual(clean_markup(u'', xml_doc=True, remove_tags=False),
- u'')
-
- self.assertEqual(clean_markup(u'http://stephen.digivate2:8080/uploads/suppliers/4/11931_Picture 001.jpg', xml_doc=True),
- u'http://stephen.digivate2:8080/uploads/suppliers/4/11931_Picture 001.jpg')
- self.assertEqual(clean_markup(u' wood ', xml_doc=True),
- u'wood')
- self.assertEqual(clean_markup(u' wood ', strip=False, xml_doc=True),
- u' wood ')
- self.assertEqual(clean_markup(u'woodiron', xml_doc=True),
- u'wood iron')
-
- self.assertEqual(clean_markup(u'hi inside cdata ]]>]]>', xml_doc=True),
- u'hi This is a inside cdata Another CDATA with tags')
-
- self.assertEqual(clean_markup(u'hi inside cdata ]]>]]>', xml_doc=True, remove_tags=False),
- u'hiThis is a inside cdata Another CDATA with tags ')
-
- self.assertEqual(clean_markup(u'hi inside cdata ]]>]]>', remove_cdata=False, xml_doc=True,remove_tags=False),
- u'hi inside cdata ]]>]]>')
-
- self.assertEqual(clean_markup(u'hi inside cdata ]]>]]>', remove_cdata=False, xml_doc=True),
- u'hi inside cdata ]]> ]]>')
-
- self.assertEqual(clean_markup(u' inside cdata ]]>', remove_cdata=False, xml_doc=True),
- u' inside cdata ]]>')
-
- def test_unquote_html(self):
- self.assertEqual(unquote_html(u'As low as £534,456.34!'),
- 'As low as £534,456.34!')
- self.assertEqual(unquote_html('As low as £534,456.34!'),
- 'As low as £534,456.34!')
-
- def test_read_decimal(self):
- self.assertEqual(read_decimal('asdf 234,234.45sdf '),
- Decimal("234234.45"))
- self.assertEqual(read_decimal('asdf 2234 sdf '),
- Decimal("2234"))
- self.assertEqual(read_decimal('947'),
- Decimal("947"))
- self.assertEqual(read_decimal('adsfg'),
- None)
- self.assertEqual(read_decimal('''stained, linseed oil finish, clear glas doors'''),
- None)
-
- def test_read_price(self):
- self.assertEqual(read_price(u'£549.97'),
- Decimal("549.97"))
- self.assertEqual(read_price(u'£5,499.97'),
- Decimal("5499.97"))
- self.assertEqual(read_price(u'Now: £5,499.97 for the last time'),
- Decimal("5499.97"))
- self.assertEqual(read_price(u'As low as 534,456.34!'),
- Decimal("534456.34"))
- self.assertEqual(read_price(u'As low as £534,456.34!'),
- Decimal("534456.34"))
- self.assertEqual(read_price(u'As low as £534,456.34!'),
- Decimal("534456.34"))
- self.assertEqual(read_price('asdf asdfg asdfg '),
- None)
- self.assertEqual(read_price('wrweq qewr -£12', True),
- Decimal("-12"))
- self.assertEqual(read_price('wrweq qewr £12', True),
- Decimal("12"))
- self.assertEqual(read_price('qwe ewqeq -12', True),
- Decimal("0"))
- self.assertEqual(read_price('qwe ewqeq 12', True),
- Decimal("0"))
- self.assertEqual(read_price('qwe ewqeq 12'),
- Decimal("12"))
- self.assertEqual(read_price('117.54 £'),
- Decimal("117.54"))
-
-
-if __name__ == "__main__":
- unittest.main()