def test_iter_texts(self):
extractor = CommandExtractor()
for doc in iter_html_docs(TEST_DATA_DIR):
nr_txts = 0
for _ in extractor.iter_text_lines(doc):
nr_txts += 1
self.assertEqual(NR_TEXTS[doc.url.url], nr_txts)
def test_iter_commands(self):
extractor = CommandExtractor()
for doc in iter_html_docs(TEST_DATA_DIR):
print(doc.url.url)
for (line, cmd), correct in zip(extractor.iter_commands(doc),
COMMANDS[doc.url.url]):
self.assertEqual((line, cmd), correct)
def test_get_command(self):
ext = CommandExtractor('find')
self.assertEqual(ext.get_command('$ find . -name "*.mp3"'),
'find . -name "*.mp3"')
self.assertIsNone(ext.get_command('# ls -hl'))
self.assertIsNone(ext.get_command('find a file'))
self.assertIsNone(ext.get_command('Find . -name "*.mp3"'))
def test_sudo(self):
ext = CommandExtractor('ls')
self.assertEqual(ext.get_command('# sudo ls -hl'),
'ls -hl')
ext = CommandExtractor('sudo')
self.assertEqual(
ext.get_command('# sudo -u www vi ~www/htdocs/index.html'),
'sudo -u www vi ~www/htdocs/index.html')
def test_is_command(self):
ext = CommandExtractor()
self.assertTrue(ext.is_command(
u'date -s \u201cYYYY-MM-DD HH:MM:SS\u201d'))
self.assertTrue(ext.is_command('uname -a'))
self.assertTrue(ext.is_command('git log'))
self.assertFalse(ext.is_command(''))
self.assertFalse(ext.is_command('when I use:'))
self.assertFalse(ext.is_command('is used.'))
self.assertFalse(ext.is_command('thank you!!!!'))
self.assertFalse(ext.is_command('thanks for sharing'))
self.assertFalse(ext.is_command(
'drwxr-xr-x 3 root root 4,0K maj 5 2014 home'))
self.assertFalse(ext.is_command(
'ls -al %s' % 'a'*ext.MAX_COMMAND_LENGTH))
self.assertFalse(ext.is_command(
("you can use '-c' & '-w' with wc to obtain number of characters"
"and words repectively")))
self.assertFalse(ext.is_command('250 total'))
# TODO
self.assertTrue(ext.is_command('thanx :)'))
def test_is_command_output(self):
ext = CommandExtractor()
self.assertTrue(ext.is_command_output(
'drwxr-xr-x 3 root root 4,0K maj 5 2014 home'))
self.assertFalse(ext.is_command_output('total 0'))
def test_is_command_name(self):
ext = CommandExtractor()
self.assertTrue(ext.is_command_name('ls'))
self.assertTrue(ext.is_command_name('l'))
self.assertTrue(ext.is_command_name('7z'))
self.assertTrue(ext.is_command_name('apt-get'))
self.assertFalse(ext.is_command_name(''))
self.assertFalse(ext.is_command_name('1'))
self.assertFalse(ext.is_command_name('22'))
self.assertFalse(ext.is_command_name('apt-'))
self.assertFalse(ext.is_command_name('-'))
self.assertFalse(ext.is_command_name('-ls'))
self.assertFalse(ext.is_command_name('apt get'))
self.assertFalse(ext.is_command_name('APT-GET'))
def test_has_wanted_command(self):
ext = CommandExtractor('xargs')
self.assertTrue(ext.has_wanted_command(
u'find ./music -name "*.mp3" -print0 | xargs -0 ls'))
self.assertFalse(ext.has_wanted_command(
u'find ./music -name "*.mp3" -print0 | grep xargs'))
ext = CommandExtractor(['du', 'mv'])
self.assertTrue(ext.has_wanted_command(
u'du -h -s *'))
self.assertFalse(ext.has_wanted_command(
u'ls -hl'))
ext = CommandExtractor(['git commit'])
self.assertTrue(ext.has_wanted_command(
u'git commit --amend'))
self.assertFalse(ext.has_wanted_command(
u'git pull origin master'))
def test_iter_commands(self):
extractor = CommandExtractor()
for doc in iter_html_docs(TEST_DATA_DIR):
for (line, cmd), correct in zip(extractor.iter_commands(doc),
COMMANDS[doc.url.url]):
self.assertEqual((line, cmd), correct)
