def concatenate_aiml(path='aiml-en-us-foundation-alice.v1-9.zip',
outfile='aiml-en-us-foundation-alice.v1-9.aiml'):
"""Strip trailing </aiml> tag and concatenate all valid AIML files found in the ZIP."""
path = find_data_path(path) or path
zf = zipfile.ZipFile(path)
for name in zf.namelist():
if not name.lower().endswith('.aiml'):
continue
with zf.open(name) as fin:
happyending = '#!*@!!BAD'
for i, line in enumerate(fin):
try:
line = line.decode('utf-8').strip()
except UnicodeDecodeError:
line = line.decode('ISO-8859-1').strip()
if line.lower().startswith('</aiml>') or line.lower().endswith(
'</aiml>'):
happyending = (i, line)
break
else:
pass
if happyending != (i, line):
print(
'Invalid AIML format: {}\nLast line (line number {}) was: {}\nexpected "</aiml>"'
.format(name, i, line))
def extract_aiml(path='aiml-en-us-foundation-alice.v1-9'):
""" Extract an aiml.zip file if it hasn't been already and return a list of aiml file paths """
path = find_data_path(path) or path
if os.path.isdir(path):
paths = os.listdir(path)
paths = [os.path.join(path, p) for p in paths]
else:
zf = zipfile.ZipFile(path)
paths = []
for name in zf.namelist():
if '.hg/' in name:
continue
paths.append(zf.extract(name, path=BIGDATA_PATH))
return paths
def create_brain(path='aiml-en-us-foundation-alice.v1-9.zip'):
""" Create an aiml_bot.Bot brain from an AIML zip file or directory of AIML files """
path = find_data_path(path) or path
bot = Bot()
num_templates = bot._brain.template_count
paths = extract_aiml(path=path)
for path in paths:
if not path.lower().endswith('.aiml'):
continue
try:
bot.learn(path)
except AimlParserError:
logger.error(format_exc())
logger.warn('AIML Parse Error: {}'.format(path))
num_templates = bot._brain.template_count - num_templates
logger.info(
'Loaded {} trigger-response pairs.\n'.format(num_templates))
print('Loaded {} trigger-response pairs from {} AIML files.'.format(
bot._brain.template_count, len(paths)))
return bot