def sentences(s):
"""Split the string s into a list of sentences."""
try: s+""
except: raise TypeError, "s must be a string"
s = u" ".join(splitChinese(s))
#TODO do better for normalChinese
s = normalChinese(s)
pos = 0
sentenceList = []
l = len(s)
while pos < l:
try: p = s.index('.', pos)
except: p = l+1
try: q = s.index('?', pos)
except: q = l+1
try: e = s.index('!', pos)
except: e = l+1
try: f = s.index('~', pos)
except: f = l+1
end = min(p,q,e,f)
sentenceList.append( s[pos:end].strip() )
pos = end+1
# If no sentences were found, return a one-item list containing
# the entire input string.
if len(sentenceList) == 0: sentenceList.append(s)
return sentenceList
def sentences(s):
"""Split the string s into a list of sentences."""
try:
s + ""
except:
raise TypeError, "s must be a string"
pos = 0
sentenceList = []
l = len(s)
while pos < l:
try:
p = s.index('.', pos)
except:
p = l + 1
try:
q = s.index('?', pos)
except:
q = l + 1
try:
e = s.index('!', pos)
except:
e = l + 1
end = min(p, q, e)
sentenceList.append(s[pos:end].strip())
pos = end + 1
# If no sentences were found, return a one-item list containing
# the entire input string.
if len(sentenceList) == 0: sentenceList.append(s)
# auto convert chinese
return map(lambda s: u' '.join(splitChinese(s)), sentenceList)
def _respond(self, input, sessionID):
"""Private version of respond(), does the real work."""
if len(input) == 0:
return ""
# guard against infinite recursion
inputStack = self.getPredicate(self._inputStack, sessionID)
if len(inputStack) > self._maxRecursionDepth:
if self._verboseMode:
err = "WARNING: maximum recursion depth exceeded (input='%s')" % input.encode(
self._textEncoding, 'replace')
sys.stderr.write(err)
return ""
# push the input onto the input stack
inputStack = self.getPredicate(self._inputStack, sessionID)
inputStack.append(input)
self.setPredicate(self._inputStack, inputStack, sessionID)
# run the input through the 'normal' subber
input = u' '.join(splitChinese(input))
if input.find(" ") < 0:
input = " " + input
subbedInput = u" ".join(
self._subbers['normal'].sub(input).strip().split())
# fetch the bot's previous response, to pass to the match()
# function as 'that'.
outputHistory = self.getPredicate(self._outputHistory, sessionID)
try:
that = outputHistory[-1]
except IndexError:
that = ""
subbedThat = self._subbers['normal'].sub(that)
# fetch the current topic
topic = self.getPredicate("topic", sessionID)
subbedTopic = self._subbers['normal'].sub(topic)
# Determine the final response.
response = ""
elem = self._brain.match(subbedInput, subbedThat, subbedTopic)
if elem is None:
if self._verboseMode:
err = "WARNING: No match found for input: %s\n" % input.encode(
self._textEncoding)
sys.stderr.write(err)
else:
# Process the element into a response string.
response += self._processElement(elem, sessionID).strip()
response += " "
response = response.strip()
# pop the top entry off the input stack.
inputStack = self.getPredicate(self._inputStack, sessionID)
inputStack.pop()
self.setPredicate(self._inputStack, inputStack, sessionID)
return response
def _respond(self, input, sessionID):
"""Private version of respond(), does the real work."""
if len(input) == 0:
return ""
# guard against infinite recursion
inputStack = self.getPredicate(self._inputStack, sessionID)
if len(inputStack) > self._maxRecursionDepth:
if self._verboseMode:
err = "WARNING: maximum recursion depth exceeded (input='%s')" % input.encode(self._textEncoding, 'replace')
sys.stderr.write(err)
return ""
# push the input onto the input stack
inputStack = self.getPredicate(self._inputStack, sessionID)
inputStack.append(input)
self.setPredicate(self._inputStack, inputStack, sessionID)
# run the input through the 'normal' subber
input=u' '.join(splitChinese(input))
if input.find(" ") < 0 :
input=" "+input
subbedInput = u" ".join(self._subbers['normal'].sub(input).strip().split())
# fetch the bot's previous response, to pass to the match()
# function as 'that'.
outputHistory = self.getPredicate(self._outputHistory, sessionID)
try: that = outputHistory[-1]
except IndexError: that = ""
subbedThat = self._subbers['normal'].sub(that)
# fetch the current topic
topic = self.getPredicate("topic", sessionID)
subbedTopic = self._subbers['normal'].sub(topic)
# Determine the final response.
response = ""
elem = self._brain.match(subbedInput, subbedThat, subbedTopic)
if elem is None:
if self._verboseMode:
err = "WARNING: No match found for input: %s\n" % input.encode(self._textEncoding)
sys.stderr.write(err)
else:
# Process the element into a response string.
response += self._processElement(elem, sessionID).strip()
response += " "
response = response.strip()
# pop the top entry off the input stack.
inputStack = self.getPredicate(self._inputStack, sessionID)
inputStack.pop()
self.setPredicate(self._inputStack, inputStack, sessionID)
return response
def _processSrai(self,elem, sessionID):
"""Process a <srai> AIML element.
<srai> elements recursively process their contents, and then
pass the results right back into the AIML interpreter as a new
piece of input. The results of this new input string are
returned.
"""
newInput = ""
for e in elem[2:]:
newInput += self._processElement(e, sessionID)
newInput = u' '.join(splitChinese(newInput))
return self._respond(newInput, sessionID)
def sentences(s):
"""Split the string s into a list of sentences."""
try: s+""
except: raise TypeError, "s must be a string"
pos = 0
sentenceList = []
l = len(s)
while pos < l:
try: p = s.index('.', pos)
except: p = l+1
try: q = s.index('?', pos)
except: q = l+1
try: e = s.index('!', pos)
except: e = l+1
end = min(p,q,e)
sentenceList.append( s[pos:end].strip() )
pos = end+1
# If no sentences were found, return a one-item list containing
# the entire input string.
if len(sentenceList) == 0: sentenceList.append(s)
# auto convert chinese
return map(lambda s: u' '.join(splitChinese(s)), sentenceList)
def _endElement(self, name):
"""验证AIML结束元素在当前上下文中是否有效。 如果遇到非法的结束元素，则引发AimlParserError。 """
if name == "aiml":
# </aiml> 标签只有在 InsideAiml 状态才是合法的
if self._state != self._STATE_InsideAiml:
raise AimlParserError("Unexpected </aiml> tag " +
self._location())
self._state = self._STATE_OutsideAiml
self._whitespaceBehaviorStack.pop()
elif name == "topic":
# </topic> 标签只有在InsideAiml 状态, 而且 _insideTopic 为 true才是合法的。
if self._state != self._STATE_InsideAiml or not self._insideTopic:
raise AimlParserError("Unexpected </topic> tag " +
self._location())
self._insideTopic = False
self._currentTopic = u""
elif name == "category":
# </category> 标签只有在 AfterTemplate 状态才是合法的
if self._state != self._STATE_AfterTemplate:
raise AimlParserError("Unexpected </category> tag " +
self._location())
self._state = self._STATE_InsideAiml
# 结束当前类别。 将当前 pattern/ that / topic和元素存储在类别字典中。
#【注意：这里修改了当前模式，用中文分割结果做了替换。。】
self._currentPattern = u' '.join(splitChinese(
self._currentPattern))
key = (self._currentPattern.strip(), self._currentThat.strip(),
self._currentTopic.strip())
self.categories[key] = self._elemStack[-1]
self._whitespaceBehaviorStack.pop()
elif name == "pattern":
# </pattern> 标签只有在 InsidePattern 状态才是合法的。
if self._state != self._STATE_InsidePattern:
raise AimlParserError("Unexpected </pattern> tag " +
self._location())
self._state = self._STATE_AfterPattern
elif name == "that" and self._state == self._STATE_InsideThat:
# </ that>标签只允许在<template>元素内部，或InsideThat状态下。本条款处理后一种情况。
self._state = self._STATE_AfterThat
elif name == "template":
# </template> 标签只允许在 InsideTemplate 状态出现。
if self._state != self._STATE_InsideTemplate:
raise AimlParserError("Unexpected </template> tag " +
self._location())
self._state = self._STATE_AfterTemplate
self._whitespaceBehaviorStack.pop()
elif self._state == self._STATE_InsidePattern:
# 特定的标签允许在 <pattern> 元素内部出现。
if name not in ["bot"]:
raise AimlParserError(("Unexpected </%s> tag " % name) +
self._location())
elif self._state == self._STATE_InsideThat:
# 特定的标签允许在 <that> 元素内部出现.
if name not in ["bot"]:
raise AimlParserError(("Unexpected </%s> tag " % name) +
self._location())
elif self._state == self._STATE_InsideTemplate:
# 当前模板内的元素结束。 将堆栈顶部的元素追加到下面的元素上。
elem = self._elemStack.pop()
self._elemStack[-1].append(elem)
self._whitespaceBehaviorStack.pop()
# 如果元素是一个条件，那么也可以从foundDefaultLiStack中弹出一个条目。
if elem[0] == "condition": self._foundDefaultLiStack.pop()
else:
# 意外的关闭标签
raise AimlParserError(("Unexpected </%s> tag " % name) +
self._location())
def _endElement(self, name):
"""Verify that an AIML end element is valid in the current
context.
Raises an AimlParserError if an illegal end element is encountered.
"""
if name == "aiml":
# </aiml> tags are only legal in the InsideAiml state
if self._state != self._STATE_InsideAiml:
raise AimlParserError, "Unexpected </aiml> tag "+self._location()
self._state = self._STATE_OutsideAiml
self._whitespaceBehaviorStack.pop()
elif name == "topic":
# </topic> tags are only legal in the InsideAiml state, and
# only if _insideTopic is true.
if self._state != self._STATE_InsideAiml or not self._insideTopic:
raise AimlParserError, "Unexpected </topic> tag "+self._location()
self._insideTopic = False
self._currentTopic = u""
elif name == "category":
# </category> tags are only legal in the AfterTemplate state
if self._state != self._STATE_AfterTemplate:
raise AimlParserError, "Unexpected </category> tag "+self._location()
self._state = self._STATE_InsideAiml
# End the current category. Store the current pattern/that/topic and
# element in the categories dictionary.
self._currentPattern = u' '.join(splitChinese(self._currentPattern))
for template in self._elemStack:
key = (self._currentPattern.strip(), self._currentThat.strip(),
# template[1].strip())
self._currentTopic.strip()+" "+template[1].strip())
self.categories[key] = template
logging.info("-"*80)
logging.info("ADD Pattern:"+str(key))
logging.info("ADD Template:"+str(template))
logging.info("-"*80)
# key = (self._currentPattern.strip(), self._currentThat.strip(),self._currentTopic.strip())
# self.categories[key] = self._elemStack[-1]
logging.debug("-"*80)
logging.debug("End Category:"+str(self._elemStack))
logging.debug("-"*80)
self._whitespaceBehaviorStack.pop()
elif name == "pattern":
# </pattern> tags are only legal in the InsidePattern state
if self._state != self._STATE_InsidePattern:
raise AimlParserError, "Unexpected </pattern> tag "+self._location()
self._state = self._STATE_AfterPattern
elif name == "that" and self._state == self._STATE_InsideThat:
# </that> tags are only allowed inside <template> elements or in
# the InsideThat state. This clause handles the latter case.
self._state = self._STATE_AfterThat
elif name == "template":
# </template> tags are only allowed in the InsideTemplate state.
if self._state != self._STATE_InsideTemplate:
raise AimlParserError, "Unexpected </template> tag "+self._location()
self._state = self._STATE_AfterTemplate
self._whitespaceBehaviorStack.pop()
elif self._state == self._STATE_InsidePattern:
# Certain tags are allowed inside <pattern> elements.
if name not in ["bot"]:
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()
elif self._state == self._STATE_InsideThat:
# Certain tags are allowed inside <that> elements.
if name not in ["bot"]:
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()
elif self._state == self._STATE_InsideTemplate:
# End of an element inside the current template. Append the
# element at the top of the stack onto the one beneath it.
elem = self._elemStack.pop()
self._elemStack[-1].append(elem)
self._whitespaceBehaviorStack.pop()
# If the element was a condition, pop an item off the
# foundDefaultLiStack as well.
if elem[0] == "condition": self._foundDefaultLiStack.pop()
else:
# Unexpected closing tag
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()
def _endElement(self, name):
"""Verify that an AIML end element is valid in the current
context.
Raises an AimlParserError if an illegal end element is encountered.
"""
if name == "aiml":
# </aiml> tags are only legal in the InsideAiml state
if self._state != self._STATE_InsideAiml:
raise AimlParserError, "Unexpected </aiml> tag "+self._location()
self._state = self._STATE_OutsideAiml
self._whitespaceBehaviorStack.pop()
elif name == "topic":
# </topic> tags are only legal in the InsideAiml state, and
# only if _insideTopic is true.
if self._state != self._STATE_InsideAiml or not self._insideTopic:
raise AimlParserError, "Unexpected </topic> tag "+self._location()
self._insideTopic = False
self._currentTopic = u""
elif name == "category":
# </category> tags are only legal in the AfterTemplate state
if self._state != self._STATE_AfterTemplate:
raise AimlParserError, "Unexpected </category> tag "+self._location()
self._state = self._STATE_InsideAiml
# End the current category. Store the current pattern/that/topic and
# element in the categories dictionary.
self._currentPattern = u' '.join(splitChinese(self._currentPattern))
key = (self._currentPattern.strip(), self._currentThat.strip(),self._currentTopic.strip())
self.categories[key] = self._elemStack[-1]
self._whitespaceBehaviorStack.pop()
elif name == "pattern":
# </pattern> tags are only legal in the InsidePattern state
if self._state != self._STATE_InsidePattern:
raise AimlParserError, "Unexpected </pattern> tag "+self._location()
self._state = self._STATE_AfterPattern
elif name == "that" and self._state == self._STATE_InsideThat:
# </that> tags are only allowed inside <template> elements or in
# the InsideThat state. This clause handles the latter case.
self._state = self._STATE_AfterThat
elif name == "template":
# </template> tags are only allowed in the InsideTemplate state.
if self._state != self._STATE_InsideTemplate:
raise AimlParserError, "Unexpected </template> tag "+self._location()
self._state = self._STATE_AfterTemplate
self._whitespaceBehaviorStack.pop()
elif self._state == self._STATE_InsidePattern:
# Certain tags are allowed inside <pattern> elements.
if name not in ["bot"]:
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()
elif self._state == self._STATE_InsideThat:
# Certain tags are allowed inside <that> elements.
if name not in ["bot"]:
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()
elif self._state == self._STATE_InsideTemplate:
# End of an element inside the current template. Append the
# element at the top of the stack onto the one beneath it.
elem = self._elemStack.pop()
self._elemStack[-1].append(elem)
self._whitespaceBehaviorStack.pop()
# If the element was a condition, pop an item off the
# foundDefaultLiStack as well.
if elem[0] == "condition": self._foundDefaultLiStack.pop()
else:
# Unexpected closing tag
raise AimlParserError, ("Unexpected </%s> tag " % name)+self._location()