From 9a30db227c27d8f636b52ccdfa906fce1e432112 Mon Sep 17 00:00:00 2001
From: minjae1025 <cnsanet@gmail.com>
Date: Mon, 10 Jun 2019 08:14:28 +0900
Subject: [PATCH] Add WikipediaResponseAdapter

---
 chatterbot/logic/__init__.py           |  2 +
 chatterbot/logic/wikipedia_response.py | 98 ++++++++++++++++++++++++++
 2 files changed, 100 insertions(+)
 create mode 100644 chatterbot/logic/wikipedia_response.py

diff --git a/chatterbot/logic/__init__.py b/chatterbot/logic/__init__.py
index e43167bee..0188edff2 100644
--- a/chatterbot/logic/__init__.py
+++ b/chatterbot/logic/__init__.py
@@ -4,6 +4,7 @@
 from chatterbot.logic.specific_response import SpecificResponseAdapter
 from chatterbot.logic.time_adapter import TimeLogicAdapter
 from chatterbot.logic.unit_conversion import UnitConversion
+from chatterbot.logic.wikipedia_response import WikipediaResponseAdapter
 
 __all__ = (
     'LogicAdapter',
@@ -12,4 +13,5 @@
     'SpecificResponseAdapter',
     'TimeLogicAdapter',
     'UnitConversion',
+    'WikipediaResponseAdapter'
 )
diff --git a/chatterbot/logic/wikipedia_response.py b/chatterbot/logic/wikipedia_response.py
new file mode 100644
index 000000000..adfc0bc7d
--- /dev/null
+++ b/chatterbot/logic/wikipedia_response.py
@@ -0,0 +1,98 @@
+from chatterbot.logic import LogicAdapter
+
+
+class WikipediaResponseAdapter(LogicAdapter):
+    """
+    Return a response which get from Wikipedia.
+
+    :kwargs:
+        * *lang_code* (``str``) --
+          Decide which language you want to get answers from Wikipedia.
+          The default is English Wikipedia.
+        * *cut_threshold* (``int``) --
+          If the data from Wikipedia is too long,
+          it's the minimum length to cut. The standard is dot character.
+    """
+
+    def __init__(self, chatbot, **kwargs):
+        super().__init__(chatbot, **kwargs)
+        self.lang_code = kwargs.get('lang_code', 'en')
+        self.cut_threshold = kwargs.get('cut_threshold', 2)
+        self.input_text = ''
+        self.input_search_text = ''
+        self.find_word = ''
+        self.response_statement = ''
+
+    def can_process(self, statement):
+        if not self.lang_code:
+            return False
+
+        input_search_text = statement.search_text \
+            if statement.search_text else \
+            self.chatbot.storage.tagger.get_text_index_string(statement.text)
+
+        for props in input_search_text.split(' '):
+            props_list = props.split(':')
+            if props_list[0] == "VERB":
+                self.find_word = props_list[1]
+                break
+
+        if not self.find_word:
+            return False
+
+        from requests import get
+
+        url = 'https://' + self.lang_code + '.wikipedia.org/w/api.php'
+
+        query = '?format=json&action=query&prop=extracts' \
+                + '&exlimit=1&explaintext&exintro&titles=' \
+                + self.find_word \
+                + '&redirects=true'
+
+        response = get(url=(url + query)).json()
+
+        pages = response['query']['pages']
+
+        if '-1' in pages.keys():
+            return False
+
+        key = next(iter(pages.keys()))
+        self.input_text = statement.text
+        self.input_search_text = input_search_text
+
+        response_text = pages[key]['extract']
+
+        cut_response = response_text.split('.')
+
+        if self.cut_threshold < len(cut_response):
+            self.response_statement = ''
+            for i in range(self.cut_threshold):
+                self.response_statement += cut_response[i] + '.'
+        else:
+            self.response_statement = response_text
+
+        self.response_statement = \
+            self.response_statement.replace('\n', ' ')
+
+        return True
+
+    def process(self, statement,
+                additional_response_selection_parameters=None):
+        from chatterbot.conversation import Statement
+
+        output_search_text = self.chatbot.storage.tagger.get_text_index_string(
+            self.response_statement)
+
+        response_text = Statement(
+            text=self.response_statement,
+            in_response_to=statement.text,
+            search_text=output_search_text,
+            search_in_respond_to=self.input_search_text,
+        )
+
+        if statement.text == self.input_text:
+            response_text.confidence = 1
+        else:
+            response_text.confidence = 0
+
+        return response_text