Add context aware search index

Dockerfile

 FROM python:3.8
 
-RUN pip install pdfplumber Whoosh Flask
+RUN pip install pdfplumber Whoosh Flask nltk
+RUN python -m nltk.downloader punkt
 RUN apt-get update
 RUN apt-get install -y libmagickwand-dev ghostscript
 RUN rm /etc/ImageMagick-6/policy.xml

app.py

@@ -53,7 +53,7 @@ def query():
         index = Index()
         query = request.form.get("term")
         context = request.form.get("context")  # TODO: Use context to find better results
-        result = index.search(query)
+        result = index.search(query, context)
         img_name = result["path"][7:] + "_" + str(result["page"]) + ".jpg"
         return jsonify({
             "type": "image",

context_aware_search_index.py

+from whoosh.qparser import QueryParser, OrGroup
+
+from search_index import BasicSearchIndex
+from nltk import word_tokenize
+
+
+class ContextAwareSearchIndex(BasicSearchIndex):
+    """Return fitting slides, favouring slides with the query contained in the title"""
+
+    def result_list(self, query, context):
+        term = query + ""  # Make new object
+        query = " ".join([token + "^1" for token in word_tokenize(term)])
+        context_tokens = word_tokenize(context)
+        if len(context_tokens):
+            individual_context_token_weight = str(round(1 / len(context_tokens), 3))
+            query += " " + " ".join([token + "^" + individual_context_token_weight for token in context_tokens])
+        query_parser = QueryParser("content", self.ix.schema, group=OrGroup.factory(0.9))
+        return self.ix.searcher().search(query_parser.parse(query))

search_index.py

@@ -24,12 +24,12 @@ class BasicSearchIndex:
         writer.add_document(path=path, page=page, content=content)
         writer.commit()
 
-    def result_list(self, query):
+    def result_list(self, query, context):
         query_parser = QueryParser("content", self.ix.schema, group=OrGroup.factory(0.9))
         return self.ix.searcher().search(query_parser.parse(query))
 
-    def search(self, query):
-        return self.result_list(query)[0]
+    def search(self, query, context):
+        return self.result_list(query, context)[0]
 
 
 if __name__ == "__main__":

title_focus_search_index.py

@@ -4,8 +4,8 @@ from search_index import BasicSearchIndex
 class TitleFocusSearchIndex(BasicSearchIndex):
     """Return fitting slides, favouring slides with the query contained in the title"""
 
-    def search(self, query):
-        results = self.result_list(query)
+    def search(self, query, context):
+        results = self.result_list(query, context)
         for result in results:
             if query.lower() in result["content"].split("\n")[0].lower():
                 return result