bug fixes

app.py

@@ -9,7 +9,6 @@ extractive_query = ExtractiveProposalQueries(es_host = Config.es_host, es_index
                                   reader_name_or_path = Config.reader_model_name_or_path,
                                   use_gpu = Config.use_gpu)
 
-
 def fake_search(question, retriever_top_k, reader_top_k):
   #p1_result = query.search_by_query(query = question, retriever_top_k = retriever_top_k, reader_top_k = reader_top_k, es_index = "informecomisionverdad")
   result = []
@@ -30,7 +29,6 @@ def search(question, retriever_top_k, reader_top_k):
   
   return result
 
-
 if __name__ == "__main__":
     # streamlit part starts here with title
     title = """
@@ -58,7 +56,7 @@ if __name__ == "__main__":
         # set start time
         stt = time.time()
         # retrieve top 5 documents
-        results = search(query, retriever_top_k=5, reader_top_k=3)
+        results = fake_search(query, retriever_top_k=5, reader_top_k=3)
         # set endtime
         ent = time.time()
         # measure resulting time

document_quieries.py

@@ -3,6 +3,9 @@ from haystack.nodes import BM25Retriever, FARMReader
 from haystack.document_stores import ElasticsearchDocumentStore
 from haystack.pipelines import ExtractiveQAPipeline
 import certifi
+import datetime
+import requests
+from base64 import b64encode
 
 ca_certs=certifi.where()        
 
@@ -17,7 +20,8 @@ class ExtractiveProposalQueries(DocumentQueries):
     def __init__(self, es_host: str, es_index: str, es_user, es_password, reader_name_or_path: str, use_gpu = True) -> None:
         reader = FARMReader(model_name_or_path = reader_name_or_path, use_gpu = use_gpu, num_processes=1, context_window_size=200)
         self._initialize_pipeline(es_host, es_index, es_user, es_password, reader = reader)
-        
+        self.log = Log(es_host= es_host, es_index="log", es_user = es_user, es_password= es_password)
+
     def _initialize_pipeline(self, es_host, es_index, es_user, es_password, reader = None):
         if reader is not None:
             self.reader = reader
@@ -29,8 +33,28 @@ class ExtractiveProposalQueries(DocumentQueries):
         self.pipe = ExtractiveQAPipeline(self.reader, self.retriever)
 
     def search_by_query(self, query : str, retriever_top_k: int, reader_top_k: int, es_index: str = None) :
+        self.log.write_log(query, "hfspace-informecomision")
         if es_index is not None:
             self._initialize_pipeline(self.es_host, es_index, self.es_user, self.es_password)
         params = {"Retriever": {"top_k": retriever_top_k}, "Reader": {"top_k": reader_top_k}}
         prediction = self.pipe.run( query = query, params = params)      
         return prediction["answers"]
+
+class Log():
+    
+    def __init__(self, es_host: str, es_index: str, es_user, es_password) -> None:
+        self.elastic_endpoint = f"https://{es_host}:443/{es_index}/_doc"
+        self.credentials = b64encode(b"elastic:Xmir2X1f5twb8OJbpjbiIgXt").decode("ascii")
+        self.auth_header = { 'Authorization' : 'Basic %s' %  self.credentials }
+                      
+    def write_log(self, message: str, source: str) -> None:
+        created_date = datetime.datetime.now().strftime('%Y-%m-%dT%H:%M:%SZ')
+        post_data  = {
+            "message" : message,
+            "createdDate": {
+                "date" : created_date
+            },
+            "source": source
+        }
+        r = requests.post(self.elastic_endpoint, json = post_data, headers = self.auth_header)
+        print(r.text)