cached autocorrect results and other things. resolves Unicaronas#30

gabisurita · Oct 25, 2018 · 4732768 · 4732768
1 parent ad8b31d
commit 4732768
Show file tree

Hide file tree

Showing 6 changed files with 74 additions and 29 deletions.
diff --git a/search/finder/grammar.py b/search/finder/grammar.py
@@ -1,8 +1,9 @@
-from .base import BaseFinder
 import re
 import pickle
 from collections import Counter
 from unidecode import unidecode
+from .base import BaseFinder
+from .redis import RedisFinder
 
 
 def words(text):
@@ -92,6 +93,20 @@ def _transform(self, term):
         to perform word corrections on it
         and find a known synonym that is better
         understood by the later finders
+
+        Since it takes a long time to correct each sentence, cache
+        corrected results for some time.
         """
-        corrected_term = self.correct_sentence(term.query)
+        # Use redis as cache
+        cache = RedisFinder()
+
+        # Get the cached results if available
+        cached_result = cache.get_key(term.query)
+        if cached_result is not None:
+            corrected_term = cached_result
+        else:
+            # If the corrected term is not cached, process it normally
+            corrected_term = self.correct_sentence(term.query)
+            # and then cache it
+            cache.set_key(term.query, corrected_term)
         return self.perform_transform(term, corrected_term)
diff --git a/search/finder/synonym.py b/search/finder/synonym.py
@@ -14,39 +14,43 @@ def get_synonym_map(self):
         """Get a map of synonyms
         Used on autocorrected words to find the result
         """
+        POSTO_IPIRANGA = 'posto ipiranga, unicamp'
+        SP = 'sao paulo, sp'
+        RODOVIARIA_TIETE = 'rodoviaria tiete, sao paulo'
+
         s_map = [
             # Map to posto ipiranga, Barão geraldo
-            ('unicamp', 'posto ipiranga, unicamp'),
-            ('unicamp', 'posto ipiranga unicamp'),
-            ('universidade estadual de campinas', 'posto ipiranga, unicamp'),
+            ('unicamp', POSTO_IPIRANGA),
+            ('unichamps', POSTO_IPIRANGA),
+            ('universidade estadual de campinas', POSTO_IPIRANGA),
 
-            ('posto ipiranga, unicamp', 'posto ipiranga, unicamp'),
-            ('posto ipiranga unicamp', 'posto ipiranga, unicamp'),
-            ('posto ipiranga', 'posto ipiranga, unicamp'),
-            ('posto da 1', 'posto ipiranga, unicamp'),
-            ('posto', 'posto ipiranga, unicamp'),
+            ('posto ipiranga, unicamp', POSTO_IPIRANGA),
+            ('posto ipiranga unicamp', POSTO_IPIRANGA),
+            ('posto ipiranga', POSTO_IPIRANGA),
+            ('posto da 1', POSTO_IPIRANGA),
+            ('posto', POSTO_IPIRANGA),
 
-            ('barao geraldo', 'posto ipiranga, unicamp'),
-            ('barao', 'posto ipiranga, unicamp'),
-            ('bg', 'posto ipiranga, unicamp'),
+            ('barao geraldo', POSTO_IPIRANGA),
+            ('barao', POSTO_IPIRANGA),
+            ('bg', POSTO_IPIRANGA),
 
             # Map to generic São Paulo, SP
-            ('sao paulo, sao paulo', 'sao paulo, sp'),
-            ('sao paulo sao paulo', 'sao paulo, sp'),
-            ('sao paulo, sp', 'sao paulo, sp'),
-            ('sao paulo sp', 'sao paulo, sp'),
-            ('sao paulo', 'sao paulo, sp'),
-            ('sp', 'sao paulo, sp'),
-            ('sampa', 'sao paulo, sp'),
+            ('sao paulo, sao paulo', SP),
+            ('sao paulo sao paulo', SP),
+            ('sao paulo, sp', SP),
+            ('sao paulo sp', SP),
+            ('sao paulo', SP),
+            ('sp', SP),
+            ('sampa', SP),
 
             # Map to rodoviária tietê
-            ('rodoviaria tiete, sao paulo', 'rodoviaria tiete, sao paulo'),
-            ('rodoviaria tiete sao paulo', 'rodoviaria tiete, sao paulo'),
-            ('rodoviaria tiete, sp', 'rodoviaria tiete, sao paulo'),
-            ('rodoviaria tiete sp', 'rodoviaria tiete, sao paulo'),
-            ('rodoviaria tiete', 'rodoviaria tiete, sao paulo'),
-            ('estacao tiete', 'rodoviaria tiete, sao paulo'),
-            ('tiete', 'rodoviaria tiete, sao paulo'),
+            ('rodoviaria tiete, sao paulo', RODOVIARIA_TIETE),
+            ('rodoviaria tiete sao paulo', RODOVIARIA_TIETE),
+            ('rodoviaria tiete, sp', RODOVIARIA_TIETE),
+            ('rodoviaria tiete sp', RODOVIARIA_TIETE),
+            ('rodoviaria tiete', RODOVIARIA_TIETE),
+            ('estacao tiete', RODOVIARIA_TIETE),
+            ('tiete', RODOVIARIA_TIETE),
         ]
         return OrderedDict(s_map)
 

diff --git a/third_parties/query/base.py b/third_parties/query/base.py
@@ -1,7 +1,7 @@
+from django.http import HttpRequest
 from search.term.base import BaseTerm
 from datetime import datetime
 from rest_framework.request import Request
-from django.http import HttpRequest
 
 
 class BaseSearchQuery(object):

diff --git a/third_parties/query/queries.py b/third_parties/query/queries.py
@@ -2,4 +2,6 @@
 
 
 class SearchQuery(BaseSearchQuery):
-    pass
+
+    def __str__(self):
+        return f"Query: (dest '{self.destination.query[:10]}', ori '{self.origin.query[:10]}', dt_gt '{self.datetime_gte}', dt_lt '{self.datetime_lte}', price '{self.price_lte}')"
diff --git a/third_parties/result/results.py b/third_parties/result/results.py
@@ -4,6 +4,18 @@
 class Result(BaseResult):
     """Results"""
 
+    def __str__(self):
+        return f"Result: ({len(self.items)} items)"
+
+    def __repr__(self):
+        return str(self)
+
 
 class ResultItem(BaseResultItem):
     """Result item"""
+
+    def __str__(self):
+        return f"ResultItem: (dest '{self.destination[:10]}', ori '{self.origin[:10]}', dt '{self.datetime}', price '{self.price}')"
+
+    def __repr__(self):
+        return str(self)
diff --git a/trips/api/versions/v1_0/serializers/search.py b/trips/api/versions/v1_0/serializers/search.py
@@ -1,3 +1,4 @@
+from django.utils import timezone
 from rest_framework import serializers
 from project.utils import local_versioned_url_name
 from third_parties.utils import get_search_keys
@@ -47,3 +48,14 @@ class ThirdPartyQuerySerializer(serializers.Serializer):
         required=False,
         help_text=f"Fontes das caronas. Lista separada por espaços com as opções `{', '.join(get_search_keys())}`, ou `all` para todas. `all` por padrão"
     )
+
+    def validate_datetime_gte(self, value):
+        return max(timezone.now(), value)
+
+    def validate_datetime_lte(self, value):
+        return max(timezone.now(), value)
+
+    def validate(self, data):
+        if data['datetime_lte'] < data['datetime_gte']:
+            raise serializers.ValidationError("'datetime_gte' deve ser antes que 'datetime_lte'")
+        return data