Increase read text speed

geotext/geotext.py

@@ -63,7 +63,7 @@ def _build_geodb(self):
             country_db, state_db, city_db, nationality_db,
             city_abbreviation_db, country_abbreviation_db
         )
-        self._location_length = self._get_locations_length()
+        self._max_location_length = self._get_locations_length()[0]
 
     def _get_locations_length(self):
         words_counts = set()
@@ -81,7 +81,9 @@ def _get_candidates(self, text):
         # TODO: improve this, since DB has unicode symbols in cities
         text = re.sub(r'[^\w]+', ' ', text).strip()
 
-        return CandidateDB(text).get_candidates()
+        return CandidateDB(
+            text, max_phrase_len=self._max_location_length
+        ).get_candidates()
 
     def read(self, text, min_population=0, skip_nationalities=False):
         self.text = text
@@ -147,7 +149,7 @@ def _get_locations_from_candidates(
                 city_abbrev_match.place.population >= min_population
             ):
                     cities.add(city_abbrev_match.place)
-                    candidate.is_location = True
+                    candidate.mark_as_location()
                     continue
 
             # 2
@@ -159,7 +161,7 @@ def _get_locations_from_candidates(
                 state_match.country.population >= min_population
             ):
                 states.add(state_match)
-                candidate.is_location = True
+                candidate.mark_as_location()
                 continue
 
             # 3
@@ -169,7 +171,7 @@ def _get_locations_from_candidates(
             )
             if country_match and country_match.population >= min_population:
                 countries.add(country_match)
-                candidate.is_location = True
+                candidate.mark_as_location()
                 continue
 
             # 4
@@ -182,7 +184,7 @@ def _get_locations_from_candidates(
                     nationality_match.place.population >= min_population
                 ):
                     nationalities.add(nationality_match.place)
-                    candidate.is_location = True
+                    candidate.mark_as_location()
                     continue
 
             # 5
@@ -194,14 +196,14 @@ def _get_locations_from_candidates(
                 country_abbrev_match.place.population >= min_population
             ):
                 countries.add(country_abbrev_match.place)
-                candidate.is_location = True
+                candidate.mark_as_location()
                 continue
 
             # 6
             city_match = self._geodb.city_db.search(candidate.text.lower())
             if city_match and city_match.population >= min_population:
                 cities.add(city_match)
-                candidate.is_location = True
+                candidate.mark_as_location()
                 continue
 
             # 7
@@ -213,7 +215,7 @@ def _get_locations_from_candidates(
                 state_match.country.population >= min_population
             ):
                 states.add(state_match)
-                candidate.is_location = True
+                candidate.mark_as_location()
                 continue
         return (
             tuple(countries), tuple(nationalities), tuple(states),

geotext/models/candidate.py

@@ -8,22 +8,27 @@ class Candidate(object):
     def __init__(self, text):
         self.text = text
         self.parents = set()
-        self.is_location = False  # whether this candidate is a valid location
+        self.children = set()
+        # Whether this candidate is a valid location
+        self.is_location = False
+        # Whether any of candidate parent and their parents and so on is
+        # a valid location
+        self.has_parent_location = False
 
     def add_parent(self, parent):
         self.parents.add(parent)
+        parent.children.add(self)
 
-    @property
-    def has_parent_location(self):
-        """
-        Whether any of candidate parent and their parents and so on is
-        a valid location
-        :return:
-        """
-        for parent in self.parents:
-            if parent.is_location or parent.has_parent_location:
-                return True
-        return False
+    def get_all_children(self):
+        result = self.children
+        for child in self.children:
+            result |= child.get_all_children()
+        return result
+
+    def mark_as_location(self):
+        self.is_location = True
+        for child in self.get_all_children():
+            child.has_parent_location = True
 
     def __repr__(self):
         return '{}: "{}"'.format(type(self).__name__, self.text)