Merge pull request #2425 from lonvia/tokenizer-documentation

[nominatim.git] / nominatim / tokenizer / icu_name_processor.py

diff --git a/nominatim/tokenizer/icu_name_processor.py b/nominatim/tokenizer/icu_name_processor.py
index a0f229742abef322c8914e81069232ef67410877..93d2b0ffa26b9151ccba1928c0e7d0745ce4380a 100644 (file)
--- a/nominatim/tokenizer/icu_name_processor.py
+++ b/nominatim/tokenizer/icu_name_processor.py
@@ -2,13 +2,14 @@
 Processor for names that are imported into the database based on the
 ICU library.
 """
 Processor for names that are imported into the database based on the
 ICU library.
 """

-import json
+from collections import defaultdict

 import itertools
 
 from icu import Transliterator
 import datrie
 
 from nominatim.db.properties import set_property, get_property
 import itertools
 
 from icu import Transliterator
 import datrie
 
 from nominatim.db.properties import set_property, get_property

+from nominatim.tokenizer import icu_variants as variants

 
 DBCFG_IMPORT_NORM_RULES = "tokenizer_import_normalisation"
 DBCFG_IMPORT_TRANS_RULES = "tokenizer_import_transliteration"
 
 DBCFG_IMPORT_NORM_RULES = "tokenizer_import_normalisation"
 DBCFG_IMPORT_TRANS_RULES = "tokenizer_import_transliteration"


@@ -31,20 +32,12 @@ class ICUNameProcessorRules:
         elif conn is not None:
             self.norm_rules = get_property(conn, DBCFG_IMPORT_NORM_RULES)
             self.trans_rules = get_property(conn, DBCFG_IMPORT_TRANS_RULES)
         elif conn is not None:
             self.norm_rules = get_property(conn, DBCFG_IMPORT_NORM_RULES)
             self.trans_rules = get_property(conn, DBCFG_IMPORT_TRANS_RULES)

-            self.replacements = json.loads(get_property(conn, DBCFG_IMPORT_REPLACEMENTS))
+            self.replacements = \
+                variants.unpickle_variant_set(get_property(conn, DBCFG_IMPORT_REPLACEMENTS))

             self.search_rules = get_property(conn, DBCFG_SEARCH_STD_RULES)
         else:
             assert False, "Parameter loader or conn required."
 
             self.search_rules = get_property(conn, DBCFG_SEARCH_STD_RULES)
         else:
             assert False, "Parameter loader or conn required."
 

-        # Compute the set of characters used in the replacement list.
-        # We need this later when computing the tree.
-        chars = set()
-        for full, repl in self.replacements:
-            chars.update(full)
-            for word in repl:
-                chars.update(word)
-        self.replacement_charset = ''.join(chars)
-

 
     def save_rules(self, conn):
         """ Save the rules in the property table of the given database.
 
     def save_rules(self, conn):
         """ Save the rules in the property table of the given database.


@@ -53,23 +46,39 @@ class ICUNameProcessorRules:
         """
         set_property(conn, DBCFG_IMPORT_NORM_RULES, self.norm_rules)
         set_property(conn, DBCFG_IMPORT_TRANS_RULES, self.trans_rules)
         """
         set_property(conn, DBCFG_IMPORT_NORM_RULES, self.norm_rules)
         set_property(conn, DBCFG_IMPORT_TRANS_RULES, self.trans_rules)

-        set_property(conn, DBCFG_IMPORT_REPLACEMENTS, json.dumps(self.replacements))
+        set_property(conn, DBCFG_IMPORT_REPLACEMENTS,
+                     variants.pickle_variant_set(self.replacements))

         set_property(conn, DBCFG_SEARCH_STD_RULES, self.search_rules)
 
 
 class ICUNameProcessor:
         set_property(conn, DBCFG_SEARCH_STD_RULES, self.search_rules)
 
 
 class ICUNameProcessor:

+    """ Collects the different transformation rules for normalisation of names
+        and provides the functions to aply the transformations.
+    """

 
     def __init__(self, rules):
         self.normalizer = Transliterator.createFromRules("icu_normalization",
                                                          rules.norm_rules)
         self.to_ascii = Transliterator.createFromRules("icu_to_ascii",
 
     def __init__(self, rules):
         self.normalizer = Transliterator.createFromRules("icu_normalization",
                                                          rules.norm_rules)
         self.to_ascii = Transliterator.createFromRules("icu_to_ascii",

-                                                       rules.trans_rules)
+                                                       rules.trans_rules +
+                                                       ";[:Space:]+ > ' '")

         self.search = Transliterator.createFromRules("icu_search",
                                                      rules.search_rules)
 
         self.search = Transliterator.createFromRules("icu_search",
                                                      rules.search_rules)
 

-        self.replacements = datrie.Trie(rules.replacement_charset)
-        for full, repl in rules.replacements:
-            self.replacements[full] = repl
+        # Intermediate reorder by source. Also compute required character set.
+        immediate = defaultdict(list)
+        chars = set()
+        for variant in rules.replacements:
+            if variant.source[-1] == ' ' and variant.replacement[-1] == ' ':
+                replstr = variant.replacement[:-1]
+            else:
+                replstr = variant.replacement
+            immediate[variant.source].append(replstr)
+            chars.update(variant.source)
+        # Then copy to datrie
+        self.replacements = datrie.Trie(''.join(chars))
+        for src, repllist in immediate.items():
+            self.replacements[src] = repllist

 
 
     def get_normalized(self, name):
 
 
     def get_normalized(self, name):


@@ -82,26 +91,52 @@ class ICUNameProcessor:
         """ Compute the spelling variants for the given normalized name
             and transliterate the result.
         """
         """ Compute the spelling variants for the given normalized name
             and transliterate the result.
         """

-        baseform = ' ' + norm_name + ' '
-        variants = ['']
+        baseform = '^ ' + norm_name + ' ^'
+        partials = ['']

 
         startpos = 0
         pos = 0
 
         startpos = 0
         pos = 0

+        force_space = False

         while pos < len(baseform):
             full, repl = self.replacements.longest_prefix_item(baseform[pos:],
                                                                (None, None))
             if full is not None:
                 done = baseform[startpos:pos]
         while pos < len(baseform):
             full, repl = self.replacements.longest_prefix_item(baseform[pos:],
                                                                (None, None))
             if full is not None:
                 done = baseform[startpos:pos]

-                variants = [v + done + r for v, r in itertools.product(variants, repl)]
+                partials = [v + done + r
+                            for v, r in itertools.product(partials, repl)
+                            if not force_space or r.startswith(' ')]
+                if len(partials) > 128:
+                    # If too many variants are produced, they are unlikely
+                    # to be helpful. Only use the original term.
+                    startpos = 0
+                    break

                 startpos = pos + len(full)
                 startpos = pos + len(full)

+                if full[-1] == ' ':
+                    startpos -= 1
+                    force_space = True

                 pos = startpos
             else:
                 pos += 1
                 pos = startpos
             else:
                 pos += 1

+                force_space = False

 
 

+        # No variants detected? Fast return.

         if startpos == 0:
         if startpos == 0:

-            return [self.to_ascii.transliterate(norm_name)]
+            trans_name = self.to_ascii.transliterate(norm_name).strip()
+            return [trans_name] if trans_name else []
+
+        return self._compute_result_set(partials, baseform[startpos:])
+
+
+    def _compute_result_set(self, partials, prefix):
+        results = set()
+
+        for variant in partials:
+            vname = variant + prefix
+            trans_name = self.to_ascii.transliterate(vname[1:-1]).strip()
+            if trans_name:
+                results.add(trans_name)

 
 

-        return [self.to_ascii.transliterate(v + baseform[startpos:pos]).strip() for v in variants]
+        return list(results)

 
 
     def get_search_normalized(self, name):
 
 
     def get_search_normalized(self, name):