Merge branch 'master' into parser-data

resources/dictionaries/it/post_office.txt

@@ -1 +1,2 @@
-casella postale|cp|c.p.|c.p|c p
+casella postale|cp|c.p.|c.p|c p|cpt
+presso|co|c \ o|c / o

resources/dictionaries/it/street_types.txt

@@ -1,19 +1,38 @@
 accesso|acc.so|accso
-autostrada|aut.da|autda
+alto|alt
+alzaia|alz
+androna|and
+arco|arc
+autostrada|aut.da|autda|aut
 audostradale|aut.sle|autsle|aut sle|aut.ale|autale
 autostrade|aut.de|autde
+baluardo|bdo
+banchina|bna
+bivio|biv
+borgata|bga
+borgo|b.go|bgo
 bretella|br.lla|brlla
 bretelle|br.lle|brlle
-calle|c.le|cle
+brigata|bgt
+calata|clt
+calle|cal
+campiello|cpl
 campo|c.po|cpo
-cascina|cna
+canale|cnl
+cantone|cat
+casale|cas
+cascina|c.na|cna
 cavalcavia|cavalc.a|cavalca
 chiusa|c.usa|cusa
 circonvallazione|circonv.e|circonve|crv
+contrà|con|contra'
 contrada|contr.a|contra|c.da|cda|cnt
 corso|c.so|cso
 corte|c.te|cte
-cortile|c.ile|ctile
+cortile|c.ile|ctile|c.le|cle
+cortina|ctn
+costa|cta
+crocevia|cra
 cupa|cup
 discesa|dis
 ferrata|f.rata|frata
@@ -61,7 +80,7 @@ rampa|rpa
 rampe|rpe
 regione|reg
 rio|rii
-rio tera'|rit
+rio terà|rit|rio tera'
 ripa|ri
 riva|rva
 riviera|riv

src/libpostal.c

@@ -280,6 +280,8 @@ static string_tree_t *add_string_alternatives(char *str, normalize_options_t opt
 
             token_t token;
 
+            size_t added_expansions = 0;
+
             if ((value.components & options.address_components) > 0) {
                 key->n = namespace_len;
                 for (int j = phrase.start; j < phrase.start + phrase.len; j++) {
@@ -298,8 +300,13 @@ static string_tree_t *add_string_alternatives(char *str, normalize_options_t opt
                 address_expansion_array *expansions = address_dictionary_get_expansions(key_str);
 
                 if (expansions != NULL) {
+                    
                     for (int j = 0; j < expansions->n; j++) {
                         address_expansion_t expansion = expansions->a[j];
+                        if ((expansion.address_components & options.address_components) == 0) {
+                            continue;
+                        }
+
                         if (expansion.canonical_index != NULL_CANONICAL_INDEX) {
                             char *canonical = address_dictionary_get_canonical(expansion.canonical_index);
                             char *canonical_normalized = normalize_string_utf8(canonical, normalize_string_options);
@@ -347,10 +354,15 @@ static string_tree_t *add_string_alternatives(char *str, normalize_options_t opt
                             }
                             cstring_array_terminate(tree->strings);
                         }
+
+                        added_expansions++;
                     }
 
+
                 }
-            } else {
+            }
+
+            if (added_expansions == 0) {
                 uint32_t start_index = cstring_array_start_token(tree->strings);
                 for (int j = phrase.start; j < phrase.start + phrase.len; j++) {
                     token = tokens->a[j];

src/normalize.h

@@ -9,7 +9,7 @@ The normalize module provides several options for preprocessing full strings:
 
 As well as normalizations for individual string tokens:
 
-- Replace hyphens with space e.g. "quatre-vignt" => "quatre vignt"
+- Replace hyphens with space e.g. "quatre-vingt" => "quatre vingt"
 - Delete hyphens e.g. "auto-estrada" => "autoestrada"
 - Delete final period "R." => "R"
 - Delete acronym periods: "U.S.A." => "USA"
@@ -73,4 +73,4 @@ string_tree_t *normalize_string(char *str, uint64_t options);
 string_tree_t *normalize_string_languages(char *str, uint64_t options, size_t num_languages, char **languages);
  
 
-#endif
+#endif