[dictionaries] additions to German dictionaries, including a separable prefix dictionary

data/dictionaries/de/academic_degrees/academic_degrees.txt

@@ -1,4 +1,5 @@
 diplom ingenieur|dipl ing
 diplom kaufmann|dipl kfm
 doktor de medizin|dr med
-doktor der philosophie|dr phil
+doktor der philosophie|dr phil
+magister|mag

data/dictionaries/de/company_types/company_types.txt

@@ -1,4 +1,8 @@
-aktiengesellschaft|ag
+ag|aktiengesellschaft
+eingetragener verein|e v
 firma|fa
 fussball club|fc|f c
-gmbh
+gmbh|gesellschaft mit beschrankter haftung
+gesellschaft|ges
+kg|kommanditgesellschaft
+ohg|offene handelsgesellschaft

data/dictionaries/de/concatenated_prefixes_separable/concatenated_prefixes_separable.txt

@@ -0,0 +1,5 @@
+hinter
+klein
+ober
+unter
+vorder

data/dictionaries/de/concatenated_suffixes_inseparable/concatenated_suffixes_inseparable

@@ -1,3 +1,10 @@
 berg|bg
 burg|bg
-dorf|df
+dorf|df
+kogel|kg
+kreuz|kz
+platz|pl
+siedlung|sdlg
+stiege|stg
+strasse|str
+wiese|ws

data/dictionaries/de/concatenated_suffixes_separable/concatenated_suffixes_separable.txt

@@ -4,9 +4,6 @@ burgermeister|bgm
 damm
 gasse
 hof
-kamp
-pfad
-platz|pl
-quelle|qu
-strasse|str
+quelle|qu|q
+rundwanderweg|rww
 weg|w

data/dictionaries/de/directionals/directionals.txt

@@ -0,0 +1,17 @@
+nord|n
+norden
+nordliche|nordl
+nordost|no
+nordwest|nw
+ost|o
+osten
+ostliche|ostl
+suden|s
+sudliche|sudl
+sudost|so
+sudosten
+sudwest|sw
+sudwesten
+west|w
+westen
+westliche|westl

data/dictionaries/de/level_types/level_types.txt

@@ -0,0 +1 @@
+untergeschoss|ug|u g

data/dictionaries/de/personal_titles/personal_titles.txt

@@ -1,7 +1,18 @@
 doktor|dr
 burgermeister|bgm
 direktor|dir
+frau|fr
+fraulein|fri
+heilige|hl
+heiliger
+heiliges
 herr|hr
 herrn|hrn
 ingnieur|ing
+kapitanleutnant|ka leut
+mitglied des bundestages|mdb
+mitglied des landtages|mdl
+oberburgermeister|ob|obgm
+pater
+professor|prof
 sankt|st

data/dictionaries/de/place_names/place_names.txt

@@ -19,10 +19,36 @@ hafen
 haus
 handelsakademie|hak
 handelsschule|hasch
+hauptbahnhof|hbf
+hohle
+hohere technische lehranstalt|htl|h t l
 hotel
+hutte|htt
+jagdhaus
+jagdhutte|jhtt
 jungenherberge|jh|j h
 kaffee
+kapelle|kap|kpl
+klaranlage|ka
+konzentrationslager|kz|kl
+magistratsabteilung|ma
 markt|mkt
 marktplatz|markt platz|markt pl|mkt pl|marktpl
+nationalpark|np|national park
+naturschutzgebiet|nsg
+neue mittelschule|nms
+padagogische hochschule|ph
 sportplatz|sport platz|sport pl|sportpl
-restaurant
+resevoir|res
+restaurant
+ruine
+schloss|schl
+steinbruch|stb
+technische universitat|tu|t u
+truppenubungsplatz|tupl
+universitat|uni
+universitatsbibliothek|ub|u b
+volksschule|vs
+wald
+wasserfall|wsf|wssf
+wirtshaus|wh

data/dictionaries/de/stopwords/stopwords.txt

@@ -18,18 +18,19 @@ fur
 gegenuber
 im|i
 in|i
-in der|id|i d
+in der|i d
 mit
 nach
 nachst
 neben
 ob|o
+oder|od
 uber
 und|&
 unter
 vor|v
 von|v
-von der|v d|vd
+von der|v d
 zu
 zu der
 zur

data/dictionaries/de/street_types/street_types.txt

@@ -14,7 +14,6 @@ graben|gr
 grosser
 grosse|gr|g
 grosses
-hauptbahnhof|hbf
 heiligen
 hof|h
 kamp|k
@@ -24,10 +23,10 @@ kleines
 obere|ob
 oberer|ob
 oberes
-pfad|p
+pfad
 platz|pl
 quelle|qu
-rhein|rh
+rundwanderweg|rww
 strasse|str
 untere|u
 unterer

data/dictionaries/de/synonyms/synonyms.txt

@@ -1,3 +1,14 @@
 deutsch|dt
 ehemalige|ehem
-haltestelle|hst
+gebruder|gebr
+haltestelle|hst
+hinter|hint|ht
+internationale|int
+kleine|kl
+kogel|kg
+niedere|nd
+rhein|rh
+see|s
+spitze|sp
+vordere|vd|vord
+wiese|ws

data/dictionaries/de/toponyms/toponyms.txt

@@ -1 +1,2 @@
-burgenlaendische|burgenlandische|bgld
+burgenlaendische|burgenlandische|bgld
+wiener|wr

data/dictionaries/de/unit_types/unit_types.txt

@@ -1,2 +1,7 @@
 abteilung|abt
-buro
+buro
+nummer|nr|#
+stiege|stg
+wohnung|whg
+zahl|z
+zimmer|zi

src/gazetteers.h

@@ -34,17 +34,18 @@ typedef enum dictionary_type {
 
     DICTIONARY_STREET_NAME = 10,
     DICTIONARY_STREET_TYPE = 11,
-    DICTIONARY_CONCATENATED_SEPARABLE = 12,
-    DICTIONARY_CONCATENATED_INSEPARABLE = 13,
-    DICTIONARY_DIRECTIONAL = 14,
-    DICTIONARY_QUALIFIER = 15,
-    DICTIONARY_BUILDING_TYPE = 16,
-    DICTIONARY_LEVEL = 17,
-    DICTIONARY_UNIT = 18,
-    DICTIONARY_POST_OFFICE = 19,
-    DICTIONARY_NO_ADDRESS = 20,
-    DICTIONARY_NULL = 21,
-    
+    DICTIONARY_CONCATENATED_PREFIX_SEPARABLE = 12,
+    DICTIONARY_CONCATENATED_SUFFIX_SEPARABLE = 13,
+    DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE = 14,
+    DICTIONARY_DIRECTIONAL = 15,
+    DICTIONARY_QUALIFIER = 16,
+    DICTIONARY_BUILDING_TYPE = 17,
+    DICTIONARY_LEVEL = 18,
+    DICTIONARY_UNIT = 19,
+    DICTIONARY_POST_OFFICE = 20,
+    DICTIONARY_NO_ADDRESS = 21,
+    DICTIONARY_NULL = 22,
+
     DICTIONARY_PLACE_NAME = 50,
     DICTIONARY_COMPANY_TYPE = 51,
     DICTIONARY_GIVEN_NAME = 52,
@@ -77,8 +78,9 @@ gazetteer_t gazetteers[] = {
     {"academic_degrees", DICTIONARY_ACADEMIC_DEGREE, ADDRESS_NAME},
     {"building_types", DICTIONARY_BUILDING_TYPE, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT},
     {"company_types", DICTIONARY_COMPANY_TYPE, ADDRESS_NAME},
-    {"concatenated_suffixes_inseparable", DICTIONARY_CONCATENATED_INSEPARABLE, ADDRESS_STREET},
-    {"concatenated_suffixes_separable", DICTIONARY_CONCATENATED_SEPARABLE, ADDRESS_STREET},
+    {"concatenated_prefixes_inseparable", DICTIONARY_CONCATENATED_PREFIX_SEPARABLE, ADDRESS_STREET},
+    {"concatenated_suffixes_inseparable", DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE, ADDRESS_STREET},
+    {"concatenated_suffixes_separable", DICTIONARY_CONCATENATED_SUFFIX_SEPARABLE, ADDRESS_STREET},
     {"directionals", DICTIONARY_DIRECTIONAL, ADDRESS_ANY},
     {"elisions", DICTIONARY_ELISION, ADDRESS_ANY},
     {"given_names", DICTIONARY_GIVEN_NAME, ADDRESS_STREET | ADDRESS_NAME},