wikidata-json-parser

build.gradle

@@ -0,0 +1,58 @@
+apply plugin: 'java'
+apply plugin: 'eclipse'
+apply plugin: 'application'
+
+mainClassName = 'com.filbertkm.importer.Importer'
+
+sourceCompatibility = 1.7
+targetCompatibility = 1.7
+
+def env = System.getenv()
+
+def cmdBranch = "git rev-parse --abbrev-ref HEAD"
+def procBranch = cmdBranch.execute()
+def gitBranch = procBranch.text.trim()
+
+def cmdVersion = "git rev-parse HEAD"
+def procVersion = cmdVersion.execute()
+def gitRevision = procVersion.text.trim()
+
+// basic setup for the script to run
+buildscript {
+
+	dependencies {
+		repositories {
+			mavenCentral()
+		}
+	}
+}
+
+jar {
+	manifest {
+		attributes("Main-Class": "${mainClassName}",
+			   "Git-Branch": ((null != gitBranch) ? gitBranch : ""),
+			   "Git-Revision": ((null != gitRevision) ? gitRevision : ""))
+	}
+	from { configurations.compile.collect { it.isDirectory() ? it : zipTree(it) } }
+}
+
+// instruct Gradle to look in mavenCentral repository first, then the /lib directory 
+repositories {
+	mavenCentral()
+	flatDir { dirs 'lib' }
+}
+
+dependencies {
+	compile "org.wikidata.wdtk:wdtk-datamodel:0.4.0"
+	compile "org.wikidata.wdtk:wdtk-dumpfiles:0.4.0"
+	compile "org.slf4j:slf4j-log4j12:1.7.6"
+        compile "args4j:args4j:2.32"
+	compile "org.postgresql:postgresql:9.3-1103-jdbc41"
+	compile "com.fasterxml.jackson.core:jackson-core:2.5.3"
+	compile "com.fasterxml.jackson.core:jackson-annotations:2.5.3"
+	compile "com.fasterxml.jackson.core:jackson-databind:2.5.3"
+}
+
+test {
+	jvmArgs "-XX:-UseSplitVerifier"
+}

sql/index.sql

@@ -0,0 +1,8 @@
+create index idx_label on label using btree (entity_id);
+create index idx_alias on alias using btree (entity_id);
+create index idx_description on description using btree (entity_id);
+create index idx_claim_coordinate on claim_coordinate using btree (entity_id);
+create index idx_claim_datetime on claim_datetime using btree (entity_id);
+create index idx_claim_entity on claim_entity using btree (entity_id);
+create index idx_claim_string on claim_string using btree (entity_id);
+create index idx_sitelink on sitelink using btree (entity_id);

sql/schema.sql

@@ -1,34 +1,71 @@
+CREATE TABLE label(
+	id serial PRIMARY KEY,
+	entity_id TEXT NOT NULL,
+	label_language TEXT,
+	label_text TEXT
+);
+
+CREATE TABLE alias(
+	id serial PRIMARY KEY,
+	entity_id TEXT NOT NULL,
+	alias_language TEXT,
+	alias_text TEXT
+);
+
+CREATE TABLE description(
+	id serial PRIMARY KEY,
+	entity_id TEXT NOT NULL,
+	description_language TEXT,
+	description_text TEXT
+);
+
+CREATE TABLE sitelink(
+	id serial PRIMARY KEY,
+	entity_id TEXT NOT NULL,
+	site_key TEXT,
+	page_title TEXT
+);
+
 CREATE EXTENSION IF NOT EXISTS postgis;
-CREATE EXTENSION IF NOT EXISTS hstore;
 
-CREATE TABLE coordinates(
+CREATE TABLE claim_coordinate(
 	id serial PRIMARY KEY,
 	entity_id TEXT NOT NULL,
+        property_id TEXT NOT NULL,
 	globe TEXT default NULL,
 	precision double precision,
 	latitude double precision,
 	longitude double precision
 );
 
-SELECT AddGeometryColumn ('public', 'coordinates', 'geom', 4326, 'POINT', 2);
+SELECT AddGeometryColumn ('public', 'claim_coordinate', 'geom', 4326, 'POINT', 2);
 
-CREATE TABLE value_snaks(
+CREATE TABLE claim_datetime(
 	id serial PRIMARY KEY,
-	entity_id VARCHAR,
-	values HSTORE
+	entity_id TEXT NOT NULL,
+        property_id TEXT NOT NULL,
+	calendar TEXT default NULL,
+	year text,
+	month text,
+	day text,
+	hour text,
+	minute text,
+	second text,
+	precision text,
+	tolerance_before text,
+	tolerance_after text
 );
 
-CREATE TABLE terms(
+CREATE TABLE claim_entity(
 	id serial PRIMARY KEY,
-	entity_id TEXT NOT NULL,
-	term_type TEXT,
-	term_language TEXT,
-	term_text TEXT
+	entity_id VARCHAR,
+        property_id VARCHAR,
+	value TEXT
 );
 
-CREATE TABLE descriptions(
+CREATE TABLE claim_string(
 	id serial PRIMARY KEY,
-	entity_id TEXT NOT NULL,
-	term_language TEXT,
-	term_text TEXT
+	entity_id VARCHAR,
+        property_id VARCHAR,
+	value TEXT
 );