From ac2251f199398b813f394bdb875bc2723557781c Mon Sep 17 00:00:00 2001
From: Joe Robinson <joe@lc8n.com>
Date: Fri, 19 Sep 2014 01:02:42 +0100
Subject: Created script for converting WordNet database into a simpler/faster
 form

---
 convert.py | 74 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)
 create mode 100644 convert.py

diff --git a/convert.py b/convert.py
new file mode 100644
index 0000000..9e8bba4
--- /dev/null
+++ b/convert.py
@@ -0,0 +1,74 @@
+import MySQLdb as mysql
+
+def create():
+	try:
+		con = mysql.connect('localhost', 'wordnet', 'words', 'wordnet');
+
+		with con:
+			cur = con.cursor(mysql.cursors.DictCursor)
+			cur.execute("CREATE TABLE types (id int not null auto_increment, type text, abbreviation text, primary key(id))")
+			cur.execute("INSERT INTO types (type, abbreviation) VALUES('noun', 'n')")
+			cur.execute("INSERT INTO types (type, abbreviation) VALUES('verb', 'v')")
+			cur.execute("INSERT INTO types (type, abbreviation) VALUES('adjective', 'adj')")
+			cur.execute("INSERT INTO types (type, abbreviation) VALUES('adverb', 'adv')")
+			cur.execute("CREATE TABLE definitions (id bigint not null auto_increment, word text, type_id int, sub_id int, synset_id bigint, definition text, primary key(id))")
+			cur.execute("CREATE TABLE uses (id bigint not null auto_increment, definition_id bigint, quote text, primary key(id))")
+
+	except mysql.Error, e:
+		print "Database Error %d: %s" % (e.args[0],e.args[1])
+		sys.exit(1)
+
+def select():
+	try:
+		con = mysql.connect('localhost', 'wordnet', 'words', 'wordnet');
+
+		with con:
+			cur = con.cursor(mysql.cursors.DictCursor)
+			cur.execute("SELECT lemma, pos, sensenum, synsetid, definition, sampleset from dict")
+			rows = cur.fetchall()
+			print len(rows)
+	except mysql.Error, e:
+		print "Database Error %d: %s" % (e.args[0],e.args[1])
+		sys.exit(1)
+	
+	return rows
+
+def insert(rows):
+	try:
+		con = mysql.connect('localhost', 'wordnet', 'words', 'wordnet');
+
+		with con:
+			cur = con.cursor(mysql.cursors.DictCursor)
+
+			for row in rows:
+				if row['pos'] == 'n':
+					type_id = 1
+				elif row['pos'] == 'v':
+					type_id = 2
+				elif row['pos'] == 'a' or row['pos'] == 's':
+					type_id = 3
+				elif row['pos'] == 'r':
+					type_id = 4
+
+				cur.execute("INSERT INTO definitions(word, type_id, sub_id, synset_id, definition) values(%s, %s, %s, %s, %s)", [row['lemma'], type_id, row['sensenum'], row['synsetid'], row['definition']])
+
+				row_id = cur.lastrowid
+
+				if row['sampleset'] is not None:
+					uses = row['sampleset'].split("|")
+
+					for use in uses:
+						cur.execute("INSERT INTO uses(definition_id, quote) values(%s, %s)", [row_id, use])
+
+	except mysql.Error, e:
+		print "Database Error %d: %s" % (e.args[0],e.args[1])
+		sys.exit(1)
+
+def main():
+	create()
+	items = select()
+	insert(items)
+
+if __name__ == "__main__":
+	main()
+
-- 
cgit v1.2.3