New Oracle Text script archive

text/oracle-text-scripts/Db Crawler/demo_database.sql

@@ -0,0 +1,43 @@
+-- JDBC connection string looks like:
+-- jdbc:oracle:thin:@localhost:1521:ses
+
+
+alter user scott account unlock;
+grant connect,resource to scott identified by tiger;
+connect scott/tiger
+
+create table my_table (
+ key varchar2(2000) primary key,
+ url varchar2(2000),
+ content varchar2(4000),
+ lastmodifieddate date,
+ lang varchar2(2000),
+ title varchar2(2000),
+ auth_list varchar2(2000)
+);
+
+-- remember, auth_list is our "Security Attributes" column
+
+-- the first document is visible only to the user sesuser1nt
+
+insert into my_table values ('doc1', 'doc1', 'the quick brown fox jumps over the lazy dog', sysdate, 'EN', 'Database Document One', 'sesuser1');
+
+-- the second document is visible to anyone who has the "managers" role
+
+insert into my_table values ('doc2', 'doc2', 'le renard brun rapide saute par-dessus le chien paresseux', sysdate, 'FR', 'Database Document Two', 'manager');
+
+-- The following table represents the security informat
+-- Each user can have a set of roles, plus his own username
+
+create table user_role_map (
+ username varchar2(2000),
+ rolelist varchar2(2000)
+);
+
+insert into user_role_map values (
+ 'sesuser1', 'sesuser1 superuser manager employee');
+insert into user_role_map values (
+ 'sesuser2', 'sesuser2 manager employee'
+);
+
+commit;

text/oracle-text-scripts/Db Crawler/demo_database_two_tables.sql

@@ -0,0 +1,70 @@
+-- JDBC connection string looks like:
+-- jdbc:oracle:thin:@localhost:1521:ses
+
+
+alter user scott account unlock;
+grant connect,resource to scott identified by tiger;
+connect scott/tiger
+
+drop table my_table;
+
+create table my_table (
+ key varchar2(2000) primary key,
+ url varchar2(2000),
+ content varchar2(4000),
+ lastmodifieddate date,
+ lang varchar2(2000),
+ title varchar2(2000),
+ auth_list varchar2(2000)
+);
+
+-- remember, auth_list is our "Security Attributes" column
+
+-- the first document is visible only to the user sesuser1nt
+
+insert into my_table values ('doc1', 'doc1', 'the quick brown fox jumps over the lazy dog', sysdate, 'EN', 'Database Document One', 'sesuser1');
+
+-- the second document is visible to anyone who has the "managers" role
+
+insert into my_table values ('doc2', 'doc2', 'le renard brun rapide saute par-dessus le chien paresseux', sysdate, 'FR', 'Database Document Two', 'manager');
+
+drop table my_table2;
+
+create table my_table2 (
+ key varchar2(2000) primary key,
+ url varchar2(2000),
+ content varchar2(4000),
+ lastmodifieddate date,
+ lang varchar2(2000),
+ title varchar2(2000),
+ auth_list varchar2(2000)
+);
+
+-- remember, auth_list is our "Security Attributes" column
+
+-- the first document is visible only to the user sesuser1nt
+
+insert into my_table values ('doc1_2', 'doc1_2', 'the quick brown fox jumps over the lazy dog', sysdate, 'EN', 'Database Document One', 'sesuser1 sesuser3');
+
+-- the second document is visible to anyone who has the "managers" role
+
+insert into my_table values ('doc2_2', 'doc2_2', 'le renard brun rapide saute par-dessus le chien paresseux', sysdate, 'FR', 'Database Document Two', 'manager foo');
+
+
+-- The following table represents the security informat
+-- Each user can have a set of roles, plus his own username
+
+drop table user_role_map;
+
+create table user_role_map (
+ username varchar2(2000),
+ rolelist varchar2(2000)
+);
+
+insert into user_role_map values (
+ 'sesuser1', 'sesuser1 superuser manager employee');
+insert into user_role_map values (
+ 'sesuser2', 'sesuser2 manager employee'
+);
+
+commit;

text/oracle-text-scripts/GermanCompound/decompound.sql

@@ -0,0 +1,45 @@
+-- this should be a test of german decompounding, but it doesn't seem to
+-- work
+-- Although when we index 'hauptbahnhof' we get tokens of type 9 for 
+-- haupt and bahnhof, we can't seem to search on them
+
+drop table detext;
+
+create table detext
+ ( id number(4) not null
+ , name varchar2(30)
+ )
+/
+
+insert into detext values (1, 'Hauptbahnhof');
+
+exec ctx_ddl.drop_preference('mylex')
+
+begin
+ ctx_ddl.create_preference('mylex', 'BASIC_LEXER');
+ ctx_ddl.set_attribute ( 'mylex', 'index_stems', 'GERMAN');
+ ctx_ddl.set_attribute ( 'mylex', 'composite', 'GERMAN');
+ ctx_ddl.set_attribute ( 'mylex', 'alternate_spelling', 'GERMAN');
+end;
+/
+
+exec ctx_ddl.drop_preference('myword')
+
+begin 
+ ctx_ddl.create_preference('myword', 'BASIC_WORDLIST'); 
+ ctx_ddl.set_attribute('myword','FUZZY_MATCH','GERMAN');
+end; 
+/
+
+create index detext_idx1 on detext(name) 
+ indextype is ctxsys.context 
+ parameters ('WORDLIST myword DATASTORE ctxsys.default_datastore LEXER mylex SYNC (on commit)')
+/
+
+select token_text, token_type from dr$detext_idx1$i;
+
+-- this doesn't work
+SELECT id, name, score(1) as score FROM detext WHERE contains(name, '$bahnhof', 1) > 0;
+
+-- but this does
+SELECT id, name, score(1) as score FROM detext WHERE contains(name, '$haupt', 1) > 0;

text/oracle-text-scripts/GermanCompound/german_names.sql~

@@ -0,0 +1,36 @@
+drop table emp;
+
+create table emp
+ ( id number(4) not null
+ , name varchar2(30)
+ )
+/
+
+
+insert into emp values (1, 'Hauptbahnhof');
+
+exec ctx_ddl.drop_preference('mylex')
+
+begin
+ ctx_ddl.create_preference('mylex', 'BASIC_LEXER');
+ ctx_ddl.set_attribute ( 'mylex', 'index_stems', 'GERMAN');
+ ctx_ddl.set_attribute ( 'mylex', 'composite', 'GERMAN');
+ ctx_ddl.set_attribute ( 'mylex', 'alternate_spelling', 'GERMAN');
+end;
+/
+
+exec ctx_ddl.drop_preference('myword')
+
+begin 
+ ctx_ddl.create_preference('myword', 'BASIC_WORDLIST'); 
+ ctx_ddl.set_attribute('myword','FUZZY_MATCH','GERMAN');
+end; 
+/
+
+create index emp_idx1 on emp(name) 
+ indextype is ctxsys.context 
+ parameters ('WORDLIST myword DATASTORE ctxsys.default_datastore LEXER mylex STOPLIST ctxsys.empty_stoplist SYNC (on commit)')
+/
+
+-- The ? indicates fuzzy matching
+SELECT id, name, score(1) as score FROM emp WHERE contains(name, 'strasse', 1) > 0;

text/oracle-text-scripts/KWIC_html_test/Create.sql

@@ -0,0 +1,36 @@
+create user testuser identified by testuser default tablespace users temporary tablespace temp;
+
+grant connect, resource, ctxapp to testuser;
+
+connect testuser/testuser
+
+create table EmailArchive (
+ PK NUMBER,
+ FOLDER VARCHAR2(255),
+ SUBJECT VARCHAR2(255),
+ MAILDATE DATE,
+ MAILTO VARCHAR2(255),
+ MAILFROM VARCHAR2(255),
+ MAILCC VARCHAR2(255),
+ TEXT CLOB);
+
+insert into emailarchive values (1, 'INBOX', 'The QB Fox', '10-JUL-2002','john.smith@oracle.com', 'roger.ford@oracle.com', null, 
+'The quick brown fox jumps over the lazy dog.
+The lazy dog was surprised to see a quick brown 
+fox jumping over it, and thought to itself "goodness,
+that looked like a wolf - or at least a brown fox -
+jumping over me".');
+
+insert into emailarchive values (1, 'INBOX', 'Now is the time...', '11-JUL-2002','roger.ford@oracle.com', 'john.smith@oracle.com', null, 
+'Now is the time for all good men to come to the aid of the party.
+The party is not doing very well at the moment as it has run out of
+drink, and most of the guests have gone home. What we need is
+for someone to come along with more drink, and hopefully with a 
+quick brown fox as a companion.
+');
+
+commit;
+
+create index email_text on emailarchive(text)
+indextype is ctxsys.context;
+