summaryrefslogtreecommitdiffstats
path: root/test/e_resolve.test
diff options
context:
space:
mode:
Diffstat (limited to 'test/e_resolve.test')
-rw-r--r--test/e_resolve.test135
1 files changed, 135 insertions, 0 deletions
diff --git a/test/e_resolve.test b/test/e_resolve.test
new file mode 100644
index 0000000..f4bb1a4
--- /dev/null
+++ b/test/e_resolve.test
@@ -0,0 +1,135 @@
+# 2010 November 30
+#
+# The author disclaims copyright to this source code. In place of
+# a legal notice, here is a blessing:
+#
+# May you do good and not evil.
+# May you find forgiveness for yourself and forgive others.
+# May you share freely, never taking more than you give.
+#
+#***********************************************************************
+#
+# This file implements tests to verify that the "testable statements" in
+# the lang_naming.html document are correct.
+#
+
+set testdir [file dirname $argv0]
+source $testdir/tester.tcl
+set ::testprefix e_resolve
+
+# An example database schema for testing name resolution:
+#
+set schema {
+ ATTACH 'test.db2' AS at1;
+ ATTACH 'test.db3' AS at2;
+
+ CREATE TABLE temp.n1(x, y); INSERT INTO temp.n1 VALUES('temp', 'n1');
+ CREATE TRIGGER temp.n3 AFTER INSERT ON n1 BEGIN SELECT 1; END;
+ CREATE INDEX temp.n4 ON n1(x, y);
+
+ CREATE TABLE main.n1(x, y); INSERT INTO main.n1 VALUES('main', 'n1');
+ CREATE TABLE main.n2(x, y); INSERT INTO main.n2 VALUES('main', 'n2');
+ CREATE INDEX main.n3 ON n2(y, x);
+ CREATE TRIGGER main.n4 BEFORE INSERT ON n2 BEGIN SELECT 1; END;
+
+ CREATE TABLE at1.n1(x, y); INSERT INTO at1.n1 VALUES('at1', 'n1');
+ CREATE TABLE at1.n2(x, y); INSERT INTO at1.n2 VALUES('at1', 'n2');
+ CREATE TABLE at1.n3(x, y); INSERT INTO at1.n3 VALUES('at1', 'n3');
+
+ CREATE TABLE at2.n1(x, y); INSERT INTO at2.n1 VALUES('at2', 'n1');
+ CREATE TABLE at2.n2(x, y); INSERT INTO at2.n2 VALUES('at2', 'n2');
+ CREATE TABLE at2.n3(x, y); INSERT INTO at2.n3 VALUES('at2', 'n3');
+ CREATE TABLE at2.n4(x, y); INSERT INTO at2.n4 VALUES('at2', 'n4');
+ CREATE TRIGGER at2.n4 BEFORE INSERT ON n4 BEGIN SELECT 1; END;
+}
+
+proc resolve_reopen_db {} {
+ db close
+ forcedelete test.db test.db2 test.db3
+ sqlite3 db test.db
+ db eval $::schema
+}
+
+
+
+# EVIDENCE-OF: R-33528-20612 If no database is specified as part of the
+# object reference, then SQLite searches the main, temp and all attached
+# databases for an object with a matching name. The temp database is
+# searched first, followed by the main database, followed all attached
+# databases in the order that they were attached. The reference resolves
+# to the first match found.
+#
+resolve_reopen_db
+do_execsql_test 1.1 { SELECT * FROM n1 } {temp n1}
+do_execsql_test 1.2 { SELECT * FROM n2 } {main n2}
+do_execsql_test 1.3 { SELECT * FROM n3 } {at1 n3}
+do_execsql_test 1.4 { SELECT * FROM n4 } {at2 n4}
+
+# EVIDENCE-OF: R-00634-08585 If a schema name is specified as part of an
+# object reference, it must be either "main", or "temp" or the
+# schema-name of an attached database.
+#
+# Or else it is a "no such table: xxx" error.
+#
+resolve_reopen_db
+do_execsql_test 2.1.1 { SELECT * FROM main.n1 } {main n1}
+do_execsql_test 2.1.2 { SELECT * FROM temp.n1 } {temp n1}
+do_execsql_test 2.1.3 { SELECT * FROM at1.n1 } {at1 n1}
+do_execsql_test 2.1.4 { SELECT * FROM at2.n1 } {at2 n1}
+
+do_catchsql_test 2.2 { SELECT * FROM xxx.n1 } {1 {no such table: xxx.n1}}
+
+# EVIDENCE-OF: R-17446-42210 Like other SQL identifiers, schema names
+# are case-insensitive.
+#
+resolve_reopen_db
+do_execsql_test 3.1 { SELECT * FROM MAIN.n1 } {main n1}
+do_execsql_test 3.2 { SELECT * FROM tEmP.n1 } {temp n1}
+do_execsql_test 3.3 { SELECT * FROM aT1.n1 } {at1 n1}
+do_execsql_test 3.4 { SELECT * FROM At2.n1 } {at2 n1}
+
+# EVIDENCE-OF: R-14755-58619 If a schema name is specified, then only
+# that one schema is searched for the named object.
+#
+do_catchsql_test 4.1 { SELECT * FROM temp.n2 } {1 {no such table: temp.n2}}
+do_catchsql_test 4.2 { SELECT * FROM main.n2 } {0 {main n2}}
+do_catchsql_test 4.3 { SELECT * FROM at1.n2 } {0 {at1 n2}}
+do_catchsql_test 4.4 { SELECT * FROM at2.n2 } {0 {at2 n2}}
+
+# EVIDENCE-OF: R-08951-19801 When searching database schemas for a named
+# object, objects of types that cannot be used in the context of the
+# reference are always ignored.
+#
+# In this case, "types that cannot be used" are triggers and indexes.
+# The temp and main databases both contain triggers and indexes named
+# "n3" and "n4". Database "at2" contains a trigger called "n4". And yet:
+#
+do_execsql_test 5.1 { SELECT * FROM n3 } {at1 n3}
+do_execsql_test 5.2 { SELECT * FROM n4 } {at2 n4}
+
+#-------------------------------------------------------------------------
+# EVIDENCE-OF: R-37286-42536
+#
+db close
+forcedelete test.db file.db
+sqlite3 db test.db
+do_execsql_test 6.1 {
+ ATTACH 'file.db' AS aux;
+ CREATE TABLE t1(x, y);
+ CREATE TEMP TABLE t1(x, y);
+ CREATE TABLE aux.t1(x, y);
+}
+
+do_execsql_test 6.2.0 { DROP TABLE t1 }
+do_catchsql_test 6.2.1 { SELECT * FROM temp.t1 } {1 {no such table: temp.t1}}
+do_catchsql_test 6.2.2 { SELECT * FROM main.t1 } {0 {}}
+do_catchsql_test 6.2.3 { SELECT * FROM aux.t1 } {0 {}}
+
+do_execsql_test 6.3.0 { DROP TABLE t1 }
+do_catchsql_test 6.3.1 { SELECT * FROM main.t1 } {1 {no such table: main.t1}}
+do_catchsql_test 6.3.3 { SELECT * FROM aux.t1 } {0 {}}
+
+do_execsql_test 6.4.0 { DROP TABLE t1 }
+do_catchsql_test 6.4.1 { SELECT * FROM aux.t1 } {1 {no such table: aux.t1}}
+
+finish_test