xref: /sqlite-3.40.0/test/enc.test (revision 0ea2d42a) 
1# 2002 May 24
2#
3# The author disclaims copyright to this source code.  In place of
4# a legal notice, here is a blessing:
5#
6#    May you do good and not evil.
7#    May you find forgiveness for yourself and forgive others.
8#    May you share freely, never taking more than you give.
9#
10#***********************************************************************
11# This file implements regression tests for SQLite library.  The focus of
12# this file is testing the SQLite routines used for converting between the
13# various suported unicode encodings (UTF-8, UTF-16, UTF-16le and
14# UTF-16be).
15#
16# $Id: enc.test,v 1.7 2007/05/23 16:23:09 danielk1977 Exp $
17
18set testdir [file dirname $argv0]
19source $testdir/tester.tcl
20
21# Skip this test if the build does not support multiple encodings.
22#
23ifcapable {!utf16} {
24  finish_test
25  return
26}
27
28proc do_bincmp_test {testname got expect} {
29  binary scan $expect \c* expectvals
30  binary scan $got \c* gotvals
31  do_test $testname [list set dummy $gotvals] $expectvals
32}
33
34# $utf16 is a UTF-16 encoded string. Swap each pair of bytes around
35# to change the byte-order of the string.
36proc swap_byte_order {utf16} {
37  binary scan $utf16 \c* ints
38
39  foreach {a b} $ints {
40    lappend ints2 $b
41    lappend ints2 $a
42  }
43
44  return [binary format \c* $ints2]
45}
46
47#
48# Test that the SQLite routines for converting between UTF encodings
49# produce the same results as their TCL counterparts.
50#
51# $testname is the prefix to be used for the test names.
52# $str is a string to use for testing (encoded in UTF-8, as normal for TCL).
53#
54# The test procedure is:
55# 1. Convert the string from UTF-8 to UTF-16le and check that the TCL and
56#    SQLite routines produce the same results.
57#
58# 2. Convert the string from UTF-8 to UTF-16be and check that the TCL and
59#    SQLite routines produce the same results.
60#
61# 3. Use the SQLite routines to convert the native machine order UTF-16
62#    representation back to the original UTF-8. Check that the result
63#    matches the original representation.
64#
65# 4. Add a byte-order mark to each of the UTF-16 representations and
66#    check that the SQLite routines can convert them back to UTF-8.  For
67#    byte-order mark info, refer to section 3.10 of the unicode standard.
68#
69# 5. Take the byte-order marked UTF-16 strings from step 4 and ensure
70#    that SQLite can convert them both to native byte order UTF-16
71#    strings, sans BOM.
72#
73# Coverage:
74#
75# sqlite_utf8to16be (step 2)
76# sqlite_utf8to16le (step 1)
77# sqlite_utf16to8 (steps 3, 4)
78# sqlite_utf16to16le (step 5)
79# sqlite_utf16to16be (step 5)
80#
81proc test_conversion {testname str} {
82
83  # Step 1.
84  set utf16le_sqlite3 [test_translate $str UTF8 UTF16LE]
85  set utf16le_tcl [encoding convertto unicode $str]
86  append utf16le_tcl "\x00\x00"
87  if { $::tcl_platform(byteOrder)!="littleEndian" } {
88    set utf16le_tcl [swap_byte_order $utf16le_tcl]
89  }
90  do_bincmp_test $testname.1 $utf16le_sqlite3 $utf16le_tcl
91  set utf16le $utf16le_tcl
92
93  # Step 2.
94  set utf16be_sqlite3 [test_translate $str UTF8 UTF16BE]
95  set utf16be_tcl [encoding convertto unicode $str]
96  append utf16be_tcl "\x00\x00"
97  if { $::tcl_platform(byteOrder)=="littleEndian" } {
98    set utf16be_tcl [swap_byte_order $utf16be_tcl]
99  }
100  do_bincmp_test $testname.2 $utf16be_sqlite3 $utf16be_tcl
101  set utf16be $utf16be_tcl
102
103  # Step 3.
104  if { $::tcl_platform(byteOrder)=="littleEndian" } {
105    set utf16 $utf16le
106  } else {
107    set utf16 $utf16be
108  }
109  set utf8_sqlite3 [test_translate $utf16 UTF16 UTF8]
110  do_bincmp_test $testname.3 $utf8_sqlite3 [binarize $str]
111
112  # Step 4 (little endian).
113  append utf16le_bom "\xFF\xFE" $utf16le
114  set utf8_sqlite3 [test_translate $utf16le_bom UTF16 UTF8 1]
115  do_bincmp_test $testname.4.le $utf8_sqlite3 [binarize $str]
116
117  # Step 4 (big endian).
118  append utf16be_bom "\xFE\xFF" $utf16be
119  set utf8_sqlite3 [test_translate $utf16be_bom UTF16 UTF8]
120  do_bincmp_test $testname.4.be $utf8_sqlite3 [binarize $str]
121
122  # Step 5 (little endian to little endian).
123  set utf16_sqlite3 [test_translate $utf16le_bom UTF16LE UTF16LE]
124  do_bincmp_test $testname.5.le.le $utf16_sqlite3 $utf16le
125
126  # Step 5 (big endian to big endian).
127  set utf16_sqlite3 [test_translate $utf16be_bom UTF16 UTF16BE]
128  do_bincmp_test $testname.5.be.be $utf16_sqlite3 $utf16be
129
130  # Step 5 (big endian to little endian).
131  set utf16_sqlite3 [test_translate $utf16be_bom UTF16 UTF16LE]
132  do_bincmp_test $testname.5.be.le $utf16_sqlite3 $utf16le
133
134  # Step 5 (little endian to big endian).
135  set utf16_sqlite3 [test_translate $utf16le_bom UTF16 UTF16BE]
136  do_bincmp_test $testname.5.le.be $utf16_sqlite3 $utf16be
137}
138
139translate_selftest
140
141test_conversion enc-1 "hello world"
142test_conversion enc-2 "sqlite"
143test_conversion enc-3 ""
144test_conversion enc-X "\u0100"
145test_conversion enc-4 "\u1234"
146test_conversion enc-5 "\u4321abc"
147test_conversion enc-6 "\u4321\u1234"
148test_conversion enc-7 [string repeat "abcde\u00EF\u00EE\uFFFCabc" 100]
149test_conversion enc-8 [string repeat "\u007E\u007F\u0080\u0081" 100]
150test_conversion enc-9 [string repeat "\u07FE\u07FF\u0800\u0801\uFFF0" 100]
151test_conversion enc-10 [string repeat "\uE000" 100]
152
153proc test_collate {enc zLeft zRight} {
154  return [string compare $zLeft $zRight]
155}
156add_test_collate $::DB 0 0 1
157do_test enc-11.1 {
158  execsql {
159    CREATE TABLE ab(a COLLATE test_collate, b);
160    INSERT INTO ab VALUES(CAST (X'C388' AS TEXT), X'888800');
161    INSERT INTO ab VALUES(CAST (X'C0808080808080808080808080808080808080808080808080808080808080808080808080808080808080808080808080808080808388' AS TEXT), X'888800');
162    CREATE INDEX ab_i ON ab(a, b);
163  }
164} {}
165do_test enc-11.2 {
166  set cp200 "\u00C8"
167  execsql {
168    SELECT count(*) FROM ab WHERE a = $::cp200;
169  }
170} {2}
171
172#-------------------------------------------------------------------------
173reset_db
174forcedelete test.db2
175forcedelete test.db3
176
177do_execsql_test enc-12.0 {
178  PRAGMA encoding = 'utf-8';
179  CREATE TABLE t1(a, b, c);
180  INSERT INTO t1 VALUES('a', 'b', 'c');
181  ATTACH 'test.db3' AS aux;
182  CREATE TABLE aux.t3(x, y, z);
183  INSERT INTO t3 VALUES('xxx', 'yyy', 'zzz');
184  PRAGMA encoding;
185} {UTF-8}
186
187do_test enc-12.1 {
188  sqlite3 db2 test.db2
189  db2 eval {
190    PRAGMA encoding = 'UTF-16le';
191    CREATE TABLE t2(d, e, f);
192    INSERT INTO t2 VALUES('d', 'e', 'f');
193    PRAGMA encoding;
194  }
195} {UTF-16le}
196
197do_test enc-12.2 {
198  db2 backup test.db
199  db2 close
200} {}
201
202do_catchsql_test enc-12.3 {
203  SELECT * FROM t2;
204} {1 {attached databases must use the same text encoding as main database}}
205
206db close
207sqlite3 db test.db3
208do_execsql_test enc-12.4 {
209  SELECT * FROM t3;
210  PRAGMA encoding = 'UTF-16le';
211  SELECT * FROM t3;
212} {xxx yyy zzz xxx yyy zzz}
213
214db close
215sqlite3 db test.db3
216breakpoint
217do_execsql_test enc-12.5 {
218  PRAGMA encoding = 'UTF-16le';
219  PRAGMA encoding;
220} {UTF-8}
221
222reset_db
223do_execsql_test enc-12.6 {
224  PRAGMA encoding = 'UTF-8';
225  CREATE TEMP TABLE t1(a, b, c);
226  INSERT INTO t1 VALUES('xxx', 'yyy', 'zzz');
227}
228do_test enc-12.7 {
229  sqlite3 db2 test.db2
230  db2 backup test.db
231  db2 close
232  db eval {
233    SELECT * FROM t1;
234  }
235} {xxx yyy zzz}
236do_catchsql_test enc-12.8 {
237  SELECT * FROM t2;
238  SELECT * FROM t1;
239} {1 {attached databases must use the same text encoding as main database}}
240
241db close
242sqlite3 db test.db
243do_execsql_test enc-12.9 {
244  CREATE TEMP TABLE t1(a, b, c);
245  INSERT INTO t1 VALUES('xxx', 'yyy', 'zzz');
246}
247do_execsql_test enc-12.10 {
248  SELECT * FROM t2;
249  SELECT * FROM t1;
250} {d e f xxx yyy zzz}
251
252finish_test
253