path: root/tests/utf.test

# This file contains a collection of tests for tclUtf.c
# Sourcing this file into Tcl runs the tests and generates output for
# errors.  No output means no errors were found.
#
# Copyright (c) 1997 Sun Microsystems, Inc.
#
# See the file "license.terms" for information on usage and redistribution
# of this file, and for a DISCLAIMER OF ALL WARRANTIES.
#
# SCCS: @(#) utf.test 1.7 98/01/15 18:41:53
#

if {[info procs test] != "test"} {
    source defs
}

test utf-1.1 {Tcl_UniCharToUtf: 1 byte sequences} {
    set x \x01
} [bytestring "\x01"]
test utf-1.2 {Tcl_UniCharToUtf: 2 byte sequences} {
    set x "\x00"
} [bytestring "\xc0\x80"]
test utf-1.3 {Tcl_UniCharToUtf: 2 byte sequences} {
    set x "\xe0"
} [bytestring "\xc3\xa0"]
test utf-1.4 {Tcl_UniCharToUtf: 3 byte sequences} {
    set x "\u4e4e"
} [bytestring "\xe4\xb9\x8e"]

test utf-2.1 {Tcl_UtfToUniChar: low ascii} {
    string length "abc"
} {3}
test utf-2.2 {Tcl_UtfToUniChar: naked trail bytes} {
    string length [bytestring "\x82\x83\x84"]
} {3}
test utf-2.3 {Tcl_UtfToUniChar: lead (2-byte) followed by non-trail} {
    string length [bytestring "\xC2"]
} {1}
test utf-2.4 {Tcl_UtfToUniChar: lead (2-byte) followed by trail} {
    string length [bytestring "\xC2\xa2"]
} {1}
test utf-2.5 {Tcl_UtfToUniChar: lead (3-byte) followed by non-trail} {
    string length [bytestring "\xE2"]
} {1}
test utf-2.6 {Tcl_UtfToUniChar: lead (3-byte) followed by 1 trail} {
    string length [bytestring "\xE2\xA2"]
} {2}
test utf-2.7 {Tcl_UtfToUniChar: lead (3-byte) followed by 2 trail} {
    string length [bytestring "\xE4\xb9\x8e"]
} {1}
test utf-2.8 {Tcl_UtfToUniChar: longer UTF sequences not supported} {
    string length [bytestring "\xF4\xA2\xA2\xA2"]
} {4}

test utf-3.1 {Tcl_UtfCharComplete} {
} {}

test utf-4.1 {Tcl_NumUtfChars: zero length} {
    string length ""
} {0}
test utf-4.2 {Tcl_NumUtfChars: length 1} {
    string length [bytestring "\xC2\xA2"]
} {1}
test utf-4.3 {Tcl_NumUtfChars: long string} {
    string length [bytestring "abc\xC2\xA2\xe4\xb9\x8e\uA2\u4e4e"]
} {7}

test utf-5.1 {Tcl_UtfFindFirsts} {
} {}

test utf-6.1 {Tcl_UtfNext} {
} {}

test utf-7.1 {Tcl_UtfPrev} {
} {}

test utf-8.1 {Tcl_UniCharAtIndex: index = 0} {
    string index abcd 0
} {a}
test utf-8.2 {Tcl_UniCharAtIndex: index = 0} {
    string index \u4e4e\u25a 0
} "\u4e4e"
test utf-8.3 {Tcl_UniCharAtIndex: index > 0} {
    string index abcd 2
} {c}
test utf-8.4 {Tcl_UniCharAtIndex: index > 0} {
    string index \u4e4e\u25a\xff\u543 2
} "\uff"

test utf-9.1 {Tcl_UtfAtIndex: index = 0} {
    string range abcd 0 2
} {abc}
test utf-9.2 {Tcl_UtfAtIndex: index > 0} {
    string range \u4e4e\u25a\xff\u543klmnop 1 5
} "\u25a\xff\u543kl"


test utf-10.1 {Tcl_UtfBackslash: dst == NULL} {
    set x \n
} {
}
test utf-10.2 {Tcl_UtfBackslash: \u subst} {
    set x \ua2
} [bytestring "\xc2\xa2"]
test utf-10.3 {Tcl_UtfBackslash: longer \u subst} {
    set x \u4e21
} [bytestring "\xe4\xb8\xa1"]
test utf-10.4 {Tcl_UtfBackslash: stops at first non-hex} {
    set x \u4e2k
} "[bytestring \xd3\xa2]k"
test utf-10.5 {Tcl_UtfBackslash: stops after 4 hex chars} {
    set x \u4e216
} "[bytestring \xe4\xb8\xa1]6"
proc bsCheck {char num} {
    global errNum
    test utf-10.$errNum {backslash substitution} {
	scan $char %c value
	set value
    } $num
    incr errNum
}
set errNum 6
bsCheck \b	8
bsCheck \e	101
bsCheck \f	12
bsCheck \n	10
bsCheck \r	13
bsCheck \t	9
bsCheck \v	11
bsCheck \{	123
bsCheck \}	125
bsCheck \[	91
bsCheck \]	93
bsCheck \$	36
bsCheck \ 	32
bsCheck \;	59
bsCheck \\	92
bsCheck \Ca	67
bsCheck \Ma	77
bsCheck \CMa	67
bsCheck \8a	8
bsCheck \14	12
bsCheck \141	97
bsCheck b\0	98
bsCheck \x	120
bsCheck \xa	10
bsCheck \xA	10
bsCheck \x41	65
bsCheck \x541	65
bsCheck \u	117
bsCheck \uk	117
bsCheck \u41	65
bsCheck \ua	10
bsCheck \uA	10
bsCheck \340	224
bsCheck \ua1	161
bsCheck \u4e21	20001

test utf-11.1 {Tcl_UtfToUpper} {
    string toupper {}
} {}
test utf-11.2 {Tcl_UtfToUpper} {
    string toupper abc
} ABC
test utf-11.3 {Tcl_UtfToUpper} {
    string toupper \u00e3ab
} \u00c3AB
test utf-11.4 {Tcl_UtfToUpper} {
    string toupper \u01e3ab
} \u01e2AB

test utf-12.1 {Tcl_UtfToLower} {
    string tolower {}
} {}
test utf-12.2 {Tcl_UtfToLower} {
    string tolower ABC
} abc
test utf-12.3 {Tcl_UtfToLower} {
    string tolower \u00c3AB
} \u00e3ab
test utf-12.4 {Tcl_UtfToLower} {
    string tolower \u01e2AB
} \u01e3ab

test utf-13.1 {Tcl_UtfToTitle} {
    string totitle {}
} {}
test utf-13.2 {Tcl_UtfToTitle} {
    string totitle abc
} Abc
test utf-13.3 {Tcl_UtfToTitle} {
    string totitle \u00e3ab
} \u00c3ab
test utf-13.4 {Tcl_UtfToTitle} {
    string totitle \u01f3ab
} \u01f2ab

test utf-14.1 {Tcl_UniCharToUpper, negative delta} {
    string toupper aA
} AA
test utf-14.2 {Tcl_UniCharToUpper, positive delta} {
    string toupper \u0178\u00ff
} \u0178\u0178
test utf-14.3 {Tcl_UniCharToUpper, no delta} {
    string toupper !
} !

test utf-15.1 {Tcl_UniCharToLower, negative delta} {
    string tolower aA
} aa
test utf-15.2 {Tcl_UniCharToLower, positive delta} {
    string tolower \u0178\u00ff
} \u00ff\u00ff
test utf-16.1 {Tcl_UniCharToLower, no delta} {
    string tolower !
} !

test utf-17.1 {Tcl_UniCharToTitle, add one for title} {
    string totitle \u01c4
} \u01c5
test utf-17.2 {Tcl_UniCharToTitle, subtract one for title} {
    string totitle \u01c6
} \u01c5
test utf-17.3 {Tcl_UniCharToTitle, subtract delta for title (positive)} {
    string totitle \u017f
} \u0053
test utf-17.4 {Tcl_UniCharToTitle, subtract delta for title (negative)} {
    string totitle \u00ff
} \u0178
test utf-17.5 {Tcl_UniCharToTitle, no delta} {
    string totitle !
} !

test utf-18.1 {TclUniCharLen} {
    list [regexp \\d abc456def foo] $foo
} {1 4}

test utf-19.1 {TclUniCharNcmp} {
} {}

test utf-20.1 {TclUniCharIsAlnum} {
} {}

test utf-21.1 {TclUniCharIsWordChar} {
    string wordend "xyz123_bar fg" 0
} 10
test utf-21.1 {TclUniCharIsWordChar} {
    string wordend "x\u5080z123_bar\u203c fg" 0
} 10
    
test utf-22.1 {TclUniCharIsAlpha} {
} {}

test utf-23.1 {TclUniCharIsDigit} {
} {}

test utf-23.1 {TclUniCharIsSpace} {
} {}


return