testsuite/gdb.base/charset.exp

# This testcase is part of GDB, the GNU debugger.

# Copyright 2001-2017 Free Software Foundation, Inc.

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.

# Please email any bugs, comments, and/or additions to this file to:
# bug-gdb (at) gnu.org

# Test GDB's character set support.


standard_testfile .c charset-malloc.c

if { [prepare_for_testing "failed to prepare" ${testfile} [list $srcfile $srcfile2]] } {
    return -1
}

# Parse the output from a `show charset' command.  Return the host
# and target charset as a two-element list.
proc parse_show_charset_output {testname} {
    global gdb_prompt

    gdb_expect {
        -re "The host character set is \"(.*)\"\\.\[\r\n\]+The target character set is \"(.*)\"\\.\[\r\n\]+The target wide character set is \"(.*)\"\\.\[\r\n\]+$gdb_prompt $" {
            set host_charset $expect_out(1,string)
            set target_charset $expect_out(2,string)
	    set retlist [list $host_charset $target_charset]
            pass $testname
        }
        -re "The host character set is \"(.*)\"\\.\[\r\n\]+$gdb_prompt $" {
            set host_charset $expect_out(1,string)
	    set retlist [list $host_charset]
            pass $testname
        }
        -re "The target character set is \"(.*)\"\\.\[\r\n\]+$gdb_prompt $" {
            set target_charset $expect_out(1,string)
	    set retlist [list $target_charset]
            pass $testname
        }
        -re ".*$gdb_prompt $" {
            fail $testname
        }
        timeout {
            fail "$testname (timeout)"
        }
    }

    return $retlist
}


# Try the various `show charset' commands.

send_gdb "show charset\n"
set show_charset [parse_show_charset_output "show charset"]

send_gdb "show target-charset\n"
set show_target_charset \
  [lindex [parse_show_charset_output "show target-charset"] 0]

if {[lsearch -exact $show_charset $show_target_charset] >= 0} {
    pass "check `show target-charset' against `show charset'"
} else {
    fail "check `show target-charset' against `show charset'"
}

send_gdb "show host-charset\n"
set show_host_charset \
  [lindex [parse_show_charset_output "show host-charset"] 0]

if {[lsearch -exact $show_charset $show_host_charset] >= 0} {
    pass "check `show host-charset' against `show charset'"
} else {
    fail "check `show host-charset' against `show charset'"
}

# Try a malformed `set charset'.
gdb_test "set charset" \
         "Requires an argument. Valid arguments are.*" \
         "try malformed `set charset'"

# Try using `set host-charset' on an invalid character set.
gdb_test "set host-charset my_grandma_bonnie" \
         "Undefined item: \"my_grandma_bonnie\"." \
         "try `set host-charset' with invalid charset"

# Try using `set target-charset' on an invalid character set.
gdb_test "set target-charset my_grandma_bonnie" \
         "Undefined item: \"my_grandma_bonnie\"." \
         "try `set target-charset' with invalid charset"

# A Tcl array mapping the names of all the character sets we've seen
# to "1" if the character set can be used as a host character set, or
# "0" otherwise.  We can use `array names charsets' just to get a list
# of all character sets.
array set charsets {}

proc all_charset_names {} {
    global charsets
    return [array names charsets]
}

proc valid_host_charset {charset} {
    global charsets
    return [expr {[info exists charsets($charset)] && $charsets($charset)}]
}

proc valid_target_charset {charset} {
    global charsets
    return [info exists charsets($charset)]
}

send_gdb "set host-charset\n"
gdb_expect {
    -re "Requires an argument. Valid arguments are (.*)\\.\r\n$gdb_prompt $" {
	set host_charset_list $expect_out(1,string)
	regsub -all {, } $host_charset_list {,} host_charset_list
	foreach host_charset [split $host_charset_list ","] {
	    set charsets($host_charset) 1
	}
	pass "capture valid host charsets"
    }

    -re ".*$gdb_prompt $" {
	fail "capture valid host charsets"
    }

    timeout {
	fail "(timeout) capture valid host charsets"
    }
}

# If gdb was built with a phony iconv, it will only have two character
# sets: "auto" and the default.  In this situation, this set of tests
# is pointless.
if {[llength [array names charsets]] < 3} {
    untested "fewer than 3 charsets"
    return -1
}

send_gdb "set target-charset\n"
gdb_expect {
    -re "Requires an argument. Valid arguments are (.*)\\.\r\n$gdb_prompt $" {
	set target_charset_list $expect_out(1,string)
	regsub -all {, } $target_charset_list {,} target_charset_list
	foreach target_charset [split $target_charset_list ","] {
	    if {! [info exists charsets($target_charset)]} {
		set charsets($target_charset) 0
	    }
	}
	pass "capture valid target charsets"
    }

    -re ".*$gdb_prompt $" {
	fail "capture valid target charsets"
    }

    timeout {
	fail "(timeout) capture valid target charsets"
    }
}

# We don't want to test all the charset names here, since that would
# be too many combinations.  We we pick a subset.
set charset_subset {ASCII ISO-8859-1 EBCDIC-US IBM1047}
foreach host_charset $charset_subset {
    if {[valid_host_charset $host_charset]} {

        set testname "try `set host-charset $host_charset'"
        send_gdb "set host-charset $host_charset\n"
        gdb_expect {
            -re "GDB doesn't know of any character set named.*\[\r\n]+${gdb_prompt} $" {
                # How did it get into `charsets' then?
                fail "$testname (didn't recognize name)"
            }
            -re "GDB can't use `.*' as its host character set\\.\[\r\n]+${gdb_prompt} $" {
                # Well, then why does its `charsets' entry say it can?
                fail $testname
            }
            -re "${gdb_prompt} $" {
                pass $testname
            }
            timeout {
                fail "$testname (timeout)"
            }
        }

        # Check that the command actually had its intended effect:
        # $host_charset should now be the host character set.
        send_gdb "show charset\n"
        set result [parse_show_charset_output "parse `show charset' after `set host-charset $host_charset'"]
        if {! [string compare [lindex $result 0] $host_charset]} {
            pass "check effect of `set host-charset $host_charset'"
        } else {
            fail "check effect of `set host-charset $host_charset'"
        }

        # Now try setting every possible target character set,
        # given that host charset.
        foreach target_charset $charset_subset {
	    if {![valid_target_charset $target_charset]} {
		continue
	    }
            set testname "try `set target-charset $target_charset'"
            send_gdb "set target-charset $target_charset\n"
            gdb_expect {
                -re "GDB doesn't know of any character set named.*\[\r\n]+${gdb_prompt} $" {
                    fail "$testname (didn't recognize name)"
                }
                -re "GDB can't convert from the .* character set to .*\\.\[\r\n\]+${gdb_prompt} $" {
                    # This is a serious problem.  GDB should be able to convert
                    # between any arbitrary pair of character sets.
                    fail "$testname (can't convert)"
                }
                -re "${gdb_prompt} $" {
                    pass $testname
                }
                timeout {
                    fail "$testname (timeout)"
                }
            }

            # Check that the command actually had its intended effect:
            # $target_charset should now be the target charset.
            send_gdb "show charset\n"
            set result [parse_show_charset_output "parse `show charset' after `set target-charset $target_charset'"]
            if {! [string compare $result [list $host_charset $target_charset]]} {
                pass "check effect of `set target-charset $target_charset'"
            } else {
                fail "check effect of `set target-charset $target_charset'"
            }

            # Test handling of characters in the host charset which
            # can't be translated into the target charset.  \xA2 is
            # `cent' in ISO-8859-1, which has no equivalent in ASCII.
            #
            # On some systems, the pseudo-tty through which we
            # communicate with GDB insists on stripping the high bit
            # from input characters, meaning that `cent' turns into
            # `"'.  Since ISO-8859-1 and ASCII are identical in the
            # lower 128 characters, it's tough to see how we can test
            # this behavior on such systems, so we just xfail it.
	    #
	    # Note: the \x16 (Control-V) is an escape to allow \xA2 to
	    # get past readline.
            if {! [string compare $host_charset iso-8859-1] && ! [string compare $target_charset ascii]} {

                set testname "untranslatable character in character literal"
                send_gdb "print '\x16\xA2'\n"
                gdb_expect {
                    -re "There is no character corresponding to .* in the target character set .*\\.\[\r\n\]+$gdb_prompt $" {
                        pass $testname
                    }
                    -re " = 34 '\"'\[\r\n\]+$gdb_prompt $" {
                        xfail "$testname (DejaGNU's pseudo-tty strips eighth bit)"
                    }
                    -re "$gdb_prompt $" {
                        fail $testname
                    }
                    timeout {
                        fail "$testname (timeout)"
                    }
                }

                set testname "untranslatable character in string literal"
                # If the PTTY zeros bit seven, then this turns into
                #   print """
                # which gets us a syntax error.  We don't care.
                send_gdb "print \"\x16\xA2\"\n"
                gdb_expect {
                    -re "There is no character corresponding to .* in the target character set .*\\.\[\r\n\]+$gdb_prompt $" {
                        pass $testname
                    }
                    -re "Unterminated string in expression.\[\r\n\]+$gdb_prompt $" {
                        xfail "$testname (DejaGNU's pseudo-tty strips eighth bit)"
                    }
                    -re "$gdb_prompt $" {
                        fail $testname
                    }
                    timeout {
                        fail "$testname (timeout)"
                    }
                }

                set testname "untranslatable characters in backslash escape"
                send_gdb "print '\\\x16\xA2'\n"
                gdb_expect {
                    -re "The escape sequence .* is equivalent to plain .*, which has no equivalent\[\r\n\]+in the .* character set\\.\[\r\n\]+$gdb_prompt $" {
                        pass $testname
                    }
                    -re " = 34 '\"'\[\r\n\]+$gdb_prompt $" {
                        xfail "$testname (DejaGNU's pseudo-tty strips eighth bit)"
                    }
                    -re "$gdb_prompt $" {
                        fail $testname
                    }
                    timeout {
                        fail "$testname (timeout)"
                    }
                }
            }
        }
    }
}


# Set the host character set to plain ASCII, and try actually printing
# some strings in various target character sets.  We need to run the
# test program to the point at which the strings have been
# initialized.
gdb_test "break ${srcfile}:[gdb_get_line_number "all strings initialized"]" \
         ".*Breakpoint.* at .*" \
         "set breakpoint after all strings have been initialized"
gdb_run_cmd
gdb_test "" "Breakpoint.*all strings initialized.*" "run until all strings have been initialized"

# We only try the wide character tests on machines where the wchar_t
# typedef in the test case has the right size.
set wchar_size [get_sizeof wchar_t 99]
set wchar_ok 0
if {$wchar_size == 2} {
    lappend charset_subset UTF-16
    set wchar_ok 1
} elseif {$wchar_size == 4} {
    lappend charset_subset UTF-32
    set wchar_ok 1
}

gdb_test_no_output "set host-charset ASCII"
foreach target_charset $charset_subset {
    if {![valid_target_charset $target_charset]} {
	continue
    }

    if {$target_charset == "UTF-32" || $target_charset == "UTF-16"} {
	set param target-wide-charset
	set L L
    } else {
	set param target-charset
	set L ""
    }
    gdb_test_no_output "set $param $target_charset"

    # Try printing the null character.  There seems to be a bug in
    # gdb_test that requires us to use gdb_expect here.
    send_gdb "print $L'\\0'\n"
    gdb_expect {
        -re "\\\$${decimal} = 0 $L'\\\\000'\[\r\n\]+$gdb_prompt $" {
            pass "print the null character in ${target_charset}"
        }
        -re "$gdb_prompt $" {
            fail "print the null character in ${target_charset}"
        }
        timeout {
            fail "print the null character in ${target_charset} (timeout)"
        }
    }

    # Compute the name of the variable in the test program that holds
    # a string in $target_charset.  The variable's name is the
    # character set's name, in lower-case, with all non-identifier
    # characters replaced with '_', with "_string" stuck on the end.
    if {$target_charset == "UTF-16"} {
	# We still use the utf_32_string variable -- but the size is
	# correct for UTF-16.
	set var_name utf_32_string
    } else {
	set var_name [string tolower "${target_charset}_string"]
	regsub -all -- "\[^a-z0-9_\]" $var_name "_" var_name
    }

    # Compute a regexp matching the results we expect.  This is static,
    # but it's easier than writing it out.
    regsub -all "." "abfnrtv" "(\\\\&|x)" escapes
    set uppercase "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    set lowercase "abcdefghijklmnopqrstuvwxyz"
    set digits "0123456789"
    set octal_escape "\\\\\[0-9\]+"

    send_gdb "print $var_name\n"
    # ${escapes}${uppercase}${lowercase}${digits}${octal}${octal}
    gdb_expect {
        -re ".* = $L\"(\\\\a|x)(\\\\b|x)(\\\\f|x)(\\\\n|x)(\\\\r|x)(\\\\t|x)(\\\\v|x)${uppercase}${lowercase}${digits}(${octal_escape}|x)+\"\[\r\n\]+$gdb_prompt $" {
            pass "print string in $target_charset"
        }
        -re "$gdb_prompt $" {
            fail "print string in $target_charset"
        }
        timeout {
            fail "print string in $target_charset (timeout)"
        }
    }

    # Try entering a character literal, and see if it comes back unchanged.
    gdb_test "print $L'A'" \
             " = \[0-9-\]+ $L'A'" \
             "parse character literal in ${target_charset}"

    # Check that the character literal was encoded correctly.
    gdb_test "print $L'A' == $var_name\[7\]" \
             " = 1" \
             "check value of parsed character literal in ${target_charset}"

    # Try entering a string literal, and see if it comes back unchanged.
    gdb_test "print $L\"abcdefABCDEF012345\"" \
             " = $L\"abcdefABCDEF012345\"" \
             "parse string literal in ${target_charset}"

    # Check that the string literal was encoded correctly.
    gdb_test "print $L\"q\"\[0\] == $var_name\[49\]" \
             " = 1" \
             "check value of parsed string literal in ${target_charset}"

    # Test handling of characters in the target charset which
    # can't be translated into the host charset.
    if {! [string compare $target_charset iso-8859-1]} {
        gdb_test "print iso_8859_1_string\[69\]" \
                 " = \[0-9-\]+ '\\\\242'" \
                 "print character with no equivalent in host character set"
        gdb_test "print iso_8859_1_string + 70" \
                 " = ${hex} \"\\\\242.*\"" \
                 "print string with no equivalent in host character set"
    }

    # Make sure that we don't apply the ISO-8859-1 `print_literally'
    # function to ASCII.
    if {! [string compare $target_charset ascii]} {
        gdb_test "print iso_8859_1_string\[69\]" \
                 " = \[0-9-\]+ '\\\\242'" \
                 "print ASCII unprintable character"
        gdb_test "print iso_8859_1_string + 70" \
                 " = ${hex} \"\\\\242.*\"" \
                 "print ASCII unprintable string"
    }

    # Try printing characters with backslash escape equivalents.
    set escapees {a b f n r t v}
    for {set i 0} {$i < [llength $escapees]} {incr i} {
        set escape [lindex $escapees $i]
        send_gdb "print $var_name\[$i\]\n"
        set have_escape 1
        gdb_expect {
            -re "= \[0-9-\]+ $L'\\\\${escape}'\[\r\n\]+$gdb_prompt $" {
                pass "try printing '\\${escape}' in ${target_charset}"
            }
            -re "= \[0-9-\]+ 'x'\[\r\n\]+$gdb_prompt $" {
                xfail "try printing '\\${escape}' in ${target_charset} (no such escape)"
                set have_escape 0
            }
            -re "$gdb_prompt $" {
                fail "try printing '\\${escape}' in ${target_charset}"
            }
            timeout {
                fail "try printing '\\${escape}' in ${target_charset} (timeout)"
            }
        }

        if {$have_escape} {

            # Try parsing a backslash escape in a character literal.
            gdb_test "print $L'\\${escape}' == $var_name\[$i\]" \
                     " = 1" \
                     "check value of '\\${escape}' in ${target_charset}"

            # Try parsing a backslash escape in a string literal.
            gdb_test "print $L\"\\${escape}\"\[0\] == $var_name\[$i\]" \
                     " = 1" \
                     "check value of \"\\${escape}\" in ${target_charset}"
        }
    }

    # Try printing a character escape that doesn't exist.  We should
    # get the unescaped character, in the target character set.
    gdb_test "print $L'\\q'" " = \[0-9-\]+ $L'q'" \
             "print escape that doesn't exist in $target_charset"
    gdb_test "print $L'\\q' == $var_name\[49\]" " = 1" \
             "check value of escape that doesn't exist in $target_charset"
}

# Reset the target charset.
gdb_test_no_output "set target-charset UTF-8"

# \242 is not a valid UTF-8 character.
gdb_test "print \"\\242\"" " = \"\\\\242\"" \
  "non-representable target character"

gdb_test "print '\\x'" "\\\\x escape without a following hex digit."
gdb_test "print '\\u'" "\\\\u escape without a following hex digit."
gdb_test "print '\\9'" " = \[0-9\]+ '9'"

# An octal escape can only be 3 digits.
gdb_test "print \"\\1011\"" " = \"A1\""

# Tests for wide- or unicode- strings.  L is the prefix letter to use,
# either "L" (for wide strings), "u" (for UTF-16), or "U" (for UTF-32).
# NAME is used in the test names and should be related to the prefix
# letter in some easy-to-undestand way.
proc test_wide_or_unicode {L name} {
    gdb_test "print $L\"ab\" $L\"c\"" " = $L\"abc\"" \
      "basic $name string concatenation"
    gdb_test "print $L\"ab\" \"c\"" " = $L\"abc\"" \
      "narrow and $name string concatenation"
    gdb_test "print \"ab\" $L\"c\"" " = $L\"abc\"" \
      "$name and narrow string concatenation"
    gdb_test "print $L\"\\xe\" $L\"c\"" " = $L\"\\\\016c\"" \
      "$name string concatenation with escape"
    gdb_test "print $L\"\" \"abcdef\" \"g\"" \
      "$L\"abcdefg\"" \
      "concatenate three strings with empty $name string"

    gdb_test "print $L'a'" "= \[0-9\]+ $L'a'" \
      "basic $name character"
}

if {$wchar_ok} {
    test_wide_or_unicode L wide
}

set ucs2_ok [expr {[get_sizeof char16_t 99] == 2}]

if ![valid_host_charset "UTF-16"] {
    verbose -log "Disabling UTF-16 tests."
    set ucs2_ok 0
}

if {$ucs2_ok} {
    test_wide_or_unicode u UTF-16
}

set ucs4_ok [expr {[get_sizeof char32_t 99] == 4}]
if {$ucs4_ok} {
    test_wide_or_unicode U UTF-32
}

# Test an invalid string combination.
proc test_combination {L1 name1 L2 name2} {
    gdb_test "print $L1\"abc\" $L2\"def\"" \
      "Undefined string concatenation." \
      "undefined concatenation of $name1 and $name2"
}

if {$wchar_ok && $ucs2_ok} {
    test_combination L wide u UTF-16
}
if {$wchar_ok && $ucs4_ok} {
    test_combination L wide U UTF-32
  # Regression test for a typedef to a typedef.
  gdb_test "print myvar" "= \[0-9\]+ L'A'" \
      "typedef to wchar_t"
}
if {$ucs2_ok && $ucs4_ok} {
    test_combination u UTF-16 U UTF-32
}

if {$ucs2_ok} {
    set go 1
    gdb_test_multiple "python print ('hello, world!')" \
	"verify python support for charset tests" {
	    -re "not supported.*$gdb_prompt $"	{
		unsupported "python support is disabled"
		set go 0
	    }
	    -re "$gdb_prompt $"	{}
	}

    if {$go} {
	gdb_test "print u\"abcdef\"" " = u\"abcdef\"" \
	    "set up for python printing of utf-16 string"

	gdb_test "python print (gdb.history(0).string())" "abcdef" \
	    "extract utf-16 string using python"
    }
}

# Regression test for a cleanup bug in the charset code.
gdb_test "print 'a' == 'a' || 'b' == 'b'" \
  ".* = 1" \
  "EVAL_SKIP cleanup handling regression test"


proc string_display { var_name set_prefix x_size x_type} {
  gdb_test_no_output "set ${var_name} = ${set_prefix}\"Test String\\0with zeroes\"" "assign ${var_name} with prefix ${set_prefix}"
  gdb_test "x /2${x_size}s ${var_name}" ".*\t${x_type}\"Test String\"\[\r\n\]+.*\t${x_type}\"with zeroes\"" "display String ${var_name} with x/${x_size}s"
}

if {$ucs2_ok} {
    string_display String16 u h u
    if {$wchar_size == 2} {
	string_display String16 L h u
    }
}

string_display String32 U w U
if {$wchar_size == 4} {
  string_display String32 L w U
}


foreach name {short int long} {
    # We're really just checking to make sure this doesn't give an
    # error.
    gdb_test "print ${name}_array = \"hi\"" \
	" = {.*}" \
	"assign string to $name array"
}


gdb_exit