Re: [PATCH] gdb/testsuite: work around empty substring bug in expect

[Tom de Vries <tdevries@suse.de>]

On 8/26/25 16:35, Andrew Burgess wrote:
> There is a bug in expect, see:
> 
>    https://sourceforge.net/p/expect/patches/26/
> 
> which causes empty substring matches from a regexp to instead return
> the complete input buffer.  To reproduce this bug, try this command:
> 
>    expect -c 'spawn sh -c "echo -n -e \"abc\""; \
>              expect -re "(a?)(a)(bc)"; \
> 	    puts "\n"; \
> 	    for { set i 1 } { $i < 4 } { incr i } { \
> 	      puts -nonewline "($i): \""; \
> 	      puts -nonewline $expect_out($i,string); \
> 	      puts "\"" \
> 	    }'
> 
> For a working expect the output looks like:
> 
>    spawn sh -c echo -n -e "abc"
>    abc
> 
>    (1): ""
>    (2): "a"
>    (3): "bc"
> 
> But for a broken expect the output looks like:
> 
>    spawn sh -c echo -n -e "abc"
>    abc
> 
>    (1): "abc"
>    (2): "a"
>    (3): "bc"
> 
> Notice that (1) is now returning the complete input buffer rather than
> the empty string, this is wrong.
> 
> This is not the first time this bug has impacted GDB's testsuite,
> this commit seems to be working around the same problem:
> 
>    commit e579b537353cd91cb8fac1eaeb69901d4936766f
>    Date:   Sat Aug 16 20:32:37 2025 +0200
> 
>        [gdb/testsuite] Fix TUI tests on freebsd
> 
> I recently pushed this commit:
> 
>    commit 3825c972a636852600b47c242826313f4b9963b8
>    Date:   Wed Jun 18 15:02:29 2025 +0100
> 
>        gdb: allow gdb.Color to work correctly with pagination
> 
> Which added gdb.python/py-color-pagination.exp.  Bug PR gdb/33321 was
> then created as the test was failing on some hosts.  Turns out, this
> is same expect bug.
> 
> The fix presented here is the same as for e579b537353cd91cb8, avoid
> using optional regexp substrings at the start of a regexp, and instead
> use two separate regexp patterns.  With this change in place, the test
> now passes on all hosts.
> 
> There's no change in what is being tested after this commit.
> 

Hi Andrew,

thanks for digging into this.

I've tested this patch on Tumbleweed, and it fixed the problem for me there.

The patch LGTM.

Approved-By: Tom de Vries <tdevries@suse.de>

- Tom

> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=33321
> ---
>   .../gdb.python/py-color-pagination.exp        | 36 +++++++++++++------
>   1 file changed, 25 insertions(+), 11 deletions(-)
> 
> diff --git a/gdb/testsuite/gdb.python/py-color-pagination.exp b/gdb/testsuite/gdb.python/py-color-pagination.exp
> index 3235fffe5cc..e1ca1f1ce75 100644
> --- a/gdb/testsuite/gdb.python/py-color-pagination.exp
> +++ b/gdb/testsuite/gdb.python/py-color-pagination.exp
> @@ -66,18 +66,25 @@ proc test_pagination { type mode } {
>   	    exp_continue
>   	}
>   
> -	-re "^(${::any_color}?)(${::any_color})$::str" {
> +	-re "^(${::any_color})(${::any_color})$::str" {
>   	    # After a continuation prompt GDB will restore the previous
>   	    # color, and then we immediately switch to a new color.
>   	    set restored_color $expect_out(1,string)
> -	    if { $restored_color ne ""
> -		 && $restored_color ne $expected_restore_color } {
> +	    if { $restored_color ne $expected_restore_color } {
>   		set saw_bad_color_handling true
>   	    }
>   	    set last_color $expect_out(2,string)
>   	    exp_continue
>   	}
>   
> +	-re "^(${::any_color})$::str" {
> +	    # This pattern matches printing STR in all cases that are not
> +	    # immediately after a pagination prompt.  In this case there is
> +	    # a single escape sequence to set the color.
> +	    set last_color $expect_out(1,string)
> +	    exp_continue
> +	}
> +
>   	-re "^\033\\\[${::decimal}m$::str" {
>   	    # This catches the case where the color's escape sequence has
>   	    # not been converted back into a full style.  This indicates
> @@ -86,18 +93,25 @@ proc test_pagination { type mode } {
>   	    exp_continue
>   	}
>   
> -	-re "^((?:\033\\\[m)?)$::pagination_prompt$" {
> +	-re "^\033\\\[m$::pagination_prompt$" {
>   	    # After a pagination prompt we expect GDB to restore the last
>   	    # color.
>   	    set expected_restore_color $last_color
>   
> -	    # If we didn't see a color reset sequence then the pagination
> -	    # prompt will have been printed in the wrong color, this is a
> -	    # GDB bug.
> -	    set color_reset $expect_out(1,string)
> -	    if { $color_reset eq "" } {
> -		set saw_bad_color_handling true
> -	    }
> +	    # Send '\n' to view more output.
> +	    send_gdb "\n"
> +	    exp_continue
> +	}
> +
> +	-re "^$::pagination_prompt$" {
> +	    # After a pagination prompt we expect GDB to restore the last
> +	    # color.
> +	    set expected_restore_color $last_color
> +
> +	    # If we didn't see a color reset sequence before the pagination
> +	    # prompt, then the prompt will have been printed in the wrong
> +	    # color, this is a GDB bug.
> +	    set saw_bad_color_handling true
>   
>   	    # Send '\n' to view more output.
>   	    send_gdb "\n"
> 
> base-commit: 3825c972a636852600b47c242826313f4b9963b8