Re: [PATCH v2] rebase -x: sanity check command
- From
Ævar Arnfjörð Bjarmason <avarab@gmail.com>
- Date
- Feb 13, 2019, 13:31 UTC
- Message-ID
- <87ftsrke6r.fsf@evledraar.gmail.com>
- In-Reply-To
- <20190129184327.30391-1-phillip.wood@talktalk.net>
On Tue, Jan 29 2019, Phillip Wood wrote:
Show 146 quoted lines
> From: Phillip Wood <phillip.wood@dunelm.org.uk>
>
> If the user gives an empty argument to --exec then git creates a todo
> list that it cannot parse. The rebase starts to run before erroring out
> with
>
> error: missing arguments for exec
> error: invalid line 2: exec
> You can fix this with 'git rebase --edit-todo' and then run 'git rebase --continue'.
> Or you can abort the rebase with 'git rebase --abort'.
>
> Instead check for empty commands before starting the rebase.
>
> Also check that the command does not contain any newlines as the
> todo-list format is unable to cope with multiline commands. Note that
> this changes the behavior, before this change one could do
>
> git rebase --exec='echo one
> exec echo two'
>
> and it would insert two exec lines in the todo list, now it will error
> out.
>
> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
> ---
>
> Apologies for the resend but I messed up the from address on the
> previous send and I don't think it made it to the mailing list.
>
> Thanks for the comments. I've updated the code as suggested by Dscho
> and tried to make it clearer in the commit message that this is fixing
> git creating a todo-list that it cannot parse.
>
> Range-diff against v1:
> 1: 9626549aec ! 1: 726b84c300 rebase -x: sanity check command
> @@ -2,8 +2,9 @@
>
> rebase -x: sanity check command
>
> - If the user gives an empty argument to --exec then the rebase starts to
> - run before erroring out with
> + If the user gives an empty argument to --exec then git creates a todo
> + list that it cannot parse. The rebase starts to run before erroring out
> + with
>
> error: missing arguments for exec
> error: invalid line 2: exec
> @@ -33,21 +34,16 @@
>
> +static int check_exec_cmd(const char *cmd)
> +{
> -+ int non_blank = 0;
> ++ if (strchr(cmd, '\n'))
> ++ return error(_("exec commands cannot contain newlines"));
> +
> -+ while (*cmd) {
> -+ if (*cmd == '\n')
> -+ return error(_("exec commands cannot contain newlines"));
> -+ if (!isspace(*cmd))
> -+ non_blank = 1;
> -+ cmd++;
> -+ }
> ++ /* Does the command consist purely of whitespace? */
> ++ if (!cmd[strspn(cmd, " \t\r\f\v")])
> ++ return error(_("empty exec command"));
> +
> -+ if (non_blank)
> -+ return 0;
> -+
> -+ return error(_("empty exec command"));
> ++ return 0;
> +}
> ++
> +
> int cmd_rebase(int argc, const char **argv, const char *prefix)
> {
>
> builtin/rebase.c | 17 +++++++++++++++++
> t/t3404-rebase-interactive.sh | 19 +++++++++++++++++++
> 2 files changed, 36 insertions(+)
>
> diff --git a/builtin/rebase.c b/builtin/rebase.c
> index 00de70365e..f75bd79b45 100644
> --- a/builtin/rebase.c
> +++ b/builtin/rebase.c
> @@ -793,6 +793,19 @@ static void set_reflog_action(struct rebase_options *options)
> strbuf_release(&buf);
> }
>
> +static int check_exec_cmd(const char *cmd)
> +{
> + if (strchr(cmd, '\n'))
> + return error(_("exec commands cannot contain newlines"));
> +
> + /* Does the command consist purely of whitespace? */
> + if (!cmd[strspn(cmd, " \t\r\f\v")])
> + return error(_("empty exec command"));
> +
> + return 0;
> +}
> +
> +
> int cmd_rebase(int argc, const char **argv, const char *prefix)
> {
> struct rebase_options options = {
> @@ -1130,6 +1143,10 @@ int cmd_rebase(int argc, const char **argv, const char *prefix)
> }
> }
>
> + for (i = 0; i < exec.nr; i++)
> + if (check_exec_cmd(exec.items[i].string))
> + exit(1);
> +
> if (!(options.flags & REBASE_NO_QUIET))
> argv_array_push(&options.git_am_opts, "-q");
>
> diff --git a/t/t3404-rebase-interactive.sh b/t/t3404-rebase-interactive.sh
> index 7a440e08d8..c98f64eb2d 100755
> --- a/t/t3404-rebase-interactive.sh
> +++ b/t/t3404-rebase-interactive.sh
> @@ -147,6 +147,25 @@ test_expect_success 'rebase -i with the exec command checks tree cleanness' '
> git rebase --continue
> '
>
> +test_expect_success 'rebase -x with empty command fails' '
> + test_when_finished "git rebase --abort ||:" &&
> + test_must_fail git rebase -x "" @ 2>actual &&
> + test_write_lines "error: empty exec command" >expected &&
> + test_i18ncmp expected actual &&
> + test_must_fail git rebase -x " " @ 2>actual &&
> + test_i18ncmp expected actual
> +'
> +
> +LF='
> +'
> +test_expect_success 'rebase -x with newline in command fails' '
> + test_when_finished "git rebase --abort ||:" &&
> + test_must_fail git rebase -x "a${LF}b" @ 2>actual &&
> + test_write_lines "error: exec commands cannot contain newlines" \
> + >expected &&
> + test_i18ncmp expected actual
> +'
> +
> test_expect_success 'rebase -i with exec of inexistent command' '
> git checkout master &&
> test_when_finished "git rebase --abort" &&This patch breaks the test suite (with these two new tests) under GIT_TEST_REBASE_USE_BUILTIN=false. So a 2.21.0-rc0 regression.
It would have been better to raise this before the rc period, but I just noticed this now, but we can now:
1. Skip the test under that mode 2. Fix the shell code to do the same thing 3. Just remove the shell code & GIT_TEST_REBASE_USE_BUILTIN=false mode
Maybe we should just do #3. The reason for the escape hatch was in case we had bugs, and now we've had a full release cycle, but maybe that's too early...