Re: [PATCH v2 2/5] am: improve author-script error reporting

[Junio C Hamano <gitster@pobox.com>]

Phillip Wood <phillip.wood@talktalk.net> writes:

> From: Phillip Wood <phillip.wood@dunelm.org.uk>
>
> If there are errors in a user edited author-script there was no
> indication of what was wrong. This commit adds some specific error messages
> depending on the problem. It also relaxes the requirement that the
> variables appear in a specific order in the file to match the behavior
> of 'rebase --interactive'.

That relaxing is sensible; there is no reason to insist that the
file we are reading was written by exactly the same writer as we
have.

> diff --git a/builtin/am.c b/builtin/am.c
> index b68578bc3f..d42b725273 100644
> --- a/builtin/am.c
> +++ b/builtin/am.c
> @@ -270,8 +270,11 @@ static int parse_key_value_squoted(char *buf, struct string_list *list)
>  		struct string_list_item *item;
>  		char *np;
>  		char *cp = strchr(buf, '=');
> -		if (!cp)
> -			return -1;
> +		if (!cp) {
> +			np = strchrnul(buf, '\n');
> +			return error(_("unable to parse '%.*s'"),
> +				     (int) (np - buf), buf);
> +		}

We are unable to parse because it is not of KEY='VALUE' form.  Is
that something worth reporting, e.g. "no key present in '%.*s'"?

> @@ -280,7 +283,8 @@ static int parse_key_value_squoted(char *buf, struct string_list *list)
>  		*np = '\0';
>  		cp = sq_dequote(cp);
>  		if (!cp)
> -			return -1;
> +			return error(_("unable to dequote value of '%s'"),
> +				     item->string);

At this point, item->string is what we earlier found on the left
hand side of the '=', i.e. the key.  The message makes sense.

> @@ -308,6 +312,7 @@ static int read_author_script(struct am_state *state)
>  	struct strbuf buf = STRBUF_INIT;
>  	struct string_list kv = STRING_LIST_INIT_DUP;
>  	int retval = -1; /* assume failure */
> +	int i, name_i = -2, email_i = -2, date_i = -2, err = 0;

That -2 is somewhat cute.  If we find a dup, then it will become -1
so later check for missing field would not trigger.

> @@ -326,14 +331,38 @@ static int read_author_script(struct am_state *state)
>  	if (parse_key_value_squoted(buf.buf, &kv))
>  		goto finish;
>  
> -	if (kv.nr != 3 ||
> -	    strcmp(kv.items[0].string, "GIT_AUTHOR_NAME") ||
> -	    strcmp(kv.items[1].string, "GIT_AUTHOR_EMAIL") ||
> -	    strcmp(kv.items[2].string, "GIT_AUTHOR_DATE"))
> +	for (i = 0; i < kv.nr; i++) {
> +		if (!strcmp(kv.items[i].string, "GIT_AUTHOR_NAME")) {
> +			if (name_i >= 0)
> +				name_i = error(_("'GIT_AUTHOR_NAME' already given"));
> +			else
> +				name_i = i;

However, if you have three instances of GIT_AUTHOR_NAME, then

 - the first one makes name_i point at it
 - the second one triggers an error and name_i becomes -1
 - the third one makes name_i point at it again

And name_i is not -2 so we won't give "missing" error, which is OK,
but we end up having a usable name even though we said we detected
duplicate!

You can probably compare name_i with -2 when detecting the dup to
fix it, i.e.

	if (name_i != -2)
		name_i = error("found dup");
	else
		name_i = i;

> +		} else if (!strcmp(kv.items[i].string, "GIT_AUTHOR_EMAIL")) {
> +			if (email_i >= 0)
> +				email_i = error(_("'GIT_AUTHOR_EMAIL' already given"));
> +			else
> +				email_i = i;
> +		} else if (!strcmp(kv.items[i].string, "GIT_AUTHOR_DATE")) {
> +			if (date_i >= 0)
> +				date_i = error(_("'GIT_AUTHOR_DATE' already given"));
> +			else
> +				date_i = i;
> +		} else {
> +			err = error(_("unknown variable '%s'"),
> +				    kv.items[i].string);
> +		}
> +	}
> +	if (name_i == -2)
> +		error(_("missing 'GIT_AUTHOR_NAME'"));
> +	if (email_i == -2)
> +		error(_("missing 'GIT_AUTHOR_EMAIL'"));
> +	if (date_i == -2)
> +		error(_("missing 'GIT_AUTHOR_DATE'"));
> +	if (date_i < 0 || email_i < 0 || date_i < 0 || err)
>  		goto finish;
> -	state->author_name = kv.items[0].util;
> -	state->author_email = kv.items[1].util;
> -	state->author_date = kv.items[2].util;
> +	state->author_name = kv.items[name_i].util;
> +	state->author_email = kv.items[email_i].util;
> +	state->author_date = kv.items[date_i].util;
>  	retval = 0;
>  finish:
>  	string_list_clear(&kv, !!retval);