From: Jani Nikula <jani@nikula.org>
To: John Lenz <lenz@math.uic.edu>, Tomi Ollila <tomi.ollila@iki.fi>,
	notmuch@notmuchmail.org
Subject: Re: cli: add --include-html option to notmuch show
In-Reply-To: <notmuch-web-1376754748.336647450@www.wuzzeb.org>
References: <notmuch-web-1372724382.450184839@www.wuzzeb.org>
	<m27ggj1x29.fsf@guru.guru-group.fi>
	<notmuch-web-1374719771.1588310986@www.wuzzeb.org>
	<m2zjsxs0g1.fsf@guru.guru-group.fi>
	<notmuch-web-1376754748.336647450@www.wuzzeb.org>
User-Agent: Notmuch/0.15.2+227~g40b2846 (http://notmuchmail.org) Emacs/24.3.1
	(x86_64-pc-linux-gnu)
Date: Sun, 18 Aug 2013 14:25:16 +0300
Message-ID: <87k3jjutpf.fsf@nikula.org>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: list

On Sat, 17 Aug 2013, John Lenz <lenz@math.uic.edu> wrote:
> On Sun Aug  4 14:47 -0500 2013, Tomi Ollila <tomi.ollila@iki.fi> wrote:
>> The next question is should we have new option as
>> 
>> --include-html
>> 
>> or as
>> 
>> --include-html=(true|false)
>> 
>> or even
>> 
>> --body=(true|false|text-and-html)
>> 
>> See --exclude option in http://notmuchmail.org/manpages/notmuch-search-1/
>> and --body option in http://notmuchmail.org/manpages/notmuch-show-1/
>> for comparison...
>> 
>
> I have no preference here, although I guess I would vote for
> --include-html=(true|false) since adding it to --body makes the --body
> options confusing: to make sense the body options should be
> --body=(text|text-and-html|none) but of course you can't change that and
> break the command line API.  Well, maybe you could add all three
> --body=(text|text-and-html|none)  and still accept true/false for
> compatibility.

Hi John & Tomi -

We could trivially amend the argument parser to |= the keyword values
(instead of =) to allow specifying keywords arguments multiple times on
the command line. With the keyword values specified as bit flags, we
could then have 'notmuch show --body=text --body=html ...' return both
text and html, and allow trivial future extension too.

--body=true and --body=false could be handled specially for backwards
compatibility, for example by forcing text only or no parts,
respectively. Since the default is currently text, --body=none might be
a suitable synonym for --body=false, while the boolean alternatives
could be deprecated.

A little less trivially it's also possible to support e.g. comma
separated keyword values, such as --body=text,html but I do prefer the
(implementation) simplicity of the above.

Untested patch to the argument parser below.

Cheers,
Jani.


diff --git a/command-line-arguments.c b/command-line-arguments.c
index bf9aeca..c426054 100644
--- a/command-line-arguments.c
+++ b/command-line-arguments.c
@@ -23,7 +23,10 @@ _process_keyword_arg (const notmuch_opt_desc_t *arg_desc, char next, const char
     while (keywords->name) {
 	if (strcmp (arg_str, keywords->name) == 0) {
 	    if (arg_desc->output_var) {
-		*((int *)arg_desc->output_var) = keywords->value;
+		if (arg_desc->opt_type == NOTMUCH_OPT_KEYWORD_FLAGS)
+		    *((int *)arg_desc->output_var) |= keywords->value;
+		else
+		    *((int *)arg_desc->output_var) = keywords->value;
 	    }
 	    return TRUE;
 	}
@@ -146,6 +149,7 @@ parse_option (const char *arg,
 
 	    switch (try->opt_type) {
 	    case NOTMUCH_OPT_KEYWORD:
+	    case NOTMUCH_OPT_KEYWORD_FLAGS:
 		return _process_keyword_arg (try, next, value);
 		break;
 	    case NOTMUCH_OPT_BOOLEAN:
diff --git a/command-line-arguments.h b/command-line-arguments.h
index de1734a..085a492 100644
--- a/command-line-arguments.h
+++ b/command-line-arguments.h
@@ -8,6 +8,7 @@ enum notmuch_opt_type {
     NOTMUCH_OPT_BOOLEAN,	/* --verbose              */
     NOTMUCH_OPT_INT,		/* --frob=8               */
     NOTMUCH_OPT_KEYWORD,	/* --format=raw|json|text */
+    NOTMUCH_OPT_KEYWORD_FLAGS,  /* the above with values OR'd together */
     NOTMUCH_OPT_STRING,		/* --file=/tmp/gnarf.txt  */
     NOTMUCH_OPT_POSITION	/* notmuch dump pos_arg   */
 };