[PATCH] Use split_ident_line() in parse_user()

[PATCH] Use split_ident_line() in parse_user()

[cgit]

Use Git's built-in ident line splitting algorithm instead of
reimplementing it. This does not only simplify the code but also makes
sure that cgit is consistent with Git when it comes to author parsing.

Signed-off-by: Lukas Fleischer <cgit at cryptocrack.de>
---
I would also like to get rid of the angle brackets and only add them
when displaying the information as HTML. That would also result in email
filters getting email addresses without angle brackets which I consider
to be a good thing but it breaks backwards compatibility. Any opinions
on this are welcome.

 parsing.c | 45 +++++++++++++++++----------------------------
 1 file changed, 17 insertions(+), 28 deletions(-)

diff --git a/parsing.c b/parsing.c
index 3dbd122..7d34d11 100644
--- a/parsing.c
+++ b/parsing.c
@@ -71,36 +71,25 @@ static char *substr(const char *head, const char *tail)
 
 static const char *parse_user(const char *t, char **name, char **email, unsigned long *date)
 {
-	const char *p = t;
-	int mode = 1;
+	const char *line_end = strchrnul(t, '\n');
+	struct ident_split ident;
+	unsigned email_len;
 
-	while (p && *p) {
-		if (mode == 1 && *p == '<') {
-			*name = substr(t, p - 1);
-			t = p;
-			mode++;
-		} else if (mode == 1 && *p == '\n') {
-			*name = substr(t, p);
-			p++;
-			break;
-		} else if (mode == 2 && *p == '>') {
-			*email = substr(t, p + 1);
-			t = p;
-			mode++;
-		} else if (mode == 2 && *p == '\n') {
-			*email = substr(t, p);
-			p++;
-			break;
-		} else if (mode == 3 && isdigit(*p)) {
-			*date = atol(p);
-			mode++;
-		} else if (*p == '\n') {
-			p++;
-			break;
-		}
-		p++;
+	if (!split_ident_line(&ident, t, line_end - t)) {
+		*name = substr(ident.name_begin, ident.name_end);
+
+		email_len = ident.mail_end - ident.mail_begin;
+		*email = xmalloc(strlen("<") + email_len + strlen(">\0"));
+		sprintf(*email, "<%.*s>", email_len, ident.mail_begin);
+
+		if (ident.date_begin)
+			*date = strtoul(ident.date_begin, NULL, 10);
 	}
-	return p;
+
+	if (*line_end)
+		return line_end + 1;
+	else
+		return line_end;
 }
 
 #ifdef NO_ICONV
-- 
2.1.3

[PATCH] Use split_ident_line() in parse_user()

[Jason]

On Wed, Dec 17, 2014 at 5:19 AM, Lukas Fleischer <cgit at cryptocrack.de>
wrote:

> Use Git's built-in ident line splitting algorithm instead of
> reimplementing it. This does not only simplify the code but also makes
> sure that cgit is consistent with Git when it comes to author parsing.
>

Thank heavens!

> +               email_len = ident.mail_end - ident.mail_begin;
> +               *email = xmalloc(strlen("<") + email_len + strlen(">\0"));
> +               sprintf(*email, "<%.*s>", email_len, ident.mail_begin);
> +
>

strlen(">\0") -- isn't the null superfluous? Cleanup and resubmit?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.zx2c4.com/pipermail/cgit/attachments/20141223/a2e20204/attachment.html>

[PATCH] Use split_ident_line() in parse_user()

[cgit]

On Wed, 24 Dec 2014 at 02:57:45, Jason A. Donenfeld wrote:
> On Wed, Dec 17, 2014 at 5:19 AM, Lukas Fleischer <cgit at cryptocrack.de>
> wrote:
> 
> > Use Git's built-in ident line splitting algorithm instead of
> > reimplementing it. This does not only simplify the code but also makes
> > sure that cgit is consistent with Git when it comes to author parsing.
> >
> 
> Thank heavens!
> 
> > +               email_len = ident.mail_end - ident.mail_begin;
> > +               *email = xmalloc(strlen("<") + email_len + strlen(">\0"));
> > +               sprintf(*email, "<%.*s>", email_len, ident.mail_begin);
> > +
> >
> 
> strlen(">\0") -- isn't the null superfluous? Cleanup and resubmit?

I think this is even wrong because it means the buffer won't be large
enough to hold the terminating null character. Will resubmit.

[PATCH v2] Use split_ident_line() in parse_user()

[cgit]

Use Git's built-in ident line splitting algorithm instead of
reimplementing it. This does not only simplify the code but also makes
sure that cgit is consistent with Git when it comes to author parsing.

Signed-off-by: Lukas Fleischer <cgit at cryptocrack.de>
---
 parsing.c | 45 +++++++++++++++++----------------------------
 1 file changed, 17 insertions(+), 28 deletions(-)

diff --git a/parsing.c b/parsing.c
index 3dbd122..53c29bb 100644
--- a/parsing.c
+++ b/parsing.c
@@ -71,36 +71,25 @@ static char *substr(const char *head, const char *tail)
 
 static const char *parse_user(const char *t, char **name, char **email, unsigned long *date)
 {
-	const char *p = t;
-	int mode = 1;
+	const char *line_end = strchrnul(t, '\n');
+	struct ident_split ident;
+	unsigned email_len;
 
-	while (p && *p) {
-		if (mode == 1 && *p == '<') {
-			*name = substr(t, p - 1);
-			t = p;
-			mode++;
-		} else if (mode == 1 && *p == '\n') {
-			*name = substr(t, p);
-			p++;
-			break;
-		} else if (mode == 2 && *p == '>') {
-			*email = substr(t, p + 1);
-			t = p;
-			mode++;
-		} else if (mode == 2 && *p == '\n') {
-			*email = substr(t, p);
-			p++;
-			break;
-		} else if (mode == 3 && isdigit(*p)) {
-			*date = atol(p);
-			mode++;
-		} else if (*p == '\n') {
-			p++;
-			break;
-		}
-		p++;
+	if (!split_ident_line(&ident, t, line_end - t)) {
+		*name = substr(ident.name_begin, ident.name_end);
+
+		email_len = ident.mail_end - ident.mail_begin;
+		*email = xmalloc(strlen("<") + email_len + strlen(">") + 1);
+		sprintf(*email, "<%.*s>", email_len, ident.mail_begin);
+
+		if (ident.date_begin)
+			*date = strtoul(ident.date_begin, NULL, 10);
 	}
-	return p;
+
+	if (*line_end)
+		return line_end + 1;
+	else
+		return line_end;
 }
 
 #ifdef NO_ICONV
-- 
2.2.1

[PATCH v2] Use split_ident_line() in parse_user()

[Jason]

Merged, thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.zx2c4.com/pipermail/cgit/attachments/20141224/5501f736/attachment.html>