[PATCH] lib/timeutils: Require '@' prefix for seconds since the Epoch timestamp

[PATCH] lib/timeutils: Require '@' prefix for seconds since the Epoch timestamp

[Peter Ujfalusi]

Since the seconds since the Epoch is just a number it might be better to
require special prefix to indicate the intention that the user wants to
interpret the number as seconds since the Epoch.

Use the same '@' character as prefix as used by systemd.time to make it
easier to integrate in scripts intended to be used on systems with or
without systemd.

Fix also the initial support which discarded the seconds from the converted
timestamp.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
---
Hi Karel,

I think this will be safer for the enduser. The original patch would interpret
_any_ number not matching with the other supported formats interpreted as seconds
since the Epoch, which might not something the user had in mind.

Sorry for the oversight on my side...

Kind regards,
Péter

 lib/timeutils.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/lib/timeutils.c b/lib/timeutils.c
index 6dda2e8deefb..4eaef9533ab2 100644
--- a/lib/timeutils.c
+++ b/lib/timeutils.c
@@ -181,7 +181,7 @@ int parse_timestamp(const char *t, usec_t *usec)
 	 *
 	 *   2012-09-22 16:34:22
 	 *   2012-09-22T16:34:22
-	 *   1348331662		  (seconds since the Epoch (1970-01-01 00:00 UTC))
+	 *   @1348331662	  (seconds since the Epoch (1970-01-01 00:00 UTC))
 	 *   2012-09-22 16:34	  (seconds will be set to 0)
 	 *   2012-09-22		  (time will be set to 00:00:00)
 	 *   16:34:22		  (date will be set to today)
@@ -233,7 +233,12 @@ int parse_timestamp(const char *t, usec_t *usec)
 			return r;
 
 		goto finish;
+	} else if (t[0] == '@') {
+		k = strptime(t + 1, "%s", &tm);
+		if (k && *k == 0)
+			goto finish;
 
+		return -EINVAL;
 	} else if (endswith(t, " ago")) {
 		char *z;
 
@@ -326,13 +331,6 @@ int parse_timestamp(const char *t, usec_t *usec)
 		goto finish;
 	}
 
-	tm = copy;
-	k = strptime(t, "%s", &tm);
-	if (k && *k == 0) {
-		tm.tm_sec = 0;
-		goto finish;
-	}
-
 	return -EINVAL;
 
  finish:
-- 
2.37.3

Re: [PATCH] lib/timeutils: Require '@' prefix for seconds since the Epoch timestamp

[Karel Zak]

On Wed, Sep 07, 2022 at 08:41:41AM +0300, Peter Ujfalusi wrote:
>  lib/timeutils.c | 14 ++++++--------
>  1 file changed, 6 insertions(+), 8 deletions(-)

Applied, let's use '@' prefix :-) Thanks.

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [PATCH] lib/timeutils: Require '@' prefix for seconds since the Epoch timestamp

[Péter Ujfalusi]

On 12/09/2022 15:23, Karel Zak wrote:
> On Wed, Sep 07, 2022 at 08:41:41AM +0300, Peter Ujfalusi wrote:
>>  lib/timeutils.c | 14 ++++++--------
>>  1 file changed, 6 insertions(+), 8 deletions(-)
> 
> Applied, let's use '@' prefix :-) Thanks.

I will try to look at documenting the supported time formats within
util-linux...

My guess would be to add a new file:
man-common/timestamp-formats.adoc

and link it in dmesg for example?

> 
>     Karel
> 

-- 
Péter

Re: [PATCH] lib/timeutils: Require '@' prefix for seconds since the Epoch timestamp

[Karel Zak]

On Tue, Sep 13, 2022 at 02:26:58PM +0300, Péter Ujfalusi wrote:
> 
> 
> On 12/09/2022 15:23, Karel Zak wrote:
> > On Wed, Sep 07, 2022 at 08:41:41AM +0300, Peter Ujfalusi wrote:
> >>  lib/timeutils.c | 14 ++++++--------
> >>  1 file changed, 6 insertions(+), 8 deletions(-)
> > 
> > Applied, let's use '@' prefix :-) Thanks.
> 
> I will try to look at documenting the supported time formats within
> util-linux...
> 
> My guess would be to add a new file:
> man-common/timestamp-formats.adoc
> 
> and link it in dmesg for example?

Yes, sounds good. 

We also use parse_timestamp() in rtcwake, cal and last.

    Karel


-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com