bug#14097: [PATCH v2] Add support for ISO 8601 basic format

[Mihai Capotă]

Could someone please review the new patch?

On Mon, Aug 5, 2013 at 6:51 PM, Mihai Capotă <address@hidden> wrote:
> The parser now accepts the basic format for combined date and time
> representations, which ommits the date and time separators, "-" and ":".
>
> See bug 23767 for GNU coreutils, <https://savannah.gnu.org/bugs/?23767>.
>
> * lib/parse-datetime.y: Parse combined date and time representations in
> ISO 8601 basic format.
> (set_hhmmss_iso_8601_basic_time) New function.
> (digits_iso_8601_basic_to_date) New function.
> * tests/test-parse-datetime.c: Add tests for combined date and time
> representations in ISO 8601 basic format.
> * doc/parse-datetime.texi Document support for combined date and time
> representations in ISO 8601 basic format.
>
> Signed-off-by: Mihai Capotă <address@hidden>
> ---
> On Thu, Apr 25, 2013 at 12:02 AM, Paul Eggert <address@hidden> wrote:
>> The most important thing is that the patch also needs
>> to update doc/parse-datetime.texi.
>
> Done.
>
>> On 03/30/13 12:18, Mihai Capotă wrote:
>>> +      /* not ISO 8601 time, forcing mktime error */
>>> +      pc->hour = 90;
>>
>> How does this force a mktime error?  mktime allows tm_hour == 90.
>
> I meant to say mktime_ok. I changed the code to reject input by incrementing 
> times_seen, like time_zone_hhmm.
>
>>>  datetime:
>>>      iso_8601_datetime
>>> +  | iso_8601_basic_datetime
>>>    ;
>>>
>>>  iso_8601_datetime:
>>>      iso_8601_date 'T' iso_8601_time
>>>    ;
>>>
>>> +iso_8601_basic_datetime:
>>> +    number 'T' iso_8601_basic_time
>>> +      { pc->dates_seen--; } /* already incremented in digits_to_date_time 
>>> */
>>
>> This doesn't look right.  'number' accepts all sort of things that we
>> would rather not accept here.
>
> I was trying to make use of the existing digits_to_date_time function. I 
> replaced it with tUNUMBER and a new function.
>
>> Conversely, why require ":" in times to
>> correlate with "-" in dates?  Shouldn't we accept a "-"less date along
>> with a ":"ful time, and vice versa?
>
> No, that is not allowed by the standard.
>
>> And that "dates_seen--" business
>> is a hack; can't we arrange things so that dates_seen is incremented
>> just once?
>
> The hack is gone.
>
>>> +iso_8601_basic_time:
>>> +    tUNUMBER o_zone_offset
>>> +      {
>>> +        set_hhmmss_iso_8601_basic_time (pc, $1.value, 0);
>>> +        pc->meridian = MER24;
>>> +      }
>>> +  | tUDECIMAL_NUMBER o_zone_offset
>>> +      {
>>> +        /* FIXME avoid time_t to long int cast */
>>
>> Why is the cast needed?  Also, can't the grammar be simplified
>> here, by using unsigned_seconds instead of using both
>> tUDECIMAL_NUMBER and tUNUMBER?
>
> I switched to using unsigned_seconds.
>
>  doc/parse-datetime.texi     |    9 +++++-
>  lib/parse-datetime.y        |   68 
> +++++++++++++++++++++++++++++++++++++++++--
>  tests/test-parse-datetime.c |   61 ++++++++++++++++++++++++++++++++++++++
>  3 files changed, 135 insertions(+), 3 deletions(-)
>
> diff --git a/doc/parse-datetime.texi b/doc/parse-datetime.texi
> index 6b3e973..9aa87ed 100644
> --- a/doc/parse-datetime.texi
> +++ b/doc/parse-datetime.texi
> @@ -327,7 +327,12 @@ The ISO 8601 date and time of day extended format 
> consists of an ISO
>  day.  This format is also recognized if the @samp{T} is replaced by a
>  space.
>
> -In this format, the time of day should use 24-hour notation.
> +The ISO 8601 basic format is also recognized. It is identical to the ISO 8601
> +extended format, except for omitting the @samp{-} separator in the date and 
> the
> address@hidden:} separator in the time. Only the HHMMSS format is supported 
> for the
> +time of day, the reduced accuracy HHMM and HH formats are not supported.
> +
> +In these formats, the time of day should use 24-hour notation.
>  Fractional seconds are allowed, with either comma or period preceding
>  the fraction.  ISO 8601 fractional minutes and hours are not
>  supported.  Typically, hosts support nanosecond timestamp resolution;
> @@ -339,6 +344,8 @@ Here are some examples:
>  2012-09-24T20:02:00.052-0500
>  2012-12-31T23:59:59,999999999+1100
>  1970-01-01 00:00Z
> +20120924T200200.052-0500
> +20121231T235959,999999999+1100
>  @end example
>
>  @node Day of week items
> diff --git a/lib/parse-datetime.y b/lib/parse-datetime.y
> index 4dce7fa..fa9719d 100644
> --- a/lib/parse-datetime.y
> +++ b/lib/parse-datetime.y
> @@ -257,6 +257,38 @@ digits_to_date_time (parser_control *pc, textint 
> text_int)
>      }
>  }
>
> +/* Extract into *PC the date info from a string of digits in ISO 8601 basic
> +   format, i.e., YYYYMMHH, YYYY, or YY meaning century. Note that YYYYMM is 
> not
> +   allowed to avoid confusion with YYMMHH  */
> +static void
> +digits_iso_8601_basic_to_date (parser_control *pc, textint text_int)
> +{
> +  switch (text_int.digits)
> +    {
> +    case 8:
> +      pc->day = text_int.value % 100;
> +      pc->month = (text_int.value / 100) % 100;
> +      pc->year.value = text_int.value / 10000;
> +      pc->year.digits = 4;
> +      return;
> +    case 4:
> +      pc->day = 1;
> +      pc->month = 1;
> +      pc->year.value = text_int.value;
> +      pc->year.digits = 4;
> +      return;
> +    case 2:
> +      pc->day = 1;
> +      pc->month = 1;
> +      pc->year.value = text_int.value * 100;
> +      pc->year.digits = 4;
> +      return;
> +    default:
> +      pc->dates_seen++;
> +      return;
> +    }
> +}
> +
>  /* Increment PC->rel by FACTOR * REL (FACTOR is 1 or -1).  */
>  static void
>  apply_relative_time (parser_control *pc, relative_time rel, int factor)
> @@ -282,6 +314,28 @@ set_hhmmss (parser_control *pc, long int hour, long int 
> minutes,
>    pc->seconds.tv_nsec = nsec;
>  }
>
> +/* Set PC-> hour, minutes, seconds and nanoseconds members from ISO 8601 
> basic
> +   time.  */
> +static void
> +set_hhmmss_iso_8601_basic_time (parser_control *pc, time_t integer_part,
> +                                long int fractional_part)
> +{
> +  if (integer_part / 1000000 > 0)
> +    {
> +      /* Not ISO 8601 time, arrange to reject it by incrementing
> +         pc->times_seen.*/
> +      pc->times_seen++;
> +    }
> +  else
> +    {
> +      /* FIXME support reduced accuracy times, i.e. HHMM and HH */
> +      pc->hour = integer_part / 10000;
> +      pc->minutes = (integer_part % 10000) / 100;
> +      pc->seconds.tv_sec = integer_part % 100;
> +      pc->seconds.tv_nsec = fractional_part;
> +    }
> +}
> +
>  %}
>
>  /* We want a reentrant parser, even if the TZ manipulation and the calls to
> @@ -290,8 +344,8 @@ set_hhmmss (parser_control *pc, long int hour, long int 
> minutes,
>  %parse-param { parser_control *pc }
>  %lex-param { parser_control *pc }
>
> -/* This grammar has 31 shift/reduce conflicts. */
> -%expect 31
> +/* This grammar has 33 shift/reduce conflicts. */
> +%expect 33
>
>  %union
>  {
> @@ -358,12 +412,22 @@ item:
>
>  datetime:
>      iso_8601_datetime
> +  | iso_8601_basic_datetime
>    ;
>
>  iso_8601_datetime:
>      iso_8601_date 'T' iso_8601_time
>    ;
>
> +iso_8601_basic_datetime:
> +    tUNUMBER 'T' unsigned_seconds o_zone_offset
> +      {
> +        digits_iso_8601_basic_to_date (pc, $1);
> +        set_hhmmss_iso_8601_basic_time (pc, $3.tv_sec, $3.tv_nsec);
> +        pc->meridian = MER24;
> +      }
> +  ;
> +
>  time:
>      tUNUMBER tMERIDIAN
>        {
> diff --git a/tests/test-parse-datetime.c b/tests/test-parse-datetime.c
> index 7eba9ad..c620009 100644
> --- a/tests/test-parse-datetime.c
> +++ b/tests/test-parse-datetime.c
> @@ -216,6 +216,67 @@ main (int argc _GL_UNUSED, char **argv)
>            && expected.tv_nsec == result.tv_nsec);
>
>
> +  /* ISO 8601 basic date and time of day representation,
> +     'T' separator, local time zone */
> +  p = "20110501T115518";
> +  expected.tv_sec = ref_time - gmtoff;
> +  expected.tv_nsec = 0;
> +  ASSERT (parse_datetime (&result, p, 0));
> +  LOG (p, expected, result);
> +  ASSERT (expected.tv_sec == result.tv_sec
> +          && expected.tv_nsec == result.tv_nsec);
> +
> +
> +  /* ISO 8601 basic date and time of day representation,
> +     'T' separator, UTC */
> +  p = "20110501T115518Z";
> +  expected.tv_sec = ref_time;
> +  expected.tv_nsec = 0;
> +  ASSERT (parse_datetime (&result, p, 0));
> +  LOG (p, expected, result);
> +  ASSERT (expected.tv_sec == result.tv_sec
> +          && expected.tv_nsec == result.tv_nsec);
> +
> +
> +  /* ISO 8601 basic date and time of day representation,
> +     'T' separator, w/UTC offset */
> +  p = "20110501T115518-0700";
> +  expected.tv_sec = 1304276118;
> +  expected.tv_nsec = 0;
> +  ASSERT (parse_datetime (&result, p, 0));
> +  LOG (p, expected, result);
> +  ASSERT (expected.tv_sec == result.tv_sec
> +          && expected.tv_nsec == result.tv_nsec);
> +
> +
> +  /* ISO 8601 basic date and time of day representation,
> +     'T' separator, w/hour only UTC offset */
> +  p = "20110501T115518-07";
> +  expected.tv_sec = 1304276118;
> +  expected.tv_nsec = 0;
> +  ASSERT (parse_datetime (&result, p, 0));
> +  LOG (p, expected, result);
> +  ASSERT (expected.tv_sec == result.tv_sec
> +          && expected.tv_nsec == result.tv_nsec);
> +
> +
> +  /* ISO 8601 basic date and time of day representation,
> +     'T' separator, w/hour only UTC offset, with ns */
> +  p = "20110501T115518,123456789-07";
> +  expected.tv_sec = 1304276118;
> +  expected.tv_nsec = 123456789;
> +  ASSERT (parse_datetime (&result, p, 0));
> +  LOG (p, expected, result);
> +  ASSERT (expected.tv_sec == result.tv_sec
> +          && expected.tv_nsec == result.tv_nsec);
> +
> +
> +  /* Invalid ISO 8601 basic date and time of day representation,
> +     too many digits for time */
> +  p = "20110501T11551800";
> +  ASSERT (!parse_datetime (&result, p, 0));
> +
> +
>    now.tv_sec = 4711;
>    now.tv_nsec = 1267;
>    p = "now";
> --
> 1.7.9.5
>