I'm a total newbie when it comes to Natty and Antler. Up to now, Natty has been great and has parsed dates with no problems. Recently we have started to receive a new date and time format which Natty has trouble extracting.
Mon 29 Feb 09:00:00 2016
It cannot extract the year due to it being separated from the rest of the date.
I've been trying to add my own format into DateParser, where it could pick up on this format as it does with any other.
I've made the following changes:
date_time: Added an extra rule called custom_dates which will be the new rule for my format
date_time
: (
(date)=>date (date_time_separator explicit_time)?
| explicit_time (time_date_separator date)?
| custom_dates
) -> ^(DATE_TIME date? explicit_time?)
| relative_time -> ^(DATE_TIME relative_time?)
;
custom_date: My new rule
custom_date
: relaxed_day_of_week WHITE_SPACE relaxed_day_of_month WHITE_SPACE relaxed_month (date_time_separator explicit_time)? relaxed_year
-> ^(EXPLICIT_DATE relaxed_day_of_week relaxed_day_of_month relaxed_month relaxed_year (date_time_separator explicit_time)?)
;
When I try to build Natty with my changes, it just hangs, and never finishes. The output up to that point is:
Decision can match input such as "COMMA WHITE_SPACE INT_00 INT_00" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input warning(200): com\joestelmach\natty\generated\DateParser.g:444:73: Decision can match input such as "COMMA WHITE_SPACE INT_00 {INT_13..INT_19, INT_20..INT_23}" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input warning(200): com\joestelmach\natty\generated\DateParser.g:496:45: Decision can match input such as "WHITE_SPACE IN {COMMA, WHITE_SPACE}" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input warning(200): com\joestelmach\natty\generated\DateParser.g:504:77: Decision can match input such as "WHITE_SPACE IN {COMMA, WHITE_SPACE}" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
Am I possibly going the wrong way about this? I've taken a look at the Natty and ANTLR v3 documentation but there isn't much to go on.
Thanks in advance
EDIT:
As requested in the comments below. I've added in where the first warning occurs. However what I've included above is just a small snapshot of the dozens of warnings that have been in there before I modified any code with my own rules
The first warning appears in the date_time_separator
date_time_separator
: WHITE_SPACE (AT WHITE_SPACE)?
| WHITE_SPACE? COMMA WHITE_SPACE? (AT WHITE_SPACE)?
| T
;
One observation I've made is when I changed my rule to always include the time
custom_date
: relaxed_day_of_week WHITE_SPACE relaxed_day_of_month WHITE_SPACE relaxed_month (date_time_separator explicit_time) relaxed_year
-> ^(EXPLICIT_DATE relaxed_day_of_week relaxed_day_of_month relaxed_month relaxed_year (date_time_separator explicit_time)?)
;
When I compile I receive this error:
error(202): com\joestelmach\natty\generated\DateParser.g:831:3: the decision cannot distinguish between alternative(s) 1,2 for input such as "INT_00 INT_00 INT_00 EOF"
Looking at line 831 is where the explicit_time resides. I cannot find anything on StackOverflow or otherwise as to what this error means. I assume this error means that there is some ambiguity between the two possible routes. However I don't understand why merely adding in my code should cause an error.
explicit_time_hours_minutes returns [String hours, String minutes, String ampm]
: hours (COLON | DOT)? minutes ((COLON | DOT)? seconds)? (WHITE_SPACE (meridian_indicator | (MILITARY_HOUR_SUFFIX | HOUR)))?
{$hours=$hours.text; $minutes=$minutes.text; $ampm=$meridian_indicator.text;}
-> hours minutes seconds? meridian_indicator?
| hours (WHITE_SPACE? meridian_indicator)?
{$hours=$hours.text; $ampm=$meridian_indicator.text;}
-> hours ^(MINUTES_OF_HOUR INT["0"]) meridian_indicator?
;