How to collect invalid inputs?

[avsaase]

I'm trying to write a parser for cron expressions that collects invalid inputs instead of failing to parse the expression altogether. For example, when parsing 1,5- 4 * * *, I want to record the valid 1 and invalid 5- entries for the minutes field, the valid 4 entry for the hours and the valid * for the remaining fields.

This is the grammar I put together:

all     = { "*" }
number  = { (ASCII_DIGIT)+ }
range   = { number ~ "-" ~ number }
step    = { (all | range) ~ "/" ~ number }
invalid = { ANY+ }
entry   = { (step | range | number | all | invalid) }
field   = { entry ~ ("," ~ entry)* }
ws      = { (" ")+ }
cron    = { SOI ~ #minutes = field ~ ws ~ #hours = field ~ ws ~ #days_of_month = field ~ ws ~ #months = field ~ ws ~ #days_of_week = field ~ EOI }

The problem is that the 5 gets eagerly parsed by number and the - is left for the next rule which then fails. Similarly, when parsing -1 4 * * *, the invalid rule captures the full input and the parser fails because there's nothing left to parse. Is there some way to look head in the entry rule to make sure the wrong rule doesn't capture part of the input? I feel like using the stack could be useful here but I have a hard time wrapping my head around how to actually use it.

[skittishdev]

Does this work?

WHITESPACE = _{ " " }
all     = { "*" }
number  = { (ASCII_DIGIT)+ }
range   = { number ~ "-" ~ number }
step    = { (all | range) ~ "/" ~ number }
entry   = { (step | range | number | all ) }
field   = { entry ~ (",")* }
cron    = { SOI ~ #minutes = field ~ #hours = field ~ #days_of_month = field ~ #months = field ~ #days_of_week = field ~ EOI }