regexp: greedy matching right-to-left vs left-to-right #110
Description
by stefanha:
regexp.MatchStrings() produces unexpected results which suggest that greedy
matching works right-to-left. Conventional regular expression engines seem
to match left-to-right (sed, Vim, Perl, Python, ...).
The documentation does not state the intended behavior. If it really
should work right-to-left, then a note in the documentation would help
reduce surprise.
What steps will reproduce the problem?
1. See the following test program:
package main
import "regexp"
import "fmt"
func main() {
re, _ := regexp.Compile("(.*).*");
a := re.MatchStrings("this is a test");
fmt.Printf("%#v\n", a);
}
2. The following output is produced:
[]string{"this is a test", ""}
The .* outside of a group has matched the line. I expect greedy matching
to work left-to-right, therefore the (.*) should match the line:
[]string{"this is a test", "this is a test"}
What is your $GOOS? $GOARCH?
GOOS=linux GOARCH=386
Which revision are you sync'ed to? (hg log -l 1)
3975:b51fd2d6c160
Simple comparisons:
$ echo this is a test | sed 's/\(.*\).*/\1/'
this is a test
$ python -c 'import re; print re.match("(.*).*", "this is a
test").groups()'
('this is a test',)