[post size limit dropped/2]
$ TZ=GMT diff -u irregex.scm irregex2.scm
--- irregex.scm 2020-07-12 09:41:49.754022271 +0000
+++ irregex2.scm 2020-07-12 09:45:05.067970996 +0000
@@ -3813,9 +3813,11 @@
(finish j acc))
((= j i)
;; skip one char forward if we match the empty string
- (lp (list str (+ j 1) end) (+ j 1) acc))
+ ; (lp (list str (+ j 1) end) (+ j 1) acc))
+ (lp src (+ j 1) acc))
(else
- (lp (list str j end) j acc))))))))))
+ ; (lp (list str j end) j acc))))))))))
+ (lp src j acc))))))))))
(define (irregex-fold irx kons . args)
(if (not (procedure? kons)) (error "irregex-fold: not a procedure" kons))This fixes the failing tests because the eq? of wrap-end-chunker now works properly which makes 'end*' in look-behind follow along with 'i', moving the look-behind check to the right place.
scheme@(guile-user)> (irregex-replace/all "(?<=x)a" "xa xa xa" "-")
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 0 8 | xa xa xa 0 0
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 1 8 | xa xa xa 0 1
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 2 8 | xa xa xa 0 2
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 3 8 | xa xa xa 0 3
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 4 8 | xa xa xa 0 4
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 5 8 | xa xa xa 0 5
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 6 8 | xa xa xa 0 6
look-behind ((xa xa xa 0 8) . 0) (xa xa xa 0 8) xa xa xa 7 8 | xa xa xa 0 7
$1 = "x- x- x-"
scheme@(guile-user)> (irregex-replace/all "(?<=x)a" "xa aaa x" "-")
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 0 8 | xa aaa x 0 0
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 1 8 | xa aaa x 0 1
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 2 8 | xa aaa x 0 2
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 3 8 | xa aaa x 0 3
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 4 8 | xa aaa x 0 4
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 5 8 | xa aaa x 0 5
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 6 8 | xa aaa x 0 6
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 7 8 | xa aaa x 0 7
look-behind ((xa aaa x 0 8) . 0) (xa aaa x 0 8) xa aaa x 8 8 | xa aaa x 0 8
$2 = "x- aaa x"
scheme@(guile-user)> (irregex-replace/all "(?<=[xyz])a" "xa ya za" "-")
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 0 8 | xa ya za 0 0
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 1 8 | xa ya za 0 1
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 2 8 | xa ya za 0 2
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 3 8 | xa ya za 0 3
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 4 8 | xa ya za 0 4
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 5 8 | xa ya za 0 5
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 6 8 | xa ya za 0 6
look-behind ((xa ya za 0 8) . 0) (xa ya za 0 8) xa ya za 7 8 | xa ya za 0 7
$3 = "x- y- z-"
scheme@(guile-user)> However, I am not familiar with all of irregex, only the small bits and pieces I've looked at, so it's entirely possible that this breaks something else. I do not know whether or not the chunk tampering behavior by irregex-fold/fast is relied on in some other part of the code. The shinnoid would have to chime in on that. Someone familiar with chicken scheme might also rerun the test suite with this change.
https://lists.nongnu.org/archive/html/chicken-users/2020-07/msg00005.html