From: John Benediktsson <mrjbq7@gmail.com>
Date: Tue, 9 Aug 2022 21:44:15 +0000 (-0700)
Subject: regexp: fix case-insensitive lookahead and lookbehind.
X-Git-Tag: 0.99~1153
X-Git-Url: https://gitweb.factorcode.org/gitweb.cgi?p=factor.git;a=commitdiff_plain;h=c0f0342d9ae9197401af909d86204c2d1f89559a

regexp: fix case-insensitive lookahead and lookbehind.
---

diff --git a/basis/regexp/nfa/nfa.factor b/basis/regexp/nfa/nfa.factor
index 6912e4d94c..044313f5e4 100644
--- a/basis/regexp/nfa/nfa.factor
+++ b/basis/regexp/nfa/nfa.factor
@@ -53,24 +53,12 @@ M:: star nfa-node ( node -- start end )
     s1 s3 epsilon-transition
     s2 s3 ;
 
-GENERIC: modify-epsilon ( tag -- newtag )
-! Potential off-by-one errors when lookaround nested in lookbehind
-
-M: object modify-epsilon ;
-
-: line-option ( multiline unix-lines default -- option )
-    multiline option? [
-        drop [ unix-lines option? ] 2dip swap ?
-    ] [ 2nip ] if ;
+DEFER: modify-class
 
-M: $crlf modify-epsilon
-    $unix end-of-input line-option ;
-
-M: ^crlf modify-epsilon
-    ^unix beginning-of-input line-option ;
+! Potential off-by-one errors when lookaround nested in lookbehind
 
 M: tagged-epsilon nfa-node
-    clone [ modify-epsilon ] change-tag add-simple-entry ;
+    clone [ modify-class ] change-tag add-simple-entry ;
 
 M: concatenation nfa-node
     [ first>> ] [ second>> ] bi
@@ -96,6 +84,31 @@ GENERIC: modify-class ( char-class -- char-class' )
 
 M: object modify-class ;
 
+M: concatenation modify-class
+    [ first>> ] [ second>> ] bi [ modify-class ] bi@
+    concatenation boa ;
+
+M: alternation modify-class
+    [ first>> ] [ second>> ] bi [ modify-class ] bi@
+    alternation boa ;
+
+M: lookahead modify-class
+    term>> modify-class lookahead boa ;
+
+M: lookbehind modify-class
+    term>> modify-class lookbehind boa ;
+
+: line-option ( multiline unix-lines default -- option )
+    multiline option? [
+        drop [ unix-lines option? ] 2dip swap ?
+    ] [ 2nip ] if ;
+
+M: $crlf modify-class
+    $unix end-of-input line-option ;
+
+M: ^crlf modify-class
+    ^unix beginning-of-input line-option ;
+
 M: integer modify-class
     case-insensitive option? [
         dup Letter? [
@@ -103,9 +116,6 @@ M: integer modify-class
         ] when
     ] when ;
 
-M: integer nfa-node
-    modify-class add-simple-entry ;
-
 M: primitive-class modify-class
     class>> modify-class <primitive-class> ;
 
diff --git a/basis/regexp/regexp-tests.factor b/basis/regexp/regexp-tests.factor
index 36e877a03f..cdbe7742ef 100644
--- a/basis/regexp/regexp-tests.factor
+++ b/basis/regexp/regexp-tests.factor
@@ -513,6 +513,9 @@ unit-test
 
 { 3 } [ "caba" "(?<=b)a" <regexp> first-match from>> ] unit-test
 
+{ "<" } [ "<style>" R/ <(?=STYLE\b)/i first-match >string ] unit-test
+{ "bar" } [ "foobar" R/ (?<=FOO)BAR/i first-match >string ] unit-test
+
 { t } [ "\ra" R/ .^a/ms matches? ] unit-test
 { f } [ "\ra" R/ .^a/mds matches? ] unit-test
 { t } [ "\na" R/ .^a/ms matches? ] unit-test