working on lookbehind

diff --git a/basis/regexp/nfa/nfa.factor b/basis/regexp/nfa/nfa.factor
index f070c3528bace3facf203089629f0ca50315091d..aebfea04e1ee534b68110e941cee5ad572289fea 100644 (file)
--- a/basis/regexp/nfa/nfa.factor
+++ b/basis/regexp/nfa/nfa.factor
@@ -14,6 +14,8 @@ SINGLETON: eps
 MIXIN: traversal-flag
 SINGLETON: lookahead-on INSTANCE: lookahead-on traversal-flag
 SINGLETON: lookahead-off INSTANCE: lookahead-off traversal-flag
+SINGLETON: lookbehind-on INSTANCE: lookbehind-on traversal-flag
+SINGLETON: lookbehind-off INSTANCE: lookbehind-off traversal-flag
 SINGLETON: capture-group-on INSTANCE: capture-group-on traversal-flag
 SINGLETON: capture-group-off INSTANCE: capture-group-off traversal-flag
 
@@ -143,6 +145,14 @@ M: lookahead nfa-node ( node -- )
     lookahead-off add-traversal-flag
     2 [ concatenate-nodes ] times ;
 
+M: lookbehind nfa-node ( node -- )
+    eps literal-transition add-simple-entry
+    lookbehind-on add-traversal-flag
+    term>> nfa-node
+    eps literal-transition add-simple-entry
+    lookbehind-off add-traversal-flag
+    2 [ concatenate-nodes ] times ;
+
 : construct-nfa ( regexp -- )
     [
         reset-regexp

diff --git a/basis/regexp/parser/parser.factor b/basis/regexp/parser/parser.factor
index 987d05591f3c61923384019af437cc795fcbca19..d2ed346bf2d0b03309008e03c0dd37fd8c88f6e5 100644 (file)
--- a/basis/regexp/parser/parser.factor
+++ b/basis/regexp/parser/parser.factor
@@ -230,8 +230,18 @@ ERROR: invalid-range a b ;
         [ [ nip at-most-n ] [ at-least-n ] if* ] if
     ] [ drop 0 max exactly-n ] if ;
 
+SINGLETON: beginning-of-input
+SINGLETON: end-of-input
+
+! : beginning-of-input ( -- obj ) 
 : handle-front-anchor ( -- ) front-anchor push-stack ;
-: handle-back-anchor ( -- ) back-anchor push-stack ;
+: end-of-line ( -- obj )
+    end-of-input
+    CHAR: \r <constant>
+    CHAR: \n <constant>
+    2dup 2array <concatenation> 4array <alternation> lookahead boa ;
+
+: handle-back-anchor ( -- ) end-of-line push-stack ;
 
 ERROR: bad-character-class obj ;
 ERROR: expected-posix-class ;
@@ -277,6 +287,8 @@ ERROR: unrecognized-escape char ;
     read1
     {
         { CHAR: \ [ CHAR: \ <constant> ] }
+        { CHAR: ^ [ CHAR: ^ <constant> ] }
+        { CHAR: $ [ CHAR: $ <constant> ] }
         { CHAR: - [ CHAR: - <constant> ] }
         { CHAR: { [ CHAR: { <constant> ] }
         { CHAR: } [ CHAR: } <constant> ] }
@@ -289,7 +301,6 @@ ERROR: unrecognized-escape char ;
         { CHAR: + [ CHAR: + <constant> ] }
         { CHAR: ? [ CHAR: ? <constant> ] }
         { CHAR: . [ CHAR: . <constant> ] }
-! xyzzy
         { CHAR: : [ CHAR: : <constant> ] }
         { CHAR: t [ CHAR: \t <constant> ] }
         { CHAR: n [ CHAR: \n <constant> ] }
@@ -297,8 +308,6 @@ ERROR: unrecognized-escape char ;
         { CHAR: f [ HEX: c <constant> ] }
         { CHAR: a [ HEX: 7 <constant> ] }
         { CHAR: e [ HEX: 1b <constant> ] }
-        { CHAR: $ [ CHAR: $ <constant> ] }
-        { CHAR: ^ [ CHAR: ^ <constant> ] }
 
         { CHAR: d [ digit-class ] }
         { CHAR: D [ digit-class <negation> ] }
@@ -320,16 +329,16 @@ ERROR: unrecognized-escape char ;
         ! { CHAR: G [ end of previous match ] }
         ! { CHAR: Z [ handle-end-of-input ] }
         ! { CHAR: z [ handle-end-of-input ] } ! except for terminator
-! xyzzy
-        { CHAR: 1 [ CHAR: 1 <constant> ] }
-        { CHAR: 2 [ CHAR: 2 <constant> ] }
-        { CHAR: 3 [ CHAR: 3 <constant> ] }
-        { CHAR: 4 [ CHAR: 4 <constant> ] }
-        { CHAR: 5 [ CHAR: 5 <constant> ] }
-        { CHAR: 6 [ CHAR: 6 <constant> ] }
-        { CHAR: 7 [ CHAR: 7 <constant> ] }
-        { CHAR: 8 [ CHAR: 8 <constant> ] }
-        { CHAR: 9 [ CHAR: 9 <constant> ] }
+
+        ! { CHAR: 1 [ CHAR: 1 <constant> ] }
+        ! { CHAR: 2 [ CHAR: 2 <constant> ] }
+        ! { CHAR: 3 [ CHAR: 3 <constant> ] }
+        ! { CHAR: 4 [ CHAR: 4 <constant> ] }
+        ! { CHAR: 5 [ CHAR: 5 <constant> ] }
+        ! { CHAR: 6 [ CHAR: 6 <constant> ] }
+        ! { CHAR: 7 [ CHAR: 7 <constant> ] }
+        ! { CHAR: 8 [ CHAR: 8 <constant> ] }
+        ! { CHAR: 9 [ CHAR: 9 <constant> ] }
 
         { CHAR: Q [ parse-escaped-literals ] }
         [ unrecognized-escape ]
@@ -406,6 +415,10 @@ DEFER: handle-left-bracket
 
 : parse-regexp-token ( token -- ? )
     {
+! todo: only match these at beginning/end of regexp
+        { CHAR: ^ [ handle-front-anchor t ] }
+        { CHAR: $ [ handle-back-anchor t ] }
+
         { CHAR: . [ handle-dot t ] }
         { CHAR: ( [ handle-left-parenthesis t ] }
         { CHAR: ) [ handle-right-parenthesis f ] }
@@ -415,8 +428,6 @@ DEFER: handle-left-bracket
         { CHAR: + [ handle-plus t ] }
         { CHAR: { [ handle-left-brace t ] }
         { CHAR: [ [ handle-left-bracket t ] }
-        { CHAR: ^ [ handle-front-anchor t ] }
-        { CHAR: $ [ handle-back-anchor t ] }
         { CHAR: \ [ handle-escape t ] }
         [ <constant> push-stack t ]
     } case ;

diff --git a/basis/regexp/regexp-tests.factor b/basis/regexp/regexp-tests.factor
index ab3bca9eadf7c34e516df7c119ad1ef398d5cbb4..4d25a58c39317030071fd75a5426fbf57e3bbbbe 100644 (file)
--- a/basis/regexp/regexp-tests.factor
+++ b/basis/regexp/regexp-tests.factor
@@ -285,3 +285,17 @@ IN: regexp-tests
 ! 2. (A)
 ! 3. (B(C))
 ! 4. (C) 
+
+! clear "a(?=b*)" <regexp> "ab" over match
+! clear "a(?=b*c)" <regexp> "abbbbbc" over match
+! clear "a(?=b*)" <regexp> "ab" over match
+
+! clear "^a" <regexp> "a" over match
+! clear "^a" <regexp> "\na" over match
+! clear "^a" <regexp> "\r\na" over match
+! clear "^a" <regexp> "\ra" over match
+
+! clear "a$" <regexp> "a" over match
+! clear "a$" <regexp> "a\n" over match
+! clear "a$" <regexp> "a\r" over match
+! clear "a$" <regexp> "a\r\n" over match

diff --git a/basis/regexp/regexp.factor b/basis/regexp/regexp.factor
index c4ab2675d22280c25beee5b20d0fb347065dbbda..debf94ef33793bf9d92366b2b90cc9233b99752a 100644 (file)
--- a/basis/regexp/regexp.factor
+++ b/basis/regexp/regexp.factor
@@ -128,6 +128,8 @@ IN: regexp
 : option? ( option regexp -- ? )
     options>> key? ;
 
+USE: multiline
+/*
 M: regexp pprint*
     [
         [
@@ -136,3 +138,4 @@ M: regexp pprint*
             case-insensitive swap option? [ "i" % ] when
         ] "" make
     ] keep present-text ;
+*/

diff --git a/basis/regexp/traversal/traversal.factor b/basis/regexp/traversal/traversal.factor
index 6f41b16d950b627a8407af0803225e855ae4f443..d4b43c37f013ac823d59e2d613f226254734de8c 100644 (file)
--- a/basis/regexp/traversal/traversal.factor
+++ b/basis/regexp/traversal/traversal.factor
@@ -8,9 +8,11 @@ IN: regexp.traversal
 TUPLE: dfa-traverser
     dfa-table
     traversal-flags
+    traverse-forward
     capture-groups
     { capture-group-index integer }
     lookahead-counters
+    lookbehind-counters
     last-state current-state
     text
     start-index current-index
@@ -23,10 +25,12 @@ TUPLE: dfa-traverser
         swap [ start-state>> >>current-state ] keep
         >>dfa-table
         swap >>text
+        t >>traverse-forward
         0 >>start-index
         0 >>current-index
         V{ } clone >>matches
         V{ } clone >>capture-groups
+        V{ } clone >>lookbehind-counters
         V{ } clone >>lookahead-counters ;
 
 : final-state? ( dfa-traverser -- ? )
@@ -52,8 +56,19 @@ M: lookahead-on flag-action ( dfa-traverser flag -- )
 
 M: lookahead-off flag-action ( dfa-traverser flag -- )
     drop
-    dup lookahead-counters>> pop
-    '[ _ - ] change-current-index drop ;
+    dup lookahead-counters>>
+    [ drop ] [ pop '[ _ - ] change-current-index drop ] if-empty ;
+
+M: lookbehind-on flag-action ( dfa-traverser flag -- )
+    drop
+    f >>traverse-forward
+    lookbehind-counters>> 0 swap push ;
+
+M: lookbehind-off flag-action ( dfa-traverser flag -- )
+    drop
+    t >>traverse-forward
+    dup lookbehind-counters>>
+    [ drop ] [ pop '[ _ + ] change-current-index drop ] if-empty ;
 
 : process-flags ( dfa-traverser -- )
     [ [ 1+ ] map ] change-lookahead-counters
@@ -62,7 +77,10 @@ M: lookahead-off flag-action ( dfa-traverser flag -- )
 
 : increment-state ( dfa-traverser state -- dfa-traverser )
     [
-        [ 1+ ] change-current-index dup current-state>> >>last-state
+        dup traverse-forward>>
+        [ [ 1+ ] change-current-index ]
+        [ [ 1- ] change-current-index ] if
+        dup current-state>> >>last-state
     ] dip
     first >>current-state ;