Some speed boost on typical regex usage

1 files changed, 57 insertions, 21 deletions

diff --git a/compiler/standardClient.ey b/compiler/standardClient.ey
index 3b7828f..d807d7c 100644
--- a/compiler/standardClient.ey
+++ b/compiler/standardClient.ey
@@ -8,6 +8,7 @@
 # ideas taken from http://swtch.com/~rsc/regexp/regexp3.html
 <
   0 ==:MATCH 1 ==:TERM 2 ==:JUMP 3 ==:SPLIT 4 ==:SAVE 5 ==:FIRST 6 ==:LAST
+  7 ==:FLOATBEGIN 8 ==:TERMBEGIN
 
   { ==b ==a [
     [ SPLIT 1 a len 2 add ]
@@ -267,27 +268,36 @@
     { [
       0 # pc
       [ currentCapture { 0 0 } rep ] # captures
-    ] } /newThread deff
+    ] }" /newThread deff
 
     # TODO: reconsider clist/ilist and also reconsider optimisation potential
-    { ==prog ==string
+    { ==string
       0 ==position
       string len ==maxPosition
       0 ==matched
-      < > ==matchedThread
+      [ ] ==matchedThread
 
-      prog len _ threadList ==clist
-               _ threadList ==nlist
-                 threadList ==ilist
+      clist .clear
+      nlist .clear
+      ilist .clear
 
       newThread _ ==thread clist .add
 
       0 ==pc
-      { } =*code
+      [ ] =*code
 
       ilist .|add =*iPush
       ilist .|pop =*iPop
 
+      {
+        { ilist .size }" {
+          iPop _ =thread
+                 threadGetPC _ =pc
+                               prog * =code
+          0 code codeSemantics *
+        }" loop
+      }" /runIList deffst
+
       [
         { # MATCH
           1 =matched
@@ -310,6 +320,27 @@
           position 0 eq { pc 1 add thread cloneThread iPush }" rep
         }" { # LAST
           position maxPosition eq { pc 1 add thread cloneThread iPush }" rep
+        }" { # FLOATBEGIN
+          pc 1 add thread cloneThread iPush
+          thread
+          runIList
+          position maxPosition lt {
+            nlist .add
+          }" { -- }" ? *
+        }" { # TERMBEGIN
+          1 code =*p
+          nlist .size not clist .size 1 eq and {
+            { position maxPosition lt { position string * p not }" { 0 }" ? * }" {
+              position 1 add =position
+            }" loop
+          }" rep
+
+          position maxPosition lt {
+            position string * p {
+              pc 1 add thread cloneThread nlist .add
+            }" rep
+            thread nlist .add
+          }" rep
         }"
       ] =*codeSemantics
 
@@ -324,12 +355,7 @@
           0 code codeSemantics *
           i 1 add =i
 
-          { ilist .size }" {
-            iPop _ =thread
-                   threadGetPC _ =pc
-                                 prog * =code
-            0 code codeSemantics *
-          }" loop
+          runIList
         }" loop
 
         # "Next input character ========" dump
@@ -348,7 +374,7 @@
         } loop
       } rep
       matched
-    } /execute deffst
+    }' /execute defvst
 
     parse ==prog --
     prog 0 -01 * 0 -01 * FIRST eq {
@@ -356,18 +382,28 @@
         1 prog len range { prog * } each
       ] =prog
     } {
-      [
-        [ SPLIT 3 1 ]
-        [ TERM { -- 1 }" ]
-        [ JUMP 2 neg ]
-        prog _ len dearray
-      ] =prog
+      prog 0 -01 * 0 -01 * TERM eq {
+        [
+          [ TERMBEGIN 1 0 prog * * ]
+          1 prog len range { prog * } each
+        ] =prog
+      } {
+        [
+          [ FLOATBEGIN ]
+          prog _ len dearray
+        ] =prog
+      } ? *
     } ? *
     [
       prog _ len dearray
       [ MATCH ]
     ] =prog
-    { prog execute }
+
+    prog len _ threadList ==clist
+             _ threadList ==nlist
+               threadList ==ilist
+
+    execute
   }
 > -- /enregex deffd