From f5a94313f275261d72b73734982c0f59eb1a7b93 Mon Sep 17 00:00:00 2001
From: David Schmenk <dschmenk@gmail.com>
Date: Mon, 12 Mar 2018 15:13:35 -0700
Subject: [PATCH] Add assembly helper for lexical scanner ID/Keyword match

---
 src/toolsrc/lex.pla   |  56 ++++++++++-------------
 src/toolsrc/plasm.pla | 100 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 124 insertions(+), 32 deletions(-)

diff --git a/src/toolsrc/lex.pla b/src/toolsrc/lex.pla
index b20e863..6d8f062 100644
--- a/src/toolsrc/lex.pla
+++ b/src/toolsrc/lex.pla
@@ -26,29 +26,24 @@
 //    fin
 //    return FALSE
 //end
-def keymatch
-    byte i, keypos
-    word chrptr
-
-    keypos = 0
-    while keywrds[keypos] < tknlen
-        keypos = keypos + keywrds[keypos] + 2
-    loop
-    chrptr = tknptr - 1
-    while keywrds[keypos] == tknlen
-        i = 1; while i <= tknlen and ^(chrptr + i) == keywrds[keypos + i]; i++; loop
-        //for i = 1 to tknlen
-        //    if ^(chrptr + i) <> keywrds[keypos + i]
-        //        break
-        //    fin
-        //next
-        if i > tknlen
-            return keywrds[keypos + keywrds[keypos] + 1]
-        fin
-        keypos = keypos + keywrds[keypos] + 2
-    loop
-    return ID_TKN
-end
+//def keymatch
+//    byte i, keypos
+//    word chrptr
+//
+//    keypos = 0
+//    while keywrds[keypos] < tknlen
+//        keypos = keypos + keywrds[keypos] + 2
+//    loop
+//    chrptr = tknptr - 1
+//    while keywrds[keypos] == tknlen
+//        i = 1; while i <= tknlen and ^(chrptr + i) == keywrds[keypos + i]; i++; loop
+//        if i > tknlen
+//            return keywrds[keypos + keywrds[keypos] + 1]
+//        fin
+//        keypos = keypos + keywrds[keypos] + 2
+//    loop
+//    return ID_TKN
+//end
 def scannum
     word num
     num = 0
@@ -85,23 +80,20 @@ def scan
         scanptr++
     loop
     tknptr  = scanptr
-    scanchr = toupper(^scanptr)
+    scanchr, scanptr, token = scanid(scanptr, @keywrds) //scanchr = toupper(^scanptr)
     //
     // Scan for token based on first character
     //
-    //if isalpha(scanchr)
-    if (scanchr >= 'A' and scanchr <= 'Z') or (scanchr == '_')
+    if token //if isalpha(scanchr)
         //
         // ID, either variable name or reserved word
         //
-        repeat
-            ^scanptr = scanchr
-            scanptr++
-            scanchr = toupper(^scanptr)
+        //repeat
+        //    ^scanptr = scanchr
+        //    scanptr++
+        //    scanchr = toupper(^scanptr)
         //until not isalphanum(scanchr)
-        until not ((scanchr >= 'A' and scanchr <= 'Z') or (scanchr >= '0' and scanchr <= '9' ) or (scanchr == '_'))
         tknlen = scanptr - tknptr
-        token  = keymatch
     elsif scanchr >= '0' and scanchr <= '9' // isnum()
         //
         // Decimal constant
diff --git a/src/toolsrc/plasm.pla b/src/toolsrc/plasm.pla
index 7b44939..f63106d 100644
--- a/src/toolsrc/plasm.pla
+++ b/src/toolsrc/plasm.pla
@@ -338,6 +338,106 @@ const ERR_SYNTAX        = $8000
 //
 //=====================================
 
+//
+// Lexical scanner helper for keyword/IDs
+//
+asm scanid(scanptr, keywrds)#3
+        !SOURCE    "vmsrc/plvmzp.inc"
+        LDA     ESTKL,X
+        STA     DSTL
+        LDA     ESTKH,X
+        STA     DSTH
+        LDA     ESTKL+1,X
+        STA     ESTKL,X         ; COPY OUTPUT SCANPTR
+        STA     SRCL
+        LDA     ESTKH+1,X
+        STA     ESTKH,X
+        STA     SRCH
+        DEX
+        LDA     #$00
+        STA     ESTKL,X         ; CLEAR OUTPUT TOKEN
+        STA     ESTKH,X
+        STA     ESTKH+2,X       ; CLEAR MSB OF SCANCHR
+        TAY
+        LDA     (SRC),Y
+        AND     #$7F
+        CMP     #'a'
+        BCC     +
+        CMP     #'z'+1
+        BCS     +
+        SBC     #$1F
+        STA     (SRC),Y
++       STA     ESTKL+2,X       ; SET SCANCHR
+        CMP     #'_'
+        BEQ     +
+        CMP     #'A'
+        BCC     SCANEX
+        CMP     #'Z'+1
+        BCS     SCANEX
++       LDA     #$D6            ; ID_TKN
+        STA     ESTKL,X         ; SET OUTPUT TOKEN = ID_TKN
+SCANID  INY
+        LDA     (SRC),Y
+        AND     #$7F
+        BEQ     ++
+        CMP     #'a'
+        BCC     +
+        CMP     #'z'+1
+        BCS     ++
+        SBC     #$1F
+        STA     (SRC),Y         ; COPY UPPERCASE CHAR BACK TO ^SCANPTR
+        BNE     SCANID
++       CMP     #'_'
+        BEQ     SCANID
+        CMP     #'0'
+        BCC     ++
+        CMP     #'9'+1
+        BCC     SCANID
+        CMP     #'A'
+        BCC     ++
+        CMP     #'Z'+1
+        BCC     SCANID
+++      STY     TMPL
+        TYA
+        LDY     #$00
+        CLC
+        ADC     SRCL
+        STA     ESTKL+1,X       ; UPDATE SCANPTR
+        BCC     MATCHLEN
+        INC     ESTKH+1,X
+MATCHLEN LDA    (DST),Y
+        CMP     TMPL
+        BCS     +
+        ADC     #$02
+        ADC     DSTL
+        STA     DSTL
+        BCC     MATCHLEN
+        INC     DSTH
+        BNE     MATCHLEN
++       BNE     SCANEX          ; NO KEY MATCH
+        TAY
+        DEY
+        INC     DSTL
+        BNE     MATCHKEY
+        INC     DSTH
+MATCHKEY LDA     (SRC),Y
+        CMP     (DST),Y
+        BNE     NEXTKEY
+        DEY
+        BPL     MATCHKEY
+        LDY     TMPL
+        LDA     (DST),Y
+        STA     ESTKL,X         ; SET OUTPUT TOKEN
+SCANEX  RTS
+NEXTKEY LDY     #$00
+        LDA     TMPL
+        SEC
+        ADC     DSTL
+        STA     DSTL
+        BCC     MATCHLEN
+        INC     DSTH
+        BNE     MATCHLEN
+end
 //
 // Handy functions
 //