Added support for octal character encoding in strings and literal chars. Closes #642

src/main/java/dk/camelot64/kickc/model/values/StringEncoding.java

@@ -3,8 +3,9 @@ package dk.camelot64.kickc.model.values;
 import dk.camelot64.kickc.model.CompileError;
 import kickass.nonasm.c64.CharToPetsciiConverter;
 
+import java.util.LinkedList;
 import java.util.Map;
-import java.util.PrimitiveIterator;
+import java.util.stream.Collectors;
 
 /** String encoding. */
 public enum StringEncoding {
@@ -15,8 +16,7 @@ public enum StringEncoding {
    SCREENCODE_UPPER("screencode_upper", "screencode_upper", "su", CharToPetsciiConverter.charToScreenCode_upper),
    ASCII("ascii", "ascii", "as", CharToPetsciiConverter.charToAscii),
    ATASCII("atascii", null, "at", CharToAtasciiConverter.charToAtascii),
-   SCREENCODE_ATARI("screencode_atari", null, "sa", CharToAtasciiConverter.charToScreenCodeAtari)
-   ;
+   SCREENCODE_ATARI("screencode_atari", null, "sa", CharToAtasciiConverter.charToScreenCodeAtari);
 
    /** Char value used to encode \xnn chars without a value within the chosen encoding. A char C is  encoded as CHAR_SPECIAL_VAL+C */
    public static final char CHAR_SPECIAL_VAL = 64000;
@@ -136,9 +136,9 @@ public enum StringEncoding {
     */
    public String escapeToAscii(String stringValue) {
       StringBuilder stringResult = new StringBuilder();
-      final PrimitiveIterator.OfInt escapedIterator = stringValue.chars().iterator();
-      while(escapedIterator.hasNext()) {
-         stringResult.append(escapeToAsciiFirst(escapedIterator));
+      final LinkedList<Integer> escapedChars = new LinkedList<>(stringValue.chars().boxed().collect(Collectors.toList()));
+      while(!escapedChars.isEmpty()) {
+         stringResult.append(escapeToAsciiFirst(escapedChars));
       }
       return stringResult.toString();
    }
@@ -151,13 +151,13 @@ public enum StringEncoding {
     * @param escapedCharsIterator The characters of the string to parse one char from. The iterator is moved beyond any handled chars.
     * @return The first ASCII character of the list.
     */
-   public char escapeToAsciiFirst(PrimitiveIterator.OfInt escapedCharsIterator) {
-      char stringChar = (char) escapedCharsIterator.nextInt();
+   public char escapeToAsciiFirst(LinkedList<Integer> escapedCharsIterator) {
+      char stringChar = (char) escapedCharsIterator.pop().intValue();
       if(stringChar != '\\')
          return stringChar;
       // Escape started - handle it!
-      if(!escapedCharsIterator.hasNext()) throw new CompileError("Unfinished string escape sequence at end of string");
-      char escapeChar = (char) escapedCharsIterator.nextInt();
+      if(escapedCharsIterator.isEmpty()) throw new CompileError("Unfinished string escape sequence at end of string");
+      char escapeChar = (char) escapedCharsIterator.pop().intValue();
       switch(escapeChar) {
          case 'n':
             return '\n';
@@ -165,8 +165,6 @@ public enum StringEncoding {
             return '\r';
          case 'f':
             return '\f';
-         case '0':
-            return '\0';
          case '"':
             return '"';
          case '\'':
@@ -175,10 +173,28 @@ public enum StringEncoding {
             return '\\';
          case 'x':
             String hexNum = "";
-            hexNum += (char) escapedCharsIterator.nextInt();
-            hexNum += (char) escapedCharsIterator.nextInt();
+            hexNum += (char) escapedCharsIterator.pop().intValue();
+            hexNum += (char) escapedCharsIterator.pop().intValue();
             final byte hexEncoding = (byte) Integer.parseInt(hexNum, 16);
             return charFromEncoded(hexEncoding);
+         case '0':
+         case '1':
+         case '2':
+         case '3':
+         case '4':
+         case '5':
+         case '6':
+         case '7':
+            String octalNum = "" + escapeChar;
+            while(octalNum.length() < 3) {
+               final Integer peek = escapedCharsIterator.peek();
+               if(peek != null && peek >= '0' && peek <= '7') {
+                  octalNum += (char) escapedCharsIterator.pop().intValue();
+               } else
+                  break;
+            }
+            final byte octalEncoding = (byte) Integer.parseInt(octalNum, 8);
+            return charFromEncoded(octalEncoding);
          default:
             throw new CompileError("Illegal string escape sequence \\" + escapeChar);
       }

src/main/java/dk/camelot64/kickc/parser/KickCLexer.g4

@@ -140,7 +140,7 @@ fragment NAME_CHAR : [a-zA-Z0-9_];
 
 // Strings and chars
 STRING : '"' ('\\"' | ~'"')* '"' [z]?([aps][tsmua]?)?[z]? ;
-CHAR : '\''  ('\\'(['"rfn\\]|'x'[0-9a-f][0-9a-f]) | ~'\'' ) '\'';
+CHAR : '\''  ('\\'(['"rfn\\0-9][0-9]?[0-9]?|'x'[0-9a-f][0-9a-f]) | ~'\'' ) '\'';
 
 // White space on hidden channel 1
 WS : [ \t\r\n\u00a0]+ -> channel(1);

src/test/java/dk/camelot64/kickc/test/TestProgramsFast.java

@@ -9,6 +9,7 @@ import java.io.IOException;
  */
 public class TestProgramsFast extends TestPrograms {
 
+
    @Test
    public void testTmpZpProblem() throws IOException {
       compileAndCompare("tmp-zp-problem.c");
@@ -1676,6 +1677,21 @@ public class TestProgramsFast extends TestPrograms {
       assertError("string-escapes-err-0.c", "Unfinished string escape sequence at end of string");
    }
 
+   @Test
+   public void testStringEscapes10() throws IOException {
+      compileAndCompare("string-escapes-10.c");
+   }
+
+   @Test
+   public void testStringEscapes9() throws IOException {
+      compileAndCompare("string-escapes-9.c");
+   }
+
+   @Test
+   public void testStringEscapes8() throws IOException {
+      compileAndCompare("string-escapes-8.c");
+   }
+
    @Test
    public void testStringEscapes7() throws IOException {
       compileAndCompare("string-escapes-7.c");

src/test/kc/string-escapes-10.c

@@ -0,0 +1,14 @@
+// Test octal escapes in chars
+
+char MSG1 = '\1';
+char MSG2 = '\02';
+char MSG3 = '\003';
+char MSG4 = '\377';
+char* SCREEN = (char*)0x0400;
+
+void main() {
+    SCREEN[0] = MSG1;
+    SCREEN[1] = MSG2;
+    SCREEN[2] = MSG3;
+    SCREEN[3] = MSG4;
+}

src/test/kc/string-escapes-8.c

@@ -0,0 +1,5 @@
+// Support for '\0' literal character
+char * const SCREEN = (char*)0x0400;
+void main(void) {
+    SCREEN[0] = '\0';
+}

src/test/kc/string-escapes-9.c

@@ -0,0 +1,10 @@
+// Test octal escapes in strings
+
+char MESSAGE[] = "\1\01\001\377";
+char* SCREEN = (char*)0x0400;
+
+void main() {
+    byte i=0;
+    while(MESSAGE[i])
+        SCREEN[i] = MESSAGE[i++];
+}

src/test/ref/string-escapes-10.asm

@@ -0,0 +1,31 @@
+// Test octal escapes in chars
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-10.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  .const MSG1 = 'a'
+  .const MSG2 = 'b'
+  .const MSG3 = 'c'
+  .const MSG4 = '\$ff'
+  .label SCREEN = $400
+.segment Code
+main: {
+    // SCREEN[0] = MSG1
+    lda #MSG1
+    sta SCREEN
+    // SCREEN[1] = MSG2
+    lda #MSG2
+    sta SCREEN+1
+    // SCREEN[2] = MSG3
+    lda #MSG3
+    sta SCREEN+2
+    // SCREEN[3] = MSG4
+    lda #MSG4
+    sta SCREEN+3
+    // }
+    rts
+}

src/test/ref/string-escapes-10.cfg

@@ -0,0 +1,11 @@
+
+void main()
+main: scope:[main]  from
+  [0] *SCREEN = MSG1
+  [1] *(SCREEN+1) = MSG2
+  [2] *(SCREEN+2) = MSG3
+  [3] *(SCREEN+3) = MSG4
+  to:main::@return
+main::@return: scope:[main]  from main
+  [4] return 
+  to:@return

src/test/ref/string-escapes-10.log

@@ -0,0 +1,207 @@
+Inlined call call __init
+
+CONTROL FLOW GRAPH SSA
+
+void main()
+main: scope:[main]  from __start::@1
+  SCREEN[0] = MSG1
+  SCREEN[1] = MSG2
+  SCREEN[2] = MSG3
+  SCREEN[3] = MSG4
+  to:main::@return
+main::@return: scope:[main]  from main
+  return 
+  to:@return
+
+void __start()
+__start: scope:[__start]  from
+  to:__start::__init1
+__start::__init1: scope:[__start]  from __start
+  to:__start::@1
+__start::@1: scope:[__start]  from __start::__init1
+  call main
+  to:__start::@2
+__start::@2: scope:[__start]  from __start::@1
+  to:__start::@return
+__start::@return: scope:[__start]  from __start::@2
+  return 
+  to:@return
+
+SYMBOL TABLE SSA
+__constant char MSG1 = 'a'
+__constant char MSG2 = 'b'
+__constant char MSG3 = 'c'
+__constant char MSG4 = '刺'
+__constant char *SCREEN = (char *)$400
+void __start()
+void main()
+
+Adding number conversion cast (unumber) 0 in SCREEN[0] = MSG1
+Adding number conversion cast (unumber) 1 in SCREEN[1] = MSG2
+Adding number conversion cast (unumber) 2 in SCREEN[2] = MSG3
+Adding number conversion cast (unumber) 3 in SCREEN[3] = MSG4
+Successful SSA optimization PassNAddNumberTypeConversions
+Simplifying constant pointer cast (char *) 1024
+Simplifying constant integer cast 0
+Simplifying constant integer cast 1
+Simplifying constant integer cast 2
+Simplifying constant integer cast 3
+Successful SSA optimization PassNCastSimplification
+Finalized unsigned number type (char) 0
+Finalized unsigned number type (char) 1
+Finalized unsigned number type (char) 2
+Finalized unsigned number type (char) 3
+Successful SSA optimization PassNFinalizeNumberTypeConversions
+Simplifying expression containing zero SCREEN in [0] SCREEN[0] = MSG1
+Successful SSA optimization PassNSimplifyExpressionWithZero
+Removing unused procedure __start
+Removing unused procedure block __start
+Removing unused procedure block __start::__init1
+Removing unused procedure block __start::@1
+Removing unused procedure block __start::@2
+Removing unused procedure block __start::@return
+Successful SSA optimization PassNEliminateEmptyStart
+Consolidated array index constant in *(SCREEN+1)
+Consolidated array index constant in *(SCREEN+2)
+Consolidated array index constant in *(SCREEN+3)
+Successful SSA optimization Pass2ConstantAdditionElimination
+CALL GRAPH
+
+Created 0 initial phi equivalence classes
+Coalesced down to 0 phi equivalence classes
+
+FINAL CONTROL FLOW GRAPH
+
+void main()
+main: scope:[main]  from
+  [0] *SCREEN = MSG1
+  [1] *(SCREEN+1) = MSG2
+  [2] *(SCREEN+2) = MSG3
+  [3] *(SCREEN+3) = MSG4
+  to:main::@return
+main::@return: scope:[main]  from main
+  [4] return 
+  to:@return
+
+
+VARIABLE REGISTER WEIGHTS
+void main()
+
+Initial phi equivalence classes
+Complete equivalence classes
+REGISTER UPLIFT POTENTIAL REGISTERS
+Statement [0] *SCREEN = MSG1 [ ] (  [ ] { }  ) always clobbers reg byte a 
+Statement [1] *(SCREEN+1) = MSG2 [ ] (  [ ] { }  ) always clobbers reg byte a 
+Statement [2] *(SCREEN+2) = MSG3 [ ] (  [ ] { }  ) always clobbers reg byte a 
+Statement [3] *(SCREEN+3) = MSG4 [ ] (  [ ] { }  ) always clobbers reg byte a 
+
+REGISTER UPLIFT SCOPES
+Uplift Scope [main] 
+Uplift Scope [] 
+
+Uplifting [main] best 33 combination 
+Uplifting [] best 33 combination 
+
+ASSEMBLER BEFORE OPTIMIZATION
+  // File Comments
+// Test octal escapes in chars
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-10.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .const MSG1 = 'a'
+  .const MSG2 = 'b'
+  .const MSG3 = 'c'
+  .const MSG4 = '\$ff'
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // [0] *SCREEN = MSG1 -- _deref_pbuc1=vbuc2 
+    lda #MSG1
+    sta SCREEN
+    // [1] *(SCREEN+1) = MSG2 -- _deref_pbuc1=vbuc2 
+    lda #MSG2
+    sta SCREEN+1
+    // [2] *(SCREEN+2) = MSG3 -- _deref_pbuc1=vbuc2 
+    lda #MSG3
+    sta SCREEN+2
+    // [3] *(SCREEN+3) = MSG4 -- _deref_pbuc1=vbuc2 
+    lda #MSG4
+    sta SCREEN+3
+    jmp __breturn
+    // main::@return
+  __breturn:
+    // [4] return 
+    rts
+}
+  // File Data
+
+ASSEMBLER OPTIMIZATIONS
+Removing instruction jmp __breturn
+Succesful ASM optimization Pass5NextJumpElimination
+Removing instruction __breturn:
+Succesful ASM optimization Pass5UnusedLabelElimination
+
+FINAL SYMBOL TABLE
+__constant char MSG1 = 'a'
+__constant char MSG2 = 'b'
+__constant char MSG3 = 'c'
+__constant char MSG4 = '刺'
+__constant char *SCREEN = (char *) 1024
+void main()
+
+
+
+FINAL ASSEMBLER
+Score: 30
+
+  // File Comments
+// Test octal escapes in chars
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-10.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .const MSG1 = 'a'
+  .const MSG2 = 'b'
+  .const MSG3 = 'c'
+  .const MSG4 = '\$ff'
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // SCREEN[0] = MSG1
+    // [0] *SCREEN = MSG1 -- _deref_pbuc1=vbuc2 
+    lda #MSG1
+    sta SCREEN
+    // SCREEN[1] = MSG2
+    // [1] *(SCREEN+1) = MSG2 -- _deref_pbuc1=vbuc2 
+    lda #MSG2
+    sta SCREEN+1
+    // SCREEN[2] = MSG3
+    // [2] *(SCREEN+2) = MSG3 -- _deref_pbuc1=vbuc2 
+    lda #MSG3
+    sta SCREEN+2
+    // SCREEN[3] = MSG4
+    // [3] *(SCREEN+3) = MSG4 -- _deref_pbuc1=vbuc2 
+    lda #MSG4
+    sta SCREEN+3
+    // main::@return
+    // }
+    // [4] return 
+    rts
+}
+  // File Data
+

src/test/ref/string-escapes-10.sym

@@ -0,0 +1,7 @@
+__constant char MSG1 = 'a'
+__constant char MSG2 = 'b'
+__constant char MSG3 = 'c'
+__constant char MSG4 = '刺'
+__constant char *SCREEN = (char *) 1024
+void main()
+

src/test/ref/string-escapes-6.asm

@@ -39,5 +39,5 @@ main: {
     jmp __b2
 }
 .segment Data
-  MESSAGE: .text @"a\$00bb\$00ccc\$00"
+  MESSAGE: .text "a@bb@ccc@"
   .byte 0

src/test/ref/string-escapes-8.asm

@@ -0,0 +1,18 @@
+// Support for '\0' literal character
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-8.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  .label SCREEN = $400
+.segment Code
+main: {
+    // SCREEN[0] = '\0'
+    lda #'@'
+    sta SCREEN
+    // }
+    rts
+}

src/test/ref/string-escapes-8.cfg

@@ -0,0 +1,8 @@
+
+void main()
+main: scope:[main]  from
+  [0] *SCREEN = '@'
+  to:main::@return
+main::@return: scope:[main]  from main
+  [1] return 
+  to:@return

src/test/ref/string-escapes-8.log

@@ -0,0 +1,141 @@
+
+CONTROL FLOW GRAPH SSA
+
+void main()
+main: scope:[main]  from __start
+  SCREEN[0] = '@'
+  to:main::@return
+main::@return: scope:[main]  from main
+  return 
+  to:@return
+
+void __start()
+__start: scope:[__start]  from
+  call main
+  to:__start::@1
+__start::@1: scope:[__start]  from __start
+  to:__start::@return
+__start::@return: scope:[__start]  from __start::@1
+  return 
+  to:@return
+
+SYMBOL TABLE SSA
+__constant char * const SCREEN = (char *)$400
+void __start()
+void main()
+
+Adding number conversion cast (unumber) 0 in SCREEN[0] = '@'
+Successful SSA optimization PassNAddNumberTypeConversions
+Simplifying constant pointer cast (char *) 1024
+Simplifying constant integer cast 0
+Successful SSA optimization PassNCastSimplification
+Finalized unsigned number type (char) 0
+Successful SSA optimization PassNFinalizeNumberTypeConversions
+Simplifying expression containing zero SCREEN in [0] SCREEN[0] = '@'
+Successful SSA optimization PassNSimplifyExpressionWithZero
+Removing unused procedure __start
+Removing unused procedure block __start
+Removing unused procedure block __start::@1
+Removing unused procedure block __start::@return
+Successful SSA optimization PassNEliminateEmptyStart
+CALL GRAPH
+
+Created 0 initial phi equivalence classes
+Coalesced down to 0 phi equivalence classes
+
+FINAL CONTROL FLOW GRAPH
+
+void main()
+main: scope:[main]  from
+  [0] *SCREEN = '@'
+  to:main::@return
+main::@return: scope:[main]  from main
+  [1] return 
+  to:@return
+
+
+VARIABLE REGISTER WEIGHTS
+void main()
+
+Initial phi equivalence classes
+Complete equivalence classes
+REGISTER UPLIFT POTENTIAL REGISTERS
+Statement [0] *SCREEN = '@' [ ] (  [ ] { }  ) always clobbers reg byte a 
+
+REGISTER UPLIFT SCOPES
+Uplift Scope [main] 
+Uplift Scope [] 
+
+Uplifting [main] best 15 combination 
+Uplifting [] best 15 combination 
+
+ASSEMBLER BEFORE OPTIMIZATION
+  // File Comments
+// Support for '\0' literal character
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-8.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // [0] *SCREEN = '@' -- _deref_pbuc1=vbuc2 
+    lda #'@'
+    sta SCREEN
+    jmp __breturn
+    // main::@return
+  __breturn:
+    // [1] return 
+    rts
+}
+  // File Data
+
+ASSEMBLER OPTIMIZATIONS
+Removing instruction jmp __breturn
+Succesful ASM optimization Pass5NextJumpElimination
+Removing instruction __breturn:
+Succesful ASM optimization Pass5UnusedLabelElimination
+
+FINAL SYMBOL TABLE
+__constant char * const SCREEN = (char *) 1024
+void main()
+
+
+
+FINAL ASSEMBLER
+Score: 12
+
+  // File Comments
+// Support for '\0' literal character
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-8.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // SCREEN[0] = '\0'
+    // [0] *SCREEN = '@' -- _deref_pbuc1=vbuc2 
+    lda #'@'
+    sta SCREEN
+    // main::@return
+    // }
+    // [1] return 
+    rts
+}
+  // File Data
+

src/test/ref/string-escapes-8.sym

@@ -0,0 +1,3 @@
+__constant char * const SCREEN = (char *) 1024
+void main()
+

src/test/ref/string-escapes-9.asm

@@ -0,0 +1,31 @@
+// Test octal escapes in strings
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-9.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  .label SCREEN = $400
+.segment Code
+main: {
+    ldx #0
+  __b1:
+    // while(MESSAGE[i])
+    lda MESSAGE,x
+    cmp #0
+    bne __b2
+    // }
+    rts
+  __b2:
+    // SCREEN[i] = MESSAGE[i++]
+    lda MESSAGE,x
+    sta SCREEN,x
+    // SCREEN[i] = MESSAGE[i++];
+    inx
+    jmp __b1
+}
+.segment Data
+  MESSAGE: .text @"aaa\$ff"
+  .byte 0

src/test/ref/string-escapes-9.cfg

@@ -0,0 +1,16 @@
+
+void main()
+main: scope:[main]  from
+  [0] phi()
+  to:main::@1
+main::@1: scope:[main]  from main main::@2
+  [1] main::i#2 = phi( main/0, main::@2/main::i#1 )
+  [2] if(0!=MESSAGE[main::i#2]) goto main::@2
+  to:main::@return
+main::@return: scope:[main]  from main::@1
+  [3] return 
+  to:@return
+main::@2: scope:[main]  from main::@1
+  [4] SCREEN[main::i#2] = MESSAGE[main::i#2]
+  [5] main::i#1 = ++ main::i#2
+  to:main::@1

src/test/ref/string-escapes-9.log

@@ -0,0 +1,245 @@
+Inlined call call __init
+
+CONTROL FLOW GRAPH SSA
+
+void main()
+main: scope:[main]  from __start::@1
+  main::i#0 = 0
+  to:main::@1
+main::@1: scope:[main]  from main main::@2
+  main::i#2 = phi( main/main::i#0, main::@2/main::i#1 )
+  main::$0 = 0 != MESSAGE[main::i#2]
+  if(main::$0) goto main::@2
+  to:main::@return
+main::@2: scope:[main]  from main::@1
+  main::i#3 = phi( main::@1/main::i#2 )
+  SCREEN[main::i#3] = MESSAGE[main::i#3]
+  main::i#1 = ++ main::i#3
+  to:main::@1
+main::@return: scope:[main]  from main::@1
+  return 
+  to:@return
+
+void __start()
+__start: scope:[__start]  from
+  to:__start::__init1
+__start::__init1: scope:[__start]  from __start
+  to:__start::@1
+__start::@1: scope:[__start]  from __start::__init1
+  call main
+  to:__start::@2
+__start::@2: scope:[__start]  from __start::@1
+  to:__start::@return
+__start::@return: scope:[__start]  from __start::@2
+  return 
+  to:@return
+
+SYMBOL TABLE SSA
+__constant char MESSAGE[] = "aaa刺"
+__constant char *SCREEN = (char *)$400
+void __start()
+void main()
+bool main::$0
+char main::i
+char main::i#0
+char main::i#1
+char main::i#2
+char main::i#3
+
+Adding number conversion cast (unumber) 0 in main::$0 = 0 != MESSAGE[main::i#2]
+Successful SSA optimization PassNAddNumberTypeConversions
+Simplifying constant pointer cast (char *) 1024
+Simplifying constant integer cast 0
+Successful SSA optimization PassNCastSimplification
+Finalized unsigned number type (char) 0
+Successful SSA optimization PassNFinalizeNumberTypeConversions
+Alias main::i#2 = main::i#3 
+Successful SSA optimization Pass2AliasElimination
+Simple Condition main::$0 [3] if(0!=MESSAGE[main::i#2]) goto main::@2
+Successful SSA optimization Pass2ConditionalJumpSimplification
+Constant main::i#0 = 0
+Successful SSA optimization Pass2ConstantIdentification
+Removing unused procedure __start
+Removing unused procedure block __start
+Removing unused procedure block __start::__init1
+Removing unused procedure block __start::@1
+Removing unused procedure block __start::@2
+Removing unused procedure block __start::@return
+Successful SSA optimization PassNEliminateEmptyStart
+Inlining constant with var siblings main::i#0
+Constant inlined main::i#0 = 0
+Successful SSA optimization Pass2ConstantInlining
+Adding NOP phi() at start of main
+CALL GRAPH
+
+Created 1 initial phi equivalence classes
+Coalesced [6] main::i#4 = main::i#1
+Coalesced down to 1 phi equivalence classes
+Adding NOP phi() at start of main
+
+FINAL CONTROL FLOW GRAPH
+
+void main()
+main: scope:[main]  from
+  [0] phi()
+  to:main::@1
+main::@1: scope:[main]  from main main::@2
+  [1] main::i#2 = phi( main/0, main::@2/main::i#1 )
+  [2] if(0!=MESSAGE[main::i#2]) goto main::@2
+  to:main::@return
+main::@return: scope:[main]  from main::@1
+  [3] return 
+  to:@return
+main::@2: scope:[main]  from main::@1
+  [4] SCREEN[main::i#2] = MESSAGE[main::i#2]
+  [5] main::i#1 = ++ main::i#2
+  to:main::@1
+
+
+VARIABLE REGISTER WEIGHTS
+void main()
+char main::i
+char main::i#1 // 22.0
+char main::i#2 // 18.333333333333332
+
+Initial phi equivalence classes
+[ main::i#2 main::i#1 ]
+Complete equivalence classes
+[ main::i#2 main::i#1 ]
+Allocated zp[1]:2 [ main::i#2 main::i#1 ]
+REGISTER UPLIFT POTENTIAL REGISTERS
+Statement [2] if(0!=MESSAGE[main::i#2]) goto main::@2 [ main::i#2 ] (  [ main::i#2 ] { }  ) always clobbers reg byte a 
+Removing always clobbered register reg byte a as potential for zp[1]:2 [ main::i#2 main::i#1 ]
+Statement [4] SCREEN[main::i#2] = MESSAGE[main::i#2] [ main::i#2 ] (  [ main::i#2 ] { }  ) always clobbers reg byte a 
+Statement [2] if(0!=MESSAGE[main::i#2]) goto main::@2 [ main::i#2 ] (  [ main::i#2 ] { }  ) always clobbers reg byte a 
+Statement [4] SCREEN[main::i#2] = MESSAGE[main::i#2] [ main::i#2 ] (  [ main::i#2 ] { }  ) always clobbers reg byte a 
+Potential registers zp[1]:2 [ main::i#2 main::i#1 ] : zp[1]:2 , reg byte x , reg byte y , 
+
+REGISTER UPLIFT SCOPES
+Uplift Scope [main] 40.33: zp[1]:2 [ main::i#2 main::i#1 ] 
+Uplift Scope [] 
+
+Uplifting [main] best 321 combination reg byte x [ main::i#2 main::i#1 ] 
+Uplifting [] best 321 combination 
+
+ASSEMBLER BEFORE OPTIMIZATION
+  // File Comments
+// Test octal escapes in strings
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-9.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // [1] phi from main to main::@1 [phi:main->main::@1]
+  __b1_from_main:
+    // [1] phi main::i#2 = 0 [phi:main->main::@1#0] -- vbuxx=vbuc1 
+    ldx #0
+    jmp __b1
+    // main::@1
+  __b1:
+    // [2] if(0!=MESSAGE[main::i#2]) goto main::@2 -- 0_neq_pbuc1_derefidx_vbuxx_then_la1 
+    lda MESSAGE,x
+    cmp #0
+    bne __b2
+    jmp __breturn
+    // main::@return
+  __breturn:
+    // [3] return 
+    rts
+    // main::@2
+  __b2:
+    // [4] SCREEN[main::i#2] = MESSAGE[main::i#2] -- pbuc1_derefidx_vbuxx=pbuc2_derefidx_vbuxx 
+    lda MESSAGE,x
+    sta SCREEN,x
+    // [5] main::i#1 = ++ main::i#2 -- vbuxx=_inc_vbuxx 
+    inx
+    // [1] phi from main::@2 to main::@1 [phi:main::@2->main::@1]
+  __b1_from___b2:
+    // [1] phi main::i#2 = main::i#1 [phi:main::@2->main::@1#0] -- register_copy 
+    jmp __b1
+}
+  // File Data
+.segment Data
+  MESSAGE: .text @"aaa\$ff"
+  .byte 0
+
+ASSEMBLER OPTIMIZATIONS
+Removing instruction jmp __b1
+Removing instruction jmp __breturn
+Succesful ASM optimization Pass5NextJumpElimination
+Removing instruction __b1_from_main:
+Removing instruction __breturn:
+Removing instruction __b1_from___b2:
+Succesful ASM optimization Pass5UnusedLabelElimination
+
+FINAL SYMBOL TABLE
+__constant char MESSAGE[] = "aaa刺"
+__constant char *SCREEN = (char *) 1024
+void main()
+char main::i
+char main::i#1 // reg byte x 22.0
+char main::i#2 // reg byte x 18.333333333333332
+
+reg byte x [ main::i#2 main::i#1 ]
+
+
+FINAL ASSEMBLER
+Score: 261
+
+  // File Comments
+// Test octal escapes in strings
+  // Upstart
+  // Commodore 64 PRG executable file
+.file [name="string-escapes-9.prg", type="prg", segments="Program"]
+.segmentdef Program [segments="Basic, Code, Data"]
+.segmentdef Basic [start=$0801]
+.segmentdef Code [start=$80d]
+.segmentdef Data [startAfter="Code"]
+.segment Basic
+:BasicUpstart(main)
+  // Global Constants & labels
+  .label SCREEN = $400
+.segment Code
+  // main
+main: {
+    // [1] phi from main to main::@1 [phi:main->main::@1]
+    // [1] phi main::i#2 = 0 [phi:main->main::@1#0] -- vbuxx=vbuc1 
+    ldx #0
+    // main::@1
+  __b1:
+    // while(MESSAGE[i])
+    // [2] if(0!=MESSAGE[main::i#2]) goto main::@2 -- 0_neq_pbuc1_derefidx_vbuxx_then_la1 
+    lda MESSAGE,x
+    cmp #0
+    bne __b2
+    // main::@return
+    // }
+    // [3] return 
+    rts
+    // main::@2
+  __b2:
+    // SCREEN[i] = MESSAGE[i++]
+    // [4] SCREEN[main::i#2] = MESSAGE[main::i#2] -- pbuc1_derefidx_vbuxx=pbuc2_derefidx_vbuxx 
+    lda MESSAGE,x
+    sta SCREEN,x
+    // SCREEN[i] = MESSAGE[i++];
+    // [5] main::i#1 = ++ main::i#2 -- vbuxx=_inc_vbuxx 
+    inx
+    // [1] phi from main::@2 to main::@1 [phi:main::@2->main::@1]
+    // [1] phi main::i#2 = main::i#1 [phi:main::@2->main::@1#0] -- register_copy 
+    jmp __b1
+}
+  // File Data
+.segment Data
+  MESSAGE: .text @"aaa\$ff"
+  .byte 0
+

src/test/ref/string-escapes-9.sym

@@ -0,0 +1,8 @@
+__constant char MESSAGE[] = "aaa刺"
+__constant char *SCREEN = (char *) 1024
+void main()
+char main::i
+char main::i#1 // reg byte x 22.0
+char main::i#2 // reg byte x 18.333333333333332
+
+reg byte x [ main::i#2 main::i#1 ]