Added encconv module.

CHANGELOG.md

@@ -8,6 +8,8 @@
 
 * Allowed passing non-decimal numbers to the `-D` option.
 
+* Added the `encconv` module.
+
 * Added `nullchar` constant as the null terminator for strings and `NULLCHAR` feature to define its value.
 
 * Added `vectrex` text encoding.

docs/README.md

@@ -22,6 +22,8 @@
 
 * [Literals](lang/literals.md)
 
+* [Predefined constants](lang/predefined_constants.md)
+
 * [List of text encodings and escape sequences](lang/text.md)
 
 * [Operators reference](lang/operators.md)
@@ -44,6 +46,8 @@
 
 * [Modules for reading input devices](stdlib/input.md)
 
+* [`encconv` module](stdlib/encconv.md)
+
 * [Other cross-platform modules (`keyboard`, `err`, `random`)](stdlib/other.md)
 
 * [Definitions available on only some platforms](stdlib/frequent.md)

docs/doc_index.md

@@ -46,6 +46,8 @@
 
 * [Modules for reading input devices](stdlib/input.md)
 
+* [`encconv` module](stdlib/encconv.md)
+
 * [Other cross-platform modules (`keyboard`, `err`, `random`)](stdlib/other.md)
 
 * [Definitions available on only some platforms](stdlib/frequent.md)

docs/lang/preprocessor.md

@@ -54,6 +54,8 @@ The following features are defined based on the chosen CPU and compilation optio
 `CPUFEATURE_8080`, `CPUFEATURE_8085`, `CPUFEATURE_GAMEBOY`, `CPUFEATURE_Z80`,
 `CPUFEATURE_6502_ILLEGALS`, `CPUFEATURE_8085_ILLEGALS`, `CPUFEATURE_Z80_ILLEGALS` – 1 if given instruction subset is enabled, 0 otherwise
 
+* `ENCODING_SAME` - 1 if the encodings `default` and `src` are the same, 0 otherwise.
+
 * `INIT_RW_MEMORY` – 1 if the option `ram_init_segment` is defined, 0 otherwise.
 See [the ROM vs RAM guide](../api/rom-vs-ram.md) for more information.
 

docs/stdlib/encconv.md

@@ -0,0 +1,74 @@
+[< back to index](../doc_index.md)
+
+## encconv
+
+The `encconv` module provides functions for character set conversions.
+
+All the functions are defined only for the characters that are valid in both input and output encoding.
+Unsupported characters may give arbitrary results.
+The unsupported characters are not guaranteed to roundtrip.
+
+Some encodings (e.g. PETSCII) allow for multiple encoding of the same character.
+For the input, all encodings are equivalent.
+For the output, the canonical encoding is preferred.
+
+Characters that are present in the input encoding,
+but are encoded as multiple bytes in the output encoding, are not supported.
+
+#### byte to_screencode(byte)
+
+Converts a byte from the default encoding to the screen encoding.
+
+If both encodings contain the character `¤`, then `to_screencode('¤') == '¤'scr`.
+
+Available only if one of the following is true:
+
+* the default encoding and the screen encoding are the same
+
+* the default encoding is `petscii`, the screen encoding is `petscr`, and the platform is 6502-based
+
+* the default encoding is `petsciijp`, the screen encoding is `petscrjp`, and the platform is 6502-based
+
+* the default encoding is `atascii`, the screen encoding is `atasciiscr`, and the platform is 6502-based
+
+#### byte from_screencode(byte)
+
+Converts a byte from the screen encoding to the default encoding.
+
+If both encodings contain the character `¤`, then `from_screencode('¤'scr) == '¤'`.
+
+Available only if `to_screencode` is available.
+
+#### byte petscii_to_petscr(byte)
+
+Converts a byte from PETSCII to a CBM screencode.  
+Works also for the variants used on the Japanese version of C64.  
+Control characters are converted to reverse characters, the same as in the standard quote mode.
+
+Available only on 6502-based platforms.
+
+#### byte petscr_to_petscii(byte)
+
+Converts a byte from a CBM screencode to PETSCII.  
+Works also for the variants used on the Japanese version of C64.  
+Reverse characters are interpreted as control characters or as non-reverse characters.
+
+Available only on 6502-based platforms.
+
+#### byte atascii_to_atasciiscr(byte)
+
+Converts a byte from ATASCII to a Atari screencode.  
+Control characters <$80 are converted to the graphical characters that share the ATASCII code.  
+Control characters ≥$80 are not supported.  
+
+Available only on 6502-based platforms.
+
+#### byte atasciiscr_to_atasciiscii(byte)
+
+Converts a byte from a Atari screencode to ATASCII.  
+Characters that share their ATASCII code with control characters are supported,
+but they require to be escaped with $1B to be printed.  
+Reverse characters are interpreted as non-reverse characters.
+
+Available only on 6502-based platforms.
+

include/encconv.mfk

@@ -0,0 +1,119 @@
+#if ENCODING_SAME
+
+#if ARCH_6502
+inline byte __byte_identity(byte a) { ? rts }
+#elseif ARCH_I80
+#pragma zilog_syntax
+inline byte __byte_identity(byte a) { ? ret }
+#else
+inline byte __byte_identity(byte a) = a
+#endif
+
+alias from_screencode = __byte_identity
+alias to_screencode = __byte_identity
+
+#else
+
+alias from_screencode = __from_screencode
+alias to_screencode = __to_screencode
+
+#endif
+
+
+// conversions for particular encoding pairs:
+
+#if ARCH_6502
+
+asm byte petscii_to_petscr(byte a) {
+    cmp #$20
+    bcc __petscii_to_petscr_ddRev
+    cmp #$60
+    bcc __petscii_to_petscr_dd1
+    cmp #$80
+    bcc __petscii_to_petscr_dd2
+    cmp #$a0
+    bcc __petscii_to_petscr_dd3
+    cmp #$c0
+    bcc __petscii_to_petscr_dd4
+    cmp #$ff
+    bcc __petscii_to_petscr_ddRev
+    lda #$5e
+    rts
+__petscii_to_petscr_dd2:
+    and #$5f
+    rts
+__petscii_to_petscr_dd3:
+    ora #$40
+    rts
+__petscii_to_petscr_dd4:
+    eor #$c0
+    rts
+__petscii_to_petscr_dd1:
+    and #$3f
+    rts
+__petscii_to_petscr_ddRev:
+    eor #$80
+    rts
+}
+
+asm byte petscr_to_petscii(byte a) {
+    cmp #$20
+    bcs __petscr_to_petscii_40
+    ora #$40
+    rts
+__petscr_to_petscii_40:
+    cmp #$40
+    bcs __petscr_to_petscii_60
+    rts
+__petscr_to_petscii_60:
+    cmp #$60
+    bcs __petscr_to_petscii_80
+    eor #$80
+    rts
+__petscr_to_petscii_80:
+    cmp #$80
+    bcs __petscr_to_petscii_a0
+    eor #$c0
+    rts
+__petscr_to_petscii_a0:
+    cmp #$a0
+    bcs __petscr_to_petscii_c0
+    eor #$80
+    rts
+__petscr_to_petscii_c0:
+    eor #$40
+    rts
+}
+
+asm byte atascii_to_atasciiscr(byte a) {
+    and #$7f
+    cmp #$20
+    bcs __atascii_to_atasciiscr_60
+    ora #$40
+    rts
+__atascii_to_atasciiscr_60:
+    cmp #$60
+    bcs __atascii_to_atasciiscr_end
+    sec
+    sbc #$20
+__atascii_to_atasciiscr_end:
+    rts
+}
+
+
+asm byte atasciiscr_to_atascii(byte a) {
+    and #$7f
+    cmp #$40
+    bcs __atascii_to_atasciiscr_60
+    clc
+    adc #$20
+    rts
+__atascii_to_atasciiscr_60:
+    cmp #$60
+    bcs __atascii_to_atasciiscr_end
+    and #$1f
+__atascii_to_atasciiscr_end:
+    rts
+}
+
+#endif

src/main/scala/millfork/Platform.scala

@@ -245,7 +245,9 @@ object Platform {
       debugOutputFormatName.toLowerCase(Locale.ROOT),
       log.fatal(s"Invalid label file format: `$debugOutputFormatName`"))
 
-    val builtInFeatures = builtInCpuFeatures(cpu)
+    val builtInFeatures = builtInCpuFeatures(cpu) ++ Map(
+      "ENCODING_SAME" -> toLong(codec.name == srcCodec.name)
+    )
 
     import scala.collection.JavaConverters._
     val ds = conf.getSection("define")
@@ -286,9 +288,10 @@ object Platform {
       outputStyle)
   }
 
+  @inline
+  private def toLong(b: Boolean): Long = if (b) 1L else 0L
+
   def builtInCpuFeatures(cpu: Cpu.Value): Map[String, Long] = {
-    @inline
-    def toLong(b: Boolean): Long = if (b) 1L else 0L
     Map[String, Long](
       "ARCH_6502" -> toLong(CpuFamily.forType(cpu) == CpuFamily.M6502),
       "CPU_6502" -> toLong(Set(Cpu.Mos, Cpu.StrictMos, Cpu.Ricoh, Cpu.StrictRicoh)(cpu)),

src/main/scala/millfork/parser/AbstractSourceLoadingQueue.scala

@@ -5,7 +5,7 @@ import java.nio.file.{Files, Paths}
 
 import fastparse.core.Parsed.{Failure, Success}
 import millfork.{CompilationFlag, CompilationOptions, Tarjan}
-import millfork.node.{ImportStatement, Position, Program}
+import millfork.node.{AliasDefinitionStatement, DeclarationStatement, ImportStatement, Position, Program}
 
 import scala.collection.mutable
 import scala.collection.convert.ImplicitConversionsToScala._
@@ -23,6 +23,20 @@ abstract class AbstractSourceLoadingQueue[T](val initialFilenames: List[String],
 
   def enqueueStandardModules(): Unit
 
+  def pseudoModules: List[DeclarationStatement] = {
+    val encodingConversionAliases = (options.platform.defaultCodec.name, options.platform.screenCodec.name) match {
+      case (TextCodec.Petscii.name, TextCodec.CbmScreencodes.name) |
+           (TextCodec.PetsciiJp.name, TextCodec.CbmScreencodesJp.name)=>
+        List(AliasDefinitionStatement("__from_screencode", "petscr_to_petscii", important = false),
+          AliasDefinitionStatement("__to_screencode", "petscii_to_petscr", important = false))
+      case (TextCodec.Atascii.name, TextCodec.AtasciiScreencodes.name)=>
+        List(AliasDefinitionStatement("__from_screencode", "atasciiscr_to_atascii", important = false),
+          AliasDefinitionStatement("__to_screencode", "atascii_to_atasciiscr", important = false))
+      case _ => Nil
+    }
+    encodingConversionAliases
+  }
+
   def run(): Program = {
     for {
       initialFilename <- initialFilenames
@@ -90,6 +104,7 @@ abstract class AbstractSourceLoadingQueue[T](val initialFilenames: List[String],
     }
     val parser = createParser(shortFileName, src, parentDir, featureConstants, pragmas.keySet)
     options.log.addSource(shortFileName, src.linesIterator.toIndexedSeq)
+    parsedModules.put("pseudomodule\u0000", Program(pseudoModules))
     parser.toAst match {
       case Success(prog, _) =>
         parsedModules.synchronized {