6502/6502bench
1
0
Fork 0
mirror of https://github.com/fadden/6502bench.git synced 2024-07-05 09:28:56 +00:00
Code Issues Projects Releases Wiki Activity

Merge Gen* and Asm* source files

Browse Source 
Each supported assembler has an IGenerator interface and an
IAssembler interface.  They're still two separate classes, but now
both are implemented in the same source file.  (They'll probably
stay separate classes, since the two have little interaction.)

I'm keeping the "Asm*" filename.  Seems the more natural fit.

Also, changed AssemblerInfo to try to get all assembler-specific
stuff into a single table.
This commit is contained in:
Andy McFadden 2018-10-17 13:11:26 -07:00
parent 42e5223efb
commit f81c534d25
6 changed files with 1262 additions and 1272 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

637
SourceGen/AsmGen/AsmCc65.cs
View File

@ -17,10 +17,645 @@ using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Diagnostics;
using System.IO;
using System.Text;
using Asm65;
using CommonUtil;
namespace SourceGen.AsmGen {
#region IGenerator
/// <summary>
/// Generate source code compatible with the cc65 assembler (https://github.com/cc65/cc65).
/// </summary>
public class GenCc65 : IGenerator {
private const string ASM_FILE_SUFFIX = "_cc65.S";
private const int MAX_OPERAND_LEN = 64;
// IGenerator
public DisasmProject Project { get; private set; }
// IGenerator
public Formatter SourceFormatter { get; private set; }
// IGenerator
public AppSettings Settings { get; private set; }
// IGenerator
public AssemblerQuirks Quirks { get; private set; }
// IGenerator
public LabelLocalizer Localizer { get { return mLocalizer; } }
/// <summary>
/// Working directory, i.e. where we write our output file(s).
/// </summary>
private string mWorkDirectory;
/// <summary>
/// If set, long labels get their own line.
/// </summary>
private bool mLongLabelNewLine;
/// <summary>
/// Base filename. Typically the project file name without the ".dis65" extension.
/// </summary>
private string mFileNameBase;
/// <summary>
/// StringBuilder to use when composing a line. Held here to reduce allocations.
/// </summary>
private StringBuilder mLineBuilder = new StringBuilder(100);
/// <summary>
/// Label localization helper.
/// </summary>
private LabelLocalizer mLocalizer;
/// <summary>
/// Stream to send the output to.
/// </summary>
private StreamWriter mOutStream;
/// <summary>
/// The first time we output a high-ASCII string, we generate a macro for it.
/// </summary>
private bool mHighAsciiMacroOutput;
/// <summary>
/// Holds detected version of configured assembler.
/// </summary>
private CommonUtil.Version mAsmVersion = CommonUtil.Version.NO_VERSION;
// We test against this in a few places.
private static CommonUtil.Version V2_17 = new CommonUtil.Version(2, 17);
// Pseudo-op string constants.
private static PseudoOp.PseudoOpNames sDataOpNames = new PseudoOp.PseudoOpNames() {
EquDirective = "=",
OrgDirective = ".org",
//RegWidthDirective // .a8, .a16, .i8, .i16
DefineData1 = ".byte",
DefineData2 = ".word",
DefineData3 = ".faraddr",
DefineData4 = ".dword",
DefineBigData2 = ".dbyt",
//DefineBigData3
//DefineBigData4
Fill = ".res",
//Dense // no equivalent, use .byte with comma-separated args
StrGeneric = ".byte",
//StrReverse
StrNullTerm = ".asciiz",
//StrLen8 // macro with .strlen?
//StrLen16
//StrDci
//StrDciReverse
};
// IGenerator
public void Configure(DisasmProject project, string workDirectory, string fileNameBase,
AssemblerVersion asmVersion, AppSettings settings) {
Debug.Assert(project != null);
Debug.Assert(!string.IsNullOrEmpty(workDirectory));
Debug.Assert(!string.IsNullOrEmpty(fileNameBase));
Project = project;
Quirks = new AssemblerQuirks();
if (asmVersion != null) {
// Use the actual version. If it's > 2.17 we'll try to take advantage of
// bug fixes.
mAsmVersion = asmVersion.Version;
} else {
// No assembler installed. Use 2.17.
mAsmVersion = V2_17;
}
if (mAsmVersion <= V2_17) {
// cc65 v2.17: https://github.com/cc65/cc65/issues/717
Quirks.BlockMoveArgsReversed = true;
// cc65 v2.17: https://github.com/cc65/cc65/issues/754
Quirks.NoPcRelBankWrap = true;
}
mWorkDirectory = workDirectory;
mFileNameBase = fileNameBase;
Settings = settings;
mLongLabelNewLine = Settings.GetBool(AppSettings.SRCGEN_LONG_LABEL_NEW_LINE, false);
}
// IGenerator
public List<string> GenerateSource(BackgroundWorker worker) {
List<string> pathNames = new List<string>(1);
string fileName = mFileNameBase + ASM_FILE_SUFFIX;
string pathName = Path.Combine(mWorkDirectory, fileName);
pathNames.Add(pathName);
Formatter.FormatConfig config = new Formatter.FormatConfig();
GenCommon.ConfigureFormatterFromSettings(Settings, ref config);
config.mForceAbsOpcodeSuffix = string.Empty;
config.mForceLongOpcodeSuffix = string.Empty;
config.mForceAbsOperandPrefix = "a:"; // absolute
config.mForceLongOperandPrefix = "f:"; // far
config.mEndOfLineCommentDelimiter = ";";
config.mFullLineCommentDelimiterBase = ";";
config.mBoxLineCommentDelimiter = ";";
config.mAllowHighAsciiCharConst = false;
config.mExpressionMode = Formatter.FormatConfig.ExpressionMode.Simple;
SourceFormatter = new Formatter(config);
string msg = string.Format(Properties.Resources.PROGRESS_GENERATING_FMT, pathName);
worker.ReportProgress(0, msg);
mLocalizer = new LabelLocalizer(Project);
if (!Settings.GetBool(AppSettings.SRCGEN_DISABLE_LABEL_LOCALIZATION, false)) {
mLocalizer.LocalPrefix = "@";
mLocalizer.Analyze();
}
// Use UTF-8 encoding, without a byte-order mark.
using (StreamWriter sw = new StreamWriter(pathName, false, new UTF8Encoding(false))) {
mOutStream = sw;
if (Settings.GetBool(AppSettings.SRCGEN_ADD_IDENT_COMMENT, false)) {
//if (mAsmVersion.IsValid && mAsmVersion <= V2_17) {
// OutputLine(SourceFormatter.FullLineCommentDelimiter +
// string.Format(Properties.Resources.GENERATED_FOR_VERSION,
// "cc65", mAsmVersion.ToString()));
//} else {
// OutputLine(SourceFormatter.FullLineCommentDelimiter +
// string.Format(Properties.Resources.GENERATED_FOR_LATEST, "cc65"));
//}
// Currently generating code for v2.17.
OutputLine(SourceFormatter.FullLineCommentDelimiter +
string.Format(Properties.Resources.GENERATED_FOR_VERSION,
"cc65", V2_17));
}
GenCommon.Generate(this, sw, worker);
}
mOutStream = null;
return pathNames;
}
// IGenerator
public void OutputAsmConfig() {
CpuDef cpuDef = Project.CpuDef;
string cpuStr;
if (cpuDef.Type == CpuDef.CpuType.Cpu65816) {
cpuStr = "65816";
} else if (cpuDef.Type == CpuDef.CpuType.Cpu65C02) {
cpuStr = "65C02";
} else if (cpuDef.Type == CpuDef.CpuType.Cpu6502 && cpuDef.HasUndocumented) {
cpuStr = "6502X";
} else {
cpuStr = "6502";
}
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(".setcpu"),
'\"' + cpuStr + '\"', string.Empty);
}
/// <summary>
/// Map the mnemonics we chose for undocumented opcodes to the cc65 mnemonics.
/// After switching to the Unintended Opcodes mnemonics there's almost no difference.
///
/// We don't include the double- and triple-byte NOPs here, as cc65 doesn't
/// appear to have a definition for them (as of 2.17).
/// </summary>
private static Dictionary<string, string> sUndocMap = new Dictionary<string, string>() {
{ OpName.ALR, "alr" }, // imm 0x4b
{ OpName.ANC, "anc" }, // imm 0x0b (and others)
{ OpName.ANE, "ane" }, // imm 0x8b
{ OpName.ARR, "arr" }, // imm 0x6b
{ OpName.DCP, "dcp" }, // abs 0xcf
{ OpName.ISC, "isc" }, // abs 0xef
{ OpName.JAM, "jam" }, // abs 0x02 (and others)
{ OpName.LAS, "las" }, // abs,y 0xbb
{ OpName.LAX, "lax" }, // imm 0xab; abs 0xaf
{ OpName.RLA, "rla" }, // abs 0x2f
{ OpName.RRA, "rra" }, // abs 0x6f
{ OpName.SAX, "sax" }, // abs 0x8f
{ OpName.SBX, "axs" }, //* imm 0xcb
{ OpName.SHA, "sha" }, // abs,y 0x9f
{ OpName.SHX, "shx" }, // abs,y 0x9e
{ OpName.SHY, "shy" }, // abs,x 0x9c
{ OpName.SLO, "slo" }, // abs 0x0f
{ OpName.SRE, "sre" }, // abs 0x4f
{ OpName.TAS, "tas" }, // abs,y 0x9b
};
// IGenerator
public string ReplaceMnemonic(OpDef op) {
if ((op == OpDef.OpWDM_WDM || op == OpDef.OpBRK_StackInt) && mAsmVersion <= V2_17) {
// cc65 v2.17 doesn't support WDM, and assembles BRK <arg> to opcode $05.
// https://github.com/cc65/cc65/issues/715
// https://github.com/cc65/cc65/issues/716
return null;
} else if (op.IsUndocumented) {
if (sUndocMap.TryGetValue(op.Mnemonic, out string newValue)) {
if ((op.Mnemonic == OpName.ANC && op.Opcode != 0x0b) ||
(op.Mnemonic == OpName.JAM && op.Opcode != 0x02)) {
// There are multiple opcodes for the same thing. cc65 outputs
// one specific thing, so we need to match that, and just do a hex
// dump for the others.
return null;
}
return newValue;
}
return null;
} else {
return string.Empty;
}
}
// IGenerator
public void GenerateShortSequence(int offset, int length, out string opcode,
out string operand) {
Debug.Assert(length >= 1 && length <= 4);
// Use a comma-separated list of individual hex bytes.
opcode = sDataOpNames.DefineData1;
StringBuilder sb = new StringBuilder(length * 4);
for (int i = 0; i < length; i++) {
if (i != 0) {
sb.Append(',');
}
sb.Append(SourceFormatter.FormatHexValue(Project.FileData[offset + i], 2));
}
operand = sb.ToString();
}
// IGenerator
public void OutputDataOp(int offset) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
string labelStr = string.Empty;
if (attr.Symbol != null) {
labelStr = mLocalizer.ConvLabel(attr.Symbol.Label);
}
string commentStr = SourceFormatter.FormatEolComment(Project.Comments[offset]);
string opcodeStr, operandStr;
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
int length = dfd.Length;
Debug.Assert(length > 0);
bool multiLine = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.Default:
if (length != 1) {
Debug.Assert(false);
length = 1;
}
opcodeStr = sDataOpNames.DefineData1;
int operand = RawData.GetWord(data, offset, length, false);
operandStr = formatter.FormatHexValue(operand, length * 2);
break;
case FormatDescriptor.Type.NumericLE:
opcodeStr = sDataOpNames.GetDefineData(length);
operand = RawData.GetWord(data, offset, length, false);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
break;
case FormatDescriptor.Type.NumericBE:
opcodeStr = sDataOpNames.GetDefineBigData(length);
if (opcodeStr == null) {
// Nothing defined, output as comma-separated single-byte values.
GenerateShortSequence(offset, length, out opcodeStr, out operandStr);
} else {
operand = RawData.GetWord(data, offset, length, true);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
}
break;
case FormatDescriptor.Type.Fill:
opcodeStr = sDataOpNames.Fill;
operandStr = length + "," + formatter.FormatHexValue(data[offset], 2);
break;
case FormatDescriptor.Type.Dense:
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
break;
case FormatDescriptor.Type.String:
multiLine = true;
opcodeStr = operandStr = null;
OutputString(offset, labelStr, commentStr);
break;
default:
opcodeStr = "???";
operandStr = "***";
break;
}
if (!multiLine) {
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
}
}
private void OutputDenseHex(int offset, int length, string labelStr, string commentStr) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
StringBuilder sb = new StringBuilder(MAX_OPERAND_LEN);
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.DefineData1);
int maxPerLine = MAX_OPERAND_LEN / 4;
int numChunks = (length + maxPerLine - 1) / maxPerLine;
for (int chunk = 0; chunk < numChunks; chunk++) {
int chunkStart = chunk * maxPerLine;
int chunkEnd = Math.Min((chunk + 1) * maxPerLine, length);
for (int i = chunkStart; i < chunkEnd; i++) {
if (i != chunkStart) {
sb.Append(',');
}
sb.Append(formatter.FormatHexValue(data[offset + i], 2));
}
OutputLine(labelStr, opcodeStr, sb.ToString(), commentStr);
labelStr = commentStr = string.Empty;
sb.Clear();
}
}
/// <summary>
/// Outputs formatted data in an unformatted way, because the code generator couldn't
/// figure out how to do something better.
/// </summary>
private void OutputNoJoy(int offset, int length, string labelStr, string commentStr) {
byte[] data = Project.FileData;
Debug.Assert(length > 0);
Debug.Assert(offset >= 0 && offset < data.Length);
bool singleValue = true;
byte val = data[offset];
for (int i = 1; i < length; i++) {
if (data[offset + i] != val) {
singleValue = false;
break;
}
}
if (singleValue) {
string opcodeStr = SourceFormatter.FormatPseudoOp(sDataOpNames.Fill);
string operandStr = length + "," + SourceFormatter.FormatHexValue(val, 2);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
} else {
OutputDenseHex(offset, length, labelStr, commentStr);
}
}
// IGenerator
public void OutputEquDirective(string name, string valueStr, string comment) {
OutputLine(name, SourceFormatter.FormatPseudoOp(sDataOpNames.EquDirective),
valueStr, SourceFormatter.FormatEolComment(comment));
}
// IGenerator
public void OutputOrgDirective(int address) {
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(sDataOpNames.OrgDirective),
SourceFormatter.FormatHexValue(address, 4), string.Empty);
}
// IGenerator
public void OutputRegWidthDirective(int offset, int prevM, int prevX, int newM, int newX) {
if (prevM != newM) {
string mop = (newM == 0) ? ".a16" : ".a8";
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(mop),
string.Empty, string.Empty);
}
if (prevX != newX) {
string xop = (newX == 0) ? ".i16" : ".i8";
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(xop),
string.Empty, string.Empty);
}
}
// IGenerator
public void OutputLine(string fullLine) {
mOutStream.WriteLine(fullLine);
}
// IGenerator
public void OutputLine(string label, string opcode, string operand, string comment) {
// If a label is provided, and it doesn't start with a '.' (indicating that it's
// a directive), and this isn't an EQU directive, add a ':'. Might be easier to
// just ".feature labels_without_colons", but I'm trying to do things the way
// that cc65 users will expect.
if (!string.IsNullOrEmpty(label) && label[0] != '.' &&
!String.Equals(opcode, sDataOpNames.EquDirective,
StringComparison.InvariantCultureIgnoreCase)) {
label += ':';
if (mLongLabelNewLine && label.Length >= 9) {
mOutStream.WriteLine(label);
label = string.Empty;
}
}
mLineBuilder.Clear();
TextUtil.AppendPaddedString(mLineBuilder, label, 9);
TextUtil.AppendPaddedString(mLineBuilder, opcode, 9 + 8);
TextUtil.AppendPaddedString(mLineBuilder, operand, 9 + 8 + 11);
if (string.IsNullOrEmpty(comment)) {
// Trim trailing spaces off of opcode or operand. If they want trailing
// spaces at the end of a comment, that's fine.
CommonUtil.TextUtil.TrimEnd(mLineBuilder);
} else {
mLineBuilder.Append(comment);
}
mOutStream.WriteLine(mLineBuilder.ToString());
}
private void OutputString(int offset, string labelStr, string commentStr) {
// Normal ASCII strings are straightforward: they're just part of a .byte
// directive, and can mix with anything else in the .byte.
//
// For CString we can use .asciiz, but only if the string fits on one line
// and doesn't include delimiters. For L8String and L16String we can
// define simple macros, but their use has a similar restriction. High-ASCII
// strings also require a macro.
//
// We might be able to define a macro for DCI and Reverse as well.
//
// The limitation on strings with delimiters arises because (1) I don't see a
// way to escape them within a string, and (2) the simple macro workarounds
// only take a single argument, not a comma-separated list of stuff.
//
// Some ideas here:
// https://groups.google.com/forum/#!topic/comp.sys.apple2.programmer/5Wkw8mUPcU0
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
Debug.Assert(dfd.FormatType == FormatDescriptor.Type.String);
Debug.Assert(dfd.Length > 0);
bool highAscii = false;
int leadingBytes = 0;
int trailingBytes = 0;
bool showLeading = false;
bool showTrailing = false;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Dci:
highAscii = (data[offset] & 0x80) != 0;
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.Reverse:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.DciReverse:
highAscii = (data[offset + dfd.Length - 1] & 0x80) != 0;
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.CString:
highAscii = (data[offset] & 0x80) != 0;
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.L8String:
if (dfd.Length > 1) {
highAscii = (data[offset + 1] & 0x80) != 0;
}
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.L16String:
if (dfd.Length > 2) {
highAscii = (data[offset + 2] & 0x80) != 0;
}
leadingBytes = 2;
showLeading = true;
break;
default:
Debug.Assert(false);
return;
}
char delim = '"';
StringGather gath = null;
// Run the string through so we can see if it'll fit on one line. As a minor
// optimization, we skip this step for "generic" strings, which are probably
// the most common thing.
if (dfd.FormatSubType != FormatDescriptor.SubType.None || highAscii) {
gath = new StringGather(this, labelStr, "???", commentStr, delim,
delim, StringGather.ByteStyle.CommaSep, MAX_OPERAND_LEN, true);
FeedGath(gath, data, offset, dfd.Length, leadingBytes, showLeading,
trailingBytes, showTrailing);
Debug.Assert(gath.NumLinesOutput > 0);
}
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.StrGeneric);
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
// Special case for simple short high-ASCII strings. These have no
// leading or trailing bytes. We can improve this a bit by handling
// arbitrarily long strings by simply breaking them across lines.
Debug.Assert(leadingBytes == 0);
Debug.Assert(trailingBytes == 0);
if (highAscii && gath.NumLinesOutput == 1 && !gath.HasDelimiter) {
if (!mHighAsciiMacroOutput) {
mHighAsciiMacroOutput = true;
// Output a macro for high-ASCII strings.
OutputLine(".macro", "HiAscii", "Arg", string.Empty);
OutputLine(string.Empty, ".repeat", ".strlen(Arg), I", string.Empty);
OutputLine(string.Empty, ".byte", ".strat(Arg, I) | $80", string.Empty);
OutputLine(string.Empty, ".endrep", string.Empty, string.Empty);
OutputLine(".endmacro", string.Empty, string.Empty, string.Empty);
}
opcodeStr = formatter.FormatPseudoOp("HiAscii");
highAscii = false;
}
break;
case FormatDescriptor.SubType.Dci:
case FormatDescriptor.SubType.Reverse:
case FormatDescriptor.SubType.DciReverse:
// Full configured above.
break;
case FormatDescriptor.SubType.CString:
if (gath.NumLinesOutput == 1 && !gath.HasDelimiter) {
opcodeStr = sDataOpNames.StrNullTerm;
showTrailing = false;
}
break;
case FormatDescriptor.SubType.L8String:
case FormatDescriptor.SubType.L16String:
// Implement macros?
break;
default:
Debug.Assert(false);
return;
}
if (highAscii) {
OutputNoJoy(offset, dfd.Length, labelStr, commentStr);
return;
}
// Create a new StringGather, with the final opcode choice.
gath = new StringGather(this, labelStr, opcodeStr, commentStr, delim,
delim, StringGather.ByteStyle.CommaSep, MAX_OPERAND_LEN, false);
FeedGath(gath, data, offset, dfd.Length, leadingBytes, showLeading,
trailingBytes, showTrailing);
}
/// <summary>
/// Feeds the bytes into the StringGather.
/// </summary>
private void FeedGath(StringGather gath, byte[] data, int offset, int length,
int leadingBytes, bool showLeading, int trailingBytes, bool showTrailing) {
int startOffset = offset;
int strEndOffset = offset + length - trailingBytes;
if (showLeading) {
while (leadingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
} else {
offset += leadingBytes;
}
for (; offset < strEndOffset; offset++) {
gath.WriteChar((char)(data[offset] & 0x7f));
}
while (showTrailing && trailingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
gath.Finish();
}
}
#endregion IGenerator
#region IAssembler
/// <summary>
/// Cross-assembler execution interface.
/// </summary>
public class AsmCc65 : IAssembler {
private List<string> PathNames { get; set; }
@ -118,4 +753,6 @@ namespace SourceGen.AsmGen {
cmd.Stderr, outputFile);
}
}
#endregion IAssembler
}

584
SourceGen/AsmGen/AsmMerlin32.cs
View File

@ -17,10 +17,592 @@ using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Diagnostics;
using System.IO;
using System.Text;
using Asm65;
using CommonUtil;
namespace SourceGen.AsmGen {
#region IGenerator
/// <summary>
/// Generate source code compatible with Brutal Deluxe's Merlin 32 assembler
/// (https://www.brutaldeluxe.fr/products/crossdevtools/merlin/).
/// </summary>
public class GenMerlin32 : IGenerator {
private const string ASM_FILE_SUFFIX = "_Merlin32.S";
private const int MAX_OPERAND_LEN = 64;
// IGenerator
public DisasmProject Project { get; private set; }
// IGenerator
public Formatter SourceFormatter { get; private set; }
// IGenerator
public AppSettings Settings { get; private set; }
// IGenerator
public AssemblerQuirks Quirks { get; private set; }
// IGenerator
public LabelLocalizer Localizer { get { return mLocalizer; } }
/// <summary>
/// Working directory, i.e. where we write our output file(s).
/// </summary>
private string mWorkDirectory;
/// <summary>
/// If set, long labels get their own line.
/// </summary>
private bool mLongLabelNewLine;
/// <summary>
/// Base filename. Typically the project file name without the ".dis65" extension.
/// </summary>
private string mFileNameBase;
/// <summary>
/// StringBuilder to use when composing a line. Held here to reduce allocations.
/// </summary>
private StringBuilder mLineBuilder = new StringBuilder(100);
/// <summary>
/// Label localization helper.
/// </summary>
private LabelLocalizer mLocalizer;
/// <summary>
/// Stream to send the output to.
/// </summary>
private StreamWriter mOutStream;
/// <summary>
/// Holds detected version of configured assembler.
/// </summary>
private CommonUtil.Version mAsmVersion = CommonUtil.Version.NO_VERSION;
// Semi-convenient way to hold all the interesting string constants in one place.
// Note the actual usage of the pseudo-op may not match what the main app does,
// e.g. RegWidthDirective behaves differently from "mx". I'm just trying to avoid
// having string constants scattered all over.
private static PseudoOp.PseudoOpNames sDataOpNames = new PseudoOp.PseudoOpNames() {
EquDirective = "equ",
OrgDirective = "org",
RegWidthDirective = "mx",
DefineData1 = "dfb",
DefineData2 = "dw",
DefineData3 = "adr",
DefineData4 = "adrl",
DefineBigData2 = "ddb",
//DefineBigData3
//DefineBigData4
Fill = "ds",
Dense = "hex",
StrGeneric = "asc",
StrGenericHi = "asc",
StrReverse = "rev",
StrReverseHi = "rev",
//StrNullTerm
StrLen8 = "str",
StrLen8Hi = "str",
StrLen16 = "strl",
StrLen16Hi = "strl",
StrDci = "dci",
StrDciHi = "dci",
//StrDciReverse
};
// IGenerator
public void Configure(DisasmProject project, string workDirectory, string fileNameBase,
AssemblerVersion asmVersion, AppSettings settings) {
Debug.Assert(project != null);
Debug.Assert(!string.IsNullOrEmpty(workDirectory));
Debug.Assert(!string.IsNullOrEmpty(fileNameBase));
Project = project;
Quirks = new AssemblerQuirks();
Quirks.TracksSepRepNotEmu = true;
Quirks.NoPcRelBankWrap = true;
mWorkDirectory = workDirectory;
mFileNameBase = fileNameBase;
Settings = settings;
mLongLabelNewLine = Settings.GetBool(AppSettings.SRCGEN_LONG_LABEL_NEW_LINE, false);
}
// IGenerator; executes on background thread
public List<string> GenerateSource(BackgroundWorker worker) {
List<string> pathNames = new List<string>(1);
string fileName = mFileNameBase + ASM_FILE_SUFFIX;
string pathName = Path.Combine(mWorkDirectory, fileName);
pathNames.Add(pathName);
Formatter.FormatConfig config = new Formatter.FormatConfig();
GenCommon.ConfigureFormatterFromSettings(Settings, ref config);
config.mForceAbsOpcodeSuffix = ":";
config.mForceLongOpcodeSuffix = "l";
config.mForceAbsOperandPrefix = string.Empty;
config.mForceLongOperandPrefix = string.Empty;
config.mEndOfLineCommentDelimiter = ";";
config.mFullLineCommentDelimiterBase = ";";
config.mBoxLineCommentDelimiter = string.Empty;
config.mAllowHighAsciiCharConst = true;
config.mExpressionMode = Formatter.FormatConfig.ExpressionMode.Merlin;
SourceFormatter = new Formatter(config);
string msg = string.Format(Properties.Resources.PROGRESS_GENERATING_FMT, pathName);
worker.ReportProgress(0, msg);
mLocalizer = new LabelLocalizer(Project);
if (!Settings.GetBool(AppSettings.SRCGEN_DISABLE_LABEL_LOCALIZATION, false)) {
mLocalizer.LocalPrefix = ":";
mLocalizer.Analyze();
}
// Use UTF-8 encoding, without a byte-order mark.
using (StreamWriter sw = new StreamWriter(pathName, false, new UTF8Encoding(false))) {
mOutStream = sw;
if (Settings.GetBool(AppSettings.SRCGEN_ADD_IDENT_COMMENT, false)) {
// No version-specific stuff yet. We're generating code for v1.0.
OutputLine(SourceFormatter.FullLineCommentDelimiter +
string.Format(Properties.Resources.GENERATED_FOR_VERSION,
"Merlin 32", new CommonUtil.Version(1, 0)));
}
GenCommon.Generate(this, sw, worker);
}
mOutStream = null;
return pathNames;
}
// IGenerator
public void OutputDataOp(int offset) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
string labelStr = string.Empty;
if (attr.Symbol != null) {
labelStr = mLocalizer.ConvLabel(attr.Symbol.Label);
}
string commentStr = SourceFormatter.FormatEolComment(Project.Comments[offset]);
string opcodeStr, operandStr;
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
int length = dfd.Length;
Debug.Assert(length > 0);
bool multiLine = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.Default:
if (length != 1) {
Debug.Assert(false);
length = 1;
}
opcodeStr = sDataOpNames.DefineData1;
int operand = RawData.GetWord(data, offset, length, false);
operandStr = formatter.FormatHexValue(operand, length * 2);
break;
case FormatDescriptor.Type.NumericLE:
opcodeStr = sDataOpNames.GetDefineData(length);
operand = RawData.GetWord(data, offset, length, false);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
break;
case FormatDescriptor.Type.NumericBE:
opcodeStr = sDataOpNames.GetDefineBigData(length);
if (opcodeStr == null) {
// Nothing defined, output as comma-separated single-byte values.
GenerateShortSequence(offset, length, out opcodeStr, out operandStr);
} else {
operand = RawData.GetWord(data, offset, length, true);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
}
break;
case FormatDescriptor.Type.Fill:
opcodeStr = sDataOpNames.Fill;
operandStr = length + "," + formatter.FormatHexValue(data[offset], 2);
break;
case FormatDescriptor.Type.Dense:
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
break;
case FormatDescriptor.Type.String:
multiLine = true;
opcodeStr = operandStr = null;
OutputString(offset, labelStr, commentStr);
break;
default:
opcodeStr = "???";
operandStr = "***";
break;
}
if (!multiLine) {
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
}
}
private void OutputDenseHex(int offset, int length, string labelStr, string commentStr) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
int maxPerLine = MAX_OPERAND_LEN / 2;
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.Dense);
for (int i = 0; i < length; i += maxPerLine) {
int subLen = length - i;
if (subLen > maxPerLine) {
subLen = maxPerLine;
}
string operandStr = formatter.FormatDenseHex(data, offset + i, subLen);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
labelStr = commentStr = string.Empty;
}
}
// IGenerator
public string ReplaceMnemonic(OpDef op) {
if (op.IsUndocumented) {
return null;
} else {
return string.Empty;
}
}
// IGenerator
public void GenerateShortSequence(int offset, int length, out string opcode,
out string operand) {
Debug.Assert(length >= 1 && length <= 4);
// Use a comma-separated list of individual hex bytes.
opcode = sDataOpNames.DefineData1;
StringBuilder sb = new StringBuilder(length * 4);
for (int i = 0; i < length; i++) {
if (i != 0) {
sb.Append(',');
}
sb.Append(SourceFormatter.FormatHexValue(Project.FileData[offset + i], 2));
}
operand = sb.ToString();
}
// IGenerator
public void OutputAsmConfig() {
// nothing to do
}
// IGenerator
public void OutputEquDirective(string name, string valueStr, string comment) {
OutputLine(name, SourceFormatter.FormatPseudoOp(sDataOpNames.EquDirective),
valueStr, SourceFormatter.FormatEolComment(comment));
}
// IGenerator
public void OutputOrgDirective(int address) {
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(sDataOpNames.OrgDirective),
SourceFormatter.FormatHexValue(address, 4), string.Empty);
}
// IGenerator
public void OutputRegWidthDirective(int offset, int prevM, int prevX, int newM, int newX) {
// prevM/prevX may be ambiguous for offset 0, but otherwise everything
// should be either 0 or 1.
Debug.Assert(newM == 0 || newM == 1);
Debug.Assert(newX == 0 || newX == 1);
if (offset == 0 && newM == 1 && newX == 1) {
// Assembler defaults to short regs, so we can skip this.
return;
}
OutputLine(string.Empty,
SourceFormatter.FormatPseudoOp(sDataOpNames.RegWidthDirective),
"%" + newM + newX, string.Empty);
}
// IGenerator
public void OutputLine(string fullLine) {
mOutStream.WriteLine(fullLine);
}
// IGenerator
public void OutputLine(string label, string opcode, string operand, string comment) {
// Split long label, but not on EQU directives (confuses the assembler).
if (mLongLabelNewLine && label.Length >= 9 &&
!String.Equals(opcode, sDataOpNames.EquDirective,
StringComparison.InvariantCultureIgnoreCase)) {
mOutStream.WriteLine(label);
label = string.Empty;
}
mLineBuilder.Clear();
TextUtil.AppendPaddedString(mLineBuilder, label, 9);
TextUtil.AppendPaddedString(mLineBuilder, opcode, 9 + 6);
TextUtil.AppendPaddedString(mLineBuilder, operand, 9 + 6 + 11);
if (string.IsNullOrEmpty(comment)) {
// Trim trailing spaces off of opcode or operand. If they want trailing
// spaces at the end of a comment, that's fine.
CommonUtil.TextUtil.TrimEnd(mLineBuilder);
} else {
mLineBuilder.Append(comment);
}
mOutStream.WriteLine(mLineBuilder.ToString());
}
private enum RevMode { Forward, Reverse, BlockReverse };
private void OutputString(int offset, string labelStr, string commentStr) {
// This gets complicated.
//
// For Dci, L8String, and L16String, the entire string needs to fit in the
// operand of one line. If it can't, we need to separate the length byte/word
// or inverted character out, and just dump the rest as ASCII. Computing the
// line length requires factoring delimiter character escapes. (NOTE: contrary
// to the documentation, STR and STRL do include trailing hex characters in the
// length calculation, so it's possible to escape delimiters.)
//
// For Reverse, we can span lines, but only if we emit the lines in
// backward order. Also, Merlin doesn't allow hex to be embedded in a REV
// operation, so we can't use REV if the string contains a delimiter.
//
// DciReverse is deprecated, but we can handle it as a Reverse string with a
// trailing byte on a following line.
//
// For aesthetic purposes, zero-length CString, L8String, and L16String
// should be output as DFB/DW zeroes rather than an empty string -- makes
// it easier to read.
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
Debug.Assert(dfd.FormatType == FormatDescriptor.Type.String);
Debug.Assert(dfd.Length > 0);
bool highAscii = false;
int showZeroes = 0;
int leadingBytes = 0;
int trailingBytes = 0;
bool showLeading = false;
bool showTrailing = false;
RevMode revMode = RevMode.Forward;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Dci:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Reverse:
highAscii = (data[offset] & 0x80) != 0;
revMode = RevMode.Reverse;
break;
case FormatDescriptor.SubType.DciReverse:
highAscii = (data[offset + dfd.Length - 1] & 0x80) != 0;
revMode = RevMode.Reverse;
break;
case FormatDescriptor.SubType.CString:
highAscii = (data[offset] & 0x80) != 0;
if (dfd.Length == 1) {
showZeroes = 1; // empty null-terminated string
}
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.L8String:
if (dfd.Length > 1) {
highAscii = (data[offset + 1] & 0x80) != 0;
} else {
//showZeroes = 1;
}
leadingBytes = 1;
break;
case FormatDescriptor.SubType.L16String:
if (dfd.Length > 2) {
highAscii = (data[offset + 2] & 0x80) != 0;
} else {
//showZeroes = 2;
}
leadingBytes = 2;
break;
default:
Debug.Assert(false);
return;
}
if (showZeroes != 0) {
// Empty string. Just output the length byte(s) or null terminator.
GenerateShortSequence(offset, showZeroes, out string opcode, out string operand);
OutputLine(labelStr, opcode, operand, commentStr);
return;
}
// Merlin 32 uses single-quote for low ASCII, double-quote for high ASCII. When
// quoting the delimiter we use a hexadecimal value. We need to bear in mind that
// we're forcing the characters to low ASCII, but the actual character being
// escaped might be in high ASCII. Hence delim vs. delimReplace.
char delim = highAscii ? '"' : '\'';
char delimReplace = highAscii ? ((char)(delim | 0x80)) : delim;
StringGather gath = null;
// Run the string through so we can see if it'll fit on one line. As a minor
// optimization, we skip this step for "generic" strings, which are probably
// the most common thing.
if (dfd.FormatSubType != FormatDescriptor.SubType.None) {
gath = new StringGather(this, labelStr, "???", commentStr, delim,
delimReplace, StringGather.ByteStyle.DenseHex, MAX_OPERAND_LEN, true);
FeedGath(gath, data, offset, dfd.Length, revMode, leadingBytes, showLeading,
trailingBytes, showTrailing);
Debug.Assert(gath.NumLinesOutput > 0);
}
string opcodeStr;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
break;
case FormatDescriptor.SubType.Dci:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrDciHi : sDataOpNames.StrDci;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
trailingBytes = 1;
showTrailing = true;
}
break;
case FormatDescriptor.SubType.Reverse:
if (gath.HasDelimiter) {
// can't include escaped delimiters in REV
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
revMode = RevMode.Forward;
} else if (gath.NumLinesOutput > 1) {
opcodeStr = highAscii ? sDataOpNames.StrReverseHi : sDataOpNames.StrReverse;
revMode = RevMode.BlockReverse;
} else {
opcodeStr = highAscii ? sDataOpNames.StrReverseHi : sDataOpNames.StrReverse;
Debug.Assert(revMode == RevMode.Reverse);
}
break;
case FormatDescriptor.SubType.DciReverse:
// Mostly punt -- output as ASCII with special handling for first byte.
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
revMode = RevMode.Forward;
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.CString:
//opcodeStr = sDataOpNames.StrNullTerm[highAscii ? 1 : 0];
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
break;
case FormatDescriptor.SubType.L8String:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrLen8Hi : sDataOpNames.StrLen8;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
leadingBytes = 1;
showLeading = true;
}
break;
case FormatDescriptor.SubType.L16String:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrLen16Hi : sDataOpNames.StrLen16;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
leadingBytes = 2;
showLeading = true;
}
break;
default:
Debug.Assert(false);
return;
}
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
// Create a new StringGather, with the final opcode choice.
gath = new StringGather(this, labelStr, opcodeStr, commentStr, delim,
delimReplace, StringGather.ByteStyle.DenseHex, MAX_OPERAND_LEN, false);
FeedGath(gath, data, offset, dfd.Length, revMode, leadingBytes, showLeading,
trailingBytes, showTrailing);
}
/// <summary>
/// Feeds the bytes into the StringGather.
/// </summary>
private void FeedGath(StringGather gath, byte[] data, int offset, int length,
RevMode revMode, int leadingBytes, bool showLeading,
int trailingBytes, bool showTrailing) {
int startOffset = offset;
int strEndOffset = offset + length - trailingBytes;
if (showLeading) {
while (leadingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
} else {
offset += leadingBytes;
}
if (revMode == RevMode.BlockReverse) {
const int maxPerLine = MAX_OPERAND_LEN - 2;
int numBlockLines = (length + maxPerLine - 1) / maxPerLine;
for (int chunk = 0; chunk < numBlockLines; chunk++) {
int chunkOffset = startOffset + chunk * maxPerLine;
int endOffset = chunkOffset + maxPerLine;
if (endOffset > strEndOffset) {
endOffset = strEndOffset;
}
for (int off = endOffset - 1; off >= chunkOffset; off--) {
gath.WriteChar((char)(data[off] & 0x7f));
}
}
} else {
for (; offset < strEndOffset; offset++) {
if (revMode == RevMode.Forward) {
gath.WriteChar((char)(data[offset] & 0x7f));
} else if (revMode == RevMode.Reverse) {
int posn = startOffset + (strEndOffset - offset) - 1;
gath.WriteChar((char)(data[posn] & 0x7f));
} else {
Debug.Assert(false);
}
}
}
while (showTrailing && trailingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
gath.Finish();
}
}
#endregion IGenerator
#region IAssembler
/// <summary>
/// Cross-assembler execution interface.
/// </summary>
public class AsmMerlin32 : IAssembler {
private List<string> PathNames { get; set; }
@ -107,4 +689,6 @@ namespace SourceGen.AsmGen {
cmd.Stderr, outputFile);
}
}
#endregion IAssembler
}

68
SourceGen/AsmGen/AssemblerInfo.cs
View File

@ -16,6 +16,7 @@
using System;
using System.Collections;
using System.Collections.Generic;
using System.Diagnostics;
namespace SourceGen.AsmGen {
/// <summary>
@ -30,9 +31,20 @@ namespace SourceGen.AsmGen {
public enum Id {
Unknown = 0,
Cc65,
Merlin32
Merlin32,
}
/// <summary>
/// Static information for all known assemblers.
///
/// The AsmType argument may be null. This is useful for non-cross assemblers.
/// </summary>
private static AssemblerInfo[] sInfo = new AssemblerInfo[] {
new AssemblerInfo(Id.Unknown, "???", null, null),
new AssemblerInfo(Id.Cc65, "cc65", typeof(GenCc65), typeof(AsmCc65)),
new AssemblerInfo(Id.Merlin32, "Merlin 32", typeof(GenMerlin32), typeof(AsmMerlin32)),
};
/// <summary>
/// Identifier.
/// </summary>
@ -43,19 +55,24 @@ namespace SourceGen.AsmGen {
/// </summary>
public string Name { get; private set; }
/// <summary>
/// Type of generator class.
/// </summary>
public Type GenType { get; private set; }
private AssemblerInfo(Id id, string name) {
/// <summary>
/// Type of assembler class.
/// </summary>
public Type AsmType { get; private set; }
private AssemblerInfo(Id id, string name, Type genType, Type asmType) {
AssemblerId = id;
Name = name;
GenType = genType;
AsmType = asmType;
}
// For simplicity, this is 1:1 with the Id enum.
private static AssemblerInfo[] sInfo = new AssemblerInfo[] {
new AssemblerInfo(Id.Unknown, "???"),
new AssemblerInfo(Id.Cc65, "cc65"),
new AssemblerInfo(Id.Merlin32, "Merlin 32")
};
/// <summary>
/// Returns an AssemblerInfo object for the specified id.
/// </summary>
@ -68,32 +85,29 @@ namespace SourceGen.AsmGen {
/// <summary>
/// Generator factory method.
/// </summary>
/// <param name="id">ID of assembler to return generator object for.</param>
/// <param name="id">ID of assembler to return generator instance for.</param>
/// <returns>New source generator object.</returns>
public static IGenerator GetGenerator(AssemblerInfo.Id id) {
switch (id) {
case Id.Cc65:
return new GenCc65();
case Id.Merlin32:
return new GenMerlin32();
default:
return null;
public static IGenerator GetGenerator(Id id) {
Type genType = sInfo[(int)id].GenType;
if (genType == null) {
Debug.Assert(false); // unexpected for generator
return null;
} else {
return (IGenerator)Activator.CreateInstance(genType);
}
}
/// <summary>
/// Assembler factory method.
/// </summary>
/// <param name="id">ID of assembler to return assembler object for.</param>
/// <param name="id">ID of assembler to return assembler instance for.</param>
/// <returns>New assembler interface object.</returns>
public static IAssembler GetAssembler(AssemblerInfo.Id id) {
switch (id) {
case Id.Cc65:
return new AsmCc65();
case Id.Merlin32:
return new AsmMerlin32();
default:
return null;
public static IAssembler GetAssembler(Id id) {
Type asmType = sInfo[(int)id].AsmType;
if (asmType == null) {
return null;
} else {
return (IAssembler)Activator.CreateInstance(asmType);
}
}

648
SourceGen/AsmGen/GenCc65.cs
View File

@ -1,648 +0,0 @@
/*
* Copyright 2018 faddenSoft
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Diagnostics;
using System.IO;
using System.Text;
using Asm65;
using CommonUtil;
namespace SourceGen.AsmGen {
/// <summary>
/// Generate source code compatible with the cc65 assembler (https://github.com/cc65/cc65).
/// </summary>
public class GenCc65 : IGenerator {
private const string ASM_FILE_SUFFIX = "_cc65.S";
private const int MAX_OPERAND_LEN = 64;
// IGenerator
public DisasmProject Project { get; private set; }
// IGenerator
public Formatter SourceFormatter { get; private set; }
// IGenerator
public AppSettings Settings { get; private set; }
// IGenerator
public AssemblerQuirks Quirks { get; private set; }
// IGenerator
public LabelLocalizer Localizer { get { return mLocalizer; } }
/// <summary>
/// Working directory, i.e. where we write our output file(s).
/// </summary>
private string mWorkDirectory;
/// <summary>
/// If set, long labels get their own line.
/// </summary>
private bool mLongLabelNewLine;
/// <summary>
/// Base filename. Typically the project file name without the ".dis65" extension.
/// </summary>
private string mFileNameBase;
/// <summary>
/// StringBuilder to use when composing a line. Held here to reduce allocations.
/// </summary>
private StringBuilder mLineBuilder = new StringBuilder(100);
/// <summary>
/// Label localization helper.
/// </summary>
private LabelLocalizer mLocalizer;
/// <summary>
/// Stream to send the output to.
/// </summary>
private StreamWriter mOutStream;
/// <summary>
/// The first time we output a high-ASCII string, we generate a macro for it.
/// </summary>
private bool mHighAsciiMacroOutput;
/// <summary>
/// Holds detected version of configured assembler.
/// </summary>
private CommonUtil.Version mAsmVersion = CommonUtil.Version.NO_VERSION;
// We test against this in a few places.
private static CommonUtil.Version V2_17 = new CommonUtil.Version(2, 17);
// Pseudo-op string constants.
private static PseudoOp.PseudoOpNames sDataOpNames = new PseudoOp.PseudoOpNames() {
EquDirective = "=",
OrgDirective = ".org",
//RegWidthDirective // .a8, .a16, .i8, .i16
DefineData1 = ".byte",
DefineData2 = ".word",
DefineData3 = ".faraddr",
DefineData4 = ".dword",
DefineBigData2 = ".dbyt",
//DefineBigData3
//DefineBigData4
Fill = ".res",
//Dense // no equivalent, use .byte with comma-separated args
StrGeneric = ".byte",
//StrReverse
StrNullTerm = ".asciiz",
//StrLen8 // macro with .strlen?
//StrLen16
//StrDci
//StrDciReverse
};
// IGenerator
public void Configure(DisasmProject project, string workDirectory, string fileNameBase,
AssemblerVersion asmVersion, AppSettings settings) {
Debug.Assert(project != null);
Debug.Assert(!string.IsNullOrEmpty(workDirectory));
Debug.Assert(!string.IsNullOrEmpty(fileNameBase));
Project = project;
Quirks = new AssemblerQuirks();
if (asmVersion != null) {
// Use the actual version. If it's > 2.17 we'll try to take advantage of
// bug fixes.
mAsmVersion = asmVersion.Version;
} else {
// No assembler installed. Use 2.17.
mAsmVersion = V2_17;
}
if (mAsmVersion <= V2_17) {
// cc65 v2.17: https://github.com/cc65/cc65/issues/717
Quirks.BlockMoveArgsReversed = true;
// cc65 v2.17: https://github.com/cc65/cc65/issues/754
Quirks.NoPcRelBankWrap = true;
}
mWorkDirectory = workDirectory;
mFileNameBase = fileNameBase;
Settings = settings;
mLongLabelNewLine = Settings.GetBool(AppSettings.SRCGEN_LONG_LABEL_NEW_LINE, false);
}
// IGenerator
public List<string> GenerateSource(BackgroundWorker worker) {
List<string> pathNames = new List<string>(1);
string fileName = mFileNameBase + ASM_FILE_SUFFIX;
string pathName = Path.Combine(mWorkDirectory, fileName);
pathNames.Add(pathName);
Formatter.FormatConfig config = new Formatter.FormatConfig();
GenCommon.ConfigureFormatterFromSettings(Settings, ref config);
config.mForceAbsOpcodeSuffix = string.Empty;
config.mForceLongOpcodeSuffix = string.Empty;
config.mForceAbsOperandPrefix = "a:"; // absolute
config.mForceLongOperandPrefix = "f:"; // far
config.mEndOfLineCommentDelimiter = ";";
config.mFullLineCommentDelimiterBase = ";";
config.mBoxLineCommentDelimiter = ";";
config.mAllowHighAsciiCharConst = false;
config.mExpressionMode = Formatter.FormatConfig.ExpressionMode.Simple;
SourceFormatter = new Formatter(config);
string msg = string.Format(Properties.Resources.PROGRESS_GENERATING_FMT, pathName);
worker.ReportProgress(0, msg);
mLocalizer = new LabelLocalizer(Project);
if (!Settings.GetBool(AppSettings.SRCGEN_DISABLE_LABEL_LOCALIZATION, false)) {
mLocalizer.LocalPrefix = "@";
mLocalizer.Analyze();
}
// Use UTF-8 encoding, without a byte-order mark.
using (StreamWriter sw = new StreamWriter(pathName, false, new UTF8Encoding(false))) {
mOutStream = sw;
if (Settings.GetBool(AppSettings.SRCGEN_ADD_IDENT_COMMENT, false)) {
//if (mAsmVersion.IsValid && mAsmVersion <= V2_17) {
// OutputLine(SourceFormatter.FullLineCommentDelimiter +
// string.Format(Properties.Resources.GENERATED_FOR_VERSION,
// "cc65", mAsmVersion.ToString()));
//} else {
// OutputLine(SourceFormatter.FullLineCommentDelimiter +
// string.Format(Properties.Resources.GENERATED_FOR_LATEST, "cc65"));
//}
// Currently generating code for v2.17.
OutputLine(SourceFormatter.FullLineCommentDelimiter +
string.Format(Properties.Resources.GENERATED_FOR_VERSION,
"cc65", V2_17));
}
GenCommon.Generate(this, sw, worker);
}
mOutStream = null;
return pathNames;
}
// IGenerator
public void OutputAsmConfig() {
CpuDef cpuDef = Project.CpuDef;
string cpuStr;
if (cpuDef.Type == CpuDef.CpuType.Cpu65816) {
cpuStr = "65816";
} else if (cpuDef.Type == CpuDef.CpuType.Cpu65C02) {
cpuStr = "65C02";
} else if (cpuDef.Type == CpuDef.CpuType.Cpu6502 && cpuDef.HasUndocumented) {
cpuStr = "6502X";
} else {
cpuStr = "6502";
}
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(".setcpu"),
'\"' + cpuStr + '\"', string.Empty);
}
/// <summary>
/// Map the mnemonics we chose for undocumented opcodes to the cc65 mnemonics.
/// After switching to the Unintended Opcodes mnemonics there's almost no difference.
///
/// We don't include the double- and triple-byte NOPs here, as cc65 doesn't
/// appear to have a definition for them (as of 2.17).
/// </summary>
private static Dictionary<string, string> sUndocMap = new Dictionary<string, string>() {
{ OpName.ALR, "alr" }, // imm 0x4b
{ OpName.ANC, "anc" }, // imm 0x0b (and others)
{ OpName.ANE, "ane" }, // imm 0x8b
{ OpName.ARR, "arr" }, // imm 0x6b
{ OpName.DCP, "dcp" }, // abs 0xcf
{ OpName.ISC, "isc" }, // abs 0xef
{ OpName.JAM, "jam" }, // abs 0x02 (and others)
{ OpName.LAS, "las" }, // abs,y 0xbb
{ OpName.LAX, "lax" }, // imm 0xab; abs 0xaf
{ OpName.RLA, "rla" }, // abs 0x2f
{ OpName.RRA, "rra" }, // abs 0x6f
{ OpName.SAX, "sax" }, // abs 0x8f
{ OpName.SBX, "axs" }, //* imm 0xcb
{ OpName.SHA, "sha" }, // abs,y 0x9f
{ OpName.SHX, "shx" }, // abs,y 0x9e
{ OpName.SHY, "shy" }, // abs,x 0x9c
{ OpName.SLO, "slo" }, // abs 0x0f
{ OpName.SRE, "sre" }, // abs 0x4f
{ OpName.TAS, "tas" }, // abs,y 0x9b
};
// IGenerator
public string ReplaceMnemonic(OpDef op) {
if ((op == OpDef.OpWDM_WDM || op == OpDef.OpBRK_StackInt) && mAsmVersion <= V2_17) {
// cc65 v2.17 doesn't support WDM, and assembles BRK <arg> to opcode $05.
// https://github.com/cc65/cc65/issues/715
// https://github.com/cc65/cc65/issues/716
return null;
} else if (op.IsUndocumented) {
if (sUndocMap.TryGetValue(op.Mnemonic, out string newValue)) {
if ((op.Mnemonic == OpName.ANC && op.Opcode != 0x0b) ||
(op.Mnemonic == OpName.JAM && op.Opcode != 0x02)) {
// There are multiple opcodes for the same thing. cc65 outputs
// one specific thing, so we need to match that, and just do a hex
// dump for the others.
return null;
}
return newValue;
}
return null;
} else {
return string.Empty;
}
}
// IGenerator
public void GenerateShortSequence(int offset, int length, out string opcode,
out string operand) {
Debug.Assert(length >= 1 && length <= 4);
// Use a comma-separated list of individual hex bytes.
opcode = sDataOpNames.DefineData1;
StringBuilder sb = new StringBuilder(length * 4);
for (int i = 0; i < length; i++) {
if (i != 0) {
sb.Append(',');
}
sb.Append(SourceFormatter.FormatHexValue(Project.FileData[offset + i], 2));
}
operand = sb.ToString();
}
// IGenerator
public void OutputDataOp(int offset) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
string labelStr = string.Empty;
if (attr.Symbol != null) {
labelStr = mLocalizer.ConvLabel(attr.Symbol.Label);
}
string commentStr = SourceFormatter.FormatEolComment(Project.Comments[offset]);
string opcodeStr, operandStr;
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
int length = dfd.Length;
Debug.Assert(length > 0);
bool multiLine = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.Default:
if (length != 1) {
Debug.Assert(false);
length = 1;
}
opcodeStr = sDataOpNames.DefineData1;
int operand = RawData.GetWord(data, offset, length, false);
operandStr = formatter.FormatHexValue(operand, length * 2);
break;
case FormatDescriptor.Type.NumericLE:
opcodeStr = sDataOpNames.GetDefineData(length);
operand = RawData.GetWord(data, offset, length, false);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
break;
case FormatDescriptor.Type.NumericBE:
opcodeStr = sDataOpNames.GetDefineBigData(length);
if (opcodeStr == null) {
// Nothing defined, output as comma-separated single-byte values.
GenerateShortSequence(offset, length, out opcodeStr, out operandStr);
} else {
operand = RawData.GetWord(data, offset, length, true);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
}
break;
case FormatDescriptor.Type.Fill:
opcodeStr = sDataOpNames.Fill;
operandStr = length + "," + formatter.FormatHexValue(data[offset], 2);
break;
case FormatDescriptor.Type.Dense:
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
break;
case FormatDescriptor.Type.String:
multiLine = true;
opcodeStr = operandStr = null;
OutputString(offset, labelStr, commentStr);
break;
default:
opcodeStr = "???";
operandStr = "***";
break;
}
if (!multiLine) {
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
}
}
private void OutputDenseHex(int offset, int length, string labelStr, string commentStr) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
StringBuilder sb = new StringBuilder(MAX_OPERAND_LEN);
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.DefineData1);
int maxPerLine = MAX_OPERAND_LEN / 4;
int numChunks = (length + maxPerLine - 1) / maxPerLine;
for (int chunk = 0; chunk < numChunks; chunk++) {
int chunkStart = chunk * maxPerLine;
int chunkEnd = Math.Min((chunk + 1) * maxPerLine, length);
for (int i = chunkStart; i < chunkEnd; i++) {
if (i != chunkStart) {
sb.Append(',');
}
sb.Append(formatter.FormatHexValue(data[offset + i], 2));
}
OutputLine(labelStr, opcodeStr, sb.ToString(), commentStr);
labelStr = commentStr = string.Empty;
sb.Clear();
}
}
/// <summary>
/// Outputs formatted data in an unformatted way, because the code generator couldn't
/// figure out how to do something better.
/// </summary>
private void OutputNoJoy(int offset, int length, string labelStr, string commentStr) {
byte[] data = Project.FileData;
Debug.Assert(length > 0);
Debug.Assert(offset >= 0 && offset < data.Length);
bool singleValue = true;
byte val = data[offset];
for (int i = 1; i < length; i++) {
if (data[offset + i] != val) {
singleValue = false;
break;
}
}
if (singleValue) {
string opcodeStr = SourceFormatter.FormatPseudoOp(sDataOpNames.Fill);
string operandStr = length + "," + SourceFormatter.FormatHexValue(val, 2);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
} else {
OutputDenseHex(offset, length, labelStr, commentStr);
}
}
// IGenerator
public void OutputEquDirective(string name, string valueStr, string comment) {
OutputLine(name, SourceFormatter.FormatPseudoOp(sDataOpNames.EquDirective),
valueStr, SourceFormatter.FormatEolComment(comment));
}
// IGenerator
public void OutputOrgDirective(int address) {
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(sDataOpNames.OrgDirective),
SourceFormatter.FormatHexValue(address, 4), string.Empty);
}
// IGenerator
public void OutputRegWidthDirective(int offset, int prevM, int prevX, int newM, int newX) {
if (prevM != newM) {
string mop = (newM == 0) ? ".a16" : ".a8";
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(mop),
string.Empty, string.Empty);
}
if (prevX != newX) {
string xop = (newX == 0) ? ".i16" : ".i8";
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(xop),
string.Empty, string.Empty);
}
}
// IGenerator
public void OutputLine(string fullLine) {
mOutStream.WriteLine(fullLine);
}
// IGenerator
public void OutputLine(string label, string opcode, string operand, string comment) {
// If a label is provided, and it doesn't start with a '.' (indicating that it's
// a directive), and this isn't an EQU directive, add a ':'. Might be easier to
// just ".feature labels_without_colons", but I'm trying to do things the way
// that cc65 users will expect.
if (!string.IsNullOrEmpty(label) && label[0] != '.' &&
!String.Equals(opcode, sDataOpNames.EquDirective,
StringComparison.InvariantCultureIgnoreCase)) {
label += ':';
if (mLongLabelNewLine && label.Length >= 9) {
mOutStream.WriteLine(label);
label = string.Empty;
}
}
mLineBuilder.Clear();
TextUtil.AppendPaddedString(mLineBuilder, label, 9);
TextUtil.AppendPaddedString(mLineBuilder, opcode, 9 + 8);
TextUtil.AppendPaddedString(mLineBuilder, operand, 9 + 8 + 11);
if (string.IsNullOrEmpty(comment)) {
// Trim trailing spaces off of opcode or operand. If they want trailing
// spaces at the end of a comment, that's fine.
CommonUtil.TextUtil.TrimEnd(mLineBuilder);
} else {
mLineBuilder.Append(comment);
}
mOutStream.WriteLine(mLineBuilder.ToString());
}
private void OutputString(int offset, string labelStr, string commentStr) {
// Normal ASCII strings are straightforward: they're just part of a .byte
// directive, and can mix with anything else in the .byte.
//
// For CString we can use .asciiz, but only if the string fits on one line
// and doesn't include delimiters. For L8String and L16String we can
// define simple macros, but their use has a similar restriction. High-ASCII
// strings also require a macro.
//
// We might be able to define a macro for DCI and Reverse as well.
//
// The limitation on strings with delimiters arises because (1) I don't see a
// way to escape them within a string, and (2) the simple macro workarounds
// only take a single argument, not a comma-separated list of stuff.
//
// Some ideas here:
// https://groups.google.com/forum/#!topic/comp.sys.apple2.programmer/5Wkw8mUPcU0
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
Debug.Assert(dfd.FormatType == FormatDescriptor.Type.String);
Debug.Assert(dfd.Length > 0);
bool highAscii = false;
int leadingBytes = 0;
int trailingBytes = 0;
bool showLeading = false;
bool showTrailing = false;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Dci:
highAscii = (data[offset] & 0x80) != 0;
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.Reverse:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.DciReverse:
highAscii = (data[offset + dfd.Length - 1] & 0x80) != 0;
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.CString:
highAscii = (data[offset] & 0x80) != 0;
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.L8String:
if (dfd.Length > 1) {
highAscii = (data[offset + 1] & 0x80) != 0;
}
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.L16String:
if (dfd.Length > 2) {
highAscii = (data[offset + 2] & 0x80) != 0;
}
leadingBytes = 2;
showLeading = true;
break;
default:
Debug.Assert(false);
return;
}
char delim = '"';
StringGather gath = null;
// Run the string through so we can see if it'll fit on one line. As a minor
// optimization, we skip this step for "generic" strings, which are probably
// the most common thing.
if (dfd.FormatSubType != FormatDescriptor.SubType.None || highAscii) {
gath = new StringGather(this, labelStr, "???", commentStr, delim,
delim, StringGather.ByteStyle.CommaSep, MAX_OPERAND_LEN, true);
FeedGath(gath, data, offset, dfd.Length, leadingBytes, showLeading,
trailingBytes, showTrailing);
Debug.Assert(gath.NumLinesOutput > 0);
}
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.StrGeneric);
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
// Special case for simple short high-ASCII strings. These have no
// leading or trailing bytes. We can improve this a bit by handling
// arbitrarily long strings by simply breaking them across lines.
Debug.Assert(leadingBytes == 0);
Debug.Assert(trailingBytes == 0);
if (highAscii && gath.NumLinesOutput == 1 && !gath.HasDelimiter) {
if (!mHighAsciiMacroOutput) {
mHighAsciiMacroOutput = true;
// Output a macro for high-ASCII strings.
OutputLine(".macro", "HiAscii", "Arg", string.Empty);
OutputLine(string.Empty, ".repeat", ".strlen(Arg), I", string.Empty);
OutputLine(string.Empty, ".byte", ".strat(Arg, I) | $80", string.Empty);
OutputLine(string.Empty, ".endrep", string.Empty, string.Empty);
OutputLine(".endmacro", string.Empty, string.Empty, string.Empty);
}
opcodeStr = formatter.FormatPseudoOp("HiAscii");
highAscii = false;
}
break;
case FormatDescriptor.SubType.Dci:
case FormatDescriptor.SubType.Reverse:
case FormatDescriptor.SubType.DciReverse:
// Full configured above.
break;
case FormatDescriptor.SubType.CString:
if (gath.NumLinesOutput == 1 && !gath.HasDelimiter) {
opcodeStr = sDataOpNames.StrNullTerm;
showTrailing = false;
}
break;
case FormatDescriptor.SubType.L8String:
case FormatDescriptor.SubType.L16String:
// Implement macros?
break;
default:
Debug.Assert(false);
return;
}
if (highAscii) {
OutputNoJoy(offset, dfd.Length, labelStr, commentStr);
return;
}
// Create a new StringGather, with the final opcode choice.
gath = new StringGather(this, labelStr, opcodeStr, commentStr, delim,
delim, StringGather.ByteStyle.CommaSep, MAX_OPERAND_LEN, false);
FeedGath(gath, data, offset, dfd.Length, leadingBytes, showLeading,
trailingBytes, showTrailing);
}
/// <summary>
/// Feeds the bytes into the StringGather.
/// </summary>
private void FeedGath(StringGather gath, byte[] data, int offset, int length,
int leadingBytes, bool showLeading, int trailingBytes, bool showTrailing) {
int startOffset = offset;
int strEndOffset = offset + length - trailingBytes;
if (showLeading) {
while (leadingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
} else {
offset += leadingBytes;
}
for (; offset < strEndOffset; offset++) {
gath.WriteChar((char)(data[offset] & 0x7f));
}
while (showTrailing && trailingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
gath.Finish();
}
}
}

595
SourceGen/AsmGen/GenMerlin32.cs
View File

@ -1,595 +0,0 @@
/*
* Copyright 2018 faddenSoft
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Diagnostics;
using System.IO;
using System.Text;
using Asm65;
using CommonUtil;
namespace SourceGen.AsmGen {
/// <summary>
/// Generate source code compatible with Brutal Deluxe's Merlin 32 assembler
/// (https://www.brutaldeluxe.fr/products/crossdevtools/merlin/).
/// </summary>
public class GenMerlin32 : IGenerator {
private const string ASM_FILE_SUFFIX = "_Merlin32.S";
private const int MAX_OPERAND_LEN = 64;
// IGenerator
public DisasmProject Project { get; private set; }
// IGenerator
public Formatter SourceFormatter { get; private set; }
// IGenerator
public AppSettings Settings { get; private set; }
// IGenerator
public AssemblerQuirks Quirks { get; private set; }
// IGenerator
public LabelLocalizer Localizer { get { return mLocalizer; } }
/// <summary>
/// Working directory, i.e. where we write our output file(s).
/// </summary>
private string mWorkDirectory;
/// <summary>
/// If set, long labels get their own line.
/// </summary>
private bool mLongLabelNewLine;
/// <summary>
/// Base filename. Typically the project file name without the ".dis65" extension.
/// </summary>
private string mFileNameBase;
/// <summary>
/// StringBuilder to use when composing a line. Held here to reduce allocations.
/// </summary>
private StringBuilder mLineBuilder = new StringBuilder(100);
/// <summary>
/// Label localization helper.
/// </summary>
private LabelLocalizer mLocalizer;
/// <summary>
/// Stream to send the output to.
/// </summary>
private StreamWriter mOutStream;
/// <summary>
/// Holds detected version of configured assembler.
/// </summary>
private CommonUtil.Version mAsmVersion = CommonUtil.Version.NO_VERSION;
// Semi-convenient way to hold all the interesting string constants in one place.
// Note the actual usage of the pseudo-op may not match what the main app does,
// e.g. RegWidthDirective behaves differently from "mx". I'm just trying to avoid
// having string constants scattered all over.
private static PseudoOp.PseudoOpNames sDataOpNames = new PseudoOp.PseudoOpNames() {
EquDirective = "equ",
OrgDirective = "org",
RegWidthDirective = "mx",
DefineData1 = "dfb",
DefineData2 = "dw",
DefineData3 = "adr",
DefineData4 = "adrl",
DefineBigData2 = "ddb",
//DefineBigData3
//DefineBigData4
Fill = "ds",
Dense = "hex",
StrGeneric = "asc",
StrGenericHi = "asc",
StrReverse = "rev",
StrReverseHi = "rev",
//StrNullTerm
StrLen8 = "str",
StrLen8Hi = "str",
StrLen16 = "strl",
StrLen16Hi = "strl",
StrDci = "dci",
StrDciHi = "dci",
//StrDciReverse
};
// IGenerator
public void Configure(DisasmProject project, string workDirectory, string fileNameBase,
AssemblerVersion asmVersion, AppSettings settings) {
Debug.Assert(project != null);
Debug.Assert(!string.IsNullOrEmpty(workDirectory));
Debug.Assert(!string.IsNullOrEmpty(fileNameBase));
Project = project;
Quirks = new AssemblerQuirks();
Quirks.TracksSepRepNotEmu = true;
Quirks.NoPcRelBankWrap = true;
mWorkDirectory = workDirectory;
mFileNameBase = fileNameBase;
Settings = settings;
mLongLabelNewLine = Settings.GetBool(AppSettings.SRCGEN_LONG_LABEL_NEW_LINE, false);
}
// IGenerator; executes on background thread
public List<string> GenerateSource(BackgroundWorker worker) {
List<string> pathNames = new List<string>(1);
string fileName = mFileNameBase + ASM_FILE_SUFFIX;
string pathName = Path.Combine(mWorkDirectory, fileName);
pathNames.Add(pathName);
Formatter.FormatConfig config = new Formatter.FormatConfig();
GenCommon.ConfigureFormatterFromSettings(Settings, ref config);
config.mForceAbsOpcodeSuffix = ":";
config.mForceLongOpcodeSuffix = "l";
config.mForceAbsOperandPrefix = string.Empty;
config.mForceLongOperandPrefix = string.Empty;
config.mEndOfLineCommentDelimiter = ";";
config.mFullLineCommentDelimiterBase = ";";
config.mBoxLineCommentDelimiter = string.Empty;
config.mAllowHighAsciiCharConst = true;
config.mExpressionMode = Formatter.FormatConfig.ExpressionMode.Merlin;
SourceFormatter = new Formatter(config);
string msg = string.Format(Properties.Resources.PROGRESS_GENERATING_FMT, pathName);
worker.ReportProgress(0, msg);
mLocalizer = new LabelLocalizer(Project);
if (!Settings.GetBool(AppSettings.SRCGEN_DISABLE_LABEL_LOCALIZATION, false)) {
mLocalizer.LocalPrefix = ":";
mLocalizer.Analyze();
}
// Use UTF-8 encoding, without a byte-order mark.
using (StreamWriter sw = new StreamWriter(pathName, false, new UTF8Encoding(false))) {
mOutStream = sw;
if (Settings.GetBool(AppSettings.SRCGEN_ADD_IDENT_COMMENT, false)) {
// No version-specific stuff yet. We're generating code for v1.0.
OutputLine(SourceFormatter.FullLineCommentDelimiter +
string.Format(Properties.Resources.GENERATED_FOR_VERSION,
"Merlin 32", new CommonUtil.Version(1, 0)));
}
GenCommon.Generate(this, sw, worker);
}
mOutStream = null;
return pathNames;
}
// IGenerator
public void OutputDataOp(int offset) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
string labelStr = string.Empty;
if (attr.Symbol != null) {
labelStr = mLocalizer.ConvLabel(attr.Symbol.Label);
}
string commentStr = SourceFormatter.FormatEolComment(Project.Comments[offset]);
string opcodeStr, operandStr;
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
int length = dfd.Length;
Debug.Assert(length > 0);
bool multiLine = false;
switch (dfd.FormatType) {
case FormatDescriptor.Type.Default:
if (length != 1) {
Debug.Assert(false);
length = 1;
}
opcodeStr = sDataOpNames.DefineData1;
int operand = RawData.GetWord(data, offset, length, false);
operandStr = formatter.FormatHexValue(operand, length * 2);
break;
case FormatDescriptor.Type.NumericLE:
opcodeStr = sDataOpNames.GetDefineData(length);
operand = RawData.GetWord(data, offset, length, false);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
break;
case FormatDescriptor.Type.NumericBE:
opcodeStr = sDataOpNames.GetDefineBigData(length);
if (opcodeStr == null) {
// Nothing defined, output as comma-separated single-byte values.
GenerateShortSequence(offset, length, out opcodeStr, out operandStr);
} else {
operand = RawData.GetWord(data, offset, length, true);
operandStr = PseudoOp.FormatNumericOperand(formatter, Project.SymbolTable,
mLocalizer.LabelMap, dfd, operand, length, false);
}
break;
case FormatDescriptor.Type.Fill:
opcodeStr = sDataOpNames.Fill;
operandStr = length + "," + formatter.FormatHexValue(data[offset], 2);
break;
case FormatDescriptor.Type.Dense:
multiLine = true;
opcodeStr = operandStr = null;
OutputDenseHex(offset, length, labelStr, commentStr);
break;
case FormatDescriptor.Type.String:
multiLine = true;
opcodeStr = operandStr = null;
OutputString(offset, labelStr, commentStr);
break;
default:
opcodeStr = "???";
operandStr = "***";
break;
}
if (!multiLine) {
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
}
}
private void OutputDenseHex(int offset, int length, string labelStr, string commentStr) {
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
int maxPerLine = MAX_OPERAND_LEN / 2;
string opcodeStr = formatter.FormatPseudoOp(sDataOpNames.Dense);
for (int i = 0; i < length; i += maxPerLine) {
int subLen = length - i;
if (subLen > maxPerLine) {
subLen = maxPerLine;
}
string operandStr = formatter.FormatDenseHex(data, offset + i, subLen);
OutputLine(labelStr, opcodeStr, operandStr, commentStr);
labelStr = commentStr = string.Empty;
}
}
// IGenerator
public string ReplaceMnemonic(OpDef op) {
if (op.IsUndocumented) {
return null;
} else {
return string.Empty;
}
}
// IGenerator
public void GenerateShortSequence(int offset, int length, out string opcode,
out string operand) {
Debug.Assert(length >= 1 && length <= 4);
// Use a comma-separated list of individual hex bytes.
opcode = sDataOpNames.DefineData1;
StringBuilder sb = new StringBuilder(length * 4);
for (int i = 0; i < length; i++) {
if (i != 0) {
sb.Append(',');
}
sb.Append(SourceFormatter.FormatHexValue(Project.FileData[offset + i], 2));
}
operand = sb.ToString();
}
// IGenerator
public void OutputAsmConfig() {
// nothing to do
}
// IGenerator
public void OutputEquDirective(string name, string valueStr, string comment) {
OutputLine(name, SourceFormatter.FormatPseudoOp(sDataOpNames.EquDirective),
valueStr, SourceFormatter.FormatEolComment(comment));
}
// IGenerator
public void OutputOrgDirective(int address) {
OutputLine(string.Empty, SourceFormatter.FormatPseudoOp(sDataOpNames.OrgDirective),
SourceFormatter.FormatHexValue(address, 4), string.Empty);
}
// IGenerator
public void OutputRegWidthDirective(int offset, int prevM, int prevX, int newM, int newX) {
// prevM/prevX may be ambiguous for offset 0, but otherwise everything
// should be either 0 or 1.
Debug.Assert(newM == 0 || newM == 1);
Debug.Assert(newX == 0 || newX == 1);
if (offset == 0 && newM == 1 && newX == 1) {
// Assembler defaults to short regs, so we can skip this.
return;
}
OutputLine(string.Empty,
SourceFormatter.FormatPseudoOp(sDataOpNames.RegWidthDirective),
"%" + newM + newX, string.Empty);
}
// IGenerator
public void OutputLine(string fullLine) {
mOutStream.WriteLine(fullLine);
}
// IGenerator
public void OutputLine(string label, string opcode, string operand, string comment) {
// Split long label, but not on EQU directives (confuses the assembler).
if (mLongLabelNewLine && label.Length >= 9 &&
!String.Equals(opcode, sDataOpNames.EquDirective,
StringComparison.InvariantCultureIgnoreCase)) {
mOutStream.WriteLine(label);
label = string.Empty;
}
mLineBuilder.Clear();
TextUtil.AppendPaddedString(mLineBuilder, label, 9);
TextUtil.AppendPaddedString(mLineBuilder, opcode, 9 + 6);
TextUtil.AppendPaddedString(mLineBuilder, operand, 9 + 6 + 11);
if (string.IsNullOrEmpty(comment)) {
// Trim trailing spaces off of opcode or operand. If they want trailing
// spaces at the end of a comment, that's fine.
CommonUtil.TextUtil.TrimEnd(mLineBuilder);
} else {
mLineBuilder.Append(comment);
}
mOutStream.WriteLine(mLineBuilder.ToString());
}
private enum RevMode { Forward, Reverse, BlockReverse };
private void OutputString(int offset, string labelStr, string commentStr) {
// This gets complicated.
//
// For Dci, L8String, and L16String, the entire string needs to fit in the
// operand of one line. If it can't, we need to separate the length byte/word
// or inverted character out, and just dump the rest as ASCII. Computing the
// line length requires factoring delimiter character escapes. (NOTE: contrary
// to the documentation, STR and STRL do include trailing hex characters in the
// length calculation, so it's possible to escape delimiters.)
//
// For Reverse, we can span lines, but only if we emit the lines in
// backward order. Also, Merlin doesn't allow hex to be embedded in a REV
// operation, so we can't use REV if the string contains a delimiter.
//
// DciReverse is deprecated, but we can handle it as a Reverse string with a
// trailing byte on a following line.
//
// For aesthetic purposes, zero-length CString, L8String, and L16String
// should be output as DFB/DW zeroes rather than an empty string -- makes
// it easier to read.
Formatter formatter = SourceFormatter;
byte[] data = Project.FileData;
Anattrib attr = Project.GetAnattrib(offset);
FormatDescriptor dfd = attr.DataDescriptor;
Debug.Assert(dfd != null);
Debug.Assert(dfd.FormatType == FormatDescriptor.Type.String);
Debug.Assert(dfd.Length > 0);
bool highAscii = false;
int showZeroes = 0;
int leadingBytes = 0;
int trailingBytes = 0;
bool showLeading = false;
bool showTrailing = false;
RevMode revMode = RevMode.Forward;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Dci:
highAscii = (data[offset] & 0x80) != 0;
break;
case FormatDescriptor.SubType.Reverse:
highAscii = (data[offset] & 0x80) != 0;
revMode = RevMode.Reverse;
break;
case FormatDescriptor.SubType.DciReverse:
highAscii = (data[offset + dfd.Length - 1] & 0x80) != 0;
revMode = RevMode.Reverse;
break;
case FormatDescriptor.SubType.CString:
highAscii = (data[offset] & 0x80) != 0;
if (dfd.Length == 1) {
showZeroes = 1; // empty null-terminated string
}
trailingBytes = 1;
showTrailing = true;
break;
case FormatDescriptor.SubType.L8String:
if (dfd.Length > 1) {
highAscii = (data[offset + 1] & 0x80) != 0;
} else {
//showZeroes = 1;
}
leadingBytes = 1;
break;
case FormatDescriptor.SubType.L16String:
if (dfd.Length > 2) {
highAscii = (data[offset + 2] & 0x80) != 0;
} else {
//showZeroes = 2;
}
leadingBytes = 2;
break;
default:
Debug.Assert(false);
return;
}
if (showZeroes != 0) {
// Empty string. Just output the length byte(s) or null terminator.
GenerateShortSequence(offset, showZeroes, out string opcode, out string operand);
OutputLine(labelStr, opcode, operand, commentStr);
return;
}
// Merlin 32 uses single-quote for low ASCII, double-quote for high ASCII. When
// quoting the delimiter we use a hexadecimal value. We need to bear in mind that
// we're forcing the characters to low ASCII, but the actual character being
// escaped might be in high ASCII. Hence delim vs. delimReplace.
char delim = highAscii ? '"' : '\'';
char delimReplace = highAscii ? ((char)(delim | 0x80)) : delim;
StringGather gath = null;
// Run the string through so we can see if it'll fit on one line. As a minor
// optimization, we skip this step for "generic" strings, which are probably
// the most common thing.
if (dfd.FormatSubType != FormatDescriptor.SubType.None) {
gath = new StringGather(this, labelStr, "???", commentStr, delim,
delimReplace, StringGather.ByteStyle.DenseHex, MAX_OPERAND_LEN, true);
FeedGath(gath, data, offset, dfd.Length, revMode, leadingBytes, showLeading,
trailingBytes, showTrailing);
Debug.Assert(gath.NumLinesOutput > 0);
}
string opcodeStr;
switch (dfd.FormatSubType) {
case FormatDescriptor.SubType.None:
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
break;
case FormatDescriptor.SubType.Dci:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrDciHi : sDataOpNames.StrDci;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
trailingBytes = 1;
showTrailing = true;
}
break;
case FormatDescriptor.SubType.Reverse:
if (gath.HasDelimiter) {
// can't include escaped delimiters in REV
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
revMode = RevMode.Forward;
} else if (gath.NumLinesOutput > 1) {
opcodeStr = highAscii ? sDataOpNames.StrReverseHi : sDataOpNames.StrReverse;
revMode = RevMode.BlockReverse;
} else {
opcodeStr = highAscii ? sDataOpNames.StrReverseHi : sDataOpNames.StrReverse;
Debug.Assert(revMode == RevMode.Reverse);
}
break;
case FormatDescriptor.SubType.DciReverse:
// Mostly punt -- output as ASCII with special handling for first byte.
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
revMode = RevMode.Forward;
leadingBytes = 1;
showLeading = true;
break;
case FormatDescriptor.SubType.CString:
//opcodeStr = sDataOpNames.StrNullTerm[highAscii ? 1 : 0];
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
break;
case FormatDescriptor.SubType.L8String:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrLen8Hi : sDataOpNames.StrLen8;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
leadingBytes = 1;
showLeading = true;
}
break;
case FormatDescriptor.SubType.L16String:
if (gath.NumLinesOutput == 1) {
opcodeStr = highAscii ? sDataOpNames.StrLen16Hi : sDataOpNames.StrLen16;
} else {
opcodeStr = highAscii ? sDataOpNames.StrGenericHi : sDataOpNames.StrGeneric;
leadingBytes = 2;
showLeading = true;
}
break;
default:
Debug.Assert(false);
return;
}
opcodeStr = formatter.FormatPseudoOp(opcodeStr);
// Create a new StringGather, with the final opcode choice.
gath = new StringGather(this, labelStr, opcodeStr, commentStr, delim,
delimReplace, StringGather.ByteStyle.DenseHex, MAX_OPERAND_LEN, false);
FeedGath(gath, data, offset, dfd.Length, revMode, leadingBytes, showLeading,
trailingBytes, showTrailing);
}
/// <summary>
/// Feeds the bytes into the StringGather.
/// </summary>
private void FeedGath(StringGather gath, byte[] data, int offset, int length,
RevMode revMode, int leadingBytes, bool showLeading,
int trailingBytes, bool showTrailing) {
int startOffset = offset;
int strEndOffset = offset + length - trailingBytes;
if (showLeading) {
while (leadingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
} else {
offset += leadingBytes;
}
if (revMode == RevMode.BlockReverse) {
const int maxPerLine = MAX_OPERAND_LEN - 2;
int numBlockLines = (length + maxPerLine - 1) / maxPerLine;
for (int chunk = 0; chunk < numBlockLines; chunk++) {
int chunkOffset = startOffset + chunk * maxPerLine;
int endOffset = chunkOffset + maxPerLine;
if (endOffset > strEndOffset) {
endOffset = strEndOffset;
}
for (int off = endOffset - 1; off >= chunkOffset; off--) {
gath.WriteChar((char)(data[off] & 0x7f));
}
}
} else {
for (; offset < strEndOffset; offset++) {
if (revMode == RevMode.Forward) {
gath.WriteChar((char)(data[offset] & 0x7f));
} else if (revMode == RevMode.Reverse) {
int posn = startOffset + (strEndOffset - offset) - 1;
gath.WriteChar((char)(data[posn] & 0x7f));
} else {
Debug.Assert(false);
}
}
}
while (showTrailing && trailingBytes-- > 0) {
gath.WriteByte(data[offset++]);
}
gath.Finish();
}
}
}

2
SourceGen/SourceGen.csproj
View File

@ -112,7 +112,6 @@
<Compile Include="AsmGen\AssemblerProgress.Designer.cs">
<DependentUpon>AssemblerProgress.cs</DependentUpon>
</Compile>
<Compile Include="AsmGen\GenCc65.cs" />
<Compile Include="AsmGen\GenCommon.cs" />
<Compile Include="AsmGen\GeneratorProgress.cs">
<SubType>Form</SubType>
@ -120,7 +119,6 @@
<Compile Include="AsmGen\GeneratorProgress.Designer.cs">
<DependentUpon>GeneratorProgress.cs</DependentUpon>
</Compile>
<Compile Include="AsmGen\GenMerlin32.cs" />
<Compile Include="AsmGen\IAssembler.cs" />
<Compile Include="AsmGen\IGenerator.cs" />
<Compile Include="AsmGen\LabelLocalizer.cs" />