/// <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();
}
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);
}
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();
}
private void OutputString(int offset, string labelStr, string commentStr)
{
// Normal ASCII strings are handled with a simple .text directive.
//
// CString and L8String have directives (.null, .ptext), but we can only use
// them if the string fits on one line and doesn't include delimiters.
//
// We could probably do something fancy with the character encoding options to
// make high-ASCII work nicely.
//
// We might be able to define a macro for DCI and Reverse.
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:
// TODO(someday): something fancy with encodings to handle high-ASCII text?
break;
case FormatDescriptor.SubType.Dci:
case FormatDescriptor.SubType.Reverse:
case FormatDescriptor.SubType.DciReverse:
// Fully configured above.
break;
case FormatDescriptor.SubType.CString:
if (gath.NumLinesOutput == 1 && !gath.HasDelimiter)
{
opcodeStr = sDataOpNames.StrNullTerm;
showTrailing = false;
}
break;
case FormatDescriptor.SubType.L8String:
if (gath.NumLinesOutput == 1 && !gath.HasDelimiter)
{
opcodeStr = sDataOpNames.StrLen8;
showLeading = false;
}
break;
case FormatDescriptor.SubType.L16String:
// Implement as macro?
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);
}
