public void TakeArgumentStateLogInitTest()
{
var sut = new RemoveArgumentStateLog();
var initHisto = sut.ToString(true);
const string expected = @"
Start: 0
LeadingWhitespace: 0
ScanningBackslashesOutsideQuotes: 0
QuoteAfterBackslashesOutsideQuotes: 0
EmittingBackslashesOutsideQuotes: 0
CheckForArgumentEndOutsideQuotes: 0
EmittingBackslashesOutsideQuotesGoingInsideQuotes: 0
ScanningBackslashesInsideQuotes: 0
QuoteAfterBackslashesInsideQuotes: 0
EmittingBackslashesInsideQuotes: 0
CheckForTwoQuotesAfterBackslashesInsideQuotes: 0
EmittingBackslashesInsideQuotesGoingOutsideQuotes: 0
CheckForArgumentEndInsideQuotesGoingOutsideQuotes: 0
EmittingBackslashesInsideQuotesStayingInsideQuotes: 0
EmitFinalBackslashesInsideQuotes: 0
CopyToEnd: 0
End: 0
";
Assert.Equal(expected, initHisto);
}
public void TakeArgumentTests(string input, string expectedArgument, string expectedRemainingLine)
{
var argumentStream = input.ToCharArray().AsEnumerable();
var argumentValue = new StringBuilder();
var log = new RemoveArgumentStateLog();
var remainingLine = input.RemoveArgument(argumentValue, log).Collect();
output_.WriteLine(log.ToString(true));
argumentValue.ToString().Should().Be(expectedArgument, "arg failed, input was \"{0}\"", input.ToLiteralFormat());
remainingLine.Should().Be(expectedRemainingLine, "remaining failed, input was \"{0}\"", input.ToLiteralFormat());
}
public void TakeArgumentStateLogWithLoggingTest()
{
var sut = new RemoveArgumentStateLog();
var sb = new StringBuilder();
sb.Append('x');
sut.Add("Start-a-0-x", CppArgumentLexing.ArgStates.Start, 'a', 0, sb);
sb.Append('y');
sut.Add("LeadingWhitespace-b-1-xy", CppArgumentLexing.ArgStates.LeadingWhitespace, 'b', 1, sb);
sb.Append('z');
sut.Add("EmitFinalBackslashesInsideQuotes-c-2-xyz", CppArgumentLexing.ArgStates.EmitFinalBackslashesInsideQuotes, 'c', 2, sb);
sb.Append('w');
sut.Add("LeadingWhitespace-d-3-xyzw", CppArgumentLexing.ArgStates.LeadingWhitespace, 'd', 3, sb);
var log = sut.ToString();
output_.WriteLine(log);
const string expectedLog = @"(Start-a-0-x, Start, 'a', 0, ""x"")
(LeadingWhitespace-b-1-xy, LeadingWhitespace, 'b', 1, ""xy"")
(EmitFinalBackslashesInsideQuotes-c-2-xyz, EmitFinalBackslashesInsideQuotes, 'c', 2, ""xyz"")
(LeadingWhitespace-d-3-xyzw, LeadingWhitespace, 'd', 3, ""xyzw"")
";
Assert.Equal(expectedLog, log);
var histo = sut.ToString(true);
const string expectedHisto = expectedLog + @"
Start: 1
LeadingWhitespace: 2
ScanningBackslashesOutsideQuotes: 0
QuoteAfterBackslashesOutsideQuotes: 0
EmittingBackslashesOutsideQuotes: 0
CheckForArgumentEndOutsideQuotes: 0
EmittingBackslashesOutsideQuotesGoingInsideQuotes: 0
ScanningBackslashesInsideQuotes: 0
QuoteAfterBackslashesInsideQuotes: 0
EmittingBackslashesInsideQuotes: 0
CheckForTwoQuotesAfterBackslashesInsideQuotes: 0
EmittingBackslashesInsideQuotesGoingOutsideQuotes: 0
CheckForArgumentEndInsideQuotesGoingOutsideQuotes: 0
EmittingBackslashesInsideQuotesStayingInsideQuotes: 0
EmitFinalBackslashesInsideQuotes: 1
CopyToEnd: 0
End: 0
";
Assert.Equal(expectedHisto, histo);
}
/// <summary>
/// Grab a parameter (_not_ the first) off of incoming stream, then pass
/// all the rest of the characters unchanged. Quoting rules are
/// complicated, see http://www.daviddeley.com/autohotkey/parameters/parameters.htm#WINCRULES
/// and https://msdn.microsoft.com/en-us/library/17w5ykft(v=vs.120).aspx
/// for details.
/// </summary>
/// <remarks>
/// Engineered from the D.Deley article above into a state machine.
/// </remarks>
public static IEnumerable<char> RemoveArgument(
this IEnumerable<char> source,
StringBuilder argument,
/*out*/ RemoveArgumentStateLog log = null)
{
ArgStates state = ArgStates.Start;
int nBackslashes = 0;
foreach (char c in source)
{
if (null != log) log.Add("fetchChar", state, c, nBackslashes, argument);
goto stateAction;
reinspectChar:
if (null != log) log.Add("reinspectChar", state, c, nBackslashes, argument);
goto stateAction;
stateAction:
switch (state)
{
case ArgStates.Start:
state = ArgStates.LeadingWhitespace;
goto reinspectChar;
case ArgStates.LeadingWhitespace:
if (' ' == c || '\t' == c)
{
continue;
}
else
{
state = ArgStates.ScanningBackslashesOutsideQuotes;
nBackslashes = 0;
goto reinspectChar;
}
case ArgStates.ScanningBackslashesOutsideQuotes:
if ('\\' == c)
{
nBackslashes++;
continue;
}
else if ('\"' == c)
{
state = ArgStates.QuoteAfterBackslashesOutsideQuotes;
goto reinspectChar;
}
else
{
state = ArgStates.EmittingBackslashesOutsideQuotes;
goto reinspectChar;
}
case ArgStates.QuoteAfterBackslashesOutsideQuotes:
{
bool oddBackslashes = 1 == nBackslashes % 2;
nBackslashes /= 2;
if (oddBackslashes)
{
state = ArgStates.EmittingBackslashesOutsideQuotes;
goto reinspectChar;
}
else
{
state = ArgStates.EmittingBackslashesOutsideQuotesGoingInsideQuotes;
continue;
}
}
case ArgStates.EmittingBackslashesOutsideQuotes:
argument.Append('\\', nBackslashes);
nBackslashes = 0;
state = ArgStates.CheckForArgumentEndOutsideQuotes;
goto reinspectChar;
case ArgStates.CheckForArgumentEndOutsideQuotes:
if (' ' == c || '\t' == c)
{
state = ArgStates.CopyToEnd;
goto reinspectChar;
}
else
{
argument.Append(c);
nBackslashes = 0;
state = ArgStates.ScanningBackslashesOutsideQuotes;
continue;
}
case ArgStates.EmittingBackslashesOutsideQuotesGoingInsideQuotes:
argument.Append('\\', nBackslashes);
nBackslashes = 0;
state = ArgStates.ScanningBackslashesInsideQuotes;
goto reinspectChar;
case ArgStates.ScanningBackslashesInsideQuotes:
if ('\\' == c)
{
nBackslashes++;
continue;
}
else if ('\"' == c)
{
state = ArgStates.QuoteAfterBackslashesInsideQuotes;
goto reinspectChar;
}
else
{
state = ArgStates.EmittingBackslashesInsideQuotes;
goto reinspectChar;
}
case ArgStates.QuoteAfterBackslashesInsideQuotes:
{
bool oddBackslashes = 1 == nBackslashes % 2;
nBackslashes /= 2;
if (oddBackslashes)
{
state = ArgStates.EmittingBackslashesInsideQuotes;
goto reinspectChar;
}
else
{
state = ArgStates.CheckForTwoQuotesAfterBackslashesInsideQuotes;
continue;
}
}
case ArgStates.EmittingBackslashesInsideQuotes:
argument.Append('\\', nBackslashes).Append(c);
nBackslashes = 0;
state = ArgStates.ScanningBackslashesInsideQuotes;
continue;
case ArgStates.CheckForTwoQuotesAfterBackslashesInsideQuotes:
if ('\"' == c)
{
state = ArgStates.EmittingBackslashesInsideQuotesStayingInsideQuotes;
goto reinspectChar;
}
else
{
state = ArgStates.EmittingBackslashesInsideQuotesGoingOutsideQuotes;
goto reinspectChar;
}
case ArgStates.EmittingBackslashesInsideQuotesGoingOutsideQuotes:
argument.Append('\\', nBackslashes);
nBackslashes = 0;
state = ArgStates.CheckForArgumentEndInsideQuotesGoingOutsideQuotes;
goto reinspectChar;
case ArgStates.EmittingBackslashesInsideQuotesStayingInsideQuotes:
argument.Append('\\', nBackslashes).Append('\"');
nBackslashes = 0;
state = ArgStates.ScanningBackslashesInsideQuotes;
continue;
case ArgStates.CheckForArgumentEndInsideQuotesGoingOutsideQuotes:
if (' ' == c || '\t' == c)
{
state = ArgStates.CopyToEnd;
goto reinspectChar;
}
else
{
argument.Append(c);
nBackslashes = 0;
state = ArgStates.ScanningBackslashesOutsideQuotes;
continue;
}
case ArgStates.CopyToEnd:
yield return c;
state = ArgStates.CopyToEnd;
continue;
}
}
// Flush final backslashes if any (there's only one state where this is necessary)
if (ArgStates.CheckForTwoQuotesAfterBackslashesInsideQuotes == state)
{
if (null != log) log.Add("flushing final backslashes", state, '¶', nBackslashes, argument);
argument.Append('\\', nBackslashes);
}
if (null != log) log.Add("final result", ArgStates.End, '¶', 0, argument);
}
