private TexFormula Parse(string value, ref int position, bool allowClosingDelimiter, string textStyle)
{
var formula = new TexFormula()
{
TextStyle = textStyle
};
var closedDelimiter = false;
while (position < value.Length && !(allowClosingDelimiter && closedDelimiter))
{
char ch = value[position];
if (IsWhiteSpace(ch))
{
position++;
}
else if (ch == escapeChar)
{
ProcessEscapeSequence(formula, value, ref position, allowClosingDelimiter, ref closedDelimiter);
}
else if (ch == leftGroupChar)
{
var groupValue = ReadGroup(formula, value, ref position, leftGroupChar, rightGroupChar);
var parsedGroup = Parse(groupValue, textStyle);
var innerGroupAtom = parsedGroup.RootAtom ?? new RowAtom();
var groupAtom = new TypedAtom(innerGroupAtom, TexAtomType.Ordinary, TexAtomType.Ordinary);
formula.Add(AttachScripts(formula, value, ref position, groupAtom));
}
else if (ch == rightGroupChar)
{
throw new TexParseException("Found a closing '" + rightGroupChar
+ "' without an opening '" + leftGroupChar + "'!");
}
else if (ch == superScriptChar || ch == subScriptChar || ch == primeChar)
{
if (position == 0)
{
throw new TexParseException("Every script needs a base: \""
+ superScriptChar + "\", \"" + subScriptChar + "\" and \""
+ primeChar + "\" can't be the first character!");
}
else
{
throw new TexParseException("Double scripts found! Try using more braces.");
}
}
else
{
var scriptsAtom = AttachScripts(formula, value, ref position,
ConvertCharacter(formula, value, ref position, ch));
formula.Add(scriptsAtom);
}
}
return(formula);
}
//public TexFormula Convert(System.Linq.Expressions.Expression value)
//{
// return new TexExpressionVisitor(value, this).Formula;
//}
public TexFormula Parse(string value)
{
var formula = new TexFormula();
var position = 0;
while (position < value.Length)
{
char ch = value[position];
if (IsWhiteSpace(ch))
{
position++;
}
else if (ch == escapeChar)
{
ProcessEscapeSequence(formula, value, ref position);
}
else if (ch == leftGroupChar)
{
formula.Add(AttachScripts(formula, value, ref position, Parse(ReadGroup(formula, value, ref position,
leftGroupChar, rightGroupChar)).RootAtom));
}
else if (ch == rightGroupChar)
{
throw new TexParseException("Found a closing '" + rightGroupChar
+ "' without an opening '" + leftGroupChar + "'!");
}
else if (ch == superScriptChar || ch == subScriptChar || ch == primeChar)
{
if (position == 0)
{
throw new TexParseException("Every script needs a base: \""
+ superScriptChar + "\", \"" + subScriptChar + "\" and \""
+ primeChar + "\" can't be the first character!");
}
else
{
throw new TexParseException("Double scripts found! Try using more braces.");
}
}
else
{
var scriptsAtom = AttachScripts(formula, value, ref position,
ConvertCharacter(formula, value, ref position, ch));
formula.Add(scriptsAtom);
}
}
return(formula);
}
private static TexFormula ConvertRawText(SourceSpan value, string textStyle)
{
var formula = new TexFormula {
Source = value, TextStyle = textStyle
};
var position = 0;
var initialPosition = position;
while (position < value.Length)
{
var ch = value[position];
var source = value.Segment(position, 1);
var atom = IsWhiteSpace(ch)
? (Atom) new SpaceAtom(source)
: new CharAtom(source, ch, textStyle);
position++;
formula.Add(atom, value.Segment(initialPosition, position - initialPosition));
}
return(formula);
}
private TexFormula Parse(
SourceSpan value,
ref int position,
bool allowClosingDelimiter,
string textStyle,
ICommandEnvironment environment)
{
var formula = new TexFormula {
Source = value, TextStyle = textStyle
};
var closedDelimiter = false;
var skipWhiteSpace = ShouldSkipWhiteSpace(textStyle);
var initialPosition = position;
while (position < value.Length && !(allowClosingDelimiter && closedDelimiter))
{
char ch = value[position];
var source = value.Segment(position, 1);
if (IsWhiteSpace(ch))
{
if (!skipWhiteSpace)
{
formula.Add(new SpaceAtom(source), source);
}
position++;
}
else if (ch == escapeChar)
{
ProcessEscapeSequence(
formula,
value,
ref position,
allowClosingDelimiter,
ref closedDelimiter,
environment);
}
else if (ch == leftGroupChar)
{
var groupValue = ReadElement(value, ref position);
var parsedGroup = Parse(groupValue, textStyle, environment.CreateChildEnvironment());
var innerGroupAtom = parsedGroup.RootAtom ?? new RowAtom(groupValue);
var groupAtom = new TypedAtom(
innerGroupAtom.Source,
innerGroupAtom,
TexAtomType.Ordinary,
TexAtomType.Ordinary);
var scriptsAtom = this.AttachScripts(formula, value, ref position, groupAtom, true, environment);
formula.Add(scriptsAtom, value.Segment(initialPosition, position - initialPosition));
}
else if (ch == rightGroupChar)
{
throw new TexParseException("Found a closing '" + rightGroupChar
+ "' without an opening '" + leftGroupChar + "'!");
}
else if (ch == superScriptChar || ch == subScriptChar || ch == primeChar)
{
if (position == 0)
{
throw new TexParseException("Every script needs a base: \""
+ superScriptChar + "\", \"" + subScriptChar + "\" and \""
+ primeChar + "\" can't be the first character!");
}
else
{
throw new TexParseException("Double scripts found! Try using more braces.");
}
}
else
{
var character = ConvertCharacter(formula, ref position, source, environment);
if (character != null)
{
var scriptsAtom = AttachScripts(
formula,
value,
ref position,
character,
skipWhiteSpace,
environment);
formula.Add(scriptsAtom, value.Segment(initialPosition, position));
}
}
}
return(formula);
}
private void ProcessEscapeSequence(TexFormula formula,
SourceSpan value,
ref int position,
bool allowClosingDelimiter,
ref bool closedDelimiter,
ICommandEnvironment environment)
{
var initialSrcPosition = position;
position++;
var start = position;
while (position < value.Length)
{
var ch = value[position];
var isEnd = position == value.Length - 1;
if (!char.IsLetter(ch) || isEnd)
{
// Escape sequence has ended
// Or it's a symbol. Assuming in this case it will only be a single char.
if ((isEnd && char.IsLetter(ch)) || position - start == 0)
{
position++;
}
break;
}
position++;
}
var commandSpan = value.Segment(start, position - start);
var command = commandSpan.ToString();
var formulaSource = new SourceSpan(value.Source, initialSrcPosition, commandSpan.End);
SymbolAtom symbolAtom = null;
if (SymbolAtom.TryGetAtom(commandSpan, out symbolAtom))
{
// Symbol was found.
if (symbolAtom.Type == TexAtomType.Accent)
{
var helper = new TexFormulaHelper(formula, formulaSource);
TexFormula accentFormula = ReadScript(formula, value, ref position, environment);
helper.AddAccent(accentFormula, symbolAtom.Name);
}
else if (symbolAtom.Type == TexAtomType.BigOperator)
{
var opAtom = new BigOperatorAtom(formulaSource, symbolAtom, null, null);
formula.Add(AttachScripts(formula, value, ref position, opAtom, true, environment), formulaSource);
}
else
{
formula.Add(
AttachScripts(formula, value, ref position, symbolAtom, true, environment), formulaSource);
}
}
else if (predefinedFormulas.TryGetValue(command, out var factory))
{
// Predefined formula was found.
var predefinedFormula = factory(formulaSource);
var atom = AttachScripts(formula, value, ref position, predefinedFormula.RootAtom, true, environment);
formula.Add(atom, formulaSource);
}
else if (command.Equals("nbsp"))
{
// Space was found.
var atom = AttachScripts(formula, value, ref position, new SpaceAtom(formulaSource), true, environment);
formula.Add(atom, formulaSource);
}
else if (textStyles.Contains(command))
{
// Text style was found.
SkipWhiteSpace(value, ref position);
var styledFormula = command == TexUtilities.TextStyleName
? ConvertRawText(ReadElement(value, ref position), command)
: Parse(ReadElement(value, ref position), command, environment.CreateChildEnvironment());
var source = value.Segment(start, position - start);
var atom = styledFormula.RootAtom ?? new NullAtom(source);
var commandAtom = AttachScripts(formula, value, ref position, atom, true, environment);
formula.Add(commandAtom, source);
}
else if (embeddedCommands.Contains(command) ||
environment.AvailableCommands.ContainsKey(command) ||
_commandRegistry.ContainsKey(command))
{
// Command was found.
var commandAtom = ProcessCommand(
formula,
value,
ref position,
command,
allowClosingDelimiter,
ref closedDelimiter,
environment);
if (commandAtom != null)
{
commandAtom = allowClosingDelimiter
? commandAtom
: AttachScripts(
formula,
value,
ref position,
commandAtom,
true,
environment);
var source = new SourceSpan(formulaSource.Source, formulaSource.Start, commandAtom.Source.End);
formula.Add(commandAtom, source);
}
}
else
{
// Escape sequence is invalid.
throw new TexParseException("Unknown symbol or command or predefined TeXFormula: '" + command + "'");
}
}
private void ProcessEscapeSequence(
TexFormula formula,
string value,
ref int position,
bool allowClosingDelimiter,
ref bool closedDelimiter)
{
var result = new StringBuilder();
position++;
while (position < value.Length)
{
var ch = value[position];
var isEnd = position == value.Length - 1;
if (!char.IsLetter(ch) || isEnd)
{
// Escape sequence has ended
// Or it's a symbol. Assuming in this case it will only be a single char.
if ((isEnd && char.IsLetter(ch)) || result.Length == 0)
{
result.Append(ch);
position++;
}
break;
}
result.Append(ch);
position++;
}
var command = result.ToString();
SymbolAtom symbolAtom = null;
TexFormula predefinedFormula = null;
if (SymbolAtom.TryGetAtom(command, out symbolAtom))
{
// Symbol was found.
if (symbolAtom.Type == TexAtomType.Accent)
{
TexFormulaHelper helper = new TexFormulaHelper(formula);
TexFormula accentFormula = ReadScript(formula, value, ref position);
helper.AddAccent(accentFormula, symbolAtom.Name);
}
else if (symbolAtom.Type == TexAtomType.BigOperator)
{
var opAtom = new BigOperatorAtom(symbolAtom, null, null);
formula.Add(AttachScripts(formula, value, ref position, opAtom));
}
else
{
formula.Add(AttachScripts(formula, value, ref position, symbolAtom));
}
}
else if (predefinedFormulas.TryGetValue(command, out predefinedFormula))
{
// Predefined formula was found.
formula.Add(AttachScripts(formula, value, ref position, predefinedFormula.RootAtom));
}
else if (command.Equals("nbsp"))
{
// Space was found.
formula.Add(AttachScripts(formula, value, ref position, new SpaceAtom()));
}
else if (textStyles.Contains(command))
{
// Text style was found.
SkipWhiteSpace(value, ref position);
var styledFormula = Parse(ReadGroup(formula, value, ref position, leftGroupChar, rightGroupChar), command);
if (styledFormula.RootAtom == null)
{
throw new TexParseException("Styled text can't be empty!");
}
formula.Add(AttachScripts(formula, value, ref position, styledFormula.RootAtom));
}
else if (commands.Contains(command))
{
// Command was found.
var commandAtom = ProcessCommand(
formula,
value,
ref position,
command,
allowClosingDelimiter,
ref closedDelimiter);
commandAtom = allowClosingDelimiter ?
commandAtom :
AttachScripts(
formula,
value,
ref position,
commandAtom);
formula.Add(commandAtom);
}
else
{
// Escape sequence is invalid.
throw new TexParseException("Unknown symbol or command or predefined TeXFormula: '" + command + "'");
}
}
private void ProcessEscapeSequence(TexFormula formula, string value, ref int position)
{
var result = new StringBuilder();
position++;
while (position < value.Length)
{
var ch = value[position];
var isEnd = position == value.Length - 1;
if (!char.IsLetter(ch) || isEnd)
{
// Escape sequence has ended.
if (isEnd)
{
result.Append(ch);
position++;
}
break;
}
result.Append(ch);
position++;
}
var command = result.ToString();
SymbolAtom symbolAtom = null;
TexFormula predefinedFormula = null;
try
{
symbolAtom = SymbolAtom.GetAtom(command);
}
catch (SymbolNotFoundException)
{
}
try
{
predefinedFormula = GetFormula(command);
}
catch (FormulaNotFoundException)
{
}
if (symbolAtom != null)
{
// Symbol was found.
formula.Add(AttachScripts(formula, value, ref position, symbolAtom));
}
else if (predefinedFormula != null)
{
// Predefined formula was found.
formula.Add(AttachScripts(formula, value, ref position, predefinedFormula.RootAtom));
}
else if (command.Equals("nbsp"))
{
// Space was found.
formula.Add(AttachScripts(formula, value, ref position, new SpaceAtom()));
}
else if (textStyles.Contains(command))
{
// Text style was found.
SkipWhiteSpace(value, ref position);
var styledFormula = Parse(ReadGroup(formula, value, ref position, leftGroupChar, rightGroupChar));
styledFormula.TextStyle = command;
formula.Add(AttachScripts(formula, value, ref position, styledFormula.RootAtom));
}
else if (commands.Contains(command))
{
// Command was found.
formula.Add(AttachScripts(formula, value, ref position, ProcessCommand(formula, value, ref position,
command)));
}
else
{
// Escape sequence is invalid.
throw new TexParseException("Unknown symbol or command or predefined TeXFormula: '" + command + "'");
}
}
//public TexFormula Convert(System.Linq.Expressions.Expression value)
//{
// return new TexExpressionVisitor(value, this).Formula;
//}
public TexFormula Parse(string value)
{
var formula = new TexFormula();
var position = 0;
while (position < value.Length)
{
char ch = value[position];
if (IsWhiteSpace(ch))
{
position++;
}
else if (ch == escapeChar)
{
ProcessEscapeSequence(formula, value, ref position);
}
else if (ch == leftGroupChar)
{
formula.Add(AttachScripts(formula, value, ref position, Parse(ReadGroup(formula, value, ref position,
leftGroupChar, rightGroupChar)).RootAtom));
}
else if (ch == rightGroupChar)
{
throw new TexParseException("Found a closing '" + rightGroupChar
+ "' without an opening '" + leftGroupChar + "'!");
}
else if (ch == superScriptChar || ch == subScriptChar || ch == primeChar)
{
if (position == 0)
throw new TexParseException("Every script needs a base: \""
+ superScriptChar + "\", \"" + subScriptChar + "\" and \""
+ primeChar + "\" can't be the first character!");
else
throw new TexParseException("Double scripts found! Try using more braces.");
}
else
{
var scriptsAtom = AttachScripts(formula, value, ref position,
ConvertCharacter(formula, value, ref position, ch));
formula.Add(scriptsAtom);
}
}
return formula;
}
private void ProcessEscapeSequence(
TexFormula formula,
SourceSpan value,
ref int position,
bool allowClosingDelimiter,
ref bool closedDelimiter)
{
var initialSrcPosition = position;
position++;
var start = position;
while (position < value.Length)
{
var ch = value[position];
var isEnd = position == value.Length - 1;
if (!char.IsLetter(ch) || isEnd)
{
// Escape sequence has ended
// Or it's a symbol. Assuming in this case it will only be a single char.
if ((isEnd && char.IsLetter(ch)) || position - start == 0)
{
position++;
}
break;
}
position++;
}
var commandSpan = value.Segment(start, position - start);
var command = commandSpan.ToString();
var formulaSource = new SourceSpan(value.Source, initialSrcPosition, commandSpan.End);
SymbolAtom symbolAtom = null;
if (SymbolAtom.TryGetAtom(commandSpan, out symbolAtom))
{
// Symbol was found.
if (symbolAtom.Type == TexAtomType.Accent)
{
var helper = new TexFormulaHelper(formula, formulaSource);
TexFormula accentFormula = ReadScript(formula, value, ref position);
helper.AddAccent(accentFormula, symbolAtom.Name);
}
else if (symbolAtom.Type == TexAtomType.BigOperator)
{
var opAtom = new BigOperatorAtom(formulaSource, symbolAtom, null, null);
formula.Add(this.AttachScripts(formula, value, ref position, opAtom), formulaSource);
}
else
{
formula.Add(this.AttachScripts(formula, value, ref position, symbolAtom), formulaSource);
}
}
else if (predefinedFormulas.TryGetValue(command, out var factory))
{
// Predefined formula was found.
var predefinedFormula = factory(formulaSource);
var atom = this.AttachScripts(formula, value, ref position, predefinedFormula.RootAtom);
formula.Add(atom, formulaSource);
}
else if (command.Equals("nbsp"))
{
// Space was found.
var atom = this.AttachScripts(formula, value, ref position, new SpaceAtom(formulaSource));
formula.Add(atom, formulaSource);
}
else if (textStyles.Contains(command))
{
// Text style was found.
this.SkipWhiteSpace(value, ref position);
var styledFormula = Parse(ReadGroup(formula, value, ref position, leftGroupChar, rightGroupChar), command);
if (styledFormula.RootAtom == null)
{
throw new TexParseException("Styled text can't be empty!");
}
var atom = this.AttachScripts(formula, value, ref position, styledFormula.RootAtom);
var source = new SourceSpan(formulaSource.Source, formulaSource.Start, position);
formula.Add(atom, source);
}
else if (commands.Contains(command))
{
// Command was found.
var commandAtom = this.ProcessCommand(
formula,
value,
ref position,
command,
allowClosingDelimiter,
ref closedDelimiter);
commandAtom = allowClosingDelimiter
? commandAtom
: AttachScripts(
formula,
value,
ref position,
commandAtom);
var source = new SourceSpan(formulaSource.Source, formulaSource.Start, commandAtom.Source.End);
formula.Add(commandAtom, source);
}
else
{
// this.AttachScripts(formula, value, ref position, );
position = initialSrcPosition;
var fallbackSpan = new SourceSpan(value.Source, start, commandSpan.End);
var escapeAtom =
this.AttachScripts(
formula,
value,
ref position,
new CharAtom(new SourceSpan(value.Source, position, 1), '\\'),
true); // this.ConvertCharacter(formula, ref position, new SourceSpan(value.Source,position,1));
var scriptsAtom = this.AttachScripts(
formula,
value,
ref position,
this.ConvertCharacter(formula, ref position, fallbackSpan),
true);
// formula.Add(scriptsAtom, formulaSource );
// Escape sequence is invalid.
//throw new TexParseException("Unknown symbol or command or predefined TeXFormula: '" + command + "'");
}
}
private void ProcessEscapeSequence(TexFormula formula, string value, ref int position)
{
var result = new StringBuilder();
position++;
while (position < value.Length)
{
var ch = value[position];
var isEnd = position == value.Length - 1;
if (!char.IsLetter(ch) || isEnd)
{
// Escape sequence has ended
// Or it's a symbol. Assuming in this case it will only be a single char.
if (isEnd || result.Length == 0)
{
result.Append(ch);
position++;
}
break;
}
result.Append(ch);
position++;
}
var command = result.ToString();
SymbolAtom symbolAtom = null;
TexFormula predefinedFormula = null;
try
{
symbolAtom = SymbolAtom.GetAtom(command);
}
catch (SymbolNotFoundException)
{
}
try
{
predefinedFormula = GetFormula(command);
}
catch (FormulaNotFoundException)
{
}
if (symbolAtom != null)
{
// Symbol was found.
if (symbolAtom.Type == TexAtomType.Accent)
{
TexFormulaHelper helper = new TexFormulaHelper(formula);
TexFormula accentFormula = ReadScript(formula, value, ref position);
helper.AddAccent(accentFormula, symbolAtom.Name);
}
else
{
formula.Add(AttachScripts(formula, value, ref position, symbolAtom));
}
}
else if (predefinedFormula != null)
{
// Predefined formula was found.
formula.Add(AttachScripts(formula, value, ref position, predefinedFormula.RootAtom));
}
else if (command.Equals("nbsp"))
{
// Space was found.
formula.Add(AttachScripts(formula, value, ref position, new SpaceAtom()));
}
else if (textStyles.Contains(command))
{
// Text style was found.
SkipWhiteSpace(value, ref position);
var styledFormula = Parse(ReadGroup(formula, value, ref position, leftGroupChar, rightGroupChar));
styledFormula.TextStyle = command;
formula.Add(AttachScripts(formula, value, ref position, styledFormula.RootAtom));
}
else if (commands.Contains(command))
{
// Command was found.
formula.Add(AttachScripts(formula, value, ref position, ProcessCommand(formula, value, ref position,
command)));
}
else
{
// Escape sequence is invalid.
throw new TexParseException("Unknown symbol or command or predefined TeXFormula: '" + command + "'");
}
}
