public static Compilation Compile(string source)
{
Compilation compilation = new Compilation();
compilation.Tokenization = Tokenizer.Tokenize(source, compilation);
compilation.Parsed = Parser.Parse(compilation.Tokenization, compilation);
return compilation;
}
public static List<ParserToken> Parse(List<Token> tokens, Compilation compilation)
{
List<ParserToken> parserTokens = new List<ParserToken>();
int index = 0;
while(true)
{
int max = -1;
ParserRuleId? maxId = null;
foreach(ParserRule rule in parserRules)
{
int matchLength = rule.Matches(tokens, index);
if (matchLength > max)
{
max = matchLength;
maxId = rule.Id;
}
}
if(max == -1)
{
++index;
if (index >= tokens.Count)
break;
compilation.ReportDiagnostic(tokens[index].Location, "Unparseable token.");
continue;
}
else
{
parserTokens.Add(
new ParserToken(
maxId.Value,
tokens.Skip(index).Take(max)
)
);
index += max;
if (index >= tokens.Count)
break;
}
}
return parserTokens;
}
public static List<Token> Tokenize(string source, Compilation compilation)
{
int index = 0;
Location location = new Location()
{
Column = 1,
Line = 1
};
List<Token> tokenization = new List<Token>();
while(true)
{
if (index >= source.Length)
break;
int max = -1;
TokenPattern maxPattern = default(TokenPattern);
foreach(TokenPattern pattern in tokens)
{
int current = pattern.Matches(source, index);
if (current > max)
{
max = current;
maxPattern = pattern;
}
}
if (maxPattern.Matches == null)
{
compilation.ReportDiagnostic(location, "No matching token.");
if(source[index] == '\n')
{
location.Column = 1;
++location.Line;
}
else
{
++location.Column;
}
index++;
continue;
}
tokenization.Add(
new Token(
maxPattern.Id,
source.Substring(index, max),
location
)
);
for(int i = 0; i < max; ++i)
{
if (source[index + i] == '\n')
{
location.Column = 1;
++location.Line;
}
else
{
++location.Column;
}
}
index += max;
}
return tokenization;
}
