projects/dil: trunk/src/dil/Token.d comparison

comparison trunk/src/dil/Token.d @ 485:ea8c7459f1c4

Changed a lot of things in the Lexer. Newlines are tokenized now, instead of being treated as whitespace. Newline tokens store location info as well, which make quite a few functions unnecessary. Added a static method getLocation() which returns a Location instance for any given token. This will also be very useful for finding the location of AST nodes (through Node.begin,) which is needed for reporting parser and semantic errors and emitting documentation. Removed rescanNewlines(), LocState, getState(), restoreState(), evaluateHashLine() and updateErrorLoc(). Added isUnicodeNewlineChar(), isUnicodeNewline(), isNewline(), isNewlineEnd(), isEndOfLine(), scanNewline(), getLocation() and error_(). Replaced some clunky expressions with isascii(), isNewlineEnd(), isEndOfLine(), isUnicodeNewline(), isUnicodeNewlineChar(). Fix in scanNormalStringLiteral(): scanPostfix() must be before label Lreturn. Fixed Lexer unittest. Fix in parseDeclarationDefinitionsBlock(): 'token' should be 'begin'. Added method isMultiline() to Token and added documentation comments.: File name too long abort: file /home/aziz/dil/trunk/Changed a lot of things in the Lexer. Newlines are tokenized now, instead of being treated as whitespace. Newline tokens store location info as well, which make quite a few functions unnecessary. Added a static method getLocation() which returns a Location instance for any given token. This will also be very useful for finding the location of AST nodes (through Node.begin,) which is needed for reporting parser and semantic errors and emitting documentation. Removed rescanNewlines(), LocState, getState(), restoreState(), evaluateHashLine() and updateErrorLoc(). Added isUnicodeNewlineChar(), isUnicodeNewline(), isNewline(), isNewlineEnd(), isEndOfLine(), scanNewline(), getLocation() and error_(). Replaced some clunky expressions with isascii(), isNewlineEnd(), isEndOfLine(), isUnicodeNewline(), isUnicodeNewlineChar(). Fix in scanNormalStringLiteral(): scanPostfix() must be before label Lreturn. Fixed Lexer unittest. Fix in parseDeclarationDefinitionsBlock(): 'token' should be 'begin'. Added method isMultiline() to Token and added documentation comments.

author	Aziz K?ksal <aziz.koeksal@gmail.com>
date	Fri, 30 Nov 2007 20:17:29 +0100
parents	9c69615a4876
children	bccca748d745

comparison

equal deleted inserted replaced

-:265c0b655f18
+:ea8c7459f1c4
 module dil.Token;
 import common;
 import tango.stdc.stdlib : malloc, free;
 import tango.core.Exception;
-struct Position
-{
-size_t loc;
-size_t col;
-}
 enum TOK : ushort
 {
 Invalid,
 /// Flag for whitespace tokens that must be ignored in the parsing phase.
 Illegal  = 1 | Whitespace,
 Comment  = 2 | Whitespace,
 Shebang  = 3 | Whitespace,
 HashLine = 4 | Whitespace,
 Filespec = 5 | Whitespace,
-Empty    = 6,
+Newline  = 6 | Whitespace,
+Empty    = 7,
-Identifier = 7,
+Identifier = 8,
 String,
 CharLiteral, WCharLiteral, DCharLiteral,
 // Special tokens
 FILE,
 alias TOK.Abstract KeywordsBegin;
 alias TOK.With KeywordsEnd;
 struct Token
 {
-TOK type;
+TOK type; /// The type of the token.
-//   Position pos;
+/// Pointers to the next and previous tokens (doubly-linked list.)
 Token* next, prev;
 char* ws;    /// Start of whitespace characters before token. Null if no WS.
 char* start; /// Start of token in source text.
 char* end;   /// Points one past the end of token in source text.
 union
 {
+/// For newline tokens.
 struct
 {
-Token* line_num; // #line number
+char[] filePath;
-Token* line_filespec; // #line number filespec
+uint lineNum;
-}
+uint lineNum_hline;
+}
+/// For #line tokens.
 struct
 {
+Token* tokLineNum; /// #line number
+Token* tokLineFilespec; /// #line number filespec
+}
+/// For string tokens.
+struct
+{
 string str;
-char pf; /// Postfix 'c', 'w' or 'd'
+char pf; /// Postfix 'c', 'w', 'd' or 0 for none.
 version(D2)
 Token* tok_str; /// Points to the contents of a token string stored as a
 /// doubly linked list. The last token is always '}' or
 /// EOF in case end of source text is "q{" EOF.
 }
 static string toString(TOK tok)
 {
 return tokToString[tok];
 }
+/++
+Returns true if this is a token which can have newlines in it.
+These can be any string literal except for escape literals
+and block and nested comments.
++/
+bool isMultiline()
+{
+return type == TOK.String && start[0] != '\\' ||
+type == TOK.Comment && start[1] != '/';
+}
+/// Returns true if this is a keyword token.
 bool isKeyword()
 {
 return KeywordsBegin <= type && type <= KeywordsEnd;
 }
+/// Returns true if this is a whitespace token.
 bool isWhitespace()
 {
 return !!(type & TOK.Whitespace);
 }
+/// Returns true if this is a special token.
 bool isSpecialToken()
 {
 return *start == '_' && type != TOK.Identifier;
 }
 version(D2)
 {
+/// Returns true if this is a token string literal.
 bool isTokenStringLiteral()
 {
 return type == TOK.String && tok_str !is null;
 }
 }
 }
 void destructHashLineToken()
 {
 assert(type == TOK.HashLine);
-delete line_num;
+delete tokLineNum;
-delete line_filespec;
+delete tokLineFilespec;
 }
 version(D2)
 {
 void destructTokenStringLiteral()
 }
 }
 }
 }
-const string[] tokToString = [
+/// A table mapping each TOK to a string.
+private const string[] tokToString = [
 "Invalid",
 "Illegal",
 "Comment",
 "#! /shebang/",
 "#line",
 `"filespec"`,
+"Newline",
 "Empty",
 "Identifier",
 "String",
 "CharLiteral", "WCharLiteral", "DCharLiteral",
 "{",
 "}",
 ".", "..", "...",
-"Unordered",
+"!<>=", // Unordered
-"UorE",
+"!<>",  // UorE
-"UorG",
+"!<=",  // UorG
-"UorGorE",
+"!<",   // UorGorE
-"UorL",
+"!>=",  // UorL
-"UorLorE",
+"!>",   // UorLorE
-"LorEorG",
+"<>=",  // LorEorG
-"LorG",
+"<>",   // LorG
 "=", "==", "!=", "!",
 "<=", "<",
 ">=", ">",
 "<<=", "<<",

Mercurial > projects > dil

comparison trunk/src/dil/Token.d @ 485:ea8c7459f1c4