static void readTokenFull (tokenInfo *const token, boolean include_newlines, vString *const repr)
{
int c;
int i;
token->type = TOKEN_UNDEFINED;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
getNextChar:
i = 0;
do
{
c = fileGetc ();
i++;
}
while (c == '\t' || c == ' ' ||
((c == '\r' || c == '\n') && ! include_newlines));
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
if (repr)
{
if (i > 1)
vStringPut (repr, ' ');
vStringPut (repr, c);
}
switch (c)
{
case EOF: token->type = TOKEN_EOF; break;
case '(': token->type = TOKEN_OPEN_PAREN; break;
case ')': token->type = TOKEN_CLOSE_PAREN; break;
case ';': token->type = TOKEN_SEMICOLON; break;
case ',': token->type = TOKEN_COMMA; break;
case '.': token->type = TOKEN_PERIOD; break;
case ':': token->type = TOKEN_COLON; break;
case '{': token->type = TOKEN_OPEN_CURLY; break;
case '}': token->type = TOKEN_CLOSE_CURLY; break;
case '=': token->type = TOKEN_EQUAL_SIGN; break;
case '[': token->type = TOKEN_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '+':
case '-':
{
int d = fileGetc ();
if (d == c) /* ++ or -- */
token->type = TOKEN_POSTFIX_OPERATOR;
else
{
fileUngetc (d);
token->type = TOKEN_BINARY_OPERATOR;
}
break;
}
case '*':
case '%':
case '?':
case '>':
case '<':
case '^':
case '|':
case '&':
token->type = TOKEN_BINARY_OPERATOR;
break;
case '\r':
case '\n':
/* This isn't strictly correct per the standard, but following the
* real rules means understanding all statements, and that's not
* what the parser currently does. What we do here is a guess, by
* avoiding inserting semicolons that would make the statement on
* the left invalid. Hopefully this should not have false negatives
* (e.g. should not miss insertion of a semicolon) but might have
* false positives (e.g. it will wrongfully emit a semicolon for the
* newline in "foo\n+bar").
* This should however be mostly harmless as we only deal with
* newlines in specific situations where we know a false positive
* wouldn't hurt too bad. */
switch (LastTokenType)
{
/* these cannot be the end of a statement, so hold the newline */
case TOKEN_EQUAL_SIGN:
case TOKEN_COLON:
case TOKEN_PERIOD:
case TOKEN_FORWARD_SLASH:
case TOKEN_BINARY_OPERATOR:
/* and these already end one, no need to duplicate it */
case TOKEN_SEMICOLON:
case TOKEN_COMMA:
case TOKEN_CLOSE_CURLY:
case TOKEN_OPEN_CURLY:
include_newlines = FALSE; /* no need to recheck */
goto getNextChar;
break;
default:
token->type = TOKEN_SEMICOLON;
}
break;
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
if (repr)
{
vStringCat (repr, token->string);
vStringPut (repr, c);
}
break;
case '\\':
c = fileGetc ();
if (c != '\\' && c != '"' && !isspace (c))
fileUngetc (c);
token->type = TOKEN_CHARACTER;
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '/':
{
int d = fileGetc ();
if ( (d != '*') && /* is this the start of a comment? */
(d != '/') ) /* is a one line comment? */
{
fileUngetc (d);
switch (LastTokenType)
{
case TOKEN_CHARACTER:
case TOKEN_IDENTIFIER:
case TOKEN_STRING:
case TOKEN_CLOSE_CURLY:
case TOKEN_CLOSE_PAREN:
case TOKEN_CLOSE_SQUARE:
token->type = TOKEN_FORWARD_SLASH;
break;
default:
token->type = TOKEN_REGEXP;
parseRegExp ();
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
}
}
else
{
if (repr) /* remove the / we added */
repr->buffer[--repr->length] = 0;
if (d == '*')
{
do
{
skipToCharacter ('*');
c = fileGetc ();
if (c == '/')
break;
else
fileUngetc (c);
} while (c != EOF && c != '\0');
goto getNextChar;
}
else if (d == '/') /* is this the start of a comment? */
{
skipToCharacter ('\n');
/* if we care about newlines, put it back so it is seen */
if (include_newlines)
fileUngetc ('\n');
goto getNextChar;
}
}
break;
}
case '#':
/* skip shebang in case of e.g. Node.js scripts */
if (token->lineNumber > 1)
token->type = TOKEN_UNDEFINED;
else if ((c = fileGetc ()) != '!')
{
fileUngetc (c);
token->type = TOKEN_UNDEFINED;
}
else
{
skipToCharacter ('\n');
goto getNextChar;
}
break;
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
token->keyword = analyzeToken (token->string);
if (isKeyword (token, KEYWORD_NONE))
token->type = TOKEN_IDENTIFIER;
else
token->type = TOKEN_KEYWORD;
if (repr && vStringLength (token->string) > 1)
vStringCatS (repr, vStringValue (token->string) + 1);
}
break;
}
LastTokenType = token->type;
}
static void readToken (tokenInfo *const token)
{
int c;
token->type = TOKEN_UNDEFINED;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
getNextChar:
do
{
c = fileGetc ();
}
while (c == '\t' || c == ' ' || c == '\n');
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
switch (c)
{
case EOF: longjmp (Exception, (int)ExceptionEOF); break;
case '(': token->type = TOKEN_OPEN_PAREN; break;
case ')': token->type = TOKEN_CLOSE_PAREN; break;
case ';': token->type = TOKEN_SEMICOLON; break;
case ',': token->type = TOKEN_COMMA; break;
case '.': token->type = TOKEN_PERIOD; break;
case ':': token->type = TOKEN_COLON; break;
case '{': token->type = TOKEN_OPEN_CURLY; break;
case '}': token->type = TOKEN_CLOSE_CURLY; break;
case '=': token->type = TOKEN_EQUAL_SIGN; break;
case '[': token->type = TOKEN_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '\\':
c = fileGetc ();
if (c != '\\' && c != '"' && !isspace (c))
fileUngetc (c);
token->type = TOKEN_CHARACTER;
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '/':
{
int d = fileGetc ();
if ( (d != '*') && /* is this the start of a comment? */
(d != '/') ) /* is a one line comment? */
{
fileUngetc (d);
switch (LastTokenType)
{
case TOKEN_CHARACTER:
case TOKEN_KEYWORD:
case TOKEN_IDENTIFIER:
case TOKEN_STRING:
case TOKEN_CLOSE_CURLY:
case TOKEN_CLOSE_PAREN:
case TOKEN_CLOSE_SQUARE:
token->type = TOKEN_FORWARD_SLASH;
break;
default:
token->type = TOKEN_REGEXP;
parseRegExp ();
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
}
}
else
{
if (d == '*')
{
do
{
skipToCharacter ('*');
c = fileGetc ();
if (c == '/')
break;
else
fileUngetc (c);
} while (c != EOF && c != '\0');
goto getNextChar;
}
else if (d == '/') /* is this the start of a comment? */
{
skipToCharacter ('\n');
goto getNextChar;
}
}
break;
}
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
token->keyword = analyzeToken (token->string);
if (isKeyword (token, KEYWORD_NONE))
token->type = TOKEN_IDENTIFIER;
else
token->type = TOKEN_KEYWORD;
}
break;
}
LastTokenType = token->type;
}
static void readTokenFull (tokenInfo *const token, boolean include_newlines, vString *const repr)
{
int c;
int i;
boolean newline_encountered = FALSE;
/* if we've got a token held back, emit it */
if (NextToken)
{
copyToken (token, NextToken, FALSE);
deleteToken (NextToken);
NextToken = NULL;
return;
}
token->type = TOKEN_UNDEFINED;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
getNextChar:
i = 0;
do
{
c = fileGetc ();
if (include_newlines && (c == '\r' || c == '\n'))
newline_encountered = TRUE;
i++;
}
while (c == '\t' || c == ' ' || c == '\r' || c == '\n');
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
if (repr)
{
if (i > 1)
vStringPut (repr, ' ');
vStringPut (repr, c);
}
switch (c)
{
case EOF: token->type = TOKEN_EOF; break;
case '(': token->type = TOKEN_OPEN_PAREN; break;
case ')': token->type = TOKEN_CLOSE_PAREN; break;
case ';': token->type = TOKEN_SEMICOLON; break;
case ',': token->type = TOKEN_COMMA; break;
case '.': token->type = TOKEN_PERIOD; break;
case ':': token->type = TOKEN_COLON; break;
case '{': token->type = TOKEN_OPEN_CURLY; break;
case '}': token->type = TOKEN_CLOSE_CURLY; break;
case '=': token->type = TOKEN_EQUAL_SIGN; break;
case '[': token->type = TOKEN_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '+':
case '-':
{
int d = fileGetc ();
if (d == c) /* ++ or -- */
token->type = TOKEN_POSTFIX_OPERATOR;
else
{
fileUngetc (d);
token->type = TOKEN_BINARY_OPERATOR;
}
break;
}
case '*':
case '%':
case '?':
case '>':
case '<':
case '^':
case '|':
case '&':
token->type = TOKEN_BINARY_OPERATOR;
break;
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
if (repr)
{
vStringCat (repr, token->string);
vStringPut (repr, c);
}
break;
case '`':
token->type = TOKEN_TEMPLATE_STRING;
parseTemplateString (token->string);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
if (repr)
{
vStringCat (repr, token->string);
vStringPut (repr, c);
}
break;
case '\\':
c = fileGetc ();
if (c != '\\' && c != '"' && !isspace (c))
fileUngetc (c);
token->type = TOKEN_CHARACTER;
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '/':
{
int d = fileGetc ();
if ( (d != '*') && /* is this the start of a comment? */
(d != '/') ) /* is a one line comment? */
{
fileUngetc (d);
switch (LastTokenType)
{
case TOKEN_CHARACTER:
case TOKEN_IDENTIFIER:
case TOKEN_STRING:
case TOKEN_TEMPLATE_STRING:
case TOKEN_CLOSE_CURLY:
case TOKEN_CLOSE_PAREN:
case TOKEN_CLOSE_SQUARE:
token->type = TOKEN_BINARY_OPERATOR;
break;
default:
token->type = TOKEN_REGEXP;
parseRegExp ();
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
}
}
else
{
if (repr) /* remove the / we added */
repr->buffer[--repr->length] = 0;
if (d == '*')
{
do
{
fileSkipToCharacter ('*');
c = fileGetc ();
if (c == '/')
break;
else
fileUngetc (c);
} while (c != EOF && c != '\0');
goto getNextChar;
}
else if (d == '/') /* is this the start of a comment? */
{
fileSkipToCharacter ('\n');
/* if we care about newlines, put it back so it is seen */
if (include_newlines)
fileUngetc ('\n');
goto getNextChar;
}
}
break;
}
case '#':
/* skip shebang in case of e.g. Node.js scripts */
if (token->lineNumber > 1)
token->type = TOKEN_UNDEFINED;
else if ((c = fileGetc ()) != '!')
{
fileUngetc (c);
token->type = TOKEN_UNDEFINED;
}
else
{
fileSkipToCharacter ('\n');
goto getNextChar;
}
break;
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
token->keyword = analyzeToken (token->string, Lang_js);
if (isKeyword (token, KEYWORD_NONE))
token->type = TOKEN_IDENTIFIER;
else
token->type = TOKEN_KEYWORD;
if (repr && vStringLength (token->string) > 1)
vStringCatS (repr, vStringValue (token->string) + 1);
}
break;
}
if (include_newlines && newline_encountered)
{
/* This isn't strictly correct per the standard, but following the
* real rules means understanding all statements, and that's not
* what the parser currently does. What we do here is a guess, by
* avoiding inserting semicolons that would make the statement on
* the left or right obviously invalid. Hopefully this should not
* have false negatives (e.g. should not miss insertion of a semicolon)
* but might have false positives (e.g. it will wrongfully emit a
* semicolon sometimes, i.e. for the newline in "foo\n(bar)").
* This should however be mostly harmless as we only deal with
* newlines in specific situations where we know a false positive
* wouldn't hurt too bad. */
/* these already end a statement, so no need to duplicate it */
#define IS_STMT_SEPARATOR(t) ((t) == TOKEN_SEMICOLON || \
(t) == TOKEN_EOF || \
(t) == TOKEN_COMMA || \
(t) == TOKEN_CLOSE_CURLY || \
(t) == TOKEN_OPEN_CURLY)
/* these cannot be the start or end of a statement */
#define IS_BINARY_OPERATOR(t) ((t) == TOKEN_EQUAL_SIGN || \
(t) == TOKEN_COLON || \
(t) == TOKEN_PERIOD || \
(t) == TOKEN_BINARY_OPERATOR)
if (! IS_STMT_SEPARATOR(LastTokenType) &&
! IS_STMT_SEPARATOR(token->type) &&
! IS_BINARY_OPERATOR(LastTokenType) &&
! IS_BINARY_OPERATOR(token->type) &&
/* these cannot be followed by a semicolon */
! (LastTokenType == TOKEN_OPEN_PAREN ||
LastTokenType == TOKEN_OPEN_SQUARE))
{
/* hold the token... */
Assert (NextToken == NULL);
NextToken = newToken ();
copyToken (NextToken, token, FALSE);
/* ...and emit a semicolon instead */
token->type = TOKEN_SEMICOLON;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
if (repr)
vStringPut (token->string, '\n');
}
#undef IS_STMT_SEPARATOR
#undef IS_BINARY_OPERATOR
}
LastTokenType = token->type;
}
