void QScriptCompletionTaskPrivate::completeScriptExpression()
{
int pos = cursorPosition;
if ((pos > 0) && contents.at(pos-1).isNumber()) {
// completion of numbers is pointless
emitFinished();
return;
}
while ((pos > 0) && isIdentChar(contents.at(pos-1)))
--pos;
int pos2 = cursorPosition - 1;
while ((pos2+1 < contents.size()) && isIdentChar(contents.at(pos2+1)))
++pos2;
QString ident = contents.mid(pos, pos2 - pos + 1);
position = pos;
QStringList path;
path.append(ident);
while ((pos > 0) && (contents.at(pos-1) == QLatin1Char('.'))) {
--pos;
pos2 = pos;
while ((pos > 0) && isIdentChar(contents.at(pos-1)))
--pos;
path.prepend(contents.mid(pos, pos2 - pos));
}
length = path.last().length();
type = QScriptCompletionTask::ScriptIdentifierCompletion;
QScriptDebuggerJob *job = new QScriptCompleteExpressionJob(frameIndex, path, this, commandScheduler);
jobScheduler->scheduleJob(job);
}
/*!
\reimp
*/
bool QScriptDebuggerCodeView::event(QEvent *e)
{
Q_D(QScriptDebuggerCodeView);
if (e->type() == QEvent::ToolTip) {
if (d->editor->executionLineNumber() == -1)
return false;
QHelpEvent *he = static_cast<QHelpEvent*>(e);
QPoint pt = he->pos();
pt.rx() -= d->editor->extraAreaWidth();
pt.ry() -= 8;
QTextCursor cursor = d->editor->cursorForPosition(pt);
QTextBlock block = cursor.block();
QString contents = block.text();
if (contents.isEmpty())
return false;
int linePosition = cursor.position() - block.position();
linePosition -= 3;
if (linePosition < 0)
linePosition = 0;
// ### generalize -- same as in completiontask
int pos = linePosition;
if ((pos > 0) && contents.at(pos-1).isNumber()) {
// tooltips for numbers is pointless
return false;
}
while ((pos > 0) && isIdentChar(contents.at(pos-1)))
--pos;
if ((pos > 0) && ((contents.at(pos-1) == QLatin1Char('\''))
|| (contents.at(pos-1) == QLatin1Char('\"')))) {
// ignore string literals
return false;
}
int pos2 = linePosition;
while ((pos2 < contents.size()-1) && isIdentChar(contents.at(pos2+1)))
++pos2;
QString ident = contents.mid(pos, pos2 - pos + 1);
QStringList path;
path.append(ident);
while ((pos > 0) && (contents.at(pos-1) == QLatin1Char('.'))) {
--pos;
pos2 = pos;
while ((pos > 0) && isIdentChar(contents.at(pos-1)))
--pos;
path.prepend(contents.mid(pos, pos2 - pos));
}
if (!path.isEmpty()) {
int lineNumber = cursor.blockNumber() + d->editor->baseLineNumber();
emit toolTipRequest(he->globalPos(), lineNumber, path);
}
}
return false;
}
/* Read a C identifier beginning with "firstChar" and places it into
* "name".
*/
static void parseIdentifier (vString *const string, const int firstChar)
{
int c = firstChar;
Assert (isIdentChar (c));
do
{
vStringPut (string, c);
c = fileGetc ();
} while (isIdentChar (c));
vStringTerminate (string);
fileUngetc (c); /* unget non-identifier character */
}
QvBool
QvName::isIdentStartChar(char c)
{
if (isdigit(c)) return FALSE;
return isIdentChar(c);
}
GvBool
GvName::isNodeNameStartChar(char c,GvBool vrml2)
{
if (vrml2) return(isIdentStartChar(c,vrml2));
// VRML 1.0
// ////// if (isdigit(c)) return FALSE;
return isIdentChar(c,vrml2);
}
static void parseIdentifier (vString *const string, const int firstChar)
{
int c = firstChar;
do
{
vStringPut (string, (char) c);
c = getcFromInputFile ();
} while (isIdentChar (c));
ungetcToInputFile (c);
}
static void parseIdentifier (vString *const string, const int firstChar)
{
int c = firstChar;
do
{
vStringPut (string, c);
c = getcFromInputFile ();
} while (isIdentChar (c));
vStringTerminate (string);
ungetcToInputFile (c); /* always unget, LF might add a semicolon */
}
/* Read a VHDL identifier beginning with "firstChar" and place it into "name".
*/
static void parseIdentifier (vString * const string, const int firstChar)
{
int c = firstChar;
Assert (isIdentChar1 (c));
do
{
vStringPut (string, c);
c = getcFromInputFile ();
} while (isIdentChar (c));
if (!isspace (c))
ungetcToInputFile (c); /* unget non-identifier character */
}
GvBool
GvName::isNodeNameChar(char c,GvBool vrml2)
{
if (vrml2) return(isIdentChar(c,vrml2));
// VRML 1.0
if (isalnum(c)) return TRUE;
// (0x00 .. 0x1F or 0x7F) iscntl(c) removed 7.10.96
if ((strchr(badCharacters, c) != NULL) || isspace(c))
return FALSE;
return TRUE;
}
void processLine(char *line)
{
char *p;
int quoteType;
static char fnm[300];
char *fname;
int nn;
p = line;
while(isspace(*p)) p++;
if (!*p) goto addToMaster;
// see if the first thing on the line is an include directive
if (*p=='.') p++;
if (strnicmp(p, "include", 7)==0 && !isIdentChar(p[7]))
{
p += 7;
// Capture the file name
while(isspace(*p)) p++;
if (*p=='"') { quoteType = '"'; p++; }
else if (*p=='<') { quoteType = '>'; p++; }
else quoteType = ' ';
nn = 0;
do {
fnm[nn] = *p;
p++; nn++;
if (quoteType==' ' && isspace(*p)) break;
else if (*p == quoteType) break;
else if (*p=='\n') break;
} while(nn < sizeof(fnm)/sizeof(char));
fnm[nn] = '\0';
fname = strdup(fnm);
processFile(fname,1);
free(fname);
return;
}
// Not an include directive, then just copy the line to the master buffer.
addToMaster:
strcpy(&masterFile[mfndx], line);
mfndx += strlen(line);
}
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;
}
GvBool
GvName::isIdentStartChar(char c,GvBool vrml2)
{
if (isdigit(c)) return FALSE;
return isIdentChar(c,vrml2);
}
static void findShTags (void)
{
vString *name = vStringNew ();
const unsigned char *line;
vString *hereDocDelimiter = NULL;
boolean hereDocIndented = FALSE;
boolean (* check_char)(int);
while ((line = fileReadLine ()) != NULL)
{
const unsigned char* cp = line;
shKind found_kind = K_NOTHING;
if (hereDocDelimiter)
{
if (hereDocIndented)
{
while (*cp == '\t')
cp++;
}
if (strcmp ((const char *) cp, vStringValue (hereDocDelimiter)) == 0)
{
vStringDelete (hereDocDelimiter);
hereDocDelimiter = NULL;
}
continue;
}
while (*cp != '\0')
{
/* jump over whitespace */
while (isspace ((int)*cp))
cp++;
/* jump over strings */
if (*cp == '"')
cp = skipDoubleString (cp);
else if (*cp == '\'')
cp = skipSingleString (cp);
/* jump over comments */
else if (*cp == '#')
break;
/* jump over here-documents */
else if (cp[0] == '<' && cp[1] == '<')
{
const unsigned char *start, *end;
boolean trimEscapeSequences = FALSE;
boolean quoted = FALSE;
cp += 2;
/* an optional "-" strips leading tabulations from the heredoc lines */
if (*cp != '-')
hereDocIndented = FALSE;
else
{
hereDocIndented = TRUE;
cp++;
}
while (isspace (*cp))
cp++;
start = end = cp;
/* the delimiter can be surrounded by quotes */
if (*cp == '"')
{
start++;
end = cp = skipDoubleString (cp);
/* we need not to worry about variable substitution, they
* don't happen in heredoc delimiter definition */
trimEscapeSequences = TRUE;
quoted = TRUE;
}
else if (*cp == '\'')
{
start++;
end = cp = skipSingleString (cp);
quoted = TRUE;
}
else
{
while (isIdentChar ((int) *cp))
cp++;
end = cp;
}
if (end > start || quoted)
{
/* The input may be broken as a shell script but we need to avoid
memory leaking. */
if (hereDocDelimiter)
vStringClear(hereDocDelimiter);
else
hereDocDelimiter = vStringNew ();
for (; end > start; start++)
{
if (trimEscapeSequences && *start == '\\')
start++;
vStringPut (hereDocDelimiter, *start);
}
}
}
if (strncmp ((const char*) cp, "function", (size_t) 8) == 0 &&
isspace ((int) cp [8]))
{
found_kind = K_FUNCTION;
cp += 8;
}
else if (strncmp ((const char*) cp, "alias", (size_t) 5) == 0 &&
isspace ((int) cp [5]))
{
found_kind = K_ALIAS;
cp += 5;
}
else if (cp [0] == '.'
&& isspace((int) cp [1]))
{
found_kind = K_SOURCE;
++cp;
}
else if (strncmp ((const char*) cp, "source", (size_t) 6) == 0
&& isspace((int) cp [6]))
{
found_kind = K_SOURCE;
cp += 6;
}
if (found_kind != K_NOTHING)
while (isspace ((int) *cp))
++cp;
// Get the name of the function, alias or file to be read by source
check_char = isIdentChar;
if (found_kind == K_SOURCE)
check_char = isFileChar;
if (! check_char ((int) *cp))
{
found_kind = K_NOTHING;
if (*cp != '\0')
++cp;
continue;
}
while (check_char ((int) *cp))
{
vStringPut (name, (int) *cp);
++cp;
}
vStringTerminate (name);
while (isspace ((int) *cp))
++cp;
if ((found_kind != K_SOURCE)
&& *cp == '(')
{
++cp;
while (isspace ((int) *cp))
++cp;
if (*cp == ')')
{
found_kind = K_FUNCTION;
++cp;
}
}
if (found_kind != K_NOTHING)
{
makeSimpleTag (name, ShKinds, found_kind);
found_kind = K_NOTHING;
}
vStringClear (name);
}
}
vStringDelete (name);
if (hereDocDelimiter)
vStringDelete (hereDocDelimiter);
}
/**
* evallabel replace "...%2i..." occurences in label with value of i
*/
static const char * evalLabel (const char* src, Tree visited, Tree localValEnv)
{
//std::cerr << "Eval Label : " << src;
int state = 0; // current state
std::string dst; // label once evaluated
std::string ident; // current identifier
std::string format; // current format
while (state != -1) {
char c = *src++;
if (state == 0) {
if (c == 0) {
state = -1;
} else if (c == '%') {
ident = "";
format = "";
state = 1;
} else {
dst+=c;
state = 0;
}
} else if (state == 1) {
if (c == 0) {
// fin et pas d'indentifiant, abandon
dst += '%';
dst += format;
state = -1;
} else if (isDigitChar(c)) {
format += c;
state = 1;
} else if (isIdentChar(c)) {
ident += c;
state = 2;
} else {
// caractere de ponctuation et pas d'indentifiant, abandon
dst += '%';
dst += format;
src--;
state = 0;
}
} else if (state == 2) {
if (isIdentChar(c)) {
ident += c;
state = 2;
} else {
writeIdentValue(dst, format, ident, visited, localValEnv);
src--;
state = 0;
}
} else {
stringstream error;
error << "internal error in evallabel : undefined state " << state << std::endl;
throw faustexception(error.str());
}
}
const char* val = strdup(dst.c_str());
//std::cerr << " ===> " << val << std::endl;
return val;
}
static void readToken (tokenInfo *const token)
{
int c;
token->type = TOKEN_UNDEFINED;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
getNextChar:
if (! InPhp)
{
c = findPhpStart ();
if (c != EOF)
InPhp = TRUE;
}
else
c = getcFromInputFile ();
c = skipWhitespaces (c);
token->lineNumber = getInputLineNumber ();
token->filePosition = getInputFilePosition ();
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_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '&': token->type = TOKEN_AMPERSAND; break;
case '\\': token->type = TOKEN_BACKSLASH; break;
case '=':
{
int d = getcFromInputFile ();
if (d == '=' || d == '>')
token->type = TOKEN_OPERATOR;
else
{
ungetcToInputFile (d);
token->type = TOKEN_EQUAL_SIGN;
}
break;
}
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getInputLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '<':
{
int d = getcFromInputFile ();
if (d == '/')
{
/* </script[:white:]*> */
if (tolower ((d = getcFromInputFile ())) == 's' &&
tolower ((d = getcFromInputFile ())) == 'c' &&
tolower ((d = getcFromInputFile ())) == 'r' &&
tolower ((d = getcFromInputFile ())) == 'i' &&
tolower ((d = getcFromInputFile ())) == 'p' &&
tolower ((d = getcFromInputFile ())) == 't' &&
(d = skipWhitespaces (getcFromInputFile ())) == '>')
{
InPhp = FALSE;
goto getNextChar;
}
else
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
}
else if (d == '<' && (d = getcFromInputFile ()) == '<')
{
token->type = TOKEN_STRING;
parseHeredoc (token->string);
}
else
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
break;
}
case '#': /* comment */
skipSingleComment ();
goto getNextChar;
break;
case '+':
case '-':
case '*':
case '%':
{
int d = getcFromInputFile ();
if (d != '=' && ! (c == '-' && d == '>'))
ungetcToInputFile (d);
token->type = TOKEN_OPERATOR;
break;
}
case '/': /* division or comment start */
{
int d = getcFromInputFile ();
if (d == '/') /* single-line comment */
{
skipSingleComment ();
goto getNextChar;
}
else if (d == '*')
{
do
{
c = skipToCharacter ('*');
if (c != EOF)
{
c = getcFromInputFile ();
if (c == '/')
break;
else
ungetcToInputFile (c);
}
} while (c != EOF && c != '\0');
goto getNextChar;
}
else
{
if (d != '=')
ungetcToInputFile (d);
token->type = TOKEN_OPERATOR;
}
break;
}
case '$': /* variable start */
{
int d = getcFromInputFile ();
if (! isIdentChar (d))
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
else
{
parseIdentifier (token->string, d);
token->type = TOKEN_VARIABLE;
}
break;
}
case '?': /* maybe the end of the PHP chunk */
{
int d = getcFromInputFile ();
if (d == '>')
{
InPhp = FALSE;
goto getNextChar;
}
else
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
break;
}
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
token->keyword = analyzeToken (token->string, getInputLanguage ());
if (token->keyword == KEYWORD_NONE)
token->type = TOKEN_IDENTIFIER;
else
token->type = TOKEN_KEYWORD;
}
break;
}
if (token->type == TOKEN_SEMICOLON ||
token->type == TOKEN_OPEN_CURLY ||
token->type == TOKEN_CLOSE_CURLY)
{
/* reset current statement details on statement end, and when entering
* a deeper scope.
* it is a bit ugly to do this in readToken(), but it makes everything
* a lot simpler. */
CurrentStatement.access = ACCESS_UNDEFINED;
CurrentStatement.impl = IMPL_UNDEFINED;
}
}
/* reads an HereDoc or a NowDoc (the part after the <<<).
* <<<[ \t]*(ID|'ID'|"ID")
* ...
* ID;?
*
* note that:
* 1) starting ID must be immediately followed by a newline;
* 2) closing ID is the same as opening one;
* 3) closing ID must be immediately followed by a newline or a semicolon
* then a newline.
*
* Example of a *single* valid heredoc:
* <<< FOO
* something
* something else
* FOO this is not an end
* FOO; this isn't either
* FOO; # neither this is
* FOO;
* # previous line was the end, but the semicolon wasn't required
*/
static void parseHeredoc (vString *const string)
{
int c;
unsigned int len;
char delimiter[64]; /* arbitrary limit, but more is crazy anyway */
int quote = 0;
do
{
c = getcFromInputFile ();
}
while (c == ' ' || c == '\t');
if (c == '\'' || c == '"')
{
quote = c;
c = getcFromInputFile ();
}
for (len = 0; len < ARRAY_SIZE (delimiter) - 1; len++)
{
if (! isIdentChar (c))
break;
delimiter[len] = (char) c;
c = getcFromInputFile ();
}
delimiter[len] = 0;
if (len == 0) /* no delimiter, give up */
goto error;
if (quote)
{
if (c != quote) /* no closing quote for quoted identifier, give up */
goto error;
c = getcFromInputFile ();
}
if (c != '\r' && c != '\n') /* missing newline, give up */
goto error;
do
{
c = getcFromInputFile ();
if (c != '\r' && c != '\n')
vStringPut (string, (char) c);
else
{
/* new line, check for a delimiter right after */
int nl = c;
int extra = EOF;
c = getcFromInputFile ();
for (len = 0; c != 0 && (c - delimiter[len]) == 0; len++)
c = getcFromInputFile ();
if (delimiter[len] != 0)
ungetcToInputFile (c);
else
{
/* line start matched the delimiter, now check whether there
* is anything after it */
if (c == '\r' || c == '\n')
{
ungetcToInputFile (c);
break;
}
else if (c == ';')
{
int d = getcFromInputFile ();
if (d == '\r' || d == '\n')
{
/* put back the semicolon since it's not part of the
* string. we can't put back the newline, but it's a
* whitespace character nobody cares about it anyway */
ungetcToInputFile (';');
break;
}
else
{
/* put semicolon in the string and continue */
extra = ';';
ungetcToInputFile (d);
}
}
}
/* if we are here it wasn't a delimiter, so put everything in the
* string */
vStringPut (string, (char) nl);
vStringNCatS (string, delimiter, len);
if (extra != EOF)
vStringPut (string, (char) extra);
}
}
while (c != EOF);
vStringTerminate (string);
return;
error:
ungetcToInputFile (c);
}
/*
* Copies the characters forming an identifier from *cp into
* name, leaving *cp pointing to the character after the identifier.
*/
static rubyKind parseIdentifier (
const unsigned char** cp, vString* name, rubyKind kind)
{
/* Method names are slightly different to class and variable names.
* A method name may optionally end with a question mark, exclamation
* point or equals sign. These are all part of the name.
* A method name may also contain a period if it's a singleton method.
*/
boolean had_sep = FALSE;
const char* also_ok;
if (kind == K_METHOD)
{
also_ok = ".?!=";
}
else if (kind == K_SINGLETON)
{
also_ok = "?!=";
}
else
{
also_ok = "";
}
skipWhitespace (cp);
/* Check for an anonymous (singleton) class such as "class << HTTP". */
if (kind == K_CLASS && **cp == '<' && *(*cp + 1) == '<')
{
return K_UNDEFINED;
}
/* Check for operators such as "def []=(key, val)". */
if (kind == K_METHOD || kind == K_SINGLETON)
{
if (parseRubyOperator (name, cp))
{
return kind;
}
}
/* Copy the identifier into 'name'. */
while (**cp != 0 && (**cp == ':' || isIdentChar (**cp) || charIsIn (**cp, also_ok)))
{
char last_char = **cp;
if (last_char == ':')
had_sep = TRUE;
else
{
if (had_sep)
{
vStringPut (name, SCOPE_SEPARATOR);
had_sep = FALSE;
}
vStringPut (name, last_char);
}
++*cp;
if (kind == K_METHOD)
{
/* Recognize singleton methods. */
if (last_char == '.')
{
vStringTerminate (name);
vStringClear (name);
return parseIdentifier (cp, name, K_SINGLETON);
}
}
if (kind == K_METHOD || kind == K_SINGLETON)
{
/* Recognize characters which mark the end of a method name. */
if (charIsIn (last_char, "?!="))
{
break;
}
}
}
return kind;
}
static void readToken (tokenInfo *const token)
{
int c;
token->type = TOKEN_UNDEFINED;
vStringClear (token->string);
getNextChar:
c = getcFromInputFile ();
c = skipWhitespaces (c);
token->lineNumber = getInputLineNumber ();
token->filePosition = getInputFilePosition ();
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_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '=': token->type = TOKEN_EQUAL_SIGN; break;
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getInputLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '<':
{
int d = getcFromInputFile ();
if (d == '#')
{
/* <# ... #> multiline comment */
do
{
c = skipToCharacterInInputFile ('#');
if (c != EOF)
{
c = getcFromInputFile ();
if (c == '>')
break;
else
ungetcToInputFile (c);
}
} while (c != EOF);
goto getNextChar;
}
else
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
break;
}
case '#': /* comment */
skipSingleComment ();
goto getNextChar;
break;
case '+':
case '-':
case '*':
case '/':
case '%':
{
int d = getcFromInputFile ();
if (d != '=')
ungetcToInputFile (d);
token->type = TOKEN_OPERATOR;
break;
}
case '$': /* variable start */
{
int d = getcFromInputFile ();
if (! isIdentChar (d))
{
ungetcToInputFile (d);
token->type = TOKEN_UNDEFINED;
}
else
{
parseIdentifier (token->string, d);
token->type = TOKEN_VARIABLE;
}
break;
}
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
if (isTokenFunction (token->string))
token->type = TOKEN_KEYWORD;
else
token->type = TOKEN_IDENTIFIER;
}
break;
}
}
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;
}
/*
* Extract a list of features from a table.
*/
static List *
analyzeTable(const char * table)
{
static char fileName[MAXSTRING];
char ** resolved;
List * features = NULL;
FileInfo info;
int k;
resolved = resolveTable(table, NULL);
if (resolved == NULL)
{
logMessage(LOG_ERROR, "Cannot resolve table '%s'", table);
return NULL;
}
sprintf(fileName, "%s", *resolved);
k = 0;
for (k = 0; resolved[k]; k++)
free(resolved[k]);
free(resolved);
if (k > 1)
{
logMessage(LOG_ERROR, "Table '%s' resolves to more than one file", table);
return NULL;
}
info.fileName = fileName;
info.encoding = noEncoding;
info.status = 0;
info.lineNumber = 0;
if ((info.in = fopen(info.fileName, "rb")))
{
while (getALine(&info))
{
if (info.linelen == 0);
else if (info.line[0] == '#')
{
if (info.linelen >= 2 && info.line[1] == '+')
{
widechar * key = NULL;
widechar * val = NULL;
size_t keySize = 0;
size_t valSize = 0;
info.linepos = 2;
if (info.linepos < info.linelen && isIdentChar(info.line[info.linepos]))
{
key = &info.line[info.linepos];
keySize = 1;
info.linepos++;
while (info.linepos < info.linelen && isIdentChar(info.line[info.linepos]))
{
keySize++;
info.linepos++;
}
if (info.linepos < info.linelen && info.line[info.linepos] == ':')
{
info.linepos++;
while (info.linepos < info.linelen
&& (info.line[info.linepos] == ' ' || info.line[info.linepos] == '\t'))
info.linepos++;
if (info.linepos < info.linelen && isIdentChar(info.line[info.linepos]))
{
val = &info.line[info.linepos];
valSize = 1;
info.linepos++;
while (info.linepos < info.linelen && isIdentChar(info.line[info.linepos]))
{
valSize++;
info.linepos++;
}
}
else
goto compile_error;
}
if (info.linepos == info.linelen)
{
char * k = widestrToStr(key, keySize);
char * v = val ? widestrToStr(val, valSize) : NULL;
Feature f = feature_new(k, v);
logMessage(LOG_DEBUG, "Table has feature '%s:%s'", f.key, f.val);
features = list_conj(features, memcpy(malloc(sizeof(f)), &f, sizeof(f)),
NULL, NULL, (void (*)(void *))feature_free);
free(k);
free(v);
}
else
goto compile_error;
}
else
goto compile_error;
}
}
else
break;
}
fclose(info.in);
}
else
logMessage (LOG_ERROR, "Cannot open table '%s'", info.fileName);
return list_sort(features, (int (*)(void *, void *))cmpKeys);
compile_error:
if (info.linepos < info.linelen)
logMessage(LOG_ERROR, "Unexpected character '%c' on line %d, column %d",
info.line[info.linepos],
info.lineNumber,
info.linepos);
else
logMessage(LOG_ERROR, "Unexpected newline on line %d", info.lineNumber);
list_free(features);
return NULL;
}
/*
* Parse a table query into a list of features. Features defined first get a
* higher importance.
*/
static List *
parseQuery(const char * query)
{
List * features = NULL;
const char * key = NULL;
const char * val = NULL;
size_t keySize = 0;
size_t valSize = 0;
const char * c;
int pos = 0;
while (1)
{
c = &query[pos++];
if (*c == ' ' || *c == '\t' || *c == '\n' | *c == '\0')
{
if (key)
{
char * k = strndup(key, keySize);
char * v = val ? strndup(val, valSize) : NULL;
FeatureWithImportance f = { feature_new(k, v), 0 };
logMessage(LOG_DEBUG, "Query has feature '%s:%s'", f.feature.key, f.feature.val);
features = list_conj(features, memcpy(malloc(sizeof(f)), &f, sizeof(f)),
NULL, NULL, (void (*)(void *))feature_free);
free(k);
free(v);
key = val = NULL;
keySize = valSize = 0;
}
if (*c == '\0')
break;
}
else if (*c == ':')
{
if (!key || val)
goto compile_error;
else
{
c = &query[pos++];
if (isIdentChar(*c))
{
val = c;
valSize = 1;
}
else
goto compile_error;
}
}
else if (isIdentChar(*c))
{
if (val)
valSize++;
else if (key)
keySize++;
else
{
key = c;
keySize = 1;
}
}
else
goto compile_error;
}
int k = 1;
List * l;
for (l = features; l; l = l->tail)
{
FeatureWithImportance * f = l->head;
f->importance = k++;
}
return list_sort(features, (int (*)(void *, void *))cmpKeys);
compile_error:
logMessage(LOG_ERROR, "Unexpected character '%c' at position %d", c, pos);
list_free(features);
return NULL;
}
static void readTokenFull (tokenInfo *const token,
boolean includeStringRepr)
{
int c;
token->type = TOKEN_UNDEFINED;
vStringClear (token->string);
do
c = getcFromInputFile ();
while (c == '\t' || c == ' ' || c == '\r' || c == '\n');
token->lineNumber = getInputLineNumber ();
token->filePosition = getInputFilePosition ();
switch (c)
{
case EOF: token->type = TOKEN_EOF; break;
case '[': token->type = TOKEN_OPEN_SQUARE; break;
case ']': token->type = TOKEN_CLOSE_SQUARE; break;
case '{': token->type = TOKEN_OPEN_CURLY; break;
case '}': token->type = TOKEN_CLOSE_CURLY; break;
case ':': token->type = TOKEN_COLON; break;
case ',': token->type = TOKEN_COMMA; break;
case '"':
{
boolean escaped = FALSE;
token->type = TOKEN_STRING;
while (TRUE)
{
c = getcFromInputFile ();
/* we don't handle unicode escapes but they are safe */
if (escaped)
escaped = FALSE;
else if (c == '\\')
escaped = TRUE;
else if (c >= 0x00 && c <= 0x1F)
break; /* break on invalid, unescaped, control characters */
else if (c == '"' || c == EOF)
break;
if (includeStringRepr)
vStringPut (token->string, c);
}
vStringTerminate (token->string);
break;
}
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
do
{
vStringPut (token->string, c);
c = getcFromInputFile ();
}
while (c != EOF && isIdentChar (c));
vStringTerminate (token->string);
ungetcToInputFile (c);
switch (lookupKeyword (vStringValue (token->string), Lang_json))
{
case KEYWORD_true: token->type = TOKEN_TRUE; break;
case KEYWORD_false: token->type = TOKEN_FALSE; break;
case KEYWORD_null: token->type = TOKEN_NULL; break;
default: token->type = TOKEN_NUMBER; break;
}
}
break;
}
}
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;
}
static boolean notIdentChar (int c)
{
return ! isIdentChar (c);
}
static void readToken (tokenInfo *const token)
{
int c;
token->type = TOKEN_UNDEFINED;
token->keyword = KEYWORD_NONE;
vStringClear (token->string);
getNextChar:
do
{
c = fileGetc ();
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
}
while (c == '\t' || c == ' ' || c == '\n');
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 '?': token->type = TOKEN_QUESTION_MARK; break;
case '*': token->type = TOKEN_STAR; break;
case '\'':
case '"':
token->type = TOKEN_STRING;
parseString (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
break;
case '\\':
/*
* All Tex tags start with a backslash.
* Check if the next character is an alpha character
* else it is not a potential tex tag.
*/
c = fileGetc ();
if (! isalpha (c))
fileUngetc (c);
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;
}
break;
case '%':
fileSkipToCharacter ('\n'); /* % are single line comments */
goto getNextChar;
break;
default:
if (! isIdentChar (c))
token->type = TOKEN_UNDEFINED;
else
{
parseIdentifier (token->string, c);
token->lineNumber = getSourceLineNumber ();
token->filePosition = getInputFilePosition ();
token->type = TOKEN_IDENTIFIER;
}
break;
}
}
static void findRubyTags (void)
{
const unsigned char *line;
boolean inMultiLineComment = FALSE;
nesting = nestingLevelsNew (0);
/* FIXME: this whole scheme is wrong, because Ruby isn't line-based.
* You could perfectly well write:
*
* def
* method
* puts("hello")
* end
*
* if you wished, and this function would fail to recognize anything.
*/
while ((line = readLineFromInputFile ()) != NULL)
{
const unsigned char *cp = line;
/* if we expect a separator after a while, for, or until statement
* separators are "do", ";" or newline */
boolean expect_separator = FALSE;
if (canMatch (&cp, "=begin", isWhitespace))
{
inMultiLineComment = TRUE;
continue;
}
if (canMatch (&cp, "=end", isWhitespace))
{
inMultiLineComment = FALSE;
continue;
}
if (inMultiLineComment)
continue;
skipWhitespace (&cp);
/* Avoid mistakenly starting a scope for modifiers such as
*
* return if <exp>
*
* FIXME: this is fooled by code such as
*
* result = if <exp>
* <a>
* else
* <b>
* end
*
* FIXME: we're also fooled if someone does something heinous such as
*
* puts("hello") \
* unless <exp>
*/
if (canMatchKeyword (&cp, "for") ||
canMatchKeyword (&cp, "until") ||
canMatchKeyword (&cp, "while"))
{
expect_separator = TRUE;
enterUnnamedScope ();
}
else if (canMatchKeyword (&cp, "case") ||
canMatchKeyword (&cp, "if") ||
canMatchKeyword (&cp, "unless"))
{
enterUnnamedScope ();
}
/*
* "module M", "class C" and "def m" should only be at the beginning
* of a line.
*/
if (canMatchKeyword (&cp, "module"))
{
readAndEmitTag (&cp, K_MODULE);
}
else if (canMatchKeyword (&cp, "class"))
{
readAndEmitTag (&cp, K_CLASS);
}
else if (canMatchKeyword (&cp, "def"))
{
rubyKind kind = K_METHOD;
NestingLevel *nl = nestingLevelsGetCurrent (nesting);
tagEntryInfo *e = getEntryOfNestingLevel (nl);
/* if the def is inside an unnamed scope at the class level, assume
* it's from a singleton from a construct like this:
*
* class C
* class << self
* def singleton
* ...
* end
* end
* end
*/
if (e && (e->kind - RubyKinds) == K_CLASS && strlen (e->name) == 0)
kind = K_SINGLETON;
readAndEmitTag (&cp, kind);
}
while (*cp != '\0')
{
/* FIXME: we don't cope with here documents,
* or regular expression literals, or ... you get the idea.
* Hopefully, the restriction above that insists on seeing
* definitions at the starts of lines should keep us out of
* mischief.
*/
if (inMultiLineComment || isspace (*cp))
{
++cp;
}
else if (*cp == '#')
{
/* FIXME: this is wrong, but there *probably* won't be a
* definition after an interpolated string (where # doesn't
* mean 'comment').
*/
break;
}
else if (canMatchKeyword (&cp, "begin"))
{
enterUnnamedScope ();
}
else if (canMatchKeyword (&cp, "do"))
{
if (! expect_separator)
enterUnnamedScope ();
else
expect_separator = FALSE;
}
else if (canMatchKeyword (&cp, "end") && nesting->n > 0)
{
/* Leave the most recent scope. */
nestingLevelsPop (nesting);
}
else if (*cp == '"')
{
/* Skip string literals.
* FIXME: should cope with escapes and interpolation.
*/
do {
++cp;
} while (*cp != 0 && *cp != '"');
if (*cp == '"')
cp++; /* skip the last found '"' */
}
else if (*cp == ';')
{
++cp;
expect_separator = FALSE;
}
else if (*cp != '\0')
{
do
++cp;
while (isIdentChar (*cp));
}
}
}
nestingLevelsFree (nesting);
}
void processSegments()
{
char *pinptr;
int segment;
if (verbose)
printf("Processing segments.\r\n");
inptr = &masterFile[0];
pinptr = inptr;
codendx = 0;
datandx = 0;
rodatandx = 0;
tlsndx = 0;
bssndx = 0;
memset(codebuf,0,sizeof(codebuf));
memset(databuf,0,sizeof(databuf));
memset(rodatabuf,0,sizeof(rodatabuf));
memset(tlsbuf,0,sizeof(tlsbuf));
memset(bssbuf,0,sizeof(bssbuf));
while (*inptr) {
SkipSpaces();
if (*inptr=='.') inptr++;
if ((strnicmp(inptr,"code",4)==0) && !isIdentChar(inptr[4])) {
segment = codeseg;
}
else if ((strnicmp(inptr,"data",4)==0) && !isIdentChar(inptr[4])) {
segment = dataseg;
}
else if ((strnicmp(inptr,"rodata",6)==0) && !isIdentChar(inptr[6])) {
segment = rodataseg;
}
else if ((strnicmp(inptr,"tls",3)==0) && !isIdentChar(inptr[3])) {
segment = tlsseg;
}
else if ((strnicmp(inptr,"bss",3)==0) && !isIdentChar(inptr[3])) {
segment = bssseg;
}
ScanToEOL();
inptr++;
switch(segment) {
case codeseg:
strncpy(&codebuf[codendx], pinptr, inptr-pinptr);
codendx += inptr-pinptr;
break;
case dataseg:
strncpy(&databuf[datandx], pinptr, inptr-pinptr);
datandx += inptr-pinptr;
break;
case rodataseg:
strncpy(&rodatabuf[rodatandx], pinptr, inptr-pinptr);
rodatandx += inptr-pinptr;
break;
case tlsseg:
strncpy(&tlsbuf[tlsndx], pinptr, inptr-pinptr);
tlsndx += inptr-pinptr;
break;
case bssseg:
strncpy(&bssbuf[bssndx], pinptr, inptr-pinptr);
bssndx += inptr-pinptr;
break;
}
pinptr = inptr;
}
memset(masterFile,0,sizeof(masterFile));
strcpy(masterFile, codebuf);
strcat(masterFile, rodatabuf);
strcat(masterFile, databuf);
strcat(masterFile, bssbuf);
strcat(masterFile, tlsbuf);
if (debug) {
FILE *fp;
fp = fopen("a64-segments.asm", "w");
if (fp) {
fwrite(masterFile, 1, strlen(masterFile), fp);
fclose(fp);
}
}
}
