/*
* _peek:
* @self: A valid %GSDLTokenizer.
* @result: (out): The location to store the resulting %gunichar.
* @err: (out) (allow-none): Location to store any %GError, or %NULL.
*
* Looks at the next UTF-8 character from the input. Will return %EOF at the end of the input.
*
* Returns: Whether the peek succeeded. This should always be checked, as _peek() may or may not
* have to read from the input.
*/
static bool _peek(GSDLTokenizer *self, gunichar *result, GError **err) {
if (!self->peek_avail) {
if (self->stringbuf) {
if (*self->stringbuf) {
self->peeked = *(self->stringbuf++);
} else {
self->stringbuf = NULL;
self->peeked = EOF;
}
self->peek_avail = true;
} else {
if (self->channel == NULL) return false;
switch (g_io_channel_read_unichar(self->channel, &(self->peeked), err)) {
case G_IO_STATUS_ERROR:
self->channel = NULL;
self->peek_avail = false;
return false;
case G_IO_STATUS_EOF:
self->peeked = EOF;
g_io_channel_shutdown(self->channel, FALSE, NULL);
g_io_channel_unref(self->channel);
self->channel = NULL;
case G_IO_STATUS_AGAIN:
case G_IO_STATUS_NORMAL:
default:
self->peek_avail = true;
}
}
}
*result = self->peeked;
return true;
}
static VALUE
rg_readchar(VALUE self)
{
gunichar thechar;
GError* err = NULL;
GIOStatus status = g_io_channel_read_unichar(_SELF(self), &thechar, &err);
ioc_error(status, err);
return UINT2NUM(thechar);
}
static VALUE
rg_getc(VALUE self)
{
gunichar thechar;
GError* err = NULL;
VALUE ret;
GIOStatus status = g_io_channel_read_unichar(_SELF(self), &thechar, &err);
if (status == G_IO_STATUS_EOF){
ret = Qnil;
} else {
ioc_error(status, err);
ret = UINT2NUM(thechar);
}
return ret;
}
/*
* _read:
* @self: A valid %GSDLTokenizer.
* @result: (out): The location to store the resulting %gunichar.
* @err: (out) (allow-none): Location to store any %GError, or %NULL.
*
* Reads a single UTF-8 character from the input. Will return %EOF once at the end of the input.
*
* Returns: Whether the read succeeded.
*/
static bool _read(GSDLTokenizer *self, gunichar *result, GError **err) {
if (self->peek_avail) {
*result = self->peeked;
self->peek_avail = false;
} else if (self->stringbuf) {
if (!*self->stringbuf) {
self->stringbuf = NULL;
*result = EOF;
} else {
*result = *(self->stringbuf++);
}
return true;
} else {
if (G_UNLIKELY(!self->channel)) return false;
switch (g_io_channel_read_unichar(self->channel, result, err)) {
case G_IO_STATUS_ERROR:
self->channel = NULL;
self->peek_avail = false;
return false;
case G_IO_STATUS_EOF:
self->peek_avail = false;
*result = EOF;
g_io_channel_shutdown(self->channel, FALSE, NULL);
g_io_channel_unref(self->channel);
self->channel = NULL;
return true;
case G_IO_STATUS_AGAIN:
case G_IO_STATUS_NORMAL:
break;
}
}
if (*result == '\n') {
self->line++;
self->col = 1;
} else {
self->col++;
}
return true;
}
static VALUE
rg_each_char(VALUE self)
{
if (!rb_block_given_p()) {
rb_raise(rb_eArgError, "called without a block");
}
while (TRUE){
gunichar thechar;
GError* err = NULL;
GIOStatus status = g_io_channel_read_unichar(_SELF(self), &thechar, &err);
if (status == G_IO_STATUS_EOF){
break;
} else {
ioc_error(status, err);
rb_yield(UINT2NUM(thechar));
}
}
return self;
}
int main (int argc, char **argv)
{
enum ParserState state = 0;
gunichar in;
gboolean init = FALSE;
/* used to parse attribute */
gint attributes = 0;
/* Input/output */
FILE *input = stdin;
FILE *output = stdout;
/* Error */
GError *error = NULL;
/* init the glib system */
g_type_init();
/* needed to get the right charset from g_get_charset */
setlocale(LC_ALL, "");
/* Get charset */
g_get_charset(&input_charset);
/* parse options */
parse_cmd_options(&argc, &argv);
/* Handle input file */
if(input_file && strcmp(input_file, "-")) {
input = g_fopen(input_file, "r");
EXCEPTION(input == NULL, "Failed to open: %s: %s\n", input_file, strerror(errno));
}
/* Handle output file */
if(output_file) {
output = g_fopen(output_file, "w");
EXCEPTION(output == NULL, "Failed to open: %s: %s\n", output_file, strerror(errno));
}
/* Create channel for input */
GIOChannel *chan = g_io_channel_unix_new(fileno(input));
/* Set channel encoding */
g_io_channel_set_encoding(chan, input_charset,&error);
EXCEPTION(error != NULL, "Failed to set input encoding: %s\n", error->message);
/* Read input */
while(g_io_channel_read_unichar(chan, &in, &error) == G_IO_STATUS_NORMAL && error == NULL)
{
if(!init) {
/* Output html header */
print_page_header(output);
/* Convert and print body */
fprintf(output, "<pre style='font-family:monospace'>\n");
init = TRUE;
}
/* If we hit the escape character, go into 'attribute parsing' mode */
if(in == ESCAPE_CHAR) {
state = PARSE_ATTRIBUTE;
/* reset */
attributes = 0;
}
/* if we are in attribute parsing mode, parse attribute */
else if(state == PARSE_ATTRIBUTE) {
if(in == '[') {
/* Begin of attributes */
state = PARSE_COLOR_ATTRIBUTE;
}else {
WARNING("Unknown Escape sequence found: %i\n", in);
state = PARSE_NORMAL;
}
} else if(state == PARSE_COLOR_ATTRIBUTE) {
if (in == ';') { /* End of element */
process_attribute(output, attributes);
attributes = 0;
} else if(in == 'm') { /* end of attribute */
process_attribute(output, attributes);
state = PARSE_NORMAL;
} else if(in >= '0' && in <= '9') {
attributes *= 10;
attributes += in-'0';
}else if (in == 'h' || in == 'l' ) {
WARNING("Unsupported attribute found: %i\n", in);
state = PARSE_NORMAL;
}
continue;
} else if (state == PARSE_NORMAL) {
/* special chars (htmlspecialchars php doc) */
if(in == '"') fputs(""", output);
else if(in == '\'') fputs("'", output);
else if(in == '&') fputs("&", output);
else if(in == '<') fputs("<", output);
else if(in == '>') fputs(">", output);
/* ascii values stay ascii*/
else if(in >= 0 && in <= 177)
fprintf(output, "%c", (char)in);
/* Rest we encode in utf8 */
else fprintf(output, "&#%i;", in);
}
}
EXCEPTION(error != NULL, "Failed to read input character: %s\n", error->message);
if(init) {
/* Close open tags */
process_attribute(output, 0);
fprintf(output,"\n </pre>\n");
print_page_footer(output);
}
/* free input channel */
g_io_channel_unref(chan);
/* close i/o */
if(input != stdin) fclose(input);
if(output != stdout) fclose(output);
return EXIT_SUCCESS;
}
