void print_ansi(char *buf, size_t buflen, buffered_output_t *output)
{
unsigned int font_style = 0;
unsigned int ansi_state;
unsigned int ansi_val[STRLEN];
int ival = 0;
size_t i;
char *ptr = buf;
char *ansi_begin;
char *ansi_end;
if (ptr == NULL)
return;
STATE_ZERO(ansi_state);
bzero(ansi_val, sizeof(ansi_val));
for (i = 0; i < buflen; i++)
{
if (STATE_ISSET(ansi_state, STATE_NEW_LINE))
{
STATE_CLR(ansi_state, STATE_NEW_LINE);
if (i < (buflen - 1) && (buf[i] == ':' && buf[i+1] == ' '))
{
STATE_SET(ansi_state, STATE_QUOTE_LINE);
if (STATE_ISSET(ansi_state, STATE_FONT_SET))
output->output("</font>", 7, output);
/* set quoted line styles */
STYLE_SET(font_style, FONT_STYLE_QUOTE);
STYLE_SET_FG(font_style, FONT_COLOR_QUOTE);
STYLE_CLR_BG(font_style);
print_font_style(font_style, output);
output->output(&buf[i], 1, output);
STATE_SET(ansi_state, STATE_FONT_SET);
STATE_CLR(ansi_state, STATE_ESC_SET);
/* clear ansi_val[] array */
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
continue;
}
else
STATE_CLR(ansi_state, STATE_QUOTE_LINE);
}
if (i < (buflen - 1) && (buf[i] == 0x1b && buf[i+1] == '['))
{
if (STATE_ISSET(ansi_state, STATE_ESC_SET))
{
/* *[*[ or *[13;24*[ */
size_t len;
ansi_end = &buf[i - 1];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
}
STATE_SET(ansi_state, STATE_ESC_SET);
ansi_begin = &buf[i];
i++; /* skip the next '[' character */
}
else if (buf[i] == '\n')
{
if (STATE_ISSET(ansi_state, STATE_ESC_SET))
{
/* *[\n or *[13;24\n */
size_t len;
ansi_end = &buf[i - 1];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
STATE_CLR(ansi_state, STATE_ESC_SET);
}
if (STATE_ISSET(ansi_state, STATE_QUOTE_LINE))
{
/* end of a quoted line */
output->output("</font>", 7, output);
STYLE_CLR(font_style, FONT_STYLE_QUOTE);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
output->output("<br />\n", 7, output);
STATE_CLR(ansi_state, STATE_QUOTE_LINE);
STATE_SET(ansi_state, STATE_NEW_LINE);
}
else
{
if (STATE_ISSET(ansi_state, STATE_ESC_SET))
{
if (buf[i] == 'm')
{
/* *[0;1;4;31m */
if (STATE_ISSET(ansi_state, STATE_FONT_SET))
{
output->output("</font>", 7, output);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
if (i < buflen - 1)
{
generate_font_style(&font_style, ansi_val, ival+1);
if (STATE_ISSET(ansi_state, STATE_QUOTE_LINE))
STYLE_SET(font_style, FONT_STYLE_QUOTE);
print_font_style(font_style, output);
STATE_SET(ansi_state, STATE_FONT_SET);
STATE_CLR(ansi_state, STATE_ESC_SET);
/*STYLE_ZERO(font_style);*/
/* clear ansi_val[] array */
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
}
}
else if (isalpha(buf[i]))
{
/* *[23;32H */
/* ignore it */
STATE_CLR(ansi_state, STATE_ESC_SET);
STYLE_ZERO(font_style);
/* clear ansi_val[] array */
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
continue;
}
else if (buf[i] == ';')
{
if (ival < sizeof(ansi_val) - 1)
{
ival ++; /* go to next ansi_val[] element */
ansi_val[ival] = 0;
}
}
else if (buf[i] >= '0' && buf[i] <= '9')
{
ansi_val[ival] *= 10;
ansi_val[ival] += (buf[i] - '0');
}
else
{
/* *[1;32/XXXX or *[* or *[[ */
/* not a valid ANSI string, just output it */
size_t len;
ansi_end = &buf[i];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
STATE_CLR(ansi_state, STATE_ESC_SET);
/* clear ansi_val[] array */
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
}
}
else
print_raw_ansi(&buf[i], 1, output);
}
}
if (STATE_ISSET(ansi_state, STATE_FONT_SET))
{
output->output("</font>", 7, output);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
flush_buffer(output);
}
int
token_next (struct tokenizer *t, char **tok, enum token_type *type)
{
enum token_type etype = TOKEN_TYPE_SYMBOL;
dstring_trunc (&t->tok);
for (;;) {
switch (chars_get (t)) {
case 0:
if (STATE_ISSET (STATE_STRING)) return 0;
if (STATE_ISSET (STATE_ESCAPE)) return 0;
if (STATE_ISSET (STATE_COMMENT)) return 0;
etype = TOKEN_TYPE_EOF;
goto END;
case -1: return 0;
default: break;
}
switch (t->ch_cur) {
case ';':
if (STATE_ISSET (STATE_STRING) || STATE_ISSET (STATE_ESCAPE)) {
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
break;
}
if (STATE_ISSET (STATE_COMMENT)) break;
STATE_SET (STATE_COMMENT);
break;
case '\n':
if (STATE_ISSET (STATE_COMMENT)) {
STATE_UNSET (STATE_COMMENT);
break;
}
t->line++;
case ' ':
case '\t':
if (STATE_ISSET (STATE_COMMENT)) break;
if (STATE_ISSET (STATE_ESCAPE)) STATE_UNSET (STATE_ESCAPE);
if (STATE_ISSET (STATE_STRING)) {
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
} else {
if (t->tok.len) goto END;
}
break;
case '"':
if (STATE_ISSET (STATE_COMMENT)) break;
if (STATE_ISSET (STATE_STRING)) {
if (STATE_ISSET (STATE_ESCAPE)) {
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
STATE_UNSET (STATE_ESCAPE);
} else {
STATE_UNSET (STATE_STRING);
goto END;
}
} else {
STATE_SET (STATE_STRING);
etype = TOKEN_TYPE_STRING;
}
break;
case '\\':
if (STATE_ISSET (STATE_COMMENT)) break;
if (STATE_ISSET (STATE_STRING)) {
if (STATE_ISSET (STATE_ESCAPE)) {
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
STATE_UNSET (STATE_ESCAPE);
} else {
STATE_SET (STATE_ESCAPE);
}
}
break;
case '(':
if (STATE_ISSET (STATE_COMMENT)) break;
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
if (!STATE_ISSET (STATE_STRING)) {
etype = TOKEN_TYPE_PAREN_OPEN;
goto END;
}
break;
case ')':
if (STATE_ISSET (STATE_COMMENT)) break;
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
if (!STATE_ISSET (STATE_STRING)) {
etype = TOKEN_TYPE_PAREN_CLOSE;
goto END;
}
break;
default:
if (STATE_ISSET (STATE_COMMENT)) break;
if (STATE_ISSET (STATE_ESCAPE)) STATE_UNSET (STATE_ESCAPE);
if (!dstring_catb (&t->tok, &t->ch_cur, 1)) return 0;
switch (t->ch_next) {
case ')':
case '(':
case '"':
if (!STATE_ISSET (STATE_STRING)) goto END;
default: break;
}
break;
}
}
END:
dstring_0 (&t->tok);
*tok = t->tok.s;
*type = etype;
return 1;
}
static void output_ansi_nforum(char *buf, size_t buflen,
buffered_output_t * output, zval *attachment)
{
unsigned int font_style = 0;
unsigned int ansi_state;
unsigned int ansi_val[STRLEN];
int ival = 0;
size_t i;
bool att = false;
char *ansi_begin = NULL;
char *ansi_end;
size_t article_len = buflen;
if (buf == NULL)
return;
STATE_ZERO(ansi_state);
bzero(ansi_val, sizeof(ansi_val));
for (i = 0; i < article_len; i++) {
if (buf[i] == '\0') { //assume ATTACHMENT_PAD[0] is '\0'
if (article_len - i >= ATTACHMENT_SIZE + sizeof(int) + 2) {
if (!memcmp(buf + i, ATTACHMENT_PAD, ATTACHMENT_SIZE)) {
att = true;
break;
}
}
}
if (STATE_ISSET(ansi_state, STATE_NEW_LINE)) {
STATE_CLR(ansi_state, STATE_NEW_LINE);
if (i < (buflen - 1) && !STATE_ISSET(ansi_state,STATE_TEX_SET) && (buf[i] == ':' && buf[i + 1] == ' ')) {
STATE_SET(ansi_state, STATE_QUOTE_LINE);
if (STATE_ISSET(ansi_state, STATE_FONT_SET))
BUFFERED_OUTPUT(output, "</font>", 7);
/*
* set quoted line styles
*/
STYLE_SET(font_style, FONT_STYLE_QUOTE);
STYLE_SET_FG(font_style, FONT_COLOR_QUOTE);
STYLE_CLR_BG(font_style);
print_font_style(font_style, output);
BUFFERED_OUTPUT(output, &buf[i], 1);
STATE_SET(ansi_state, STATE_FONT_SET);
STATE_CLR(ansi_state, STATE_ESC_SET);
/*
* clear ansi_val[] array
*/
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
continue;
} else
STATE_CLR(ansi_state, STATE_QUOTE_LINE);
}
/*
* is_tex 情况下，\[upload 优先匹配 \[ 而不是 [upload
* is_tex 情况下应该还有一个问题是 *[\[ 等，不过暂时不管了 - atppp
*/
if (i < (buflen - 1) && !STATE_ISSET(ansi_state,STATE_TEX_SET) && (buf[i] == 0x1b && buf[i + 1] == '[')) {
if (STATE_ISSET(ansi_state, STATE_ESC_SET)) {
/*
*[*[ or *[13;24*[ */
size_t len;
ansi_end = &buf[i - 1];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
}
STATE_SET(ansi_state, STATE_ESC_SET);
ansi_begin = &buf[i];
i++; /* skip the next '[' character */
} else if (buf[i] == '\n') {
if (STATE_ISSET(ansi_state, STATE_ESC_SET)) {
/*
*[\n or *[13;24\n */
size_t len;
ansi_end = &buf[i - 1];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
STATE_CLR(ansi_state, STATE_ESC_SET);
}
if (STATE_ISSET(ansi_state, STATE_QUOTE_LINE)) {
/*
* end of a quoted line
*/
BUFFERED_OUTPUT(output, " </font>", 8);
STYLE_CLR(font_style, FONT_STYLE_QUOTE);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
if (!STATE_ISSET(ansi_state,STATE_TEX_SET)) {
BUFFERED_OUTPUT(output, " <br /> ", 8);
}
STATE_CLR(ansi_state, STATE_QUOTE_LINE);
STATE_SET(ansi_state, STATE_NEW_LINE);
} else {
if (STATE_ISSET(ansi_state, STATE_ESC_SET)) {
if (buf[i] == 'm') {
/*
*[0;1;4;31m */
if (STATE_ISSET(ansi_state, STATE_FONT_SET)) {
BUFFERED_OUTPUT(output, "</font>", 7);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
if (i < buflen - 1) {
generate_font_style(&font_style, ansi_val, ival + 1);
if (STATE_ISSET(ansi_state, STATE_QUOTE_LINE))
STYLE_SET(font_style, FONT_STYLE_QUOTE);
print_font_style(font_style, output);
STATE_SET(ansi_state, STATE_FONT_SET);
STATE_CLR(ansi_state, STATE_ESC_SET);
/*
* STYLE_ZERO(font_style);
*/
/*
* clear ansi_val[] array
*/
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
}
} else if (isalpha(buf[i])) {
/*
*[23;32H */
/*
* ignore it
*/
STATE_CLR(ansi_state, STATE_ESC_SET);
STYLE_ZERO(font_style);
/*
* clear ansi_val[] array
*/
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
continue;
} else if (buf[i] == ';') {
if (ival < sizeof(ansi_val) - 1) {
ival++; /* go to next ansi_val[] element */
ansi_val[ival] = 0;
}
} else if (buf[i] >= '0' && buf[i] <= '9') {
ansi_val[ival] *= 10;
ansi_val[ival] += (buf[i] - '0');
} else {
/*
*[1;32/XXXX or *[* or *[[ */
/*
* not a valid ANSI string, just output it
*/
size_t len;
ansi_end = &buf[i];
len = ansi_end - ansi_begin + 1;
print_raw_ansi(ansi_begin, len, output);
STATE_CLR(ansi_state, STATE_ESC_SET);
/*
* clear ansi_val[] array
*/
bzero(ansi_val, sizeof(ansi_val));
ival = 0;
}
} else
print_raw_ansi(&buf[i], 1, output);
}
}
if (STATE_ISSET(ansi_state, STATE_FONT_SET)) {
BUFFERED_OUTPUT(output, "</font>", 7);
STATE_CLR(ansi_state, STATE_FONT_SET);
}
if (att) { //attachment
char *cur_ptr, *attachfilename, *attachptr;
long attach_len, attach_pos, newlen;
off_t ptrlen;
zval *item;
cur_ptr = buf + i;
ptrlen = article_len - i;
while (ptrlen > 0) {
if (((attachfilename = checkattach(cur_ptr, ptrlen,
&attach_len, &attachptr)) == NULL)) {
break;
}
attach_pos = attachfilename - buf;
newlen = attachptr - cur_ptr + attach_len;
cur_ptr += newlen;
ptrlen -= newlen;
if (ptrlen < 0) break;
MAKE_STD_ZVAL(item);
array_init(item);
add_assoc_string(item, "name", attachfilename, 1);
add_assoc_long(item, "size", attach_len);
add_assoc_long(item, "pos", attach_pos);
add_next_index_zval(attachment, item);
}
}
BUFFERED_FLUSH(output);
}