Octstr *html_to_sms(Octstr *html)
{
long i, len;
int c;
Octstr *sms;
sms = octstr_create("");
len = octstr_len(html);
i = 0;
while (i < len) {
c = octstr_get_char(html, i);
switch (c) {
case '<':
if (html_comment_begins(html, i))
skip_html_comment(html, &i);
else
skip_html_tag(html, &i);
break;
case '&':
convert_html_entity(sms, html, &i);
break;
default:
octstr_append_char(sms, c);
++i;
break;
}
}
octstr_shrink_blanks(sms);
octstr_strip_blanks(sms);
return sms;
}
static int next_token(parser_ctx_t *ctx, void *lval)
{
do {
skip_spaces(ctx);
if(ctx->ptr == ctx->end)
return tEOF;
}while(skip_comment(ctx) || skip_html_comment(ctx));
if(isalphaW(*ctx->ptr)) {
int ret = check_keywords(ctx, lval);
if(ret)
return ret;
return parse_identifier(ctx, lval);
}
if(isdigitW(*ctx->ptr))
return parse_numeric_literal(ctx, lval);
switch(*ctx->ptr) {
case '{':
case '(':
case ')':
case '[':
case ']':
case ';':
case ',':
case '~':
case '?':
case ':':
return *ctx->ptr++;
case '}':
*(const WCHAR**)lval = ctx->ptr++;
return '}';
case '.':
if(++ctx->ptr < ctx->end && isdigitW(*ctx->ptr))
return parse_double_literal(ctx, 0, lval);
return '.';
case '<':
if(++ctx->ptr == ctx->end) {
*(int*)lval = EXPR_LESS;
return tRelOper;
}
switch(*ctx->ptr) {
case '=': /* <= */
ctx->ptr++;
*(int*)lval = EXPR_LESSEQ;
return tRelOper;
case '<': /* << */
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* <<= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNLSHIFT;
return tAssignOper;
}
*(int*)lval = EXPR_LSHIFT;
return tShiftOper;
default: /* < */
*(int*)lval = EXPR_LESS;
return tRelOper;
}
case '>':
if(++ctx->ptr == ctx->end) { /* > */
*(int*)lval = EXPR_GREATER;
return tRelOper;
}
switch(*ctx->ptr) {
case '=': /* >= */
ctx->ptr++;
*(int*)lval = EXPR_GREATEREQ;
return tRelOper;
case '>': /* >> */
if(++ctx->ptr < ctx->end) {
if(*ctx->ptr == '=') { /* >>= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNRSHIFT;
return tAssignOper;
}
if(*ctx->ptr == '>') { /* >>> */
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* >>>= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNRRSHIFT;
return tAssignOper;
}
*(int*)lval = EXPR_RRSHIFT;
return tRelOper;
}
}
*(int*)lval = EXPR_RSHIFT;
return tShiftOper;
default:
*(int*)lval = EXPR_GREATER;
return tRelOper;
}
case '+':
ctx->ptr++;
if(ctx->ptr < ctx->end) {
switch(*ctx->ptr) {
case '+': /* ++ */
ctx->ptr++;
return tINC;
case '=': /* += */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNADD;
return tAssignOper;
}
}
return '+';
case '-':
ctx->ptr++;
if(ctx->ptr < ctx->end) {
switch(*ctx->ptr) {
case '-': /* -- or --> */
ctx->ptr++;
if(ctx->is_html && ctx->nl && ctx->ptr < ctx->end && *ctx->ptr == '>') {
ctx->ptr++;
return tHTMLCOMMENT;
}
return tDEC;
case '=': /* -= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNSUB;
return tAssignOper;
}
}
return '-';
case '*':
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* *= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNMUL;
return tAssignOper;
}
return '*';
case '%':
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* %= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNMOD;
return tAssignOper;
}
return '%';
case '&':
if(++ctx->ptr < ctx->end) {
switch(*ctx->ptr) {
case '=': /* &= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNAND;
return tAssignOper;
case '&': /* && */
ctx->ptr++;
return tANDAND;
}
}
return '&';
case '|':
if(++ctx->ptr < ctx->end) {
switch(*ctx->ptr) {
case '=': /* |= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNOR;
return tAssignOper;
case '|': /* || */
ctx->ptr++;
return tOROR;
}
}
return '|';
case '^':
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* ^= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNXOR;
return tAssignOper;
}
return '^';
case '!':
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* != */
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* !== */
ctx->ptr++;
*(int*)lval = EXPR_NOTEQEQ;
return tEqOper;
}
*(int*)lval = EXPR_NOTEQ;
return tEqOper;
}
return '!';
case '=':
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* == */
if(++ctx->ptr < ctx->end && *ctx->ptr == '=') { /* === */
ctx->ptr++;
*(int*)lval = EXPR_EQEQ;
return tEqOper;
}
*(int*)lval = EXPR_EQ;
return tEqOper;
}
return '=';
case '/':
if(++ctx->ptr < ctx->end) {
if(*ctx->ptr == '=') { /* /= */
ctx->ptr++;
*(int*)lval = EXPR_ASSIGNDIV;
return kDIVEQ;
}
}
return '/';
case '\"':
case '\'':
return parse_string_literal(ctx, lval, *ctx->ptr);
case '_':
case '$':
return parse_identifier(ctx, lval);
case '@':
return '@';
}
WARN("unexpected char '%c' %d\n", *ctx->ptr, *ctx->ptr);
return 0;
}
