comparison libpurple/util.c @ 29693:7925bb7f2aa7

propagate from branch 'im.pidgin.pidgin' (head 92e101e3698042e7600729bc09dec9e28f81de8f) to branch 'im.pidgin.cpw.attention_ui' (head 563c35d20efcc234a74567ebb88c059d10ce1e09)

29692:fb99a0067812

		state_encoded_text, state_question4, state_equal2 = state_start
	} encoded_word_state_t;
	encoded_word_state_t state = state_start;
	const char *cur, *mark;
	const char *charset0 = NULL, *encoding0 = NULL, *encoded_text0 = NULL;
	char *n, *new;

	/* token can be any CHAR (supposedly ISO8859-1/ISO2022), not just ASCII */
	#define token_char_p(c) \
		(c != ' ' && !iscntrl(c) && !strchr("()<>@,;:\"/[]?.=", c))

	/* But encoded-text must be ASCII; alas, isascii() may not exist */
	#define encoded_text_char_p(c) \
		((c & 0x80) == 0 && c != '?' && c != ' ' && isgraph(c))

	#define RECOVER_MARKED_TEXT strncpy(n, mark, cur - mark + 1); \
		n += cur - mark + 1

	g_return_val_if_fail(str != NULL, NULL);

	/* NOTE: Assuming that we need just strlen(str)+1 *may* be wrong.
	 * It would be wrong if one byte (in some unknown encoding) could
	 * expand to >=4 bytes of UTF-8; I don't know if there are such things.
	 */
	n = new = g_malloc(strlen(str) + 1);

	/* Here we will be looking for encoded words and if they seem to be
	 * valid then decode them.
	 * They are of this form: =?charset?encoding?text?=
	 */

		switch (state) {
		case state_equal1:
			if (*cur == '?') {
				state = state_question1;
			} else {
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question1:
			if (token_char_p(*cur)) {
				charset0 = cur;
				state = state_charset;
			} else { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_charset:
			if (*cur == '?') {
				state = state_question2;
			} else if (!token_char_p(*cur)) { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question2:
			if (token_char_p(*cur)) {
				encoding0 = cur;
				state = state_encoding;
			} else { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_encoding:
			if (*cur == '?') {
				state = state_question3;
			} else if (!token_char_p(*cur)) { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question3:
			if (encoded_text_char_p(*cur)) {

				state = state_encoded_text;
			} else if (*cur == '?') { /* empty string */
				encoded_text0 = cur;
				state = state_question4;
			} else { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_encoded_text:
			if (*cur == '?') {
				state = state_question4;
			} else if (!encoded_text_char_p(*cur)) {
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question4:
			if (*cur == '=') { /* Got the whole encoded-word */

				if (decoded) {
					gsize len;
					char *converted = g_convert((const gchar *)decoded, dec_len, "utf-8", charset, NULL, &len, NULL);

					if (converted) {
						n = strncpy(n, converted, len) + len;
						g_free(converted);
					}
					g_free(decoded);
				}
				g_free(charset);
				g_free(encoding);
				g_free(encoded_text);
				state = state_equal2; /* Restart the FSM */
			} else { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		default:
			if (*cur == '=') {
				mark = cur;
				state = state_equal1;
			} else {
				/* Some unencoded text. */
				*n = *cur;
				n += 1;
			}
			break;
		} /* switch */
	} /* for */

	if (state != state_start) {
		RECOVER_MARKED_TEXT;
	}
	*n = '\0';

	return new;
}


/**************************************************************************
 * Date/Time Functions

		header_len = strlen(header);

	/* Note: data is _not_ nul-terminated.  */
	if (data_len > header_len) {
		if (header[0] == '\n')
			p = (g_strncasecmp(data, header + 1, header_len - 1) == 0) ? data : NULL;
		if (!p)
			p = purple_strcasestr(data, header);
		if (p)
			p += header_len;
	}

static gboolean
content_is_chunked(const char *data, size_t data_len)
{
	const char *p = find_header_content(data, data_len, "\nTransfer-Encoding: ", sizeof("\nTransfer-Encoding: ") - 1);
	if (p && g_strncasecmp(p, "chunked", 7) == 0)
		return TRUE;

	return FALSE;
}

29693:7925bb7f2aa7

		state_encoded_text, state_question4, state_equal2 = state_start
	} encoded_word_state_t;
	encoded_word_state_t state = state_start;
	const char *cur, *mark;
	const char *charset0 = NULL, *encoding0 = NULL, *encoded_text0 = NULL;
	GString *new;

	/* token can be any CHAR (supposedly ISO8859-1/ISO2022), not just ASCII */
	#define token_char_p(c) \
		(c != ' ' && !iscntrl(c) && !strchr("()<>@,;:\"/[]?.=", c))

	/* But encoded-text must be ASCII; alas, isascii() may not exist */
	#define encoded_text_char_p(c) \
		((c & 0x80) == 0 && c != '?' && c != ' ' && isgraph(c))

	g_return_val_if_fail(str != NULL, NULL);

	new = g_string_new(NULL);

	/* Here we will be looking for encoded words and if they seem to be
	 * valid then decode them.
	 * They are of this form: =?charset?encoding?text?=
	 */

		switch (state) {
		case state_equal1:
			if (*cur == '?') {
				state = state_question1;
			} else {
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_question1:
			if (token_char_p(*cur)) {
				charset0 = cur;
				state = state_charset;
			} else { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_charset:
			if (*cur == '?') {
				state = state_question2;
			} else if (!token_char_p(*cur)) { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_question2:
			if (token_char_p(*cur)) {
				encoding0 = cur;
				state = state_encoding;
			} else { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_encoding:
			if (*cur == '?') {
				state = state_question3;
			} else if (!token_char_p(*cur)) { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_question3:
			if (encoded_text_char_p(*cur)) {

				state = state_encoded_text;
			} else if (*cur == '?') { /* empty string */
				encoded_text0 = cur;
				state = state_question4;
			} else { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_encoded_text:
			if (*cur == '?') {
				state = state_question4;
			} else if (!encoded_text_char_p(*cur)) {
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		case state_question4:
			if (*cur == '=') { /* Got the whole encoded-word */

				if (decoded) {
					gsize len;
					char *converted = g_convert((const gchar *)decoded, dec_len, "utf-8", charset, NULL, &len, NULL);

					if (converted) {
						g_string_append_len(new, converted, len);
						g_free(converted);
					}
					g_free(decoded);
				}
				g_free(charset);
				g_free(encoding);
				g_free(encoded_text);
				state = state_equal2; /* Restart the FSM */
			} else { /* This should never happen */
				g_string_append_len(new, mark, cur - mark + 1);
				state = state_start;
			}
			break;
		default:
			if (*cur == '=') {
				mark = cur;
				state = state_equal1;
			} else {
				/* Some unencoded text. */
				g_string_append_c(new, *cur);
			}
			break;
		} /* switch */
	} /* for */

	if (state != state_start)
		g_string_append_len(new, mark, cur - mark + 1);

	return g_string_free(new, FALSE);;
}


/**************************************************************************
 * Date/Time Functions

		header_len = strlen(header);

	/* Note: data is _not_ nul-terminated.  */
	if (data_len > header_len) {
		if (header[0] == '\n')
			p = (g_ascii_strncasecmp(data, header + 1, header_len - 1) == 0) ? data : NULL;
		if (!p)
			p = purple_strcasestr(data, header);
		if (p)
			p += header_len;
	}

static gboolean
content_is_chunked(const char *data, size_t data_len)
{
	const char *p = find_header_content(data, data_len, "\nTransfer-Encoding: ", sizeof("\nTransfer-Encoding: ") - 1);
	if (p && g_ascii_strncasecmp(p, "chunked", 7) == 0)
		return TRUE;

	return FALSE;
}