comparison src/util.c @ 8976:6f21aa413b18

[gaim-migrate @ 9750] "This patch fixes two relatively minor problems for the new gaim_mime_decode_field code: 1. Use g_return_val_if_fail to check for NULL. (Problem discovered by shx) 2. Added code to handle the boundary case =?charset?encoding??= (i.e., no encoded-text) Some comments also have been changed/added for clarification." --Ambrose C. LI committer: Tailor Script <[e-mail]>

8975:574393ccd705

	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, not necessarily 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

	/* NOTE: Assuming that we need just strlen(str)+1 may be wrong */
	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?=

			}
			break;
		case state_charset:
			if (*cur == '?') {
				state = state_question2;
			} else if (!token_char_p(*cur)) {
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question2:

			}
			break;
		case state_encoding:
			if (*cur == '?') {
				state = state_question3;
			} else if (!token_char_p(*cur)) {
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;
		case state_question3:
			if (encoded_text_char_p(*cur)) {
				encoded_text0 = cur;
				state = state_encoded_text;
			} else { /* This should never happen */
				RECOVER_MARKED_TEXT;
				state = state_start;
			}
			break;

8976:6f21aa413b18

	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?=

			}
			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:

			}
			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)) {
				encoded_text0 = 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;