Attachment #129063 for bug #190015

View | Details | Raw Unified | Return to bug 190015
Collapse All | Expand All

Lines 1874-1882 Link Here

(-)wget-1.10.2/src/http.c (-1 / +15 lines)
1874	else	1874	else
1875	logputs (LOG_VERBOSE, "\n");	1875	logputs (LOG_VERBOSE, "\n");
1876	}	1876	}
1877	}	1877	}
		1878	int is_acceptable_content_type = 1;
		1879	if(type) is_acceptable_content_type = acceptable_content_type(type);
1878	xfree_null (type);	1880	xfree_null (type);
1879	type = NULL; /* We don't need it any more. */	1881	type = NULL; /* We don't need it any more. */
		1882	if(!is_acceptable_content_type)
		1883	{
		1884	CLOSE_FINISH (sock);
		1885	return CTYPEREJECTED;
		1886	}
1880		1887
1881	/* Return if we have no intention of further downloading. */	1888	/* Return if we have no intention of further downloading. */
1882	if (!(*dt & RETROKF) \|\| head_only)	1889	if (!(*dt & RETROKF) \|\| head_only)
Lines 2241-2246 Link Here
2241	free_hstat (&hstat);	2248	free_hstat (&hstat);
2242	xfree_null (dummy);	2249	xfree_null (dummy);
2243	return err;	2250	return err;
		2251	case CTYPEREJECTED:
		2252	logprintf (LOG_VERBOSE, _("Not downloading `%s', "
		2253	"content type rejected. -- not retrieving.\n"),
		2254	url_string (u, 1));
		2255	free_hstat (&hstat);
		2256	xfree_null (dummy);
		2257	return err;
2244	case FWRITEERR: case FOPENERR:	2258	case FWRITEERR: case FOPENERR:
2245	/* Another fatal error. */	2259	/* Another fatal error. */
2246	logputs (LOG_VERBOSE, "\n");	2260	logputs (LOG_VERBOSE, "\n");

Lines 132-137 Link Here

(-)wget-1.10.2/src/init.c (+2 lines)
132	{ "checkcertificate", &opt.check_cert, cmd_boolean },	132	{ "checkcertificate", &opt.check_cert, cmd_boolean },
133	#endif	133	#endif
134	{ "connecttimeout", &opt.connect_timeout, cmd_time },	134	{ "connecttimeout", &opt.connect_timeout, cmd_time },
		135	{ "contentaccepts", &opt.content_accepts, cmd_vector },
		136	{ "contentrejects", &opt.content_rejects, cmd_vector },
135	{ "continue", &opt.always_rest, cmd_boolean },	137	{ "continue", &opt.always_rest, cmd_boolean },
136	{ "convertlinks", &opt.convert_links, cmd_boolean },	138	{ "convertlinks", &opt.convert_links, cmd_boolean },
137	{ "cookies", &opt.cookies, cmd_boolean },	139	{ "cookies", &opt.cookies, cmd_boolean },

Lines 169-174 Link Here

(-)wget-1.10.2/src/main.c (+9 lines)
169	{ "connect-timeout", 0, OPT_VALUE, "connecttimeout", -1 },	169	{ "connect-timeout", 0, OPT_VALUE, "connecttimeout", -1 },
170	{ "continue", 'c', OPT_BOOLEAN, "continue", -1 },	170	{ "continue", 'c', OPT_BOOLEAN, "continue", -1 },
171	{ "convert-links", 'k', OPT_BOOLEAN, "convertlinks", -1 },	171	{ "convert-links", 'k', OPT_BOOLEAN, "convertlinks", -1 },
		172	{ "content-type", 0, OPT_VALUE, "contentaccepts", -1 },
		173	{ "content-type-exclude", 0, OPT_VALUE, "contentrejects", -1 },
172	{ "cookies", 0, OPT_BOOLEAN, "cookies", -1 },	174	{ "cookies", 0, OPT_BOOLEAN, "cookies", -1 },
173	{ "cut-dirs", 0, OPT_VALUE, "cutdirs", -1 },	175	{ "cut-dirs", 0, OPT_VALUE, "cutdirs", -1 },
174	{ IF_DEBUG ("debug"), 'd', OPT_BOOLEAN, "debug", -1 },	176	{ IF_DEBUG ("debug"), 'd', OPT_BOOLEAN, "debug", -1 },
Lines 541-546 Link Here
541	--post-data=STRING use the POST method; send STRING as the data.\n"),	543	--post-data=STRING use the POST method; send STRING as the data.\n"),
542	N_("\	544	N_("\
543	--post-file=FILE use the POST method; send contents of FILE.\n"),	545	--post-file=FILE use the POST method; send contents of FILE.\n"),
		546	N_("\
		547	--content-type=LIST comma-separated list of accepted content-types.\n"),
		548	N_("\
		549	--content-type-exclude=LIST comma-separated list of rejected content-types.\n"),
544	"\n",	550	"\n",
545		551
546	#ifdef HAVE_SSL	552	#ifdef HAVE_SSL
Lines 942-947 Link Here
942	else	948	else
943	status = retrieve_url (*t, &filename, &redirected_URL, NULL, &dt);	949	status = retrieve_url (*t, &filename, &redirected_URL, NULL, &dt);
944		950
		951	if(status == CTYPEREJECTED)
		952	logprintf (LOG_VERBOSE, _("Not downloading, content type rejected.\n"));
		953
945	if (opt.delete_after && file_exists_p(filename))	954	if (opt.delete_after && file_exists_p(filename))
946	{	955	{
947	DEBUGP (("Removing file due to --delete-after in main():\n"));	956	DEBUGP (("Removing file due to --delete-after in main():\n"));

Lines 57-62 Link Here

(-)wget-1.10.2/src/options.h (+3 lines)
57		57
58	int spider; /* Is Wget in spider mode? */	58	int spider; /* Is Wget in spider mode? */
59		59
		60	char *content_accepts; / List of content-type patterns to accept. */
		61	char *content_rejects; / List of content-type patterns to reject. */
		62
60	char *accepts; / List of patterns to accept. */	63	char *accepts; / List of patterns to accept. */
61	char *rejects; / List of patterns to reject. */	64	char *rejects; / List of patterns to reject. */
62	char *excludes; / List of excluded FTP directories. */	65	char *excludes; / List of excluded FTP directories. */

Lines 317-322 Link Here

(-)wget-1.10.2/src/recur.c (+3 lines)
317	}	317	}
318	}	318	}
319		319
		320	if(status == CTYPEREJECTED)
		321	logprintf (LOG_VERBOSE, _("Not downloading, content type rejected.\n"));
		322
320	/* If the downloaded document was HTML, parse it and enqueue the	323	/* If the downloaded document was HTML, parse it and enqueue the
321	links it contains. */	324	links it contains. */
322		325




  return 1;
}

/* Determine whether a content type is acceptable to be downloaded, according to
 * commandline options. */
int
acceptable_content_type (const char *type)
{
   if (opt.content_accepts)
     {
       if (opt.content_rejects)
   return (in_acclist ((const char *const *)opt.content_accepts, type, 1)
       && !in_acclist ((const char *const *)opt.content_rejects, type, 1));
       else
   return in_acclist ((const char *const *)opt.content_accepts, type, 1);
     }
   else if (opt.content_rejects)
     return !in_acclist ((const char *const *)opt.content_rejects, type, 1);
   return 1;
}

/* Compare S1 and S2 frontally; S2 must begin with S1.  E.g. if S1 is
   `/something', frontcmp() will return 1 only if S2 begins with
   `/something'.  Otherwise, 0 is returned.  */

Lines 79-84 Link Here

(-)wget-1.10.2/src/utils.h (+1 lines)
79	char file_merge PARAMS ((const char , const char *));	79	char file_merge PARAMS ((const char , const char *));
80		80
81	int acceptable PARAMS ((const char *));	81	int acceptable PARAMS ((const char *));
		82	int acceptable_content_type (const char *);
82	int accdir PARAMS ((const char *s, enum accd));	83	int accdir PARAMS ((const char *s, enum accd));
83	char suffix PARAMS ((const char s));	84	char suffix PARAMS ((const char s));
84	int match_tail PARAMS ((const char , const char , int));	85	int match_tail PARAMS ((const char , const char , int));

Lines 294-300 Link Here

(-)wget-1.10.2/src/wget.h (-1 / +1 lines)
294	CONTNOTSUPPORTED, RETRUNNEEDED, RETRFINISHED, READERR, TRYLIMEXC,	294	CONTNOTSUPPORTED, RETRUNNEEDED, RETRFINISHED, READERR, TRYLIMEXC,
295	URLBADPATTERN, FILEBADFILE, RANGEERR, RETRBADPATTERN,	295	URLBADPATTERN, FILEBADFILE, RANGEERR, RETRBADPATTERN,
296	RETNOTSUP, ROBOTSOK, NOROBOTS, PROXERR, AUTHFAILED,	296	RETNOTSUP, ROBOTSOK, NOROBOTS, PROXERR, AUTHFAILED,
297	QUOTEXC, WRITEFAILED, SSLINITFAILED	297	QUOTEXC, WRITEFAILED, SSLINITFAILED, CTYPEREJECTED
298	} uerr_t;	298	} uerr_t;
299		299
300	#endif /* WGET_H */	300	#endif /* WGET_H */

Return to bug 190015

Lines 668-673 Link Here

(-)wget-1.10.2/src/utils.c (+18 lines)
668	return 1;	668	return 1;
669	}	669	}
670		670
		671	/* Determine whether a content type is acceptable to be downloaded, according to
		672	* commandline options. */
		673	int
		674	acceptable_content_type (const char *type)
		675	{
		676	if (opt.content_accepts)
		677	{
		678	if (opt.content_rejects)
		679	return (in_acclist ((const char const )opt.content_accepts, type, 1)
		680	&& !in_acclist ((const char const )opt.content_rejects, type, 1));
		681	else
		682	return in_acclist ((const char const )opt.content_accepts, type, 1);
		683	}
		684	else if (opt.content_rejects)
		685	return !in_acclist ((const char const )opt.content_rejects, type, 1);
		686	return 1;
		687	}
		688
671	/* Compare S1 and S2 frontally; S2 must begin with S1. E.g. if S1 is	689	/* Compare S1 and S2 frontally; S2 must begin with S1. E.g. if S1 is
672	`/something', frontcmp() will return 1 only if S2 begins with	690	`/something', frontcmp() will return 1 only if S2 begins with
673	`/something'. Otherwise, 0 is returned. */	691	`/something'. Otherwise, 0 is returned. */