Fossil SCM

Improvements and simplifications to anti-robot defenses.

drh 2025-08-16 13:59 trunk merge

Commit 16b33097fe60eac71f84a7ca28362d36da76a1ff5748bb132fc1bcdf73ca4476

Parent d546932976888ae…

13 files changed -1 +12 -8 +1 -2 +1 -1 +4 -9 +11 -96 +2 -1 +109 -60 +20 -37 +1 -46 +1 -1 +1 +1 -1

~ src/browse.c ~ src/captcha.c ~ src/diff.c ~ src/diffcmd.c ~ src/info.c ~ src/login.c ~ src/main.c ~ src/robot.c ~ src/setup.c ~ src/style.c ~ src/tar.c ~ src/timeline.c ~ src/zip.c

M src/browse.c

-1

		--- src/browse.c
		+++ src/browse.c
		@@ -1162,11 +1162,10 @@
1162	1162	int showId = PB("showid");
1163	1163	Stmt q1, q2;
1164	1164	double baseTime;
1165	1165	login_check_credentials();
1166	1166	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1167		- if( exclude_spiders(0) ) return;
1168	1167	zName = P("name");
1169	1168	if( zName==0 ) zName = "tip";
1170	1169	rid = symbolic_name_to_rid(zName, "ci");
1171	1170	if( rid==0 ){
1172	1171	fossil_fatal("not a valid check-in: %s", zName);
1173	1172

	--- src/browse.c
	+++ src/browse.c
	@@ -1162,11 +1162,10 @@
1162	int showId = PB("showid");
1163	Stmt q1, q2;
1164	double baseTime;
1165	login_check_credentials();
1166	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1167	if( exclude_spiders(0) ) return;
1168	zName = P("name");
1169	if( zName==0 ) zName = "tip";
1170	rid = symbolic_name_to_rid(zName, "ci");
1171	if( rid==0 ){
1172	fossil_fatal("not a valid check-in: %s", zName);
1173

	--- src/browse.c
	+++ src/browse.c
	@@ -1162,11 +1162,10 @@
1162	int showId = PB("showid");
1163	Stmt q1, q2;
1164	double baseTime;
1165	login_check_credentials();
1166	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }

1167	zName = P("name");
1168	if( zName==0 ) zName = "tip";
1169	rid = symbolic_name_to_rid(zName, "ci");
1170	if( rid==0 ){
1171	fossil_fatal("not a valid check-in: %s", zName);
1172

M src/captcha.c

+12 -8

		--- src/captcha.c
		+++ src/captcha.c
		@@ -744,11 +744,11 @@
744	744	(void)exclude_spiders(1);
745	745	@ <hr><p>The captcha is shown above. Add a name=HEX query parameter
746	746	@ to see how HEX would be rendered in the current captcha font.
747	747	@ <h2>Debug/Testing Values:</h2>
748	748	@ <ul>
749		- @ <li> g.isHuman = %d(g.isHuman)
	749	+ @ <li> g.isRobot = %d(g.isRobot)
750	750	@ <li> g.zLogin = %h(g.zLogin)
751	751	@ <li> login_cookie_welformed() = %d(login_cookie_wellformed())
752	752	@ <li> captcha_is_correct(1) = %d(captcha_is_correct(1)).
753	753	@ </ul>
754	754	style_finish_page();
		@@ -759,10 +759,18 @@
759	759	@ %s(captcha_render(zPw))
760	760	@ </pre>
761	761	style_finish_page();
762	762	}
763	763	}
	764	+
	765	+/*
	766	+** WEBPAGE: honeypot
	767	+** This page is a honeypot for spiders and bots.
	768	+*/
	769	+void honeypot_page(void){
	770	+ (void)exclude_spiders(0);
	771	+}
764	772
765	773	/*
766	774	** Check to see if the current request is coming from an agent that
767	775	** self-identifies as a spider.
768	776	**
		@@ -776,27 +784,23 @@
776	784	** If the bTest argument is non-zero, then show the captcha regardless of
777	785	** how the agent identifies. This is used for testing only.
778	786	*/
779	787	int exclude_spiders(int bTest){
780	788	if( !bTest ){
781		- if( g.isHuman ) return 0; /* This user has already proven human */
782	789	if( g.zLogin!=0 ) return 0; /* Logged in. Consider them human */
783	790	if( login_cookie_wellformed() ){
784	791	/* Logged into another member of the login group */
785	792	return 0;
786	793	}
787	794	}
788	795
789	796	/* This appears to be a spider. Offer the captcha */
790	797	style_set_current_feature("captcha");
791		- style_header("I think you are a robot");
	798	+ style_header("Captcha");
792	799	style_submenu_enable(0);
793	800	@ <form method='POST' action='%R/ityaar'>
794		- @ <p>You seem like a robot.
795		- @
796		- @ <p>If you are human, you can prove that by solving the captcha below,
797		- @ after which you will be allowed to proceed.
	801	+ @ <h2>Prove that you are human:
798	802	if( bTest ){
799	803	@ <input type="hidden" name="istest" value="1">
800	804	}
801	805	captcha_generate(3);
802	806	@ </form>
		@@ -830,11 +834,11 @@
830	834	}
831	835	cgi_append_header("X-Robot: 0\r\n");
832	836	}
833	837	login_redirect_to_g();
834	838	}else{
835		- g.isHuman = 0;
	839	+ g.isRobot = 1;
836	840	(void)exclude_spiders(bTest);
837	841	if( bTest ){
838	842	@ <hr><p>Wrong code. Try again
839	843	style_finish_page();
840	844	}
841	845

	--- src/captcha.c
	+++ src/captcha.c
	@@ -744,11 +744,11 @@
744	(void)exclude_spiders(1);
745	@ <hr><p>The captcha is shown above. Add a name=HEX query parameter
746	@ to see how HEX would be rendered in the current captcha font.
747	@ <h2>Debug/Testing Values:</h2>
748	@ <ul>
749	@ <li> g.isHuman = %d(g.isHuman)
750	@ <li> g.zLogin = %h(g.zLogin)
751	@ <li> login_cookie_welformed() = %d(login_cookie_wellformed())
752	@ <li> captcha_is_correct(1) = %d(captcha_is_correct(1)).
753	@ </ul>
754	style_finish_page();
	@@ -759,10 +759,18 @@
759	@ %s(captcha_render(zPw))
760	@ </pre>
761	style_finish_page();
762	}
763	}








764
765	/*
766	** Check to see if the current request is coming from an agent that
767	** self-identifies as a spider.
768	**
	@@ -776,27 +784,23 @@
776	** If the bTest argument is non-zero, then show the captcha regardless of
777	** how the agent identifies. This is used for testing only.
778	*/
779	int exclude_spiders(int bTest){
780	if( !bTest ){
781	if( g.isHuman ) return 0; /* This user has already proven human */
782	if( g.zLogin!=0 ) return 0; /* Logged in. Consider them human */
783	if( login_cookie_wellformed() ){
784	/* Logged into another member of the login group */
785	return 0;
786	}
787	}
788
789	/* This appears to be a spider. Offer the captcha */
790	style_set_current_feature("captcha");
791	style_header("I think you are a robot");
792	style_submenu_enable(0);
793	@ <form method='POST' action='%R/ityaar'>
794	@ <p>You seem like a robot.
795	@
796	@ <p>If you are human, you can prove that by solving the captcha below,
797	@ after which you will be allowed to proceed.
798	if( bTest ){
799	@ <input type="hidden" name="istest" value="1">
800	}
801	captcha_generate(3);
802	@ </form>
	@@ -830,11 +834,11 @@
830	}
831	cgi_append_header("X-Robot: 0\r\n");
832	}
833	login_redirect_to_g();
834	}else{
835	g.isHuman = 0;
836	(void)exclude_spiders(bTest);
837	if( bTest ){
838	@ <hr><p>Wrong code. Try again
839	style_finish_page();
840	}
841

	--- src/captcha.c
	+++ src/captcha.c
	@@ -744,11 +744,11 @@
744	(void)exclude_spiders(1);
745	@ <hr><p>The captcha is shown above. Add a name=HEX query parameter
746	@ to see how HEX would be rendered in the current captcha font.
747	@ <h2>Debug/Testing Values:</h2>
748	@ <ul>
749	@ <li> g.isRobot = %d(g.isRobot)
750	@ <li> g.zLogin = %h(g.zLogin)
751	@ <li> login_cookie_welformed() = %d(login_cookie_wellformed())
752	@ <li> captcha_is_correct(1) = %d(captcha_is_correct(1)).
753	@ </ul>
754	style_finish_page();
	@@ -759,10 +759,18 @@
759	@ %s(captcha_render(zPw))
760	@ </pre>
761	style_finish_page();
762	}
763	}
764
765	/*
766	** WEBPAGE: honeypot
767	** This page is a honeypot for spiders and bots.
768	*/
769	void honeypot_page(void){
770	(void)exclude_spiders(0);
771	}
772
773	/*
774	** Check to see if the current request is coming from an agent that
775	** self-identifies as a spider.
776	**
	@@ -776,27 +784,23 @@
784	** If the bTest argument is non-zero, then show the captcha regardless of
785	** how the agent identifies. This is used for testing only.
786	*/
787	int exclude_spiders(int bTest){
788	if( !bTest ){

789	if( g.zLogin!=0 ) return 0; /* Logged in. Consider them human */
790	if( login_cookie_wellformed() ){
791	/* Logged into another member of the login group */
792	return 0;
793	}
794	}
795
796	/* This appears to be a spider. Offer the captcha */
797	style_set_current_feature("captcha");
798	style_header("Captcha");
799	style_submenu_enable(0);
800	@ <form method='POST' action='%R/ityaar'>
801	@ <h2>Prove that you are human:



802	if( bTest ){
803	@ <input type="hidden" name="istest" value="1">
804	}
805	captcha_generate(3);
806	@ </form>
	@@ -830,11 +834,11 @@
834	}
835	cgi_append_header("X-Robot: 0\r\n");
836	}
837	login_redirect_to_g();
838	}else{
839	g.isRobot = 1;
840	(void)exclude_spiders(bTest);
841	if( bTest ){
842	@ <hr><p>Wrong code. Try again
843	style_finish_page();
844	}
845

M src/diff.c

+1 -2

		--- src/diff.c
		+++ src/diff.c
		@@ -3788,12 +3788,11 @@
3788	3788	int bBlame = g.zPath[0]!='a';/* True for BLAME output. False for ANNOTATE. */
3789	3789
3790	3790	/* Gather query parameters */
3791	3791	login_check_credentials();
3792	3792	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
3793		- if( exclude_spiders(0) ) return;
3794		- if( robot_squelch(990) ) return;
	3793	+ if( robot_restrict("annotate") ) return;
3795	3794	fossil_nice_default();
3796	3795	zFilename = P("filename");
3797	3796	zRevision = PD("checkin",0);
3798	3797	zOrigin = P("origin");
3799	3798	zLimit = P("limit");
3800	3799

	--- src/diff.c
	+++ src/diff.c
	@@ -3788,12 +3788,11 @@
3788	int bBlame = g.zPath[0]!='a';/* True for BLAME output. False for ANNOTATE. */
3789
3790	/* Gather query parameters */
3791	login_check_credentials();
3792	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
3793	if( exclude_spiders(0) ) return;
3794	if( robot_squelch(990) ) return;
3795	fossil_nice_default();
3796	zFilename = P("filename");
3797	zRevision = PD("checkin",0);
3798	zOrigin = P("origin");
3799	zLimit = P("limit");
3800

	--- src/diff.c
	+++ src/diff.c
	@@ -3788,12 +3788,11 @@
3788	int bBlame = g.zPath[0]!='a';/* True for BLAME output. False for ANNOTATE. */
3789
3790	/* Gather query parameters */
3791	login_check_credentials();
3792	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
3793	if( robot_restrict("annotate") ) return;

3794	fossil_nice_default();
3795	zFilename = P("filename");
3796	zRevision = PD("checkin",0);
3797	zOrigin = P("origin");
3798	zLimit = P("limit");
3799

M src/diffcmd.c

+1 -1

		--- src/diffcmd.c
		+++ src/diffcmd.c
		@@ -1522,11 +1522,11 @@
1522	1522	DiffConfig DCfg;
1523	1523	cgi_check_for_malice();
1524	1524	login_check_credentials();
1525	1525	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1526	1526	if( zFrom==0 \|\| zTo==0 ) fossil_redirect_home();
1527		- if( robot_squelch(800) ) return;
	1527	+ if( robot_restrict("diff") ) return;
1528	1528
1529	1529	fossil_nice_default();
1530	1530	cgi_set_content_type("text/plain");
1531	1531	diff_config_init(&DCfg, DIFF_VERBOSE);
1532	1532	diff_two_versions(zFrom, zTo, &DCfg, 0);
1533	1533

	--- src/diffcmd.c
	+++ src/diffcmd.c
	@@ -1522,11 +1522,11 @@
1522	DiffConfig DCfg;
1523	cgi_check_for_malice();
1524	login_check_credentials();
1525	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1526	if( zFrom==0 \|\| zTo==0 ) fossil_redirect_home();
1527	if( robot_squelch(800) ) return;
1528
1529	fossil_nice_default();
1530	cgi_set_content_type("text/plain");
1531	diff_config_init(&DCfg, DIFF_VERBOSE);
1532	diff_two_versions(zFrom, zTo, &DCfg, 0);
1533

	--- src/diffcmd.c
	+++ src/diffcmd.c
	@@ -1522,11 +1522,11 @@
1522	DiffConfig DCfg;
1523	cgi_check_for_malice();
1524	login_check_credentials();
1525	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1526	if( zFrom==0 \|\| zTo==0 ) fossil_redirect_home();
1527	if( robot_restrict("diff") ) return;
1528
1529	fossil_nice_default();
1530	cgi_set_content_type("text/plain");
1531	diff_config_init(&DCfg, DIFF_VERBOSE);
1532	diff_two_versions(zFrom, zTo, &DCfg, 0);
1533

M src/info.c

+4 -9

		--- src/info.c
		+++ src/info.c
		@@ -1421,11 +1421,11 @@
1421	1421	Blob qpGlob; /* glob= query parameter for generated links */
1422	1422	int bInvert = PB("inv");
1423	1423
1424	1424	login_check_credentials();
1425	1425	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1426		- if( robot_squelch(950) ) return;
	1426	+ if( robot_restrict("diff") ) return;
1427	1427	login_anonymous_available();
1428	1428	fossil_nice_default();
1429	1429	blob_init(&qp, 0, 0);
1430	1430	blob_init(&qpGlob, 0, 0);
1431	1431	diffType = preferred_diff_type();
		@@ -1975,11 +1975,11 @@
1975	1975	int verbose = PB("verbose");
1976	1976	DiffConfig DCfg;
1977	1977
1978	1978	login_check_credentials();
1979	1979	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1980		- if( robot_squelch(800) ) return;
	1980	+ if( robot_restrict("diff") ) return;
1981	1981	diff_config_init(&DCfg, 0);
1982	1982	diffType = preferred_diff_type();
1983	1983	if( P("from") && P("to") ){
1984	1984	v1 = artifact_from_ci_and_filename("from");
1985	1985	v2 = artifact_from_ci_and_filename("to");
		@@ -2416,15 +2416,15 @@
2416	2416	object_description(rid, objdescFlags, 0, &downloadName);
2417	2417	style_submenu_element("Download", "%R/raw/%s?at=%T",
2418	2418	zUuid, file_tail(blob_str(&downloadName)));
2419	2419	@ <hr>
2420	2420	content_get(rid, &content);
2421		- if( !g.isHuman ){
	2421	+ if( blob_size(&content)>100000 ){
2422	2422	/* Prevent robots from running hexdump on megabyte-sized source files
2423	2423	** and there by eating up lots of CPU time and bandwidth. There is
2424	2424	** no good reason for a robot to need a hexdump. */
2425		- @ <p>A hex dump of this file is not available.
	2425	+ @ <p>A hex dump of this file is not available because it is too large.
2426	2426	@ Please download the raw binary file and generate a hex dump yourself.</p>
2427	2427	}else{
2428	2428	@ <blockquote><pre>
2429	2429	hexdump(&content);
2430	2430	@ </pre></blockquote>
		@@ -2702,24 +2702,19 @@
2702	2702	HQuery url;
2703	2703	char *zCIUuid = 0;
2704	2704	int isSymbolicCI = 0; /* ci= exists and is a symbolic name, not a hash */
2705	2705	int isBranchCI = 0; /* ci= refers to a branch name */
2706	2706	char *zHeader = 0;
2707		- int iCost;
2708	2707
2709	2708	login_check_credentials();
2710	2709	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
2711	2710	cgi_check_for_malice();
2712	2711	style_set_current_feature("artifact");
2713	2712	if( fossil_strcmp(g.zPath, "docfile")==0 ){
2714	2713	isFile = 1;
2715	2714	docOnly = 1;
2716	2715	}
2717		- iCost = 200;
2718		- if( isFile ) iCost += 100;
2719		- if( zCI ) iCost += 100;
2720		- if( robot_squelch(iCost) ) return;
2721	2716
2722	2717	/* Capture and normalize the name= and ci= query parameters */
2723	2718	if( zName==0 ){
2724	2719	zName = P("filename");
2725	2720	if( zName==0 ){
2726	2721

	--- src/info.c
	+++ src/info.c
	@@ -1421,11 +1421,11 @@
1421	Blob qpGlob; /* glob= query parameter for generated links */
1422	int bInvert = PB("inv");
1423
1424	login_check_credentials();
1425	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1426	if( robot_squelch(950) ) return;
1427	login_anonymous_available();
1428	fossil_nice_default();
1429	blob_init(&qp, 0, 0);
1430	blob_init(&qpGlob, 0, 0);
1431	diffType = preferred_diff_type();
	@@ -1975,11 +1975,11 @@
1975	int verbose = PB("verbose");
1976	DiffConfig DCfg;
1977
1978	login_check_credentials();
1979	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1980	if( robot_squelch(800) ) return;
1981	diff_config_init(&DCfg, 0);
1982	diffType = preferred_diff_type();
1983	if( P("from") && P("to") ){
1984	v1 = artifact_from_ci_and_filename("from");
1985	v2 = artifact_from_ci_and_filename("to");
	@@ -2416,15 +2416,15 @@
2416	object_description(rid, objdescFlags, 0, &downloadName);
2417	style_submenu_element("Download", "%R/raw/%s?at=%T",
2418	zUuid, file_tail(blob_str(&downloadName)));
2419	@ <hr>
2420	content_get(rid, &content);
2421	if( !g.isHuman ){
2422	/* Prevent robots from running hexdump on megabyte-sized source files
2423	** and there by eating up lots of CPU time and bandwidth. There is
2424	** no good reason for a robot to need a hexdump. */
2425	@ <p>A hex dump of this file is not available.
2426	@ Please download the raw binary file and generate a hex dump yourself.</p>
2427	}else{
2428	@ <blockquote><pre>
2429	hexdump(&content);
2430	@ </pre></blockquote>
	@@ -2702,24 +2702,19 @@
2702	HQuery url;
2703	char *zCIUuid = 0;
2704	int isSymbolicCI = 0; /* ci= exists and is a symbolic name, not a hash */
2705	int isBranchCI = 0; /* ci= refers to a branch name */
2706	char *zHeader = 0;
2707	int iCost;
2708
2709	login_check_credentials();
2710	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
2711	cgi_check_for_malice();
2712	style_set_current_feature("artifact");
2713	if( fossil_strcmp(g.zPath, "docfile")==0 ){
2714	isFile = 1;
2715	docOnly = 1;
2716	}
2717	iCost = 200;
2718	if( isFile ) iCost += 100;
2719	if( zCI ) iCost += 100;
2720	if( robot_squelch(iCost) ) return;
2721
2722	/* Capture and normalize the name= and ci= query parameters */
2723	if( zName==0 ){
2724	zName = P("filename");
2725	if( zName==0 ){
2726

	--- src/info.c
	+++ src/info.c
	@@ -1421,11 +1421,11 @@
1421	Blob qpGlob; /* glob= query parameter for generated links */
1422	int bInvert = PB("inv");
1423
1424	login_check_credentials();
1425	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1426	if( robot_restrict("diff") ) return;
1427	login_anonymous_available();
1428	fossil_nice_default();
1429	blob_init(&qp, 0, 0);
1430	blob_init(&qpGlob, 0, 0);
1431	diffType = preferred_diff_type();
	@@ -1975,11 +1975,11 @@
1975	int verbose = PB("verbose");
1976	DiffConfig DCfg;
1977
1978	login_check_credentials();
1979	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
1980	if( robot_restrict("diff") ) return;
1981	diff_config_init(&DCfg, 0);
1982	diffType = preferred_diff_type();
1983	if( P("from") && P("to") ){
1984	v1 = artifact_from_ci_and_filename("from");
1985	v2 = artifact_from_ci_and_filename("to");
	@@ -2416,15 +2416,15 @@
2416	object_description(rid, objdescFlags, 0, &downloadName);
2417	style_submenu_element("Download", "%R/raw/%s?at=%T",
2418	zUuid, file_tail(blob_str(&downloadName)));
2419	@ <hr>
2420	content_get(rid, &content);
2421	if( blob_size(&content)>100000 ){
2422	/* Prevent robots from running hexdump on megabyte-sized source files
2423	** and there by eating up lots of CPU time and bandwidth. There is
2424	** no good reason for a robot to need a hexdump. */
2425	@ <p>A hex dump of this file is not available because it is too large.
2426	@ Please download the raw binary file and generate a hex dump yourself.</p>
2427	}else{
2428	@ <blockquote><pre>
2429	hexdump(&content);
2430	@ </pre></blockquote>
	@@ -2702,24 +2702,19 @@
2702	HQuery url;
2703	char *zCIUuid = 0;
2704	int isSymbolicCI = 0; /* ci= exists and is a symbolic name, not a hash */
2705	int isBranchCI = 0; /* ci= refers to a branch name */
2706	char *zHeader = 0;

2707
2708	login_check_credentials();
2709	if( !g.perm.Read ){ login_needed(g.anon.Read); return; }
2710	cgi_check_for_malice();
2711	style_set_current_feature("artifact");
2712	if( fossil_strcmp(g.zPath, "docfile")==0 ){
2713	isFile = 1;
2714	docOnly = 1;
2715	}




2716
2717	/* Capture and normalize the name= and ci= query parameters */
2718	if( zName==0 ){
2719	zName = P("filename");
2720	if( zName==0 ){
2721

M src/login.c

+11 -96

		--- src/login.c
		+++ src/login.c
		@@ -1303,98 +1303,10 @@
1303	1303	}
1304	1304	fossil_free(zDecode);
1305	1305	return uid;
1306	1306	}
1307	1307
1308		-/*
1309		-** SETTING: robot-restrict width=40 block-text
1310		-** The VALUE of this setting is a list of GLOB patterns that match
1311		-** pages for which complex HTTP requests from robots should be disallowed.
1312		-** The recommended value for this setting is:
1313		-**
1314		-** timeline,vdiff,fdiff,annotate,blame
1315		-**
1316		-*/
1317		-
1318		-/*
1319		-** Check to see if the current HTTP request is a complex request that
1320		-** is coming from a robot and if access should restricted for such robots.
1321		-** For the purposes of this module, a "complex request" is an HTTP
1322		-** request with one or more query parameters other than "name".
1323		-**
1324		-** If this routine determines that robots should be restricted, then
1325		-** this routine publishes a redirect to the honeypot and exits without
1326		-** returning to the caller.
1327		-**
1328		-** This routine believes that this is a complex request is coming from
1329		-** a robot if all of the following are true:
1330		-**
1331		-** * The user is "nobody".
1332		-** * Either the REFERER field of the HTTP header is missing or empty,
1333		-** or the USERAGENT field of the HTTP header suggests that
1334		-** the request as coming from a robot.
1335		-** * There are one or more query parameters other than "name".
1336		-**
1337		-** Robot restrictions are governed by settings.
1338		-**
1339		-** robot-restrict The value is a list of GLOB patterns for pages
1340		-** that should restrict robot access. No restrictions
1341		-** are applied if this setting is undefined or is
1342		-** an empty string.
1343		-*/
1344		-void login_restrict_robot_access(void){
1345		- const char *zGlob;
1346		- int isMatch = 1;
1347		- int nQP; /* Number of query parameters other than name= */
1348		- if( g.zLogin!=0 ) return;
1349		- zGlob = db_get("robot-restrict",0);
1350		- if( zGlob==0 \|\| zGlob[0]==0 ) return;
1351		- if( g.isHuman ){
1352		- const char *zReferer;
1353		- const char *zAccept;
1354		- const char *zBr;
1355		- zReferer = P("HTTP_REFERER");
1356		- if( zReferer && zReferer[0]!=0 ) return;
1357		-
1358		- /* Robots typically do not accept the brotli encoding, at least not
1359		- ** at the time of this writing (2025-04-01), but standard web-browser
1360		- ** all generally do accept brotli. So if brotli is accepted,
1361		- ** assume we are not talking to a robot. We might want to revisit this
1362		- ** heuristic in the future...
1363		- */
1364		- if( (zAccept = P("HTTP_ACCEPT_ENCODING"))!=0
1365		- && (zBr = strstr(zAccept,"br"))!=0
1366		- && !fossil_isalnum(zBr[2])
1367		- && (zBr==zAccept \|\| !fossil_isalnum(zBr[-1]))
1368		- ){
1369		- return;
1370		- }
1371		- }
1372		- nQP = cgi_qp_count();
1373		- if( nQP<1 ) return;
1374		- isMatch = glob_multi_match(zGlob, g.zPath);
1375		- if( !isMatch ) return;
1376		-
1377		- /* Check for exceptions to the restriction on the number of query
1378		- ** parameters. */
1379		- zGlob = db_get("robot-restrict-qp",0);
1380		- if( zGlob && zGlob[0] ){
1381		- char *zPath = mprintf("%s/%d", g.zPath, nQP);
1382		- isMatch = glob_multi_match(zGlob, zPath);
1383		- fossil_free(zPath);
1384		- if( isMatch ) return;
1385		- }
1386		-
1387		- /* If we reach this point, it means we have a situation where we
1388		- ** want to restrict the activity of a robot.
1389		- */
1390		- g.isHuman = 0;
1391		- (void)exclude_spiders(0);
1392		- cgi_reply();
1393		- fossil_exit(0);
1394		-}
1395		-
1396	1308	/*
1397	1309	** When this routine is called, we know that the request does not
1398	1310	** have a login on the present repository. This routine checks to
1399	1311	** see if their login cookie might be for another member of the
1400	1312	** login-group.
		@@ -1429,11 +1341,11 @@
1429	1341	**
1430	1342	** g.userUid Database USER.UID value. Might be -1 for "nobody"
1431	1343	** g.zLogin Database USER.LOGIN value. NULL for user "nobody"
1432	1344	** g.perm Permissions granted to this user
1433	1345	** g.anon Permissions that would be available to anonymous
1434		-** g.isHuman True if the user is human, not a spider or robot
	1346	+** g.isRobot True if the client is known to be a spider or robot
1435	1347	** g.perm Populated based on user account's capabilities
1436	1348	**
1437	1349	*/
1438	1350	void login_check_credentials(void){
1439	1351	int uid = 0; /* User id */
		@@ -1470,11 +1382,11 @@
1470	1382	uid = db_int(0, "SELECT uid FROM user WHERE cap LIKE '%%s%%'");
1471	1383	}
1472	1384	g.zLogin = db_text("?", "SELECT login FROM user WHERE uid=%d", uid);
1473	1385	zCap = "sxy";
1474	1386	g.noPswd = 1;
1475		- g.isHuman = 1;
	1387	+ g.isRobot = 0;
1476	1388	zSeed = db_text("??", "SELECT uid\|\|quote(login)\|\|quote(pw)\|\|quote(cookie)"
1477	1389	" FROM user WHERE uid=%d", uid);
1478	1390	login_create_csrf_secret(zSeed);
1479	1391	fossil_free(zSeed);
1480	1392	}
		@@ -1604,12 +1516,15 @@
1604	1516	login_create_csrf_secret("none");
1605	1517	}
1606	1518
1607	1519	login_set_uid(uid, zCap);
1608	1520
1609		- /* Maybe restrict access to robots */
1610		- login_restrict_robot_access();
	1521	+ /* Maybe restrict access by robots */
	1522	+ if( g.zLogin==0 && robot_restrict(g.zPath) ){
	1523	+ cgi_reply();
	1524	+ fossil_exit(0);
	1525	+ }
1611	1526	}
1612	1527
1613	1528	/*
1614	1529	** Set the current logged in user to be uid. zCap is precomputed
1615	1530	** (override) capabilities. If zCap==0, then look up the capabilities
		@@ -1644,15 +1559,15 @@
1644	1559	g.userUid = uid;
1645	1560	if( fossil_strcmp(g.zLogin,"nobody")==0 ){
1646	1561	g.zLogin = 0;
1647	1562	}
1648	1563	if( PB("isrobot") ){
1649		- g.isHuman = 0;
	1564	+ g.isRobot = 1;
1650	1565	}else if( g.zLogin==0 ){
1651		- g.isHuman = isHuman(P("HTTP_USER_AGENT"));
	1566	+ g.isRobot = !isHuman(P("HTTP_USER_AGENT"));
1652	1567	}else{
1653		- g.isHuman = 1;
	1568	+ g.isRobot = 0;
1654	1569	}
1655	1570
1656	1571	/* Set the capabilities */
1657	1572	login_replace_capabilities(zCap, 0);
1658	1573
		@@ -1662,11 +1577,11 @@
1662	1577	** enabled for this repository and make appropriate adjustments to the
1663	1578	** permission flags if it is. This should be done before the permissions
1664	1579	** are (potentially) copied to the anonymous permission set; otherwise,
1665	1580	** those will be out-of-sync.
1666	1581	*/
1667		- if( zCap[0] && !g.perm.Hyperlink && g.isHuman ){
	1582	+ if( zCap[0] && !g.perm.Hyperlink && !g.isRobot ){
1668	1583	int autoLink = db_get_int("auto-hyperlink",1);
1669	1584	if( autoLink==1 ){
1670	1585	g.jsHref = 1;
1671	1586	g.perm.Hyperlink = 1;
1672	1587	}else if( autoLink==2 ){
1673	1588

	--- src/login.c
	+++ src/login.c
	@@ -1303,98 +1303,10 @@
1303	}
1304	fossil_free(zDecode);
1305	return uid;
1306	}
1307
1308	/*
1309	** SETTING: robot-restrict width=40 block-text
1310	** The VALUE of this setting is a list of GLOB patterns that match
1311	** pages for which complex HTTP requests from robots should be disallowed.
1312	** The recommended value for this setting is:
1313	**
1314	** timeline,vdiff,fdiff,annotate,blame
1315	**
1316	*/
1317
1318	/*
1319	** Check to see if the current HTTP request is a complex request that
1320	** is coming from a robot and if access should restricted for such robots.
1321	** For the purposes of this module, a "complex request" is an HTTP
1322	** request with one or more query parameters other than "name".
1323	**
1324	** If this routine determines that robots should be restricted, then
1325	** this routine publishes a redirect to the honeypot and exits without
1326	** returning to the caller.
1327	**
1328	** This routine believes that this is a complex request is coming from
1329	** a robot if all of the following are true:
1330	**
1331	** * The user is "nobody".
1332	** * Either the REFERER field of the HTTP header is missing or empty,
1333	** or the USERAGENT field of the HTTP header suggests that
1334	** the request as coming from a robot.
1335	** * There are one or more query parameters other than "name".
1336	**
1337	** Robot restrictions are governed by settings.
1338	**
1339	** robot-restrict The value is a list of GLOB patterns for pages
1340	** that should restrict robot access. No restrictions
1341	** are applied if this setting is undefined or is
1342	** an empty string.
1343	*/
1344	void login_restrict_robot_access(void){
1345	const char *zGlob;
1346	int isMatch = 1;
1347	int nQP; /* Number of query parameters other than name= */
1348	if( g.zLogin!=0 ) return;
1349	zGlob = db_get("robot-restrict",0);
1350	if( zGlob==0 \|\| zGlob[0]==0 ) return;
1351	if( g.isHuman ){
1352	const char *zReferer;
1353	const char *zAccept;
1354	const char *zBr;
1355	zReferer = P("HTTP_REFERER");
1356	if( zReferer && zReferer[0]!=0 ) return;
1357
1358	/* Robots typically do not accept the brotli encoding, at least not
1359	** at the time of this writing (2025-04-01), but standard web-browser
1360	** all generally do accept brotli. So if brotli is accepted,
1361	** assume we are not talking to a robot. We might want to revisit this
1362	** heuristic in the future...
1363	*/
1364	if( (zAccept = P("HTTP_ACCEPT_ENCODING"))!=0
1365	&& (zBr = strstr(zAccept,"br"))!=0
1366	&& !fossil_isalnum(zBr[2])
1367	&& (zBr==zAccept \|\| !fossil_isalnum(zBr[-1]))
1368	){
1369	return;
1370	}
1371	}
1372	nQP = cgi_qp_count();
1373	if( nQP<1 ) return;
1374	isMatch = glob_multi_match(zGlob, g.zPath);
1375	if( !isMatch ) return;
1376
1377	/* Check for exceptions to the restriction on the number of query
1378	** parameters. */
1379	zGlob = db_get("robot-restrict-qp",0);
1380	if( zGlob && zGlob[0] ){
1381	char *zPath = mprintf("%s/%d", g.zPath, nQP);
1382	isMatch = glob_multi_match(zGlob, zPath);
1383	fossil_free(zPath);
1384	if( isMatch ) return;
1385	}
1386
1387	/* If we reach this point, it means we have a situation where we
1388	** want to restrict the activity of a robot.
1389	*/
1390	g.isHuman = 0;
1391	(void)exclude_spiders(0);
1392	cgi_reply();
1393	fossil_exit(0);
1394	}
1395
1396	/*
1397	** When this routine is called, we know that the request does not
1398	** have a login on the present repository. This routine checks to
1399	** see if their login cookie might be for another member of the
1400	** login-group.
	@@ -1429,11 +1341,11 @@
1429	**
1430	** g.userUid Database USER.UID value. Might be -1 for "nobody"
1431	** g.zLogin Database USER.LOGIN value. NULL for user "nobody"
1432	** g.perm Permissions granted to this user
1433	** g.anon Permissions that would be available to anonymous
1434	** g.isHuman True if the user is human, not a spider or robot
1435	** g.perm Populated based on user account's capabilities
1436	**
1437	*/
1438	void login_check_credentials(void){
1439	int uid = 0; /* User id */
	@@ -1470,11 +1382,11 @@
1470	uid = db_int(0, "SELECT uid FROM user WHERE cap LIKE '%%s%%'");
1471	}
1472	g.zLogin = db_text("?", "SELECT login FROM user WHERE uid=%d", uid);
1473	zCap = "sxy";
1474	g.noPswd = 1;
1475	g.isHuman = 1;
1476	zSeed = db_text("??", "SELECT uid\|\|quote(login)\|\|quote(pw)\|\|quote(cookie)"
1477	" FROM user WHERE uid=%d", uid);
1478	login_create_csrf_secret(zSeed);
1479	fossil_free(zSeed);
1480	}
	@@ -1604,12 +1516,15 @@
1604	login_create_csrf_secret("none");
1605	}
1606
1607	login_set_uid(uid, zCap);
1608
1609	/* Maybe restrict access to robots */
1610	login_restrict_robot_access();



1611	}
1612
1613	/*
1614	** Set the current logged in user to be uid. zCap is precomputed
1615	** (override) capabilities. If zCap==0, then look up the capabilities
	@@ -1644,15 +1559,15 @@
1644	g.userUid = uid;
1645	if( fossil_strcmp(g.zLogin,"nobody")==0 ){
1646	g.zLogin = 0;
1647	}
1648	if( PB("isrobot") ){
1649	g.isHuman = 0;
1650	}else if( g.zLogin==0 ){
1651	g.isHuman = isHuman(P("HTTP_USER_AGENT"));
1652	}else{
1653	g.isHuman = 1;
1654	}
1655
1656	/* Set the capabilities */
1657	login_replace_capabilities(zCap, 0);
1658
	@@ -1662,11 +1577,11 @@
1662	** enabled for this repository and make appropriate adjustments to the
1663	** permission flags if it is. This should be done before the permissions
1664	** are (potentially) copied to the anonymous permission set; otherwise,
1665	** those will be out-of-sync.
1666	*/
1667	if( zCap[0] && !g.perm.Hyperlink && g.isHuman ){
1668	int autoLink = db_get_int("auto-hyperlink",1);
1669	if( autoLink==1 ){
1670	g.jsHref = 1;
1671	g.perm.Hyperlink = 1;
1672	}else if( autoLink==2 ){
1673

	--- src/login.c
	+++ src/login.c
	@@ -1303,98 +1303,10 @@
1303	}
1304	fossil_free(zDecode);
1305	return uid;
1306	}
1307
























































































1308	/*
1309	** When this routine is called, we know that the request does not
1310	** have a login on the present repository. This routine checks to
1311	** see if their login cookie might be for another member of the
1312	** login-group.
	@@ -1429,11 +1341,11 @@
1341	**
1342	** g.userUid Database USER.UID value. Might be -1 for "nobody"
1343	** g.zLogin Database USER.LOGIN value. NULL for user "nobody"
1344	** g.perm Permissions granted to this user
1345	** g.anon Permissions that would be available to anonymous
1346	** g.isRobot True if the client is known to be a spider or robot
1347	** g.perm Populated based on user account's capabilities
1348	**
1349	*/
1350	void login_check_credentials(void){
1351	int uid = 0; /* User id */
	@@ -1470,11 +1382,11 @@
1382	uid = db_int(0, "SELECT uid FROM user WHERE cap LIKE '%%s%%'");
1383	}
1384	g.zLogin = db_text("?", "SELECT login FROM user WHERE uid=%d", uid);
1385	zCap = "sxy";
1386	g.noPswd = 1;
1387	g.isRobot = 0;
1388	zSeed = db_text("??", "SELECT uid\|\|quote(login)\|\|quote(pw)\|\|quote(cookie)"
1389	" FROM user WHERE uid=%d", uid);
1390	login_create_csrf_secret(zSeed);
1391	fossil_free(zSeed);
1392	}
	@@ -1604,12 +1516,15 @@
1516	login_create_csrf_secret("none");
1517	}
1518
1519	login_set_uid(uid, zCap);
1520
1521	/* Maybe restrict access by robots */
1522	if( g.zLogin==0 && robot_restrict(g.zPath) ){
1523	cgi_reply();
1524	fossil_exit(0);
1525	}
1526	}
1527
1528	/*
1529	** Set the current logged in user to be uid. zCap is precomputed
1530	** (override) capabilities. If zCap==0, then look up the capabilities
	@@ -1644,15 +1559,15 @@
1559	g.userUid = uid;
1560	if( fossil_strcmp(g.zLogin,"nobody")==0 ){
1561	g.zLogin = 0;
1562	}
1563	if( PB("isrobot") ){
1564	g.isRobot = 1;
1565	}else if( g.zLogin==0 ){
1566	g.isRobot = !isHuman(P("HTTP_USER_AGENT"));
1567	}else{
1568	g.isRobot = 0;
1569	}
1570
1571	/* Set the capabilities */
1572	login_replace_capabilities(zCap, 0);
1573
	@@ -1662,11 +1577,11 @@
1577	** enabled for this repository and make appropriate adjustments to the
1578	** permission flags if it is. This should be done before the permissions
1579	** are (potentially) copied to the anonymous permission set; otherwise,
1580	** those will be out-of-sync.
1581	*/
1582	if( zCap[0] && !g.perm.Hyperlink && !g.isRobot ){
1583	int autoLink = db_get_int("auto-hyperlink",1);
1584	if( autoLink==1 ){
1585	g.jsHref = 1;
1586	g.perm.Hyperlink = 1;
1587	}else if( autoLink==2 ){
1588

M src/main.c

+2 -1

		--- src/main.c
		+++ src/main.c
		@@ -233,11 +233,12 @@
233	233	* applicable when using SEE on Windows or Linux. */
234	234	#endif
235	235	int useLocalauth; /* No login required if from 127.0.0.1 */
236	236	int noPswd; /* Logged in without password (on 127.0.0.1) */
237	237	int userUid; /* Integer user id */
238		- int isHuman; /* True if access by a human, not a spider or bot */
	238	+ int isRobot; /* True if the client is definitely a robot. False
	239	+ ** negatives are common for this flag */
239	240	int comFmtFlags; /* Zero or more "COMMENT_PRINT_*" bit flags, should be
240	241	** accessed through get_comment_format(). */
241	242	const char zSockName; / Name of the unix-domain socket file */
242	243	const char zSockMode; / File permissions for unix-domain socket */
243	244	const char zSockOwner; / Owner, or owner:group for unix-domain socket */
244	245

	--- src/main.c
	+++ src/main.c
	@@ -233,11 +233,12 @@
233	* applicable when using SEE on Windows or Linux. */
234	#endif
235	int useLocalauth; /* No login required if from 127.0.0.1 */
236	int noPswd; /* Logged in without password (on 127.0.0.1) */
237	int userUid; /* Integer user id */
238	int isHuman; /* True if access by a human, not a spider or bot */

239	int comFmtFlags; /* Zero or more "COMMENT_PRINT_*" bit flags, should be
240	** accessed through get_comment_format(). */
241	const char zSockName; / Name of the unix-domain socket file */
242	const char zSockMode; / File permissions for unix-domain socket */
243	const char zSockOwner; / Owner, or owner:group for unix-domain socket */
244

	--- src/main.c
	+++ src/main.c
	@@ -233,11 +233,12 @@
233	* applicable when using SEE on Windows or Linux. */
234	#endif
235	int useLocalauth; /* No login required if from 127.0.0.1 */
236	int noPswd; /* Logged in without password (on 127.0.0.1) */
237	int userUid; /* Integer user id */
238	int isRobot; /* True if the client is definitely a robot. False
239	** negatives are common for this flag */
240	int comFmtFlags; /* Zero or more "COMMENT_PRINT_*" bit flags, should be
241	** accessed through get_comment_format(). */
242	const char zSockName; / Name of the unix-domain socket file */
243	const char zSockMode; / File permissions for unix-domain socket */
244	const char zSockOwner; / Owner, or owner:group for unix-domain socket */
245

M src/robot.c

+109 -60

		--- src/robot.c
		+++ src/robot.c
		@@ -22,19 +22,11 @@
22	22	#include "config.h"
23	23	#include "robot.h"
24	24	#include <assert.h>
25	25	#include <time.h>
26	26
27		-/*
28		-** SETTING: robot-squelch width=10 default=200
29		-** The VALUE of is an integer between 0 and 1000 that determines how
30		-** readily Fossil will squelch requests from robots. A value of 0
31		-** means "never squelch requests". A value of 1000 means "always
32		-** squelch requests from user 'nobody'". For values greater than 0
33		-** and less than 1000, the decision to squelch is based on a variety
34		-** of heuristics, but is more likely to occur the larger the number.
35		-*/
	27	+#define POW_COOKIE "fossil-proofofwork"
36	28
37	29	/*
38	30	** Rewrite the current page with a robot squelch captcha and return 1.
39	31	**
40	32	** Or, if valid proof-of-work is present as either a query parameter or
		@@ -71,14 +63,14 @@
71	63	h2 = (h2 % 900000000) + 100000000;
72	64
73	65	/* If there is already a proof-of-work cookie with this value
74	66	** that means that the user agent has already authenticated.
75	67	*/
76		- z = P("fossil-proofofwork");
	68	+ z = P(POW_COOKIE);
77	69	if( z
78	70	&& (atoi(z)==h1 \|\| atoi(z)==h2)
79		- && !cgi_is_qp("fossil-proofofwork") ){
	71	+ && !cgi_is_qp(POW_COOKIE) ){
80	72	return 0;
81	73	}
82	74
83	75	/* Check for a proof query parameter. If found, that means that
84	76	** the captcha has just now passed, so set the proof-of-work cookie
		@@ -86,83 +78,140 @@
86	78	*/
87	79	z = P("proof");
88	80	if( z
89	81	&& (atoi(z)==h1 \|\| atoi(z)==h2)
90	82	){
91		- cgi_set_cookie("fossil-proofofwork",z,"/",900);
	83	+ cgi_set_cookie(POW_COOKIE,z,"/",900);
92	84	return 0;
93	85	}
94	86	cgi_tag_query_parameter("proof");
95	87
96	88	/* Ask the client to present proof-of-work */
97	89	cgi_reset_content();
98	90	cgi_set_content_type("text/html");
99		- style_header("Captcha");
100		- @ <h1>Prove That You Are Human</h1>
	91	+ style_header("Browser Verification");
	92	+ @ <h1 id="x1">Checking to see if you are a robot<span id="x2"></span></h1>
101	93	@ <form method="GET">
102		- @ <p>Press the button below</p><p>
	94	+ @ <p id="x3" style="visibility:hidden;">\
	95	+ @ Press <input type="submit" id="x5" value="Ok" focus> to continue</p>
103	96	cgi_query_parameters_to_hidden();
104		- @ <input id="vx" type="hidden" name="proof" value="0">
105		- @ <input id="cx" type="submit" value="Wait..." disabled>
	97	+ @ <input id="x4" type="hidden" name="proof" value="0">
106	98	@ </form>
107	99	@ <script nonce='%s(style_nonce())'>
108		- @ function Nhtot1520(x){return document.getElementById(x);}
109		- @ function Aoxlxzajv(h){\
110		- @ Nhtot1520("vx").value=h;\
111		- @ Nhtot1520("cx").value="Ok";\
112		- @ Nhtot1520("cx").disabled=false;\
113		- @ }
114		- @ function Vhcnyarsm(h,a){\
115		- @ if(a>0){setTimeout(Vhcnyarsm,1,h+a,a-1);}else{Aoxlxzajv(h);}\
116		- @ }
117		- k = 200 + h2%99;
	100	+ @ function aaa(x){return document.getElementById(x);}
	101	+ @ function bbb(h,a){
	102	+ @ aaa("x4").value=h
	103	+ @ if((a%%75)==0){
	104	+ @ aaa("x2").textContent=aaa("x2").textContent+".";
	105	+ @ }
	106	+ @ if(a>0){
	107	+ @ setTimeout(bbb,1,h+a,a-1);
	108	+ @ }else{
	109	+ @ aaa("x3").style.visibility="visible";
	110	+ @ aaa("x2").textContent="";
	111	+ @ aaa("x1").textContent="All clear";
	112	+ @ aaa("x5").focus();
	113	+ @ }
	114	+ @ }
	115	+ k = 800 + h2%99;
118	116	h2 = (k*k + k)/2;
119		- @ setTimeout(function(){Vhcnyarsm(%u(h1-h2),%u(k));},10);
	117	+ @ setTimeout(function(){bbb(%u(h1-h2),%u(k));},10);
120	118	@ </script>
121	119	style_finish_page();
122	120	return 1;
123	121	}
124	122
	123	+/*
	124	+** SETTING: robot-restrict width=40 block-text
	125	+** The VALUE of this setting is a list of GLOB patterns that match
	126	+** pages for which complex HTTP requests from unauthenicated clients
	127	+** should be disallowed. "Unauthenticated" means the user is "nobody".
	128	+** The recommended value for this setting is:
	129	+**
	130	+** timelineX,diff,annotate,zip,fileage,file
	131	+**
	132	+** The "diff" tag covers all diffing pages such as /vdiff, /fdiff, and
	133	+** /vpatch. The "annotate" tag also covers /blame and /praise. "zip"
	134	+** also covers /tarball and /sqlar. If a tag has an "X" character appended,
	135	+** then it only applies if query parameters are such that the page is
	136	+** particularly difficult to compute.
	137	+**
	138	+** In all other case, the tag should exactly match the page name.
	139	+*/
125	140
126	141	/*
127		-** WEBPAGE functions can invoke this routine with an argument
128		-** that is between 0 and 1000. Based on that argument, and on
129		-** other factors, this routine decides whether or not to squelch
130		-** the request. "Squelch" in this context, means to require the
131		-** client to show proof-of-work before the request is processed.
132		-** The idea here is to prevent server overload due to excess robot
133		-** traffic. If a robot (or any client application really) wants us
134		-** to spend a lot of CPU computing some result for it, then it needs
135		-** to first demonstrate good faith by doing some make-work for us.
136		-**
137		-** This routine returns true for a squelch and false if the original
138		-** request should go through.
139		-**
140		-** The input parameter is an estimate of how much CPU time
141		-** and bandwidth is needed to compute a response. The higher the
142		-** value of this parameter, the more likely this routine is to squelch
143		-** the page. A value of zero means "never squelch". A value of
144		-** 1000 means always squelch if the user is "nobody".
145		-**
146		-** Squelching only happens if the user is "nobody". If the request
147		-** comes from any other user, including user "anonymous", the request
148		-** is never squelched.
149		-*/
150		-int robot_squelch(int n){
151		- const char *zToken;
152		- int iSquelch;
153		- assert( n>=0 && n<=1000 );
154		- if( g.zLogin ) return 0; /* Logged in users always get through */
155		- if( n==0 ) return 0; /* Squelch is completely disabled */
	142	+** Return the default restriction GLOB
	143	+*/
	144	+const char *robot_restrict_default(void){
	145	+ return "timelineX,diff,annotate,zip,fileage,file";
	146	+}
	147	+/*
	148	+** Check to see if the page named in the argument is on the
	149	+** robot-restrict list. If it is on the list and if the user
	150	+** is "nobody" then bring up a captcha to test to make sure that
	151	+** client is not a robot.
	152	+**
	153	+** This routine returns true if a captcha was rendered and if subsequent
	154	+** page generation should be aborted. It returns false if the page
	155	+** should not be restricted and should be rendered normally.
	156	+*/
	157	+int robot_restrict(const char *zPage){
	158	+ const char *zGlob;
	159	+ const char *zToken;
	160	+ if( g.zLogin ) return 0; /* Logged in users always get through */
	161	+ zGlob = db_get("robot-restrict",robot_restrict_default());
	162	+ if( zGlob==0 \|\| zGlob[0]==0 ) return 0;
	163	+ if( !glob_multi_match(zGlob, zPage) ) return 0;
156	164	zToken = P("token");
157	165	if( zToken!=0
158	166	&& db_exists("SELECT 1 FROM config WHERE name='token-%q'", zToken)
159	167	){
160	168	return 0; /* There is a valid token= query parameter */
161	169	}
162		- iSquelch = db_get_int("robot-squelch",200);
163		- if( iSquelch<=0 ) return 0;
164		- if( n+iSquelch>=1000 && robot_proofofwork() ){
	170	+ if( robot_proofofwork() ){
165	171	return 1;
166	172	}
167	173	return 0;
168	174	}
	175	+
	176	+
	177	+/*
	178	+** WEBPAGE: test-robotck
	179	+**
	180	+** Run the robot_restrict() function using the value of the "name="
	181	+** query parameter as an argument. Used for testing the robot_restrict()
	182	+** logic.
	183	+**
	184	+** Whenever this page is successfully rendered (when it doesn't go to
	185	+** the captcha) it deletes the proof-of-work cookie. So reloading the
	186	+** page will reset the cookie and restart the verification.
	187	+*/
	188	+void robot_restrict_test_page(void){
	189	+ const char *zName = P("name");
	190	+ const char *zP1 = P("proof");
	191	+ const char *zP2 = P(POW_COOKIE);
	192	+ const char *z;
	193	+ if( zName==0 \|\| zName[0]==0 ) zName = g.zPath;
	194	+ login_check_credentials();
	195	+ if( !g.perm.Admin ){ login_needed(0); return; }
	196	+ g.zLogin = 0;
	197	+ if( robot_restrict(zName) ) return;
	198	+ style_set_current_feature("test");
	199	+ style_header("robot_restrict() test");
	200	+ @ <h1>Captcha passed</h1>
	201	+ @
	202	+ @ <p>
	203	+ if( zP1 && zP1[0] ){
	204	+ @ proof=%h(zP1)<br>
	205	+ }
	206	+ if( zP2 && zP2[0] ){
	207	+ @ fossil_proofofwork=%h(zP2)<br>
	208	+ cgi_set_cookie(POW_COOKIE,"",0,-1);
	209	+ }
	210	+ z = db_get("robot-restrict",robot_restrict_default());
	211	+ if( z && z[0] ){
	212	+ @ robot-restrict=%h(z)</br>
	213	+ }
	214	+ @ </p>
	215	+ @ <p><a href="%R/test-robotck/%h(zName)">Retry</a>
	216	+ style_finish_page();
	217	+}
169	218

	--- src/robot.c
	+++ src/robot.c
	@@ -22,19 +22,11 @@
22	#include "config.h"
23	#include "robot.h"
24	#include <assert.h>
25	#include <time.h>
26
27	/*
28	** SETTING: robot-squelch width=10 default=200
29	** The VALUE of is an integer between 0 and 1000 that determines how
30	** readily Fossil will squelch requests from robots. A value of 0
31	** means "never squelch requests". A value of 1000 means "always
32	** squelch requests from user 'nobody'". For values greater than 0
33	** and less than 1000, the decision to squelch is based on a variety
34	** of heuristics, but is more likely to occur the larger the number.
35	*/
36
37	/*
38	** Rewrite the current page with a robot squelch captcha and return 1.
39	**
40	** Or, if valid proof-of-work is present as either a query parameter or
	@@ -71,14 +63,14 @@
71	h2 = (h2 % 900000000) + 100000000;
72
73	/* If there is already a proof-of-work cookie with this value
74	** that means that the user agent has already authenticated.
75	*/
76	z = P("fossil-proofofwork");
77	if( z
78	&& (atoi(z)==h1 \|\| atoi(z)==h2)
79	&& !cgi_is_qp("fossil-proofofwork") ){
80	return 0;
81	}
82
83	/* Check for a proof query parameter. If found, that means that
84	** the captcha has just now passed, so set the proof-of-work cookie
	@@ -86,83 +78,140 @@
86	*/
87	z = P("proof");
88	if( z
89	&& (atoi(z)==h1 \|\| atoi(z)==h2)
90	){
91	cgi_set_cookie("fossil-proofofwork",z,"/",900);
92	return 0;
93	}
94	cgi_tag_query_parameter("proof");
95
96	/* Ask the client to present proof-of-work */
97	cgi_reset_content();
98	cgi_set_content_type("text/html");
99	style_header("Captcha");
100	@ <h1>Prove That You Are Human</h1>
101	@ <form method="GET">
102	@ <p>Press the button below</p><p>

103	cgi_query_parameters_to_hidden();
104	@ <input id="vx" type="hidden" name="proof" value="0">
105	@ <input id="cx" type="submit" value="Wait..." disabled>
106	@ </form>
107	@ <script nonce='%s(style_nonce())'>
108	@ function Nhtot1520(x){return document.getElementById(x);}
109	@ function Aoxlxzajv(h){\
110	@ Nhtot1520("vx").value=h;\
111	@ Nhtot1520("cx").value="Ok";\
112	@ Nhtot1520("cx").disabled=false;\
113	@ }
114	@ function Vhcnyarsm(h,a){\
115	@ if(a>0){setTimeout(Vhcnyarsm,1,h+a,a-1);}else{Aoxlxzajv(h);}\
116	@ }
117	k = 200 + h2%99;






118	h2 = (k*k + k)/2;
119	@ setTimeout(function(){Vhcnyarsm(%u(h1-h2),%u(k));},10);
120	@ </script>
121	style_finish_page();
122	return 1;
123	}
124

















125
126	/*
127	** WEBPAGE functions can invoke this routine with an argument
128	** that is between 0 and 1000. Based on that argument, and on
129	** other factors, this routine decides whether or not to squelch
130	** the request. "Squelch" in this context, means to require the
131	** client to show proof-of-work before the request is processed.
132	** The idea here is to prevent server overload due to excess robot
133	** traffic. If a robot (or any client application really) wants us
134	** to spend a lot of CPU computing some result for it, then it needs
135	** to first demonstrate good faith by doing some make-work for us.
136	**
137	** This routine returns true for a squelch and false if the original
138	** request should go through.
139	**
140	** The input parameter is an estimate of how much CPU time
141	** and bandwidth is needed to compute a response. The higher the
142	** value of this parameter, the more likely this routine is to squelch
143	** the page. A value of zero means "never squelch". A value of
144	** 1000 means always squelch if the user is "nobody".
145	**
146	** Squelching only happens if the user is "nobody". If the request
147	** comes from any other user, including user "anonymous", the request
148	** is never squelched.
149	*/
150	int robot_squelch(int n){
151	const char *zToken;
152	int iSquelch;
153	assert( n>=0 && n<=1000 );
154	if( g.zLogin ) return 0; /* Logged in users always get through */
155	if( n==0 ) return 0; /* Squelch is completely disabled */
156	zToken = P("token");
157	if( zToken!=0
158	&& db_exists("SELECT 1 FROM config WHERE name='token-%q'", zToken)
159	){
160	return 0; /* There is a valid token= query parameter */
161	}
162	iSquelch = db_get_int("robot-squelch",200);
163	if( iSquelch<=0 ) return 0;
164	if( n+iSquelch>=1000 && robot_proofofwork() ){
165	return 1;
166	}
167	return 0;
168	}











































169

	--- src/robot.c
	+++ src/robot.c
	@@ -22,19 +22,11 @@
22	#include "config.h"
23	#include "robot.h"
24	#include <assert.h>
25	#include <time.h>
26
27	#define POW_COOKIE "fossil-proofofwork"








28
29	/*
30	** Rewrite the current page with a robot squelch captcha and return 1.
31	**
32	** Or, if valid proof-of-work is present as either a query parameter or
	@@ -71,14 +63,14 @@
63	h2 = (h2 % 900000000) + 100000000;
64
65	/* If there is already a proof-of-work cookie with this value
66	** that means that the user agent has already authenticated.
67	*/
68	z = P(POW_COOKIE);
69	if( z
70	&& (atoi(z)==h1 \|\| atoi(z)==h2)
71	&& !cgi_is_qp(POW_COOKIE) ){
72	return 0;
73	}
74
75	/* Check for a proof query parameter. If found, that means that
76	** the captcha has just now passed, so set the proof-of-work cookie
	@@ -86,83 +78,140 @@
78	*/
79	z = P("proof");
80	if( z
81	&& (atoi(z)==h1 \|\| atoi(z)==h2)
82	){
83	cgi_set_cookie(POW_COOKIE,z,"/",900);
84	return 0;
85	}
86	cgi_tag_query_parameter("proof");
87
88	/* Ask the client to present proof-of-work */
89	cgi_reset_content();
90	cgi_set_content_type("text/html");
91	style_header("Browser Verification");
92	@ <h1 id="x1">Checking to see if you are a robot<span id="x2"></span></h1>
93	@ <form method="GET">
94	@ <p id="x3" style="visibility:hidden;">\
95	@ Press <input type="submit" id="x5" value="Ok" focus> to continue</p>
96	cgi_query_parameters_to_hidden();
97	@ <input id="x4" type="hidden" name="proof" value="0">

98	@ </form>
99	@ <script nonce='%s(style_nonce())'>
100	@ function aaa(x){return document.getElementById(x);}
101	@ function bbb(h,a){
102	@ aaa("x4").value=h
103	@ if((a%%75)==0){
104	@ aaa("x2").textContent=aaa("x2").textContent+".";
105	@ }
106	@ if(a>0){
107	@ setTimeout(bbb,1,h+a,a-1);
108	@ }else{
109	@ aaa("x3").style.visibility="visible";
110	@ aaa("x2").textContent="";
111	@ aaa("x1").textContent="All clear";
112	@ aaa("x5").focus();
113	@ }
114	@ }
115	k = 800 + h2%99;
116	h2 = (k*k + k)/2;
117	@ setTimeout(function(){bbb(%u(h1-h2),%u(k));},10);
118	@ </script>
119	style_finish_page();
120	return 1;
121	}
122
123	/*
124	** SETTING: robot-restrict width=40 block-text
125	** The VALUE of this setting is a list of GLOB patterns that match
126	** pages for which complex HTTP requests from unauthenicated clients
127	** should be disallowed. "Unauthenticated" means the user is "nobody".
128	** The recommended value for this setting is:
129	**
130	** timelineX,diff,annotate,zip,fileage,file
131	**
132	** The "diff" tag covers all diffing pages such as /vdiff, /fdiff, and
133	** /vpatch. The "annotate" tag also covers /blame and /praise. "zip"
134	** also covers /tarball and /sqlar. If a tag has an "X" character appended,
135	** then it only applies if query parameters are such that the page is
136	** particularly difficult to compute.
137	**
138	** In all other case, the tag should exactly match the page name.
139	*/
140
141	/*
142	** Return the default restriction GLOB
143	*/
144	const char *robot_restrict_default(void){
145	return "timelineX,diff,annotate,zip,fileage,file";
146	}
147	/*
148	** Check to see if the page named in the argument is on the
149	** robot-restrict list. If it is on the list and if the user
150	** is "nobody" then bring up a captcha to test to make sure that
151	** client is not a robot.
152	**
153	** This routine returns true if a captcha was rendered and if subsequent
154	** page generation should be aborted. It returns false if the page
155	** should not be restricted and should be rendered normally.
156	*/
157	int robot_restrict(const char *zPage){
158	const char *zGlob;
159	const char *zToken;
160	if( g.zLogin ) return 0; /* Logged in users always get through */
161	zGlob = db_get("robot-restrict",robot_restrict_default());
162	if( zGlob==0 \|\| zGlob[0]==0 ) return 0;
163	if( !glob_multi_match(zGlob, zPage) ) return 0;







164	zToken = P("token");
165	if( zToken!=0
166	&& db_exists("SELECT 1 FROM config WHERE name='token-%q'", zToken)
167	){
168	return 0; /* There is a valid token= query parameter */
169	}
170	if( robot_proofofwork() ){


171	return 1;
172	}
173	return 0;
174	}
175
176
177	/*
178	** WEBPAGE: test-robotck
179	**
180	** Run the robot_restrict() function using the value of the "name="
181	** query parameter as an argument. Used for testing the robot_restrict()
182	** logic.
183	**
184	** Whenever this page is successfully rendered (when it doesn't go to
185	** the captcha) it deletes the proof-of-work cookie. So reloading the
186	** page will reset the cookie and restart the verification.
187	*/
188	void robot_restrict_test_page(void){
189	const char *zName = P("name");
190	const char *zP1 = P("proof");
191	const char *zP2 = P(POW_COOKIE);
192	const char *z;
193	if( zName==0 \|\| zName[0]==0 ) zName = g.zPath;
194	login_check_credentials();
195	if( !g.perm.Admin ){ login_needed(0); return; }
196	g.zLogin = 0;
197	if( robot_restrict(zName) ) return;
198	style_set_current_feature("test");
199	style_header("robot_restrict() test");
200	@ <h1>Captcha passed</h1>
201	@
202	@ <p>
203	if( zP1 && zP1[0] ){
204	@ proof=%h(zP1)<br>
205	}
206	if( zP2 && zP2[0] ){
207	@ fossil_proofofwork=%h(zP2)<br>
208	cgi_set_cookie(POW_COOKIE,"",0,-1);
209	}
210	z = db_get("robot-restrict",robot_restrict_default());
211	if( z && z[0] ){
212	@ robot-restrict=%h(z)</br>
213	}
214	@ </p>
215	@ <p><a href="%R/test-robotck/%h(zName)">Retry</a>
216	style_finish_page();
217	}
218

M src/setup.c

+20 -37

		--- src/setup.c
		+++ src/setup.c
		@@ -495,20 +495,29 @@
495	495	@
496	496	@ <form action="%R/setup_robot" method="post"><div>
497	497	login_insert_csrf_secret();
498	498	@ <input type="submit" name="submit" value="Apply Changes"></p>
499	499	@ <hr>
500		- entry_attribute("Robot Squelch", 6, "robot-squelch", "rsq", "200", 0);
501		- @ <p>The "squelch" setting determines how aggressive Fossil is about
502		- @ trying to weed out robots using captchas. Squelch only applies to
503		- @ expensive requests from user "nobody". The higher the squelch setting,
504		- @ the more likely the request is to generate a captcha instead of the
505		- @ requested page. Squelch can be any integer between 0 and 1000.
506		- @ 0 means squelch is disabled and all requests go through without a
507		- @ captcha. 1000 means every expensive request from user "nobody" gets
508		- @ a captcha.
509		- @ (Property: "robot-squelch")</p>
	500	+ @ <p><b>Do not allow robots access to these pages.</b>
	501	+ @ <p> If the page name matches the GLOB pattern of this setting, and the
	502	+ @ users is "nobody", and the client has not previously passed a captcha
	503	+ @ test to show that it is not a robot, then the page is not displayed.
	504	+ @ A captcha test is is rendered instead.
	505	+ @ The recommended value for this setting is:
	506	+ @ <p>
	507	+ @ &emsp;&emsp;&emsp;<tt>%h(robot_restrict_default())</tt>
	508	+ @ <p>
	509	+ @ The "diff" tag covers all diffing pages such as /vdiff, /fdiff, and
	510	+ @ /vpatch. The "annotate" tag covers /annotate and also /blame and
	511	+ @ /praise. The "zip" covers itself and also /tarball and /sqlar. If a
	512	+ @ tag has an "X" character appended, then it only applies if query
	513	+ @ parameters are such that the page is particularly difficult to compute.
	514	+ @ In all other case, the tag should exactly match the page name.
	515	+ @ (Property: robot-restrict)
	516	+ @ <br>
	517	+ textarea_attribute("", 2, 80,
	518	+ "robot-restrict", "rbrestrict", robot_restrict_default(), 0);
510	519
511	520	@ <hr>
512	521	addAutoHyperlinkSettings();
513	522
514	523	@ <hr>
		@@ -520,37 +529,11 @@
520	529	@ computations here. Set this to 0.0 to disable the load average limit.
521	530	@ This limit is only enforced on Unix servers. On Linux systems,
522	531	@ access to the /proc virtual filesystem is required, which means this limit
523	532	@ might not work inside a chroot() jail.
524	533	@ (Property: "max-loadavg")</p>
525		-
526		- @ <hr>
527		- @ <p><b>Do not allow robots to make complex requests
528		- @ against the following pages.</b>
529		- @ <p> A "complex request" is an HTTP request that has one or more query
530		- @ parameters. Some robots will spend hours juggling around query parameters
531		- @ or even forging fake query parameters in an effort to discover new
532		- @ behavior or to find an SQL injection opportunity or similar. This can
533		- @ waste hours of CPU time and gigabytes of bandwidth on the server. A
534		- @ suggested value for this setting is:
535		- @ "<tt>timeline,*diff,vpatch,annotate,blame,praise,dir,tree</tt>".
536		- @ (Property: robot-restrict)
537		- @ <br>
538		- textarea_attribute("", 2, 80,
539		- "robot-restrict", "rbrestrict", "", 0);
540		- @ <br> The following comma-separated GLOB pattern allows for exceptions
541		- @ in the maximum number of query parameters before a request is considered
542		- @ complex. If this GLOB pattern exists and is non-empty and if it
543		- @ matches against the pagename followed by "/" and the number of query
544		- @ parameters, then the request is allowed through. For example, the
545		- @ suggested pattern of "timeline/[012]" allows the /timeline page to
546		- @ pass with up to 2 query parameters besides "name".
547		- @ (Property: robot-restrict-qp)
548		- @ <br>
549		- textarea_attribute("", 2, 80,
550		- "robot-restrict-qp", "rbrestrictqp", "", 0);
551		-
	534	+ @
552	535	@ <hr>
553	536	@ <p><input type="submit" name="submit" value="Apply Changes"></p>
554	537	@ </div></form>
555	538	db_end_transaction(0);
556	539	style_finish_page();
557	540

	--- src/setup.c
	+++ src/setup.c
	@@ -495,20 +495,29 @@
495	@
496	@ <form action="%R/setup_robot" method="post"><div>
497	login_insert_csrf_secret();
498	@ <input type="submit" name="submit" value="Apply Changes"></p>
499	@ <hr>
500	entry_attribute("Robot Squelch", 6, "robot-squelch", "rsq", "200", 0);
501	@ <p>The "squelch" setting determines how aggressive Fossil is about
502	@ trying to weed out robots using captchas. Squelch only applies to
503	@ expensive requests from user "nobody". The higher the squelch setting,
504	@ the more likely the request is to generate a captcha instead of the
505	@ requested page. Squelch can be any integer between 0 and 1000.
506	@ 0 means squelch is disabled and all requests go through without a
507	@ captcha. 1000 means every expensive request from user "nobody" gets
508	@ a captcha.
509	@ (Property: "robot-squelch")</p>









510
511	@ <hr>
512	addAutoHyperlinkSettings();
513
514	@ <hr>
	@@ -520,37 +529,11 @@
520	@ computations here. Set this to 0.0 to disable the load average limit.
521	@ This limit is only enforced on Unix servers. On Linux systems,
522	@ access to the /proc virtual filesystem is required, which means this limit
523	@ might not work inside a chroot() jail.
524	@ (Property: "max-loadavg")</p>
525
526	@ <hr>
527	@ <p><b>Do not allow robots to make complex requests
528	@ against the following pages.</b>
529	@ <p> A "complex request" is an HTTP request that has one or more query
530	@ parameters. Some robots will spend hours juggling around query parameters
531	@ or even forging fake query parameters in an effort to discover new
532	@ behavior or to find an SQL injection opportunity or similar. This can
533	@ waste hours of CPU time and gigabytes of bandwidth on the server. A
534	@ suggested value for this setting is:
535	@ "<tt>timeline,*diff,vpatch,annotate,blame,praise,dir,tree</tt>".
536	@ (Property: robot-restrict)
537	@ <br>
538	textarea_attribute("", 2, 80,
539	"robot-restrict", "rbrestrict", "", 0);
540	@ <br> The following comma-separated GLOB pattern allows for exceptions
541	@ in the maximum number of query parameters before a request is considered
542	@ complex. If this GLOB pattern exists and is non-empty and if it
543	@ matches against the pagename followed by "/" and the number of query
544	@ parameters, then the request is allowed through. For example, the
545	@ suggested pattern of "timeline/[012]" allows the /timeline page to
546	@ pass with up to 2 query parameters besides "name".
547	@ (Property: robot-restrict-qp)
548	@ <br>
549	textarea_attribute("", 2, 80,
550	"robot-restrict-qp", "rbrestrictqp", "", 0);
551
552	@ <hr>
553	@ <p><input type="submit" name="submit" value="Apply Changes"></p>
554	@ </div></form>
555	db_end_transaction(0);
556	style_finish_page();
557

	--- src/setup.c
	+++ src/setup.c
	@@ -495,20 +495,29 @@
495	@
496	@ <form action="%R/setup_robot" method="post"><div>
497	login_insert_csrf_secret();
498	@ <input type="submit" name="submit" value="Apply Changes"></p>
499	@ <hr>
500	@ <p><b>Do not allow robots access to these pages.</b>
501	@ <p> If the page name matches the GLOB pattern of this setting, and the
502	@ users is "nobody", and the client has not previously passed a captcha
503	@ test to show that it is not a robot, then the page is not displayed.
504	@ A captcha test is is rendered instead.
505	@ The recommended value for this setting is:
506	@ <p>
507	@ &emsp;&emsp;&emsp;<tt>%h(robot_restrict_default())</tt>
508	@ <p>
509	@ The "diff" tag covers all diffing pages such as /vdiff, /fdiff, and
510	@ /vpatch. The "annotate" tag covers /annotate and also /blame and
511	@ /praise. The "zip" covers itself and also /tarball and /sqlar. If a
512	@ tag has an "X" character appended, then it only applies if query
513	@ parameters are such that the page is particularly difficult to compute.
514	@ In all other case, the tag should exactly match the page name.
515	@ (Property: robot-restrict)
516	@ <br>
517	textarea_attribute("", 2, 80,
518	"robot-restrict", "rbrestrict", robot_restrict_default(), 0);
519
520	@ <hr>
521	addAutoHyperlinkSettings();
522
523	@ <hr>
	@@ -520,37 +529,11 @@
529	@ computations here. Set this to 0.0 to disable the load average limit.
530	@ This limit is only enforced on Unix servers. On Linux systems,
531	@ access to the /proc virtual filesystem is required, which means this limit
532	@ might not work inside a chroot() jail.
533	@ (Property: "max-loadavg")</p>
534	@


























535	@ <hr>
536	@ <p><input type="submit" name="submit" value="Apply Changes"></p>
537	@ </div></form>
538	db_end_transaction(0);
539	style_finish_page();
540

M src/style.c

+1 -46

		--- src/style.c
		+++ src/style.c
		@@ -1390,55 +1390,10 @@
1390	1390	*/
1391	1391	void page_test_env(void){
1392	1392	webpage_error("");
1393	1393	}
1394	1394
1395		-/*
1396		-** WEBPAGE: honeypot
1397		-** This page is a honeypot for spiders and bots.
1398		-*/
1399		-void honeypot_page(void){
1400		- unsigned int uSeed = captcha_seed();
1401		- const char *zDecoded = captcha_decode(uSeed, 0);
1402		- int bAutoCaptcha = db_get_boolean("auto-captcha", 0);
1403		- char *zCaptcha = captcha_render(zDecoded);
1404		- style_header("I think you are a robot");
1405		- @ <p>You seem like a robot.</p>
1406		- @
1407		- @ <p>Is that incorrect? Are you really human?
1408		- @ If so, please prove it by transcribing the captcha text
1409		- @ into the entry box below and pressing "Submit".
1410		- @ <form action="%R/login" method="post">
1411		- @ <input type="hidden" id="u" name="u" value="anonymous">
1412		- @ <p>
1413		- @ Captcha: <input type="text" id="p" name="p" value="">
1414		- @ <input type="submit" name="in" value="Submit">
1415		- @
1416		- @ <p>Alternatively, you can <a href="%R/login">log in</a> using an
1417		- @ existing userid.
1418		- @
1419		- @ <p><input type="hidden" name="cs" value="%u(uSeed)">
1420		- @ <div class="captcha"><table class="captcha"><tr><td>\
1421		- @ <pre class="captcha">
1422		- @ %h(zCaptcha)
1423		- @ </pre></td></tr></table>
1424		- if( bAutoCaptcha ) {
1425		- @ <input type="button" value="Fill out captcha" id='autofillButton' \
1426		- @ data-af='%s(zDecoded)'>
1427		- builtin_request_js("login.js");
1428		- }
1429		- @ </div>
1430		- free(zCaptcha);
1431		- @
1432		- @ <p>We regret this inconvenience. However, robots have become so
1433		- @ prolific and so aggressive that they will soak up too much CPU time
1434		- @ and network bandwidth on our servers if allowed to run unchecked.
1435		- @ Your cooperation in demonstrating that you are human is
1436		- @ appreciated.
1437		- style_finish_page();
1438		-}
1439		-
1440	1395	/*
1441	1396	** Webpages that encounter an error due to missing or incorrect
1442	1397	** query parameters can jump to this routine to render an error
1443	1398	** message screen.
1444	1399	**
		@@ -1487,11 +1442,11 @@
1487	1442	@ g.zHttpsURL = %h(g.zHttpsURL)<br>
1488	1443	@ g.zTop = %h(g.zTop)<br>
1489	1444	@ g.zPath = %h(g.zPath)<br>
1490	1445	@ g.userUid = %d(g.userUid)<br>
1491	1446	@ g.zLogin = %h(g.zLogin)<br>
1492		- @ g.isHuman = %d(g.isHuman)<br>
	1447	+ @ g.isRobot = %d(g.isRobot)<br>
1493	1448	@ g.jsHref = %d(g.jsHref)<br>
1494	1449	if( g.zLocalRoot ){
1495	1450	@ g.zLocalRoot = %h(g.zLocalRoot)<br>
1496	1451	}else{
1497	1452	@ g.zLocalRoot = <i>none</i><br>
1498	1453

	--- src/style.c
	+++ src/style.c
	@@ -1390,55 +1390,10 @@
1390	*/
1391	void page_test_env(void){
1392	webpage_error("");
1393	}
1394
1395	/*
1396	** WEBPAGE: honeypot
1397	** This page is a honeypot for spiders and bots.
1398	*/
1399	void honeypot_page(void){
1400	unsigned int uSeed = captcha_seed();
1401	const char *zDecoded = captcha_decode(uSeed, 0);
1402	int bAutoCaptcha = db_get_boolean("auto-captcha", 0);
1403	char *zCaptcha = captcha_render(zDecoded);
1404	style_header("I think you are a robot");
1405	@ <p>You seem like a robot.</p>
1406	@
1407	@ <p>Is that incorrect? Are you really human?
1408	@ If so, please prove it by transcribing the captcha text
1409	@ into the entry box below and pressing "Submit".
1410	@ <form action="%R/login" method="post">
1411	@ <input type="hidden" id="u" name="u" value="anonymous">
1412	@ <p>
1413	@ Captcha: <input type="text" id="p" name="p" value="">
1414	@ <input type="submit" name="in" value="Submit">
1415	@
1416	@ <p>Alternatively, you can <a href="%R/login">log in</a> using an
1417	@ existing userid.
1418	@
1419	@ <p><input type="hidden" name="cs" value="%u(uSeed)">
1420	@ <div class="captcha"><table class="captcha"><tr><td>\
1421	@ <pre class="captcha">
1422	@ %h(zCaptcha)
1423	@ </pre></td></tr></table>
1424	if( bAutoCaptcha ) {
1425	@ <input type="button" value="Fill out captcha" id='autofillButton' \
1426	@ data-af='%s(zDecoded)'>
1427	builtin_request_js("login.js");
1428	}
1429	@ </div>
1430	free(zCaptcha);
1431	@
1432	@ <p>We regret this inconvenience. However, robots have become so
1433	@ prolific and so aggressive that they will soak up too much CPU time
1434	@ and network bandwidth on our servers if allowed to run unchecked.
1435	@ Your cooperation in demonstrating that you are human is
1436	@ appreciated.
1437	style_finish_page();
1438	}
1439
1440	/*
1441	** Webpages that encounter an error due to missing or incorrect
1442	** query parameters can jump to this routine to render an error
1443	** message screen.
1444	**
	@@ -1487,11 +1442,11 @@
1487	@ g.zHttpsURL = %h(g.zHttpsURL)<br>
1488	@ g.zTop = %h(g.zTop)<br>
1489	@ g.zPath = %h(g.zPath)<br>
1490	@ g.userUid = %d(g.userUid)<br>
1491	@ g.zLogin = %h(g.zLogin)<br>
1492	@ g.isHuman = %d(g.isHuman)<br>
1493	@ g.jsHref = %d(g.jsHref)<br>
1494	if( g.zLocalRoot ){
1495	@ g.zLocalRoot = %h(g.zLocalRoot)<br>
1496	}else{
1497	@ g.zLocalRoot = <i>none</i><br>
1498

	--- src/style.c
	+++ src/style.c
	@@ -1390,55 +1390,10 @@
1390	*/
1391	void page_test_env(void){
1392	webpage_error("");
1393	}
1394













































1395	/*
1396	** Webpages that encounter an error due to missing or incorrect
1397	** query parameters can jump to this routine to render an error
1398	** message screen.
1399	**
	@@ -1487,11 +1442,11 @@
1442	@ g.zHttpsURL = %h(g.zHttpsURL)<br>
1443	@ g.zTop = %h(g.zTop)<br>
1444	@ g.zPath = %h(g.zPath)<br>
1445	@ g.userUid = %d(g.userUid)<br>
1446	@ g.zLogin = %h(g.zLogin)<br>
1447	@ g.isRobot = %d(g.isRobot)<br>
1448	@ g.jsHref = %d(g.jsHref)<br>
1449	if( g.zLocalRoot ){
1450	@ g.zLocalRoot = %h(g.zLocalRoot)<br>
1451	}else{
1452	@ g.zLocalRoot = <i>none</i><br>
1453

M src/tar.c

+1 -1

		--- src/tar.c
		+++ src/tar.c
		@@ -760,11 +760,11 @@
760	760	Blob tarball; /* Tarball accumulated here */
761	761	const char *z;
762	762
763	763	login_check_credentials();
764	764	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
765		- if( robot_squelch(900) ) return;
	765	+ if( robot_restrict("zip") ) return;
766	766	fossil_nice_default();
767	767	zName = fossil_strdup(PD("name",""));
768	768	z = P("r");
769	769	if( z==0 ) z = P("uuid");
770	770	if( z==0 ) z = tar_uuid_from_name(&zName);
771	771

	--- src/tar.c
	+++ src/tar.c
	@@ -760,11 +760,11 @@
760	Blob tarball; /* Tarball accumulated here */
761	const char *z;
762
763	login_check_credentials();
764	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
765	if( robot_squelch(900) ) return;
766	fossil_nice_default();
767	zName = fossil_strdup(PD("name",""));
768	z = P("r");
769	if( z==0 ) z = P("uuid");
770	if( z==0 ) z = tar_uuid_from_name(&zName);
771

	--- src/tar.c
	+++ src/tar.c
	@@ -760,11 +760,11 @@
760	Blob tarball; /* Tarball accumulated here */
761	const char *z;
762
763	login_check_credentials();
764	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
765	if( robot_restrict("zip") ) return;
766	fossil_nice_default();
767	zName = fossil_strdup(PD("name",""));
768	z = P("r");
769	if( z==0 ) z = P("uuid");
770	if( z==0 ) z = tar_uuid_from_name(&zName);
771

M src/timeline.c

		--- src/timeline.c
		+++ src/timeline.c
		@@ -1830,10 +1830,11 @@
1830	1830	\|\| (bisectLocal && !g.perm.Setup)
1831	1831	){
1832	1832	login_needed(g.anon.Read && g.anon.RdTkt && g.anon.RdWiki);
1833	1833	return;
1834	1834	}
	1835	+ if( zBefore && robot_restrict("timelineX") ) return;
1835	1836	if( !bisectLocal ){
1836	1837	etag_check(ETAG_QUERY\|ETAG_COOKIE\|ETAG_DATA\|ETAG_CONFIG, 0);
1837	1838	}
1838	1839	cookie_read_parameter("y","y");
1839	1840	zType = P("y");
1840	1841

	--- src/timeline.c
	+++ src/timeline.c
	@@ -1830,10 +1830,11 @@
1830	\|\| (bisectLocal && !g.perm.Setup)
1831	){
1832	login_needed(g.anon.Read && g.anon.RdTkt && g.anon.RdWiki);
1833	return;
1834	}

1835	if( !bisectLocal ){
1836	etag_check(ETAG_QUERY\|ETAG_COOKIE\|ETAG_DATA\|ETAG_CONFIG, 0);
1837	}
1838	cookie_read_parameter("y","y");
1839	zType = P("y");
1840

	--- src/timeline.c
	+++ src/timeline.c
	@@ -1830,10 +1830,11 @@
1830	\|\| (bisectLocal && !g.perm.Setup)
1831	){
1832	login_needed(g.anon.Read && g.anon.RdTkt && g.anon.RdWiki);
1833	return;
1834	}
1835	if( zBefore && robot_restrict("timelineX") ) return;
1836	if( !bisectLocal ){
1837	etag_check(ETAG_QUERY\|ETAG_COOKIE\|ETAG_DATA\|ETAG_CONFIG, 0);
1838	}
1839	cookie_read_parameter("y","y");
1840	zType = P("y");
1841

M src/zip.c

+1 -1

		--- src/zip.c
		+++ src/zip.c
		@@ -1012,11 +1012,11 @@
1012	1012	int eType = ARCHIVE_ZIP; /* Type of archive to generate */
1013	1013	char zType; / Human-readable archive type */
1014	1014
1015	1015	login_check_credentials();
1016	1016	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
1017		- if( robot_squelch(900) ) return;
	1017	+ if( robot_restrict("zip") ) return;
1018	1018	if( fossil_strcmp(g.zPath, "sqlar")==0 ){
1019	1019	eType = ARCHIVE_SQLAR;
1020	1020	zType = "SQL";
1021	1021	/* For some reason, SQL-archives are like catnip for robots. So
1022	1022	** don't allow them to be downloaded by user "nobody" */
1023	1023

	--- src/zip.c
	+++ src/zip.c
	@@ -1012,11 +1012,11 @@
1012	int eType = ARCHIVE_ZIP; /* Type of archive to generate */
1013	char zType; / Human-readable archive type */
1014
1015	login_check_credentials();
1016	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
1017	if( robot_squelch(900) ) return;
1018	if( fossil_strcmp(g.zPath, "sqlar")==0 ){
1019	eType = ARCHIVE_SQLAR;
1020	zType = "SQL";
1021	/* For some reason, SQL-archives are like catnip for robots. So
1022	** don't allow them to be downloaded by user "nobody" */
1023

	--- src/zip.c
	+++ src/zip.c
	@@ -1012,11 +1012,11 @@
1012	int eType = ARCHIVE_ZIP; /* Type of archive to generate */
1013	char zType; / Human-readable archive type */
1014
1015	login_check_credentials();
1016	if( !g.perm.Zip ){ login_needed(g.anon.Zip); return; }
1017	if( robot_restrict("zip") ) return;
1018	if( fossil_strcmp(g.zPath, "sqlar")==0 ){
1019	eType = ARCHIVE_SQLAR;
1020	zType = "SQL";
1021	/* For some reason, SQL-archives are like catnip for robots. So
1022	** don't allow them to be downloaded by user "nobody" */
1023

Fossil SCM

Keyboard Shortcuts