Add a whitelist of capsules mirrored on HTTP #8

rdelaage commented 2022年01月05日 16:12:45 +01:00

Currently it is impossible to follow a link to another capsule, it could be interesting to have a list of capsule also served on HTTP (e.g. automatic list of all capsules also served by the same ergol-http instance, manual list of other capsules like https://gemini.circumlunar.space/).

Instead of being redirected to a page showing a warning message you could be redirected the http(s) version of the capsule.

Currently it is impossible to follow a link to another capsule, it could be interesting to have a list of capsule also served on HTTP (e.g. automatic list of all capsules also served by the same ergol-http instance, manual list of other capsules like https://gemini.circumlunar.space/). Instead of being redirected to a page showing a warning message you could be redirected the http(s) version of the capsule.

adele commented 2022年01月05日 18:09:55 +01:00

Does it mean ergol-http connect to these other servers through gemeini protocol, parse content and propose html version ?

Does it mean ergol-http connect to these other servers through gemeini protocol, parse content and propose html version ?

rdelaage commented 2022年01月05日 18:53:59 +01:00

No, ergol-http just know that the capsule is mirrored on http (by itself or another server) so it redirect to this http mirror. For exemple example1.net is mirrored by ergol-http and has a link to examples.org, it knows that examples.org is mirrored by Kineto so it shows a to http://examples.org

No, ergol-http just know that the capsule is mirrored on http (by itself or another server) so it redirect to this http mirror. For exemple example1.net is mirrored by ergol-http and has a link to examples.org, it knows that examples.org is mirrored by Kineto so it shows a to http://examples.org

rdelaage commented 2022年04月28日 09:01:21 +02:00

Do you think something like this is adequate ?

diff --git a/config-sample.php b/config-sample.php
index f41ea0c..5c92601 100644
--- a/config-sample.php
+++ b/config-sample.php
@@ -7,3 +7,5 @@ define('CONFIG_TYPE', 'ergol');
 // See ergol.json description on
 // https://codeberg.org/adele.work/ergol#configuration-ergol-json
 define('CONFIG_PATH', __DIR__."/../ergol.json");
+
+define('WHITELIST_DOMAINS_PATH', __DIR__.'/whitelist_domains.txt');
diff --git a/index.php b/index.php
index 33b145d..7c7f12b 100755
--- a/index.php
+++ b/index.php
@@ -131,6 +131,8 @@ if(strpos($_SERVER['HTTP_HOST'],':')!==false)
 else
 	$capsule = strtolower($_SERVER['HTTP_HOST']);
 
+$whitelist_domains = explode("\n", file_get_contents(WHITELIST_DOMAINS_PATH));
+
 $response = false;
 $response_code = 0;
 $body = false;
@@ -356,6 +358,7 @@ exit;
 
 function gmi2html($capsule, $body, $lang, $urlgem, $favicon)
 {
+	global $whitelist_domains;
 	if(isset($_SERVER['REQUEST_SCHEME'])) {
 		$scheme = $_SERVER['REQUEST_SCHEME'];
 	}
@@ -437,7 +440,12 @@ function gmi2html($capsule, $body, $lang, $urlgem, $favicon)
 					$lines[] = '<a href="'.str_replace('gemini://'.$capsule,$scheme.'://'.$_SERVER['HTTP_HOST'], $link[0]).'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>";
 				}
 				else if(str_starts_with($link[0], 'gemini://')) {
-					$lines[] = '<a href="/?qx='.urlencode($link[0]).'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>";
+					$link_domain = parse_url($link[0], PHP_URL_HOST);
+					if(in_array($link_domain, $whitelist_domains))
+						$link_href = str_replace('gemini://','http://',$link[0]);
+					else
+						$link_href = '/?qx='.urlencode($link[0]);
+					$lines[] = '<a href="'.$link_href.'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>";
 				}
 				else {
 					$lines[] = '<a href="'.$link[0].'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>";
 

The goal is to have a basic 'whitelist_domains.txt` file with the host you know that are mirrored on the web and this list is automatically completed with the host in your config file.

Do you think something like this is adequate ? ``` diff --git a/config-sample.php b/config-sample.php index f41ea0c..5c92601 100644 --- a/config-sample.php +++ b/config-sample.php @@ -7,3 +7,5 @@ define('CONFIG_TYPE', 'ergol'); // See ergol.json description on // https://codeberg.org/adele.work/ergol#configuration-ergol-json define('CONFIG_PATH', __DIR__."/../ergol.json"); + +define('WHITELIST_DOMAINS_PATH', __DIR__.'/whitelist_domains.txt'); diff --git a/index.php b/index.php index 33b145d..7c7f12b 100755 --- a/index.php +++ b/index.php @@ -131,6 +131,8 @@ if(strpos($_SERVER['HTTP_HOST'],':')!==false) else $capsule = strtolower($_SERVER['HTTP_HOST']); +$whitelist_domains = explode("\n", file_get_contents(WHITELIST_DOMAINS_PATH)); + $response = false; $response_code = 0; $body = false; @@ -356,6 +358,7 @@ exit; function gmi2html($capsule, $body, $lang, $urlgem, $favicon) { + global $whitelist_domains; if(isset($_SERVER['REQUEST_SCHEME'])) { $scheme = $_SERVER['REQUEST_SCHEME']; } @@ -437,7 +440,12 @@ function gmi2html($capsule, $body, $lang, $urlgem, $favicon) $lines[] = '<a href="'.str_replace('gemini://'.$capsule,$scheme.'://'.$_SERVER['HTTP_HOST'], $link[0]).'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>"; } else if(str_starts_with($link[0], 'gemini://')) { - $lines[] = '<a href="/?qx='.urlencode($link[0]).'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>"; + $link_domain = parse_url($link[0], PHP_URL_HOST); + if(in_array($link_domain, $whitelist_domains)) + $link_href = str_replace('gemini://','http://',$link[0]); + else + $link_href = '/?qx='.urlencode($link[0]); + $lines[] = '<a href="'.$link_href.'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>"; } else { $lines[] = '<a href="'.$link[0].'">'.htmlentities(empty($link[1])?rawurldecode($link[0]):$link[1])."</a>"; ``` The goal is to have a basic 'whitelist_domains.txt` file with the host you know that are mirrored on the web and this list is automatically completed with the host in your config file.

adele commented 2022年04月28日 10:46:34 +02:00

This seems great !

This seems great !

adele commented 2022年04月28日 10:51:14 +02:00

In fact, why having a whitelist_domains.txt ? all known hosts are in config file.

In fact, why having a whitelist_domains.txt ? all known hosts are in config file.

rdelaage commented 2022年04月28日 13:59:25 +02:00

Not really, for exemple my own server knows rdelaage.ovh and patu.re but if I want to link pollux.casa, it will show a warning because this is on an other server. So if I know that your capsule is proxied on http I can add it to whitelist_domains.txt to tell ergo-http "ok you can convert gemini://pollux.casa to http://pollux.casa, it will work".

The goal is to have a single $whitelist_domains with both the domains retrived from the config and the domains you manually added to the file.

Not really, for exemple my own server knows rdelaage.ovh and patu.re but if I want to link pollux.casa, it will show a warning because this is on an other server. So if I know that your capsule is proxied on http I can add it to whitelist_domains.txt to tell ergo-http "ok you can convert gemini://pollux.casa to http://pollux.casa, it will work". The goal is to have a single `$whitelist_domains` with both the domains retrived from the config and the domains you manually added to the file.

rdelaage commented 2022年04月28日 14:24:50 +02:00

I have opened a PR #9

I have opened a PR #9

adele commented 2022年04月28日 14:55:20 +02:00

You're right, let's go.

I think you have autorization to merge your PR, haven't you ?

You're right, let's go. I think you have autorization to merge your PR, haven't you ?

rdelaage commented 2022年04月28日 15:52:54 +02:00

Yes, this is done :)

Yes, this is done :)