Serve archived hydrillabugs.koszko.org website

4 files changed, 96 insertions, 22 deletions

diff --git a/Makefile b/Makefile
index 29705dc..4f5ef58 100644
--- a/Makefile
+++ b/Makefile
@@ -77,6 +77,7 @@ sample-malcontent:
 		"$$(realpath $@)" | \
 		make -C subrepos/hydrilla shell-with-hydrilla-only
 
+HYDRILLABUGS_HTML_DIR = $(TEST_ROOT_DIR)/var/www/hydrillabugs.koszko.org/html
 LETSENCRYPT_ETC_DIR = $(TEST_ROOT_DIR)/etc/letsencrypt
 EXIM_ETC_DIR = $(TEST_ROOT_DIR)/etc/exim
 DOVECOT_ETC_DIR = $(TEST_ROOT_DIR)/etc/dovecot
@@ -110,11 +111,21 @@ prepare-test-root: sample-malcontent
 		hydrillarepos.koszko.org/html      \
 		hydrilla.koszko.org/html/downloads \
 		hydrillarepos.koszko.org/html      \
+		hydrillabugs.koszko.org/html       \
 		; do \
 		mkdir -p $(TEST_ROOT_DIR)/var/www/"$$WWW_SUBDIR";             \
 		printf 'This is dummy file for %s :D\n' "$$WWW_SUBDIR"        \
 			> $(TEST_ROOT_DIR)/var/www/"$$WWW_SUBDIR"/index.html; \
 	done
+	printf '/index /index.html\n' > $(HYDRILLABUGS_HTML_DIR)/uri-map.txt
+	mkdir -p $(HYDRILLABUGS_HTML_DIR)/javascripts
+	printf 'console.log("dummy js");\n'                           \
+		> $(HYDRILLABUGS_HTML_DIR)'/javascripts/foo.js?31415'
+	mkdir -p $(HYDRILLABUGS_HTML_DIR)/stylesheets
+	printf 'dummy-css {\ncolor: #555;\n}\n'                         \
+		> $(HYDRILLABUGS_HTML_DIR)'/stylesheets/b?a=r&ba=z.css'
+	printf '/stylesheets/b?a=r&ba=z /stylesheets/b?a=r&ba=z.css' \
+		>> $(HYDRILLABUGS_HTML_DIR)/uri-map.txt
 	@# Prepare replacement `/var/log`
 	mkdir -p $(TEST_ROOT_DIR)/var/log
 	@# Prepare replacement `/etc/letsencrypt`
diff --git a/container.scm b/container.scm
index 9f51721..d57b565 100644
--- a/container.scm
+++ b/container.scm
@@ -232,27 +232,38 @@
  (koszko-httpd-site-conf
   (name-and-aliases (list "hydrillabugs.koszko.org" "hachettebugs.koszko.org"))
   (body
-   `(;; GDPR hack
-     ,(httpd-directive 'Header 'unset 'Set-Cookie)
+   `(,(httpd-directive 'DocumentRoot "/srv/http/hydrillabugs.koszko.org")
 
-     ,(httpd-tag 'Proxy '("*")
-                 (httpd-directive
-                  'Redirect 'permanent
-                  "/projects/hachette" "/projects/haketilo")
-                 ;; I don't remember why I added the following line so I'm
-                 ;; keeping it just in case.
-                 (httpd-directive 'RequestHeader 'unset 'Accept-Encoding))
+     ,(httpd-directive 'RewriteEngine 'On)
 
-     ,(httpd-directive
-       'ProxyPass "/projects/haketilo"
-       "http://10.207.87.1:21011/projects/hachette")
-     ,(httpd-directive
-       'ProxyPassReverse "/projects/haketilo"
-       "http://10.207.87.1:21011/projects/hachette")
+     ,(httpd-directive 'RewriteCond "%{REQUEST_METHOD}" "!GET")
+     ,(httpd-directive 'RewriteRule "^(.*)$" "/hydrillabugs-archived.html"
+                       "[L,R=301,QSD]")
+
+     ,(httpd-directive 'Alias "/hydrillabugs-archived.html"
+                       (local-file "./hydrillabugs-archived.html"))
+
+     ,(httpd-directive 'RewriteRule "/uri-map.txt" "/dummy" "[L,F]")
+
+     ;; Serve wget-archived website.
+     ,(httpd-directive 'RewriteMap 'add-ext
+                       "txt:/srv/http/hydrillabugs.koszko.org/uri-map.txt")
+
+     ,(httpd-tag 'Location (list "/javascripts")
+                 (httpd-directive 'ForceType "application/javascript"))
+
+     ,(httpd-directive 'RewriteCond "%{QUERY_STRING}" "!^$")
+     ,(httpd-directive 'RewriteCond "${add-ext:$1?%{QUERY_STRING}|NONE}"
+                       "!NONE")
+     ,(httpd-directive 'RewriteRule "^(.*)$" "${add-ext:$1?%{QUERY_STRING}}"
+                       "[L,R=301]")
+
+     ,(httpd-directive 'RewriteCond "%{QUERY_STRING}" "!^$")
+     ,(httpd-directive 'RewriteRule "^(.*)$" "$1?%{QUERY_STRING}?" "[L,QSL]")
 
-     ,(httpd-directive 'ProxyPass "/.well-known/acme-challenge/" "!")
-     ,(httpd-directive 'ProxyPass "/" "http://10.207.87.1:21011/")
-     ,(httpd-directive 'ProxyPassReverse "/" "http://10.207.87.1:21011/")))))
+     ,(httpd-directive 'RewriteCond "%{QUERY_STRING}" "^$")
+     ,(httpd-directive 'RewriteCond "${add-ext:$1|NONE}" "!NONE")
+     ,(httpd-directive 'RewriteRule "^(.*)$" "${add-ext:$1}" "[L,R=301]")))))
 
 (add-site-conf
  (koszko-httpd-site-conf
@@ -338,10 +349,10 @@
          (name "proxy_http_module")
          (file (file-append httpd "/modules/mod_proxy_http.so")))))
 
-(define %headers-module
+(define %rewrite-module
   (httpd-module
-   (name "headers_module")
-   (file (file-append httpd "/modules/mod_headers.so"))))
+   (name "rewrite_module")
+   (file (file-append httpd "/modules/mod_rewrite.so"))))
 
 ;; logio is needed for the '%O' log format directive
 (define %logio-module
@@ -372,7 +383,7 @@
       (modules `(,%cgid-module
                  ,%wsgi-module
                  ,@%proxy-http-modules
-                 ,%headers-module
+                 ,%rewrite-module
                  ,%logio-module
                  ,@%default-httpd-modules))
       (extra-config
diff --git a/guix-container.sh b/guix-container.sh
index 5d2983b..1d4d6e0 100755
--- a/guix-container.sh
+++ b/guix-container.sh
@@ -237,6 +237,7 @@ start() {
     KOSZKO_SIDELOAD_REAL="$HOST_SYSTEM_ROOT"/var/www/koszko.org/html
     HYDRILLA_HTTP_REAL="$HOST_SYSTEM_ROOT"/var/www/hydrilla.koszko.org/html
     HYDRILLAREPOS_HTTP_REAL="$HOST_SYSTEM_ROOT"/var/www/hydrillarepos.koszko.org/html
+    HYDRILLABUGS_HTTP_REAL="$HOST_SYSTEM_ROOT"/var/www/hydrillabugs.koszko.org/html
     LOG_REAL="$LOG_DIR"/container
     ETC_LETSENCRYPT_REAL="$HOST_SYSTEM_ROOT"/etc/letsencrypt
     ETC_EXIM_REAL="$HOST_SYSTEM_ROOT"/etc/exim
@@ -250,6 +251,7 @@ start() {
     KOSZKO_SIDELOAD_DIR_SHARE_OPT=--share="$KOSZKO_SIDELOAD_REAL"=/srv/http/koszko.org
     HYDRILLA_HTTP_DIR_SHARE_OPT=--share="$HYDRILLA_HTTP_REAL"=/srv/http/hydrilla.koszko.org
     HYDRILLAREPOS_HTTP_DIR_SHARE_OPT=--share="$HYDRILLAREPOS_HTTP_REAL"=/srv/http/hydrillarepos.koszko.org
+    HYDRILLABUGS_HTTP_DIR_SHARE_OPT=--share="$HYDRILLABUGS_HTTP_REAL"=/srv/http/hydrillabugs.koszko.org
     LOG_DIR_SHARE_OPT=--share="$LOG_REAL"=/var/log
     ETC_LETSENCRYPT_DIR_SHARE_OPT=--share="$ETC_LETSENCRYPT_REAL"=/etc/letsencrypt
     ETC_EXIM_DIR_SHARE_OPT=--share="$ETC_EXIM_REAL"=/etc/exim
@@ -266,6 +268,7 @@ start() {
     "$EXECUTABLE" "$KOSZKO_SIDELOAD_DIR_SHARE_OPT"                     \
                   "$HYDRILLA_HTTP_DIR_SHARE_OPT"                       \
                   "$HYDRILLAREPOS_HTTP_DIR_SHARE_OPT"                  \
+                  "$HYDRILLABUGS_HTTP_DIR_SHARE_OPT"                   \
                   "$LOG_DIR_SHARE_OPT"                                 \
                   "$ETC_LETSENCRYPT_DIR_SHARE_OPT"                     \
                   "$ETC_EXIM_DIR_SHARE_OPT"                            \
diff --git a/hydrillabugs-archived.html b/hydrillabugs-archived.html
new file mode 100644
index 0000000..360617a
--- /dev/null
+++ b/hydrillabugs-archived.html
@@ -0,0 +1,49 @@
+<!doctype html>
+<!--
+    SPDX-License-Identifier: CC0-1.0
+
+    Copyright (C) 2023 Wojtek Kosior
+
+    Available under the terms of Creative Commons Zero v1.0 Universal.
+  -->
+<html>
+  <head>
+    <title>Hydrillabugs archived</title>
+
+    <meta charset="utf-8" />
+    <meta http-equiv="Content-type" content="text/html; charset=utf-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1" />
+
+    <style>
+      body {
+          color: #555;
+      }
+
+      #main {
+          margin: auto;
+          max-width: 700px;
+      }
+    </style>
+  </head>
+
+  <body>
+    <div id="main">
+      <h1>Hydrillabugs archived</h1>
+      <p>
+        The feature you were trying to use is no longer available. The
+        Hydrillabugs issue tracker has been archived and no longer allows
+        registration, login nor upload of data. In case of questions, please
+        reach out to <a href="mailto:koszko@koszko.org">koszko@koszko.org</a>.
+      </p>
+      <script type="text/plain">
+        For anyone interested, archival was done with more or less these:
+
+        wget --timestamping --recursive --level=5 --execute robots=off --no-remove-listing --adjust-extension --page-requisites --reject-regex '^([^/]|/[^r]|/r[^e]|/re[^p]|/rep[^o]|/repo[^s]|/repos[^i]|/reposi[^t]|/reposit[^o]|/reposito[^r]|/repositor[^y]|/repository[^/])+[^0-9][.]html$' --rejected-log=./rejectlist.csv --quiet https://hydrillabugs.koszko.org
+        (cd hydrillabugs.koszko.org/; find . -type f) > filelist
+        awk '/\.(html|css)$/{match($1, @/\.(\/.*)\.(html|css)$/, arr); print arr[1] " " arr[1] "." arr[2]}' filelist > hydrillabugs.koszko.org/uri-map.txt
+
+        There are surely better ways. This is just what I came up with.
+      </script>
+    </div>
+  </body>
+</html>