add hydrilla.koszko.org virtualhost

9 files changed, 232 insertions, 18 deletions

diff --git a/.gitignore b/.gitignore
index fcd3471..2b7dbc1 100644
--- a/.gitignore
+++ b/.gitignore
@@ -9,3 +9,5 @@ log
 pidfile
 hosts
 *.touchfile
+schemas/
+sample-malcontent/
diff --git a/.gitmodules b/.gitmodules
index 06004a1..e343c8e 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -10,3 +10,6 @@
 [submodule "hydrilla-website"]
 	path = subrepos/hydrilla-website
 	url = ../hydrilla-website
+[submodule "hydrilla"]
+	path = subrepos/hydrilla
+	url = ../pydrilla
diff --git a/Makefile b/Makefile
index 7de829e..b0eb3a8 100644
--- a/Makefile
+++ b/Makefile
@@ -16,7 +16,8 @@ GUIX_TM = $(GUIX) time-machine --commit=$(GUIX_COMMIT) --
 GUIX_LOAD_PATHS = $$(                                 \
 	find subrepos/ -mindepth 1 -maxdepth 1        \
 	-exec printf ' -L %s/guix-module-dir/' {} ';' \
-	)
+	)                                             \
+	-L ./guix-module-dir/
 
 GUIX_SYS_CONTAINER = $(GUIX_TM) system container $(GUIX_LOAD_PATHS)
 
@@ -26,27 +27,31 @@ KOSZKO_ORG_WEBSITE_INFO = \
 	subrepos/koszko-org-website/src/koszko_org_website.egg-info/PKG-INFO
 HYDRILLA_WEBSITE_INFO = \
 	subrepos/hydrilla-website/src/hydrilla_website.egg-info/PKG-INFO
+HYDRILLA_INFO = subrepos/hydrilla/src/hydrilla.egg-info/PKG-INFO
+
+ALL_EGG_INFOS =                    \
+	$(KOSZKO_ORG_WEBSITE_INFO) \
+	$(HYDRILLA_WEBSITE_INFO)   \
+	$(HYDRILLA_INFO)
 
 TEST_ROOT_DIR = "/tmp/$$(pwd | sed 's|/|!|g')!!test-root"
 
-all: | container-runner.touchfile log
+all: | container-runner.touchfile log sample-malcontent
 
 $(KOSZKO_ORG_WEBSITE_INFO):
 	$(MAKE) -C subrepos/koszko-org-website dist
 $(HYDRILLA_WEBSITE_INFO):
 	$(MAKE) -C subrepos/hydrilla-website dist
+$(HYDRILLA_INFO):
+	$(MAKE) -C subrepos/hydrilla dist
 
-container-runner: | container.scm          \
-		$(KOSZKO_ORG_WEBSITE_INFO) \
-		$(HYDRILLA_WEBSITE_INFO)
-	$(GUIX_SYS_CONTAINER) container.scm -r container-runner
-	touch container-runner.touchfile
+container-runner: | container.scm $(ALL_EGG_INFOS) hydrilla-wsgi.py
+container-runner.touchfile: container.scm $(ALL_EGG_INFOS) hydrilla-wsgi.py
 
-container-runner.touchfile: container.scm  \
-		$(KOSZKO_ORG_WEBSITE_INFO) \
-		$(HYDRILLA_WEBSITE_INFO)
+container-runner container-runner.touchfile:
 	$(MAKE) clean-runner
-	$(MAKE) container-runner
+	$(GUIX_SYS_CONTAINER) container.scm -r container-runner
+	touch container-runner.touchfile
 
 hosts: hosts-extra /etc/hosts
 	cat $^ > $@
@@ -54,9 +59,16 @@ hosts: hosts-extra /etc/hosts
 log:
 	ln -sf $(TEST_ROOT_DIR)/var/log/guix-container $@
 
+sample-malcontent:
+	mkdir $@
+	printf 'hydrilla-builder -s tests/source-package-example/ -d %s' \
+		"$$(realpath $@)" | \
+		make -C subrepos/hydrilla shell-with-haketilo
+
 HYDRILLA_WEBSITE_ETC_DIR = $(TEST_ROOT_DIR)/etc/guix-container/hydrilla-website
+MALCONTENT_DIR = $(TEST_ROOT_DIR)/var/lib/hydrilla/malcontent_dirs
 
-prepare-test-root:
+prepare-test-root: sample-malcontent
 	rm -rf $(TEST_ROOT_DIR)
 	@# Prepare replacement `/var/www`
 	mkdir -p $(TEST_ROOT_DIR)/var/www/koszko.org/html
@@ -68,6 +80,10 @@ prepare-test-root:
 	printf 'test non-secret\n' > $(HYDRILLA_WEBSITE_ETC_DIR)/secret.txt
 	chgrp -R 133 $(HYDRILLA_WEBSITE_ETC_DIR)
 	chmod 640 $(HYDRILLA_WEBSITE_ETC_DIR)/secret.txt
+	@# Prepare replacement `/var/lib/hydrilla`
+	mkdir -p $(MALCONTENT_DIR)/api_v2
+	ln -sf api_v2 $(MALCONTENT_DIR)/api_v1
+	cp -r $</* $(MALCONTENT_DIR)/api_v2/
 
 GUIX_CONTAINER_FLAGS = -e ./container-runner -p ./pidfile -r $(TEST_ROOT_DIR)
 
@@ -101,12 +117,15 @@ reinstall:
 	rm -rf /usr/local/bin/guix-container
 	$(MAKE) install
 
+build-hydrilla-json-schemas:
+	$(GUIX_TM) build -L ./guix-module-dir/ hydrilla-json-schemas
+
 clean-runner:
 	rm -rf container-runner container-runner.touchfile
 
 clean: clean-runner
 	$(MAKE) -C subrepos/koszko-org-website clean
-	rm -rf log hosts
+	rm -rf log hosts schemas sample-malcontent
 
 .PHONY: all                                              \
 	clean-runner clean                               \
@@ -114,4 +133,4 @@ clean: clean-runner
 	start-container stop-container restart-container \
 	enter-container fake-client                      \
 	install reinstall                                \
-	sdists koszko-org-wesite-sdist
+	build-hydrilla-json-schemas
diff --git a/container.scm b/container.scm
index c50a75b..39163d9 100644
--- a/container.scm
+++ b/container.scm
@@ -4,13 +4,25 @@
 ;;
 ;; Available under the terms of Creative Commons Zero v1.0 Universal.
 
-(use-modules (gnu))
-(use-modules (koszko-org-website))
-(use-modules (hydrilla-website))
+(use-modules (gnu)
+             (koszko-org-website)
+             (hydrilla-website)
+             (hydrilla-json-schemas)
+             (hydrilla)
+             ;; The following 4 are needed to construct GUIX_PYTHONPATH for
+             ;; Hydrilla WSGI scripts.
+             (guix build-system python)
+             (guix packages)
+             (guix search-paths)
+             (guix modules))
 (use-package-modules web)
+(use-package-modules python)
 (use-service-modules web)
 (use-service-modules shepherd)
 
+(define %here
+  (getcwd))
+
 (define* (simple-wsgi-alias-gexp package wsgi-path #:key (aliased-path "/"))
   #~(let* ((wsgi-file (string-append #$package #$wsgi-path)))
       (format #f
@@ -102,6 +114,62 @@
          (simple-wsgi-alias-gexp hydrilla-website
                                  "/share/hydrilla-website/wsgi.py"))))
 
+(define %python-path-spec-sexp
+   (search-path-specification->sexp
+    (guix-pythonpath-search-path (package-version (default-python)))))
+
+(define %hydrilla-pythonpath-inputs
+  (cons hydrilla (map cadr (package-transitive-target-inputs hydrilla))))
+
+(define %hydrilla-pythonpath-gexp
+  (with-imported-modules (source-module-closure '((guix search-paths)))
+    #~(begin
+        (use-modules (guix search-paths))
+        (let ((evaluated-list (evaluate-search-paths
+                      (list (sexp->search-path-specification
+                             '#$%python-path-spec-sexp))
+                      '#$%hydrilla-pythonpath-inputs)))
+          (cdar evaluated-list)))))
+
+(define %hydrilla-wsgi-regex
+  "^/api_v[^/]+/((resource|mapping)/[^/]+[.]json|query|list_all)$")
+
+(define %hydrilla-virtualhost
+  (httpd-virtualhost
+   "*:80"
+   (list "\
+         ServerName hydrilla.koszko.org
+         ServerAlias www.hydrilla.koszko.org
+         ServerAdmin koszko@koszko.org
+
+         <If \"%{HTTP_HOST} != 'hydrilla.koszko.org'\">
+             Redirect permanent / http://hydrilla.koszko.org/
+         </If>
+         "
+         #~(format #f "Alias /schemas ~a\n"
+                   (string-append #$hydrilla-json-schemas
+                                  "/share/hydrilla-json-schemas"))
+         "\
+
+         DocumentRoot /var/lib/hydrilla/malcontent_dirs
+         <Location ~ \"^/api_v[^/]+/(resource|mapping)/\">
+             ForceType application/json
+         </Location>
+
+         WSGIScriptReloading On
+         "
+         (format #f "SetEnvIf Request_URI ~s MALCONENT_DIR=~a\n"
+                 "^/(api_v[0-9]+)/"
+                 "/var/lib/hydrilla/malcontent_dirs/$1")
+         #~(format #f "SetEnvIf Request_URI ~s HYDRILLA_GUIX_PYTHONPATH=~a\n"
+                   "^/api_v[0-9]+/"
+                   #$%hydrilla-pythonpath-gexp)
+         #~(format #f "WSGIScriptAliasMatch ~s ~a\n"
+                   #$%hydrilla-wsgi-regex
+                   (string-append
+                    #$(local-file (string-append %here "/hydrilla-wsgi.py"))
+                    "/$1")))))
+
 (define %wsgi-module
   (httpd-module
    (name "wsgi_module")
@@ -168,7 +236,8 @@
                                 %koszkonutek-tmp-virtualhost
                                 %hydrillabugs-virtualhost
                                 %hachettebugs-virtualhost
-                                %haketilo-virtualhost))
+                                %haketilo-virtualhost
+                                %hydrilla-virtualhost))
           (service
            (shepherd-service-type
             'dummy-network
diff --git a/guix-container.sh b/guix-container.sh
index afede73..3167ac7 100755
--- a/guix-container.sh
+++ b/guix-container.sh
@@ -84,10 +84,12 @@ start() {
     KOSZKO_SIDELOAD_REAL="$HOST_SYSTEM_ROOT"/var/www/koszko.org/html
     LOG_REAL="$LOG_DIR"/container
     ETC_REAL="$HOST_SYSTEM_ROOT"/etc/guix-container
+    VAR_HYDRILLA_REAL="$HOST_SYSTEM_ROOT"/var/lib/hydrilla
 
     HTTP_DIR_SHARE_OPT=--share="$KOSZKO_SIDELOAD_REAL"=/srv/http/koszko.org
     LOG_DIR_SHARE_OPT=--share="$LOG_REAL"=/var/log
     ETC_DIR_SHARE_OPT=--share="$ETC_REAL"=/etc
+    VAR_HYDRILLA_DIR_SHARE_OPT=--share="$VAR_HYDRILLA_REAL"=/var/lib/hydrilla
 
     mkdir --mode=700 -p "$LOG_DIR"
     mkdir --mode=700 -p "$LOG_DIR"/container
@@ -95,6 +97,7 @@ start() {
     "$EXECUTABLE" "$HTTP_DIR_SHARE_OPT"                                \
                   "$LOG_DIR_SHARE_OPT"                                 \
                   "$ETC_DIR_SHARE_OPT"                                 \
+                  "$VAR_HYDRILLA_DIR_SHARE_OPT"                        \
                   >> "$LOG_DIR"/stdout.log 2>> "$LOG_DIR"/stderr.log &
 
     GUILE_PID=$!
diff --git a/guix-module-dir/hydrilla-json-schemas.scm b/guix-module-dir/hydrilla-json-schemas.scm
new file mode 100644
index 0000000..a3a8dbe
--- /dev/null
+++ b/guix-module-dir/hydrilla-json-schemas.scm
@@ -0,0 +1,90 @@
+;; SPDX-License-Identifier: CC0-1.0
+
+;; Copyright (C) 2022 Wojtek Kosior <koszko@koszko.org>
+;;
+;; Available under the terms of Creative Commons Zero v1.0 Universal.
+
+(define-module (hydrilla-json-schemas)
+  #:use-module (guix packages)
+  #:use-module (guix gexp)
+  #:use-module (guix build-system copy)
+  #:use-module ((guix licenses) #:prefix license:)
+  #:use-module (ice-9 match)
+  #:use-module (ice-9 popen)
+  #:use-module (ice-9 rdelim)
+  #:use-module (ice-9 regex)
+  #:use-module (srfi srfi-26))
+
+(define %source-dir
+  (let* ((this-file (search-path %load-path "hydrilla-json-schemas.scm"))
+         (proj-dir (dirname (dirname this-file))))
+    (if (absolute-file-name? proj-dir)
+        proj-dir
+        (string-append (getcwd) "/" proj-dir))))
+
+(define %schema-git-locations
+  '(("d94ef4544faac662f49bed41700c9010804b2450" "-2.schema.json")
+    ("v1.0.1"                                   "-1.0.1.schema.json")
+    ("v1.0"                                     "-1.schema.json")))
+
+(define %schemas-subrepo
+  (string-append %source-dir "/subrepos/hydrilla/src/hydrilla/schemas/1.x"))
+
+(define %schemas-dest
+  (string-append %source-dir "/schemas"))
+
+(define* (run-git-in-subrepo #:rest args)
+  (apply open-pipe* (cons* OPEN_READ "git" "-C" %schemas-subrepo args)))
+
+(define (filter-locations port file-ending)
+  (let ((regex (string-append "[^[:space:]]+" (regexp-quote file-ending) "$")))
+    (filter (lambda (val) (not (unspecified? val)))
+            (map (lambda (line)
+                   (let ((match-result (string-match regex line)))
+                     (when match-result (match:substring match-result 0))))
+                 (string-split (read-string port) #\xA)))))
+
+(define (get-git-file rev filename)
+  (let* ((port (run-git-in-subrepo "show" (string-append rev ":" filename)))
+         (contents (read-string port)))
+    (close-port port)
+    contents))
+
+(define (write-schema-file filename contents)
+  (call-with-output-file (string-append %schemas-dest "/" filename)
+    (cut display contents <>)))
+
+(define (transfer-file rev filename)
+  (write-schema-file filename (get-git-file rev filename)))
+
+(define (prepare-schemas-directory)
+  (when (not (file-exists? %schemas-dest))
+    (mkdir %schemas-dest))
+  (for-each (match-lambda
+              ((rev file-ending)
+               (let* ((port (run-git-in-subrepo "ls-tree" rev))
+                      (filtered (filter-locations port file-ending)))
+                 (close-port port)
+                 (for-each (cut transfer-file rev <>) filtered))))
+            %schema-git-locations)
+  %schemas-dest)
+
+(define-public hydrilla-json-schemas
+  (package
+    (name "hydrilla-json-schemas")
+    (version "current")
+    (source
+     (local-file (prepare-schemas-directory) #:recursive? #t))
+    (build-system copy-build-system)
+    (arguments
+     '(#:install-plan '(("." "share/hydrilla-json-schemas/"
+                         #:include (".schema.json")))))
+    (home-page "https://git.koszko.org/hydrilla-json-schemas")
+    (synopsis
+     "JSON schemas used by Haketilo and Hydrilla")
+    (description "Haketilo HTTP proxy facilitates viewing of websites while
+having their original JavaScript replaced by user-provided scripts. Haketilo
+combines the functionalities of content blocker and user script manager. It can
+be used with its script repository, Hydrilla. This package makes available the
+JSON schemas describing various interfaces of Haketilo and Hydrilla.")
+    (license license:cc0)))
diff --git a/hosts-extra b/hosts-extra
index f9d1a88..0debbef 100644
--- a/hosts-extra
+++ b/hosts-extra
@@ -14,3 +14,5 @@
 10.207.87.2 www.hydrillabugs.koszko.org
 10.207.87.2 haketilo.koszko.org
 10.207.87.2 www.haketilo.koszko.org
+10.207.87.2 hydrilla.koszko.org
+10.207.87.2 www.hydrilla.koszko.org
diff --git a/hydrilla-wsgi.py b/hydrilla-wsgi.py
new file mode 100644
index 0000000..45ec248
--- /dev/null
+++ b/hydrilla-wsgi.py
@@ -0,0 +1,26 @@
+# SPDX-License-Identifier: CC0-1.0
+
+# Copyright (C) 2022 Wojtek Kosior <koszko@koszko.org>
+#
+# Available under the terms of Creative Commons Zero v1.0 Universal.
+
+import sys
+
+_application = None
+
+def application(environ, *args):
+    global _application
+
+    if _application is None:
+        sys.path.extend(environ['HYDRILLA_GUIX_PYTHONPATH'].split(':'))
+
+        from hydrilla.server import start_wsgi, config
+
+        current_config = {
+            **config.load(),
+            'malcontent_dir': environ['MALCONENT_DIR']
+        }
+
+        _application = start_wsgi(standalone_mode=False, obj=current_config)
+
+    return _application(environ, *args)
diff --git a/subrepos/hydrilla b/subrepos/hydrilla
new file mode 160000
+Subproject 0792f29ea1fcc4386246cfc26bf4b2fbe3a5f9d