Replaces importDump.php with PagePort import script

Dockerfile

@@ -1,4 +1,4 @@
 FROM debian:12.5 as base
 
 LABEL maintainers="[email protected],[email protected]"
 LABEL org.opencontainers.image.source=https://github.com/WikiTeq/Taqasta
@@ -16,7 +16,7 @@
 	APACHE_LOG_DIR=/var/log/apache2
 
 # System setup
 RUN set x; \
 	apt-get clean \
 	&& apt-get update \
 	&& apt-get --no-install-recommends install -y aptitude \
@@ -113,11 +113,11 @@
 	&& mkdir -p $MW_VOLUME
 
 # Composer
 RUN set -x; \
 	curl -sS https://getcomposer.org/installer | php -- --install-dir=/usr/local/bin --filename=composer \
 	&& composer self-update 2.1.3
 
 FROM base as core
 # MediaWiki core
 RUN set -x; \
 	git clone --depth 1 -b $MW_CORE_VERSION https://gerrit.wikimedia.org/r/mediawiki/core.git $MW_HOME \
@@ -139,7 +139,7 @@
 	cd $MW_HOME \
 	&& find . \( -name ".git" -o -name ".gitignore" -o -name ".gitmodules" -o -name ".gitattributes" \) -exec rm -rf -- {} +
 
 FROM base as skins
 # Skins
 # The Minerva Neue, MonoBook, Timeless, Vector and Vector 2022 skins are bundled into MediaWiki and do not need to be
 # separately installed.
@@ -179,7 +179,7 @@
 	cd $MW_HOME/skins \
 	&& find . \( -name ".git" -o -name ".gitignore" -o -name ".gitmodules" -o -name ".gitattributes" \) -exec rm -rf -- {} +
 
 FROM base as extensions
 # Extensions
 #
 # The following extensions are bundled into MediaWiki and do not need to be separately installed (though in some cases
@@ -546,7 +546,7 @@
 	# PagePort
 	&& git clone --single-branch -b master https://github.com/WikiTeq/PagePort.git $MW_HOME/extensions/PagePort \
 	&& cd $MW_HOME/extensions/PagePort \
-	&& git checkout -q a6b800c9b3f58c151cdda4ec2f1aa396536c3a7d
+	&& git checkout -q 65cb6908c852c5483efa5f8022b8bb97b5adb208
 
 # R
 RUN set -x; \
@@ -977,7 +977,7 @@
 	cd $MW_HOME/extensions \
 	&& find . \( -name ".git" -o -name ".gitignore" -o -name ".gitmodules" -o -name ".gitattributes" \) -exec rm -rf -- {} +
 
 FROM base as composer
 
 # Copy core, skins and extensions
 COPY --from=core $MW_HOME $MW_HOME
@@ -1012,13 +1012,13 @@
 	&& ln -s $MW_VOLUME/images $MW_HOME/images \
 	&& ln -s $MW_VOLUME/cache $MW_HOME/cache
 
 FROM base as final
 
 COPY --from=composer $MW_HOME $MW_HOME
 COPY --from=composer $MW_ORIGIN_FILES $MW_ORIGIN_FILES
 
 # Default values
 ENV MW_AUTOUPDATE=true \
 	MW_MAINTENANCE_UPDATE=0 \
 	MW_ENABLE_EMAIL=0 \
 	MW_ENABLE_USER_EMAIL=0 \
@@ -1059,7 +1059,8 @@
 	MEDIAWIKI_MAINTENANCE_AUTO_ENABLED=false \
 	MW_SENTRY_DSN="" \
 	MW_USE_CACHE_DIRECTORY=1 \
-	APACHE_REMOTE_IP_HEADER=X-Forwarded-For
+	APACHE_REMOTE_IP_HEADER=X-Forwarded-For \
+	MW_AUTO_IMPORT=1
 
 COPY _sources/configs/msmtprc /etc/
 COPY _sources/configs/mediawiki.conf /etc/apache2/sites-enabled/

_sources/canasta/DockerSettings.php

@@ -606,6 +606,10 @@ function isEnvTrue( $name ): bool {
 	$wgUseCdn = false; // make sure profile is not cached
 }
 
+if ( getenv( 'MW_AUTO_IMPORT' ) ) {
+	wfLoadExtension( 'PagePort' );
+}
+
 # Include all php files in config/settings directory
 foreach ( glob( getenv( 'MW_CONFIG_DIR' ) . '/settings/*.php' ) as $filename ) {
 	if ( is_readable( $filename ) ) {

_sources/scripts/run-maintenance-scripts.sh

@@ -361,25 +361,16 @@ run_autoupdate () {
 }
 
 run_import () {
-    # Import dumps if any
+    # Import PagePort dumps if any
     if mountpoint -q -- "$MW_IMPORT_VOLUME"; then
-        echo "Found $MW_IMPORT_VOLUME, checking for XML dump presence.."
-        local FILES_IMPORTED=0
-        for filename in $MW_IMPORT_VOLUME/*.xml; do
-            echo "Found $filename file, importing.."
-            php maintenance/importDump.php \
-                --username-prefix="" \
-                --memory-limit=max \
-                --report 1 \
-                "$filename"
-            FILES_IMPORTED=$((FILES_IMPORTED + 1))
-        done
-        if [ "$FILES_IMPORTED" -gt "0" ]; then
-            echo "Running post-import maintenance scripts.."
-            php maintenance/rebuildrecentchanges.php
-            php maintenance/initSiteStats.php
+        echo "Found $MW_IMPORT_VOLUME, running PagePort import.."
+        XML_TEST=($(find $MW_IMPORT_VOLUME -maxdepth 1 -name "*.xml"))
+        if [ ${#XML_TEST[@]} -gt 0 ]; then
+        	echo "WARNING! The directory contains XML files, did you forget to migrate the dump to PagePort?"
+        	return
         fi
-        echo "Imported $FILES_IMPORTED files!"
+        php extensions/PagePort/maintenance/importPages.php --source "$MW_IMPORT_VOLUME" --user 'Maintenance script'
+        echo "Imported completed!"
     fi
 }
 
@@ -391,7 +382,9 @@ else
 fi
 
 # Run import after install and update are completed
-run_import
+if isTrue "$MW_AUTO_IMPORT"; then
+	run_import
+fi
 
 jobrunner &
 transcoder &