spec: &spec
  x-sub-request-filters:
    - type: default
      name: http
      options:
        allow:
          - pattern: /^https?:\/\//
            forward_headers:
              user-agent: true
  title: The Change Propagation root
  paths:
    /sys/links:
      x-modules:
        - path: sys/dep_updates.js
          options:
            templates:
              mw_api:
                  uri: '{env(MEDIAWIKI_URL)}/w/api.php'
                  headers:
                    host: '{{message.meta.domain}}'
                  body:
                    formatversion: 2
    /sys/queue:
      x-modules:
        - path: sys/kafka.js
          options:
            metadata_broker_list: '{env(KAFKA_BROKER_LIST)}'
            dc_name: eqiad
            startup_delay: 0
            concurrency: 250
            templates:
              summary_definition_rerender: &summary_definition_rerender_spec
                topic: '/^(?:change-prop\.transcludes\.)?resource[-_]change$/'
                retry_limit: 2
                retry_delay: 500
                retry_on:
                  status:
                    - '5xx'
                match:
                  meta:
                    uri: '/^https?:\/\/[^\/]+\/api\/rest_v1\/page\/html\/(?<title>[^/]+)$/'
                  tags:
                    - restbase
                match_not:
                  - meta:
                      domain: '/wiktionary.org$/'
                  - meta:
                      domain: /\.wikidata\.org$/
                exec:
                  method: get
                  # Don't encode title since it should be already encoded
                  uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/summary/{{match.meta.uri.title}}'
                  query:
                    redirect: false
                  headers:
                    cache-control: no-cache

              # RESTBase update jobs
              mw_purge:
                topic: resource_change
                match:
                  meta:
                    uri: '/^https?:\/\/[^\/]+\/wiki\/(?<title>.+)$/'
                  tags:
                    - purge
                exec:
                  method: get
                  # This even comes directly from MediaWiki, so title is encoded in MW-specific way.
                  # Re-encode the title in standard `encodeURIComponent` encoding.
                  uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/html/{decode(match.meta.uri.title)}'
                  headers:
                    cache-control: no-cache
                    if-unmodified-since: '{{date(message.meta.dt)}}'
                  query:
                    redirect: false

              null_edit:
                topic: resource_change
                ignore:
                  status:
                    - 403 # Ignoring 403 since some of the pages with high number of null_edit events are blacklisted
                    - 412
                match:
                  meta:
                    uri: '/^https?:\/\/[^\/]+\/wiki\/(?<title>.+)$/'
                  tags:
                    - null_edit
                exec:
                  method: get
                  # This even comes directly from MediaWiki, so title is encoded in MW-specific way.
                  # Re-encode the title in standard `encodeURIComponent` encoding.
                  uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/html/{decode(match.meta.uri.title)}'
                  headers:
                    cache-control: no-cache
                    if-unmodified-since: '{{date(message.meta.dt)}}'
                  query:
                    redirect: false

              page_edit:
                topic: mediawiki.revision-create
                retry_on:
                  status:
                    - '5xx'
                    - 404 # Sometimes occasional 404s happen because of the mysql replication lag, so retry
                exec:
                  - method: get
                    uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/html/{message.page_title}/{{message.rev_id}}'
                    headers:
                      cache-control: no-cache
                      x-restbase-parentrevision: '{{message.rev_parent_id}}'
                      if-unmodified-since: '{{date(message.meta.dt)}}'
                    query:
                      redirect: false
                  - method: post
                    uri: '/sys/links/transcludes/{message.page_title}'
                    body: '{{globals.message}}'

              revision_visibility_change:
                topic: mediawiki.revision-visibility-change
                ignore:
                  status:
                    - 403 # When the revision is hidden 403 will be returned by RESTBase, it's a valid situation
                    - 412
                exec:
                  method: get
                  uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/revision/{{message.rev_id}}'
                  headers:
                    cache-control: no-cache
                  query:
                    redirect: false

              page_delete:
                topic: mediawiki.page-delete
                ignore:
                  status:
                    - 404 # 404 is a normal response for page deletion
                    - 412
                exec:
                  - method: get
                    uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/title/{message.page_title}'
                    headers:
                      cache-control: no-cache
                    query:
                      redirect: false

              page_restore:
                topic: mediawiki.page-undelete
                exec:
                  method: get
                  uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/title/{message.page_title}'
                  headers:
                    cache-control: no-cache
                  query:
                    redirect: false

              page_move:
                topic: mediawiki.page-move
                exec:
                  - method: get
                    uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/html/{message.page_title}/{{message.rev_id}}'
                    headers:
                      cache-control: no-cache
                      if-unmodified-since: '{{date(message.meta.dt)}}'
                    query:
                      redirect: false
                  - method: get
                    uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/title/{message.prior_state.page_title}'
                    headers:
                      cache-control: no-cache
                    query:
                      redirect: false

              on_transclusion_update:
                topic: change-prop.transcludes.resource-change
                cases:
                  - match:
                      meta:
                        schema_uri: 'resource_change/1'
                        uri: '/https?:\/\/[^\/]+\/wiki\/(?<title>.+)/'
                      tags: [ 'transcludes' ]
                    exec:
                      method: get
                      uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/html/{{match.meta.uri.title}}'
                      headers:
                        cache-control: no-cache
                        if-unmodified-since: '{{date(message.meta.dt)}}'
                        x-restbase-mode: '{{message.tags[1]}}'
                      query:
                        redirect: false
                  - match:
                      meta:
                        schema_uri: 'continue/1'
                    exec:
                      method: post
                      uri: '/sys/links/transcludes/{message.original_event.page_title}'
                      body: '{{globals.message}}'

              page_images:
                topic: mediawiki.page-properties-change
                # We don't support 'OR' in the match section, so workaround it by 2 cases with identical exec
                cases:
                  - match:
                      added_properties:
                        page_image: '/.+/' # Regex that matches anything just to check the prop is set
                    exec:
                      - method: get
                        uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/summary/{message.page_title}'
                        headers:
                          cache-control: no-cache
                        query:
                          redirect: false
                      - method: get
                        uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/mobile-sections/{message.page_title}'
                        headers:
                          cache-control: no-cache
                        query:
                          redirect: false
                  - match:
                      removed_properties:
                        page_image: '/.+/' # Regex that matches anything just to check the prop is set
                    exec:
                      - method: get
                        uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/summary/{message.page_title}'
                        headers:
                          cache-control: no-cache
                        query:
                          redirect: false
                      - method: get
                        uri: '{env(RESTBASE_URL)}/{{message.meta.domain}}/v1/page/mobile-sections/{message.page_title}'
                        headers:
                          cache-control: no-cache
                        query:
                          redirect: false

num_workers: ncpu
logging:
  name: changeprop
  level: info
services:
  - name: changeprop
    module: hyperswitch
    conf:
      port: 7272
      user_agent: SampleChangePropInstance
      spec: *spec