# This workflow identifies changes between the base and the head ref, for use in
# other workflows to decide if they should be executed.

name: Identify Changes

on:
  workflow_call:
    # These inputs allow the filter action to be able to access the correct refs for
    # comparison in changes detection, it is required as this is called from the
    # merge_group context.
    inputs:
      base_ref:
        required: false
        type: string
      head_ref:
        required: false
        type: string
      int_tests:
        required: false
        type: boolean
        default: false
      e2e_tests:
        required: false
        type: boolean
        default: false
      source:
        required: false
        type: boolean
        default: true
    outputs:
      source:
        value: ${{ jobs.source.outputs.source }}
      dependencies:
        value: ${{ jobs.source.outputs.dependencies }}
      deny:
        value: ${{ jobs.source.outputs.deny }}
      internal_events:
        value: ${{ jobs.source.outputs.internal_events }}
      cue:
        value: ${{ jobs.source.outputs.cue }}
      component_docs:
        value: ${{ jobs.source.outputs.component_docs }}
      markdown:
        value: ${{ jobs.source.outputs.markdown }}
      scripts:
        value: ${{ jobs.source.outputs.scripts }}
      website_only:
        value: ${{ jobs.source.outputs.website_only }}
      install:
        value: ${{ jobs.source.outputs.install }}
      prettier:
        value: ${{ jobs.source.outputs.prettier }}
      test-yml:
        value: ${{ jobs.source.outputs.test-yml }}
      integration-yml:
        value: ${{ jobs.source.outputs.integration-yml }}
      test-make-command-yml:
        value: ${{ jobs.source.outputs.test-make-command-yml }}
      msrv-yml:
        value: ${{ jobs.source.outputs.msrv-yml }}
      cross-yml:
        value: ${{ jobs.source.outputs.cross-yml }}
      unit_mac-yml:
        value: ${{ jobs.source.outputs.unit_mac-yml }}
      unit_windows-yml:
        value: ${{ jobs.source.outputs.unit_windows-yml }}
      k8s:
        value: ${{ jobs.source.outputs.k8s }}
      amqp:
        value: ${{ jobs.int_tests.outputs.amqp }}
      appsignal:
        value: ${{ jobs.int_tests.outputs.appsignal }}
      aws:
        value: ${{ jobs.int_tests.outputs.aws }}
      axiom:
        value: ${{ jobs.int_tests.outputs.axiom }}
      azure:
        value: ${{ jobs.int_tests.outputs.azure }}
      clickhouse:
        value: ${{ jobs.int_tests.outputs.clickhouse }}
      databend:
        value: ${{ jobs.int_tests.outputs.databend }}
      datadog-agent:
        value: ${{ jobs.int_tests.outputs.datadog-agent }}
      datadog-logs:
        value: ${{ jobs.int_tests.outputs.datadog-logs }}
      datadog-metrics:
        value: ${{ jobs.int_tests.outputs.datadog-metrics }}
      datadog-traces:
        value: ${{ jobs.int_tests.outputs.datadog-traces }}
      dnstap:
        value: ${{ jobs.int_tests.outputs.dnstap }}
      docker-logs:
        value: ${{ jobs.int_tests.outputs.docker-logs }}
      doris:
        value: ${{ jobs.int_tests.outputs.doris }}
      elasticsearch:
        value: ${{ jobs.int_tests.outputs.elasticsearch }}
      eventstoredb:
        value: ${{ jobs.int_tests.outputs.eventstoredb }}
      fluent:
        value: ${{ jobs.int_tests.outputs.fluent }}
      gcp:
        value: ${{ jobs.int_tests.outputs.gcp }}
      greptimedb:
        value: ${{ jobs.int_tests.outputs.greptimedb }}
      humio:
        value: ${{ jobs.int_tests.outputs.humio }}
      http-client:
        value: ${{ jobs.int_tests.outputs.http-client }}
      influxdb:
        value: ${{ jobs.int_tests.outputs.influxdb }}
      kafka:
        value: ${{ jobs.int_tests.outputs.kafka }}
      logstash:
        value: ${{ jobs.int_tests.outputs.logstash }}
      loki:
        value: ${{ jobs.int_tests.outputs.loki }}
      mongodb:
        value: ${{ jobs.int_tests.outputs.mongodb }}
      mqtt:
        value: ${{ jobs.int_tests.outputs.mqtt }}
      nats:
        value: ${{ jobs.int_tests.outputs.nats }}
      nginx:
        value: ${{ jobs.int_tests.outputs.nginx }}
      opentelemetry:
        value: ${{ jobs.int_tests.outputs.opentelemetry }}
      postgres:
        value: ${{ jobs.int_tests.outputs.postgres }}
      prometheus:
        value: ${{ jobs.int_tests.outputs.prometheus }}
      pulsar:
        value: ${{ jobs.int_tests.outputs.pulsar }}
      redis:
        value: ${{ jobs.int_tests.outputs.redis }}
      splunk:
        value: ${{ jobs.int_tests.outputs.splunk }}
      webhdfs:
        value: ${{ jobs.int_tests.outputs.webhdfs }}
      # e2e tests
      e2e-datadog-logs:
        value: ${{ jobs.e2e_tests.outputs.datadog-logs }}
      e2e-datadog-metrics:
        value: ${{ jobs.e2e_tests.outputs.datadog-metrics }}
      e2e-opentelemetry-logs:
        value: ${{ jobs.e2e_tests.outputs.opentelemetry-logs }}
      e2e-opentelemetry-metrics:
        value: ${{ jobs.e2e_tests.outputs.opentelemetry-metrics }}
      int-tests-any:
        value: ${{ jobs.int_tests.outputs.any }}
      e2e-tests-any:
        value: ${{ jobs.e2e_tests.outputs.any }}

# Workflow-level permissions - read access to repository contents
permissions:
  contents: read # Required to checkout code

env:
  BASE_SHA: ${{ inputs.base_ref || (github.event_name == 'merge_group' && github.event.merge_group.base_sha) || github.event.pull_request.base.sha }}
  HEAD_SHA: ${{ inputs.head_ref || (github.event_name == 'merge_group' && github.event.merge_group.head_sha) || github.event.pull_request.head.sha }}

jobs:
  # Detects changes that are not specific to integration tests
  source:
    runs-on: ubuntu-24.04
    timeout-minutes: 5
    if: ${{ inputs.source }}
    outputs:
      source: ${{ steps.filter.outputs.source }}
      dependencies: ${{ steps.filter.outputs.dependencies }}
      deny: ${{ steps.filter.outputs.deny }}
      internal_events: ${{ steps.filter.outputs.internal_events }}
      cue: ${{ steps.filter.outputs.cue }}
      component_docs: ${{ steps.filter.outputs.component_docs }}
      markdown: ${{ steps.filter.outputs.markdown }}
      scripts: ${{ steps.filter.outputs.scripts }}
      install: ${{ steps.filter.outputs.install }}
      k8s: ${{ steps.filter.outputs.k8s }}
      website_only: ${{ steps.filter.outputs.website == 'true' && steps.filter.outputs.not_website == 'false' }}
      prettier: ${{ steps.filter.outputs.prettier }}
      test-yml: ${{ steps.filter.outputs.test-yml }}
      integration-yml: ${{ steps.filter.outputs.integration-yml }}
      test-make-command-yml: ${{ steps.filter.outputs.test-make-command-yml }}
      msrv-yml: ${{ steps.filter.outputs.msrv-yml }}
      cross-yml: ${{ steps.filter.outputs.cross-yml }}
      unit_mac-yml: ${{ steps.filter.outputs.unit_mac-yml }}
      unit_windows-yml: ${{ steps.filter.outputs.unit_windows-yml }}
    steps:
      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2

      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
        id: filter
        with:
          base: ${{ env.BASE_SHA }}
          ref: ${{ env.HEAD_SHA }}
          filters: |
            source:
              - ".cargo/**"
              - "benches/**"
              - "lib/**"
              - "proto/**"
              - "scripts/**"
              - "src/**"
              - "tests/**"
              - "build.rs"
              - "Cargo.lock"
              - "Cargo.toml"
              - "Makefile"
              - "rust-toolchain.toml"
              - "vdev/**"
              - ".github/workflows/changes.yml"
            dependencies:
              - ".cargo/**"
              - 'Cargo.toml'
              - 'Cargo.lock'
              - 'rust-toolchain.toml'
              - 'Makefile'
              - 'scripts/cross/**'
              - "vdev/**"
              - ".github/workflows/changes.yml"
            deny:
              - '**/Cargo.toml'
              - 'Cargo.lock'
              - ".github/workflows/deny.yml"
            cue:
              - 'website/cue/**'
              - "vdev/**"
              - ".github/workflows/changes.yml"
            component_docs:
              - 'scripts/generate-component-docs.rb'
              - "vdev/**"
              - 'website/cue/**/*.cue'
              - 'docs/generated/**'
              # If changes to the VRL sha is made the combined generated cue file will change which
              # may cause issues
              - 'Cargo.lock'
              - ".github/workflows/changes.yml"
            markdown:
              - '**/**.md'
              - "vdev/**"
              - ".github/workflows/changes.yml"
            scripts:
              - '**/**.sh'
              - ".github/workflows/changes.yml"
            internal_events:
              - 'src/internal_events/**'
              - "vdev/**"
              - ".github/workflows/changes.yml"
            docker:
              - 'distribution/docker/**'
              - "vdev/**"
              - ".github/workflows/changes.yml"
            install:
              - ".github/workflows/install-sh.yml"
              - "distribution/install.sh"
              - ".github/workflows/changes.yml"
            k8s:
              - "src/sources/kubernetes_logs/**"
              - "lib/k8s-e2e-tests/**"
              - "lib/k8s-test-framework/**"
              - "scripts/test-e2e-kubernetes.sh"
              - "scripts/build-docker.sh"
              - "scripts/deploy-chart-test.sh"
              - ".github/workflows/k8s_e2e.yml"
              - ".github/workflows/changes.yml"
            website:
              - "website/**"
            not_website:
              - "!website/**"
            prettier:
              - "**/*.yml"
              - "**/*.yaml"
              - "**/*.js"
              - "**/*.ts"
              - "**/*.tsx"
              - "**/*.json"
              - ".prettierrc.json"
              - ".prettierignore"
              - ".github/workflows/changes.yml"
            test-yml:
              - ".github/workflows/test.yml"
              - ".github/workflows/changes.yml"
            integration-yml:
              - ".github/workflows/integration.yml"
              - ".github/workflows/changes.yml"
            test-make-command-yml:
              - ".github/workflows/test-make-command.yml"
              - ".github/workflows/changes.yml"
            msrv-yml:
              - ".github/workflows/msrv.yml"
              - ".github/workflows/changes.yml"
            cross-yml:
              - ".github/workflows/cross.yml"
              - ".github/workflows/changes.yml"
            unit_mac-yml:
              - ".github/workflows/unit_mac.yml"
              - ".github/workflows/changes.yml"
            unit_windows-yml:
              - ".github/workflows/unit_windows.yml"
              - ".github/workflows/changes.yml"

  # Detects changes that are specific to integration tests
  int_tests:
    runs-on: ubuntu-24.04
    timeout-minutes: 15
    if: ${{ inputs.int_tests }}
    permissions:
      contents: read
    outputs:
      amqp: ${{ steps.filter.outputs.amqp }}
      appsignal: ${{ steps.filter.outputs.appsignal}}
      aws: ${{ steps.filter.outputs.aws }}
      axiom: ${{ steps.filter.outputs.axiom }}
      azure: ${{ steps.filter.outputs.azure }}
      clickhouse: ${{ steps.filter.outputs.clickhouse }}
      databend: ${{ steps.filter.outputs.databend }}
      datadog-agent: ${{ steps.filter.outputs.datadog-agent }}
      datadog-logs: ${{ steps.filter.outputs.datadog-logs }}
      datadog-metrics: ${{ steps.filter.outputs.datadog-metrics }}
      datadog-traces: ${{ steps.filter.outputs.datadog-traces }}
      dnstap: ${{ steps.filter.outputs.dnstap }}
      docker-logs: ${{ steps.filter.outputs.docker-logs }}
      doris: ${{ steps.filter.outputs.doris }}
      elasticsearch: ${{ steps.filter.outputs.elasticsearch }}
      eventstoredb: ${{ steps.filter.outputs.eventstoredb }}
      fluent: ${{ steps.filter.outputs.fluent }}
      gcp: ${{ steps.filter.outputs.gcp }}
      greptimedb: ${{ steps.filter.outputs.greptimedb }}
      humio: ${{ steps.filter.outputs.humio }}
      http-client: ${{ steps.filter.outputs.http-client }}
      influxdb: ${{ steps.filter.outputs.influxdb }}
      kafka: ${{ steps.filter.outputs.kafka }}
      logstash: ${{ steps.filter.outputs.logstash }}
      loki: ${{ steps.filter.outputs.loki }}
      mongodb: ${{ steps.filter.outputs.mongodb }}
      mqtt: ${{ steps.filter.outputs.mqtt }}
      nats: ${{ steps.filter.outputs.nats }}
      nginx: ${{ steps.filter.outputs.nginx }}
      opentelemetry: ${{ steps.filter.outputs.opentelemetry }}
      postgres: ${{ steps.filter.outputs.postgres }}
      prometheus: ${{ steps.filter.outputs.prometheus }}
      pulsar: ${{ steps.filter.outputs.pulsar }}
      redis: ${{ steps.filter.outputs.redis }}
      splunk: ${{ steps.filter.outputs.splunk }}
      webhdfs: ${{ steps.filter.outputs.webhdfs }}
      any: ${{ steps.detect-changes.outputs.any }}
    steps:
      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2

      - uses: ./.github/actions/setup
        with:
          vdev: true
          mold: false
          cargo-cache: false

      # creates a yaml file that contains the filters for each integration,
      # extracted from the output of the `vdev int ci-paths` command, which
      # sources the paths from the tests/integration/.../config/test.yaml files
      - name: Create filter rules for integrations
        run: vdev int ci-paths > int_test_filters.yaml

      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
        id: filter
        with:
          base: ${{ env.BASE_SHA }}
          ref: ${{ env.HEAD_SHA }}
          filters: int_test_filters.yaml

      # This JSON hack was introduced because GitHub Actions does not support dynamic expressions in the
      # `if` section of a job or step. In return, we can reduce a lot of redundant job configuration in integration.yml.
      - name: Detect changes and generate JSON
        id: detect-changes
        run: |
          json=$(cat <<EOF
          {
            "amqp": ${{ steps.filter.outputs.amqp }},
            "appsignal": ${{ steps.filter.outputs.appsignal }},
            "aws": ${{ steps.filter.outputs.aws }},
            "axiom": ${{ steps.filter.outputs.axiom }},
            "azure": ${{ steps.filter.outputs.azure }},
            "clickhouse": ${{ steps.filter.outputs.clickhouse }},
            "databend": ${{ steps.filter.outputs.databend }},
            "datadog-agent": ${{ steps.filter.outputs.datadog-agent }},
            "datadog-logs": ${{ steps.filter.outputs.datadog-logs }},
            "datadog-metrics": ${{ steps.filter.outputs.datadog-metrics }},
            "datadog-traces": ${{ steps.filter.outputs.datadog-traces }},
            "dnstap": ${{ steps.filter.outputs.dnstap }},
            "docker-logs": ${{ steps.filter.outputs.docker-logs }},
            "doris": ${{ steps.filter.outputs.doris }},
            "elasticsearch": ${{ steps.filter.outputs.elasticsearch }},
            "eventstoredb": ${{ steps.filter.outputs.eventstoredb }},
            "fluent": ${{ steps.filter.outputs.fluent }},
            "gcp": ${{ steps.filter.outputs.gcp }},
            "greptimedb": ${{ steps.filter.outputs.greptimedb }},
            "humio": ${{ steps.filter.outputs.humio }},
            "http-client": ${{ steps.filter.outputs.http-client }},
            "influxdb": ${{ steps.filter.outputs.influxdb }},
            "kafka": ${{ steps.filter.outputs.kafka }},
            "logstash": ${{ steps.filter.outputs.logstash }},
            "loki": ${{ steps.filter.outputs.loki }},
            "mongodb": ${{ steps.filter.outputs.mongodb }},
            "mqtt": ${{ steps.filter.outputs.mqtt }},
            "nats": ${{ steps.filter.outputs.nats }},
            "nginx": ${{ steps.filter.outputs.nginx }},
            "opentelemetry": ${{ steps.filter.outputs.opentelemetry }},
            "postgres": ${{ steps.filter.outputs.postgres }},
            "prometheus": ${{ steps.filter.outputs.prometheus }},
            "pulsar": ${{ steps.filter.outputs.pulsar }},
            "redis": ${{ steps.filter.outputs.redis }},
            "splunk": ${{ steps.filter.outputs.splunk }},
            "webhdfs": ${{ steps.filter.outputs.webhdfs }}
          }
          EOF
          )

          echo "$json" > int_tests_changes.json

          # Check if at least one integration test has changes
          any_changed=$(jq -r 'to_entries | map(select(.value == true)) | length > 0' int_tests_changes.json)
          echo "any=$any_changed" >> $GITHUB_OUTPUT

      - name: Upload JSON artifact
        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
        with:
          name: int_tests_changes
          path: int_tests_changes.json

  # Detects changes that are specific to e2e tests
  e2e_tests:
    runs-on: ubuntu-24.04
    timeout-minutes: 15
    if: ${{ inputs.e2e_tests }}
    permissions:
      contents: read
    outputs:
      datadog-logs: ${{ steps.filter.outputs.datadog-logs }}
      datadog-metrics: ${{ steps.filter.outputs.datadog-metrics }}
      opentelemetry-logs: ${{ steps.filter.outputs.opentelemetry-logs }}
      opentelemetry-metrics: ${{ steps.filter.outputs.opentelemetry-metrics }}
      any: ${{ steps.detect-changes.outputs.any }}
    steps:
      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2

      - uses: ./.github/actions/setup
        with:
          vdev: true
          mold: false
          cargo-cache: false

      # creates a yaml file that contains the filters for each test,
      # extracted from the output of the `vdev int ci-paths` command, which
      # sources the paths from the tests/e2e/.../test.yaml files
      - name: Create filter rules for e2e tests
        run: vdev e2e ci-paths > int_test_filters.yaml

      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
        id: filter
        with:
          base: ${{ env.BASE_SHA }}
          ref: ${{ env.HEAD_SHA }}
          filters: int_test_filters.yaml

      # This JSON hack was introduced because GitHub Actions does not support dynamic expressions in the
      # `if` section of a job or step. In return, we can reduce a lot of redundant job configuration in integration.yml.
      - name: Detect changes and generate JSON
        id: detect-changes
        run: |
          json=$(cat <<EOF
          {
            "datadog-logs": ${{ steps.filter.outputs.datadog-logs }},
            "datadog-metrics": ${{ steps.filter.outputs.datadog-metrics }},
            "opentelemetry-logs": ${{ steps.filter.outputs.opentelemetry-logs }},
            "opentelemetry-metrics": ${{ steps.filter.outputs.opentelemetry-metrics }}
          }
          EOF
          )

          echo "$json" > e2e_tests_changes.json

          # Check if at least one e2e test has changes
          any_changed=$(jq -r 'to_entries | map(select(.value == true)) | length > 0' e2e_tests_changes.json)
          echo "any=$any_changed" >> $GITHUB_OUTPUT

      - name: Upload JSON artifact
        uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
        with:
          name: e2e_tests_changes
          path: e2e_tests_changes.json