gen-killed-in-gaza.yml

on:
  workflow_dispatch:
  push:
    branches:
      - "**"
      - "!main"
    paths:
      - "scripts/data/common/killed-in-gaza/**"
      - "scripts/data/v2/killed-in-gaza.ts"
      - ".github/workflows/gen-killed-in-gaza.yml"

jobs:
  datasets_gen_killed:
    runs-on: ubuntu-latest
    permissions:
      contents: write
      pull-requests: write
    steps:
      - uses: actions/checkout@v4
      - name: checkout main for diff
        run: |
          git fetch origin main --depth 1

      - uses: oven-sh/setup-bun@v1
        name: setup bun
        with:
          bun-version: 1.0.22

      - name: install dependencies
        run: bun install

      - name: sort dict csvs
        run: |
          bun run sort-list-csvs scripts/data/common/killed-in-gaza/data/dict_ar_en.csv
          bun run sort-list-csvs scripts/data/common/killed-in-gaza/data/dict_ar_ar.csv

      - name: run scripts & diff
        run: |
          bun run scripts/data/common/killed-in-gaza/generate_killed_list.ts
          mkdir -p ci-tmp
          bun run gen-killed
          git diff origin/main:killed-in-gaza.json killed-in-gaza.json > ci-tmp/killed-in-gaza-json.diff

      - name: upload debug artifacts
        uses: actions/upload-artifact@v4
        with:
          name: GAZA-LIST-REVIEW-ITEMS
          path: ci-tmp/*

      - uses: EndBug/add-and-commit@v9 # You can change this to use a specific version.
        name: commit changes
        id: commit
        with:
          message: "killed-in-gaza: json update"
          add: '["killed-in-gaza.json", "killed-in-gaza.min.json", "scripts/data/common/killed-in-gaza/data/**"]'
          push: true

      - name: create & upload derived API resources artifact
        run: bun run gen-derived
        env:
          TFP_SHEET_KEY: ${{ secrets.TFP_SHEET_KEY }}

      - uses: 8BitJonny/gh-get-current-pr@3.0.0
        id: PR
        with:
          github-token: ${{ github.token }}
          sha: ${{ github.event.pull_request.head.sha }}
          filterOutClosed: true
          filterOutDraft: false

      - name: build report
        id: report
        run: |
          # if our log output has lines there's issues worth noting
          if [[ $(wc -l <ci-tmp/killed-in-gaza-log.txt) -ge 2 ]]; then
            echo "logs=1" >> "$GITHUB_OUTPUT"
          fi
          if [[ $(wc -l <ci-tmp/killed-in-gaza-json.diff) -ge 1 ]]; then
            echo "diff=1" >> "$GITHUB_OUTPUT"
          fi

      - uses: actions/github-script@v7
        if: steps.PR.outputs.pr_found == 'true'
        name: add pr comment
        env:
          COMMITTED: ${{ steps.commit.outputs.committed }}
          HASLOGS: ${{ steps.report.outputs.logs }}
          HASDIFF: ${{ steps.report.outputs.diff }}
          PRNUM: ${{ steps.PR.outputs.number }}
        with:
          script: |
            let body = ''
            const changesStart = '✅'
            const missStart = '💨'

            if (process.env.COMMITTED === 'true') {
              body += `${changesStart} The JSON build scripts ran and produced JSON changes, please review them.\n\n`
            } else {
              body += `${missStart} The JSON build scripts ran and did not produce any changes to commit.\n\n`
            }

            if (process.env.HASLOGS || process.env.HASDIFF) {
              body += `The CI workflow artifacts have items you can [download and review](https://github.com/TechForPalestine/palestine-datasets/actions/runs/${context.runId}):\n`
            }

            if (process.env.HASLOGS) {
              body += '- [ ] logs detailing potential issues to resolve\n'
            }

            if (process.env.HASDIFF) {
              body += '- [ ] a diff of the JSON file changes for review\n'
            }

            const commonInputs = {
              issue_number: process.env.PRNUM,
              owner: context.repo.owner,
              repo: context.repo.repo
            }

            // delete prior comments
            const comments = await github.rest.issues.listComments(commonInputs)
            const idsToDelete = comments.data
              .filter(comment => comment.user.login.includes('github-action'))
              .map(comment => comment.id)

            // check if a recent commit had changes to review
            let recentChanges = false
            const lastCommentBody = comments.data
              .sort((a, b) => b.created_at.localeCompare(a.created_at))[0]?.body ?? ''

            if (lastCommentBody.startsWith(changesStart)) {
              recentChanges = true
            }

            if (recentChanges && body.startsWith(missStart)) {
              console.log('skipping comment, recent changes')
              return; // keep last comment if commit leads to no change
            }

            console.log(`found ${idsToDelete.length} comments to delete: ${idsToDelete.join(',')}`)

            if (idsToDelete.length) {
              await idsToDelete.reduce(async (chain, comment_id) => {
                await chain;
                await github.rest.issues.deleteComment({
                  ...commonInputs,
                  comment_id
                })
              }, Promise.resolve())
            }

            await github.rest.issues.createComment({
              ...commonInputs,
              body
            })