check-templates.yml

on:
  workflow_dispatch:
  workflow_call:
    inputs:
      r-version:
        description: 'The version of R to use'
        default: '4.1'
        required: false
        type: string
      push-templates-data:
        description: 'Push generated templates data to pharmaverseadam repo'
        default: false
        required: false
        type: boolean
      exclude-templates-data:
        description: 'List of data generated by templates to exclude (templates script will run, but results will not be push to pharmaverseadam repo) - comma seperated list'
        default: ''
        required: false
        type: string
  push:
    branches:
      - main
      - devel
      - pre-release
  pull_request:
    branches:
      - main
      - devel
      - pre-release

env:
  # branch name for PR
  source-branch: update_templates_data
  repo: pharmaverseadam
  target-branch: main
  reviewers: >
    {
      "admiralonco": {
        "reviewers": ["bundfussr", "manciniedoardo"]
      },
      "admiralophtha": {
        "reviewers": ["manciniedoardo"]
      },
      "admiral": {
        "reviewers": ["manciniedoardo", "bms63"]
      },
      "admiralvaccine": {
        "reviewers": ["ahasoplakus", "arjoon-r", "manciniedoardo"]
      }
    }
  # data folder name for templates
  templates_data_folder: ./data/

name: Check Templates

concurrency:
  group: templates-${{ github.event.pull_request.number || github.ref }}
  cancel-in-progress: true

jobs:
  get_r_version:
    name: Get R version
    runs-on: ubuntu-latest
    if: >
      !contains(github.event.commits[0].message, '[skip check_templates]')
    outputs:
      r-version: ${{ steps.normalizer.outputs.R_VERSION }}
    steps:
      - name: Normalize inputs
        id: normalizer
        run: |
          R_VERSION="${{ inputs.r-version }}"
          if [ "$R_VERSION" == "" ]
          then {
            R_VERSION="4.1"
          }
          fi
          echo "R_VERSION=$R_VERSION" >> $GITHUB_OUTPUT
        shell: bash

  check_templates:
    name: Verify
    if: >
      !contains(github.event.commits[0].message, '[skip check_templates]')
    runs-on: ubuntu-latest
    container:
      image: "ghcr.io/pharmaverse/admiralci-${{needs.get_r_version.outputs.r-version}}:latest"
    needs: get_r_version

    env:
      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
    steps:
      ##################### BEGIN boilerplate steps #####################
      - name: Get branch names
        id: branch-name
        uses: tj-actions/branch-names@v5.4

      - name: Checkout repo (PR) 🛎
        uses: actions/checkout@v3
        if: github.event_name == 'pull_request'
        with:
          ref: ${{ steps.branch-name.outputs.head_ref_branch }}
          repository: ${{ github.event.pull_request.head.repo.full_name }}

      - name: Checkout repository
        uses: actions/checkout@v3
        if: github.event_name != 'pull_request'
        with:
          ref: ${{ steps.branch-name.outputs.head_ref_branch }}
          fetch-depth: 0

      - name: Add safe directory
        run: |
          git config --global --add safe.directory "${GITHUB_WORKSPACE}" # open issue: git dubious ownership running in docker container context: https://github.com/actions/runner/issues/2033
        shell: bash

      - name: Normalize inputs
        id: normalizer
        run: |
          function remove_substring() {
              echo "${1//$2}"
          }
          BRANCH=${{ steps.branch-name.outputs.head_ref_branch }}
          if [ "${BRANCH}" == "" ]
          then {
            BRANCH=$(git name-rev HEAD | awk '{print $NF}')
          }
          fi
          if [[ $BRANCH == remotes/pull/* ]]
          then {
            PR=$(remove_substring "$BRANCH" "/merge")
            PR=$(remove_substring "$PR" "remotes/")
            PR=$(echo $PR | sed 's/pull/pulls/')
            echo "Accessing repos/${GITHUB_REPOSITORY}/${PR}"
            BRANCH=$(gh api -X GET repos/${GITHUB_REPOSITORY}/${PR} | jq -r '.head.ref')
          }
          fi
          echo "BRANCH is $BRANCH"
          R_VERSION="${{ inputs.r-version }}"
          if [ "$R_VERSION" == "" ]
          then {
            R_VERSION="4.1"
          }
          fi
          echo "R_VERSION=$R_VERSION" >> $GITHUB_OUTPUT
          echo "SD_GIT_REF=$BRANCH" >> $GITHUB_OUTPUT
        shell: bash
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

      - name: Restore cache
        uses: actions/cache@v3
        with:
          path: |
            ~/.staged.dependencies
          key: staged-deps
          restore-keys: staged-deps

      - name: Run Staged dependencies
        uses: insightsengineering/staged-dependencies-action@v1
        with:
          run-system-dependencies: false
          enable-check: false
          git-ref: ${{ steps.normalizer.outputs.SD_GIT_REF }}
          renv-restore: false
        env:
          GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
      ##################### END boilerplate steps #####################

      - name: Install package
        run: |
          R CMD build --no-build-vignettes --no-manual .
          R CMD INSTALL *.tar.gz

      - name: Run Template Scripts
        run: |
          desc <- read.dcf("DESCRIPTION")
          package_name <- as.character(desc[, "Package"])
          templates <- list.files(
            system.file("templates", package = package_name),
            pattern = "\\.R$",
            ignore.case = TRUE,
            full.names = TRUE
          )
          if (length(templates) > 0) {
            exit_codes <- purrr::map_chr(templates, function(file) {
              cmd <- sprintf("Rscript --vanilla %s", file)
              system(cmd)
            })
            if (any(exit_codes == 1L)) {
              failed_scripts <- basename(templates)[which(exit_codes == 1L)]
              err_msg <- sprintf(
                "Executing the following template scripts failed:\n%s",
                paste("-", failed_scripts, collapse = "\n")
              )
              stop(err_msg, call. = FALSE)
            }
          }
        shell: Rscript {0}

      - name: Add other data formats and suffixes, exclude data
        if: inputs.push-templates-data
        run: |
          library(readxl)
          library(zip)
          loadRData <- function(fileName){
            load(fileName)
            get(ls()[ls() != "fileName"])
          }
          print(file.path("${{ github.workspace }}", "tmp"))
          print(file.path(Sys.getenv("GITHUB_WORKSPACE"), "tmp"))
          folder_path <- file.path(Sys.getenv("GITHUB_WORKSPACE"), "tmp")
          rda_files <- list.files(path = folder_path, pattern = "\\.rda$")

          ex_files <- unlist(strsplit("${{ inputs.exclude-templates-data }}", ",")) # files to exclude

          for (rda_file in rda_files) {
            # delete file if part of inputs.exclude-templates-data
            if (gsub("\\.rda$", "", rda_file) %in% ex_files) {
              file.remove(file.path(folder_path, rda_file))
              cat("Deleted:", rda_file, "\n")
            }
            else {
              print(sprintf("converting file %s", rda_file))
              data <- loadRData(file.path(folder_path, rda_file))
              suffix <- gsub("pharmaverse/admiral", "", "${{ github.repository }}")
              if (nchar(suffix) > 0) {
                rda_file_renamed <- gsub("\\.rda$", sprintf("_%s.rda", suffix), rda_file)
                csv_file <- gsub("\\.rda$", sprintf("_%s.csv", suffix), rda_file)
                file.rename(file.path(folder_path, rda_file), file.path(folder_path, rda_file_renamed))
              }
              else {
                csv_file <- gsub("\\.rda$", ".csv", rda_file)
                rda_file_renamed <- rda_file
              }
              write.csv(data, file = file.path(folder_path, csv_file), row.names = FALSE)

              # rename content of rda file
              rda_var <- gsub("\\.rda$", "", rda_file_renamed)
              print(rda_var)
              assign(rda_var, data)
              do.call(save, list(rda_var, file = file.path(folder_path, sprintf("%s.rda", rda_var)), compress = "bzip2"))

              # create associated documentation inside R folder
              dataset_name <- gsub("\\.rda$", "", rda_file_renamed)
              doc_string <- paste(
                sprintf("#' Dataset %s", dataset_name),
                "#'",
                sprintf("#' %s dataset", dataset_name),
                "#'",
                sprintf("#' @name %s", dataset_name),
                "#' @docType data",
                sprintf("#' @format A data frame with %s columns:", ncol(data)),
                "#'   \\describe{",
                paste(sapply(names(data), function(col_name) {
                  paste(sprintf("#'     \\item{ %s }{%s}", col_name, col_name))
                }, USE.NAMES = FALSE), collapse="\n"),
                "#'   }",
                "#'", sprintf("#' @source Generated from ${{ github.repository }}."),
                "#' @references None",
                "#'", sprintf("#' @examples\n#' data(\"%s\")", dataset_name),
                sep = '\n',
                sprintf("\"%s\"", dataset_name)
              )

              doc_dir <- "datasets_doc"
              if (!file.exists(doc_dir)) {
                dir.create(doc_dir, recursive = TRUE)
              }
              writeLines(doc_string, con = file.path(doc_dir, paste0(dataset_name, ".R")))
            }
          }
        shell: Rscript {0}

      # zip templates data
      - name: zip artifacts data
        if: inputs.push-templates-data
        run: |
          find "$GITHUB_WORKSPACE/tmp" -type f \( -name "*.rda" -o -name "*.csv" \) -exec zip -j "$GITHUB_WORKSPACE/data.zip" {} \;
          find "$GITHUB_WORKSPACE/datasets_doc" -type f \( -name "*.R" \) -exec zip -j "$GITHUB_WORKSPACE/doc.zip" {} \;

      # store templates data as artifacts
      - name: Archive templates data
        if: inputs.push-templates-data
        uses: actions/upload-artifact@v3
        with:
          name: data_templates
          path: ${{ github.workspace }}/data.zip

      # store templates data as artifacts
      - name: Archive doc
        if: inputs.push-templates-data
        uses: actions/upload-artifact@v3
        with:
          name: doc_templates
          path: ${{ github.workspace }}/doc.zip

      - name: Checkout repo (PR) 🛎
        if: inputs.push-templates-data
        uses: actions/checkout@v3
        with:
          ref: main
          repository: "${{ github.repository_owner }}/${{ env.repo }}"
          fetch-depth: 2
          token: ${{ secrets.PHARMAVERSE_BOT }}

      - name: Add safe directory
        if: inputs.push-templates-data
        run: |
          git config --global --add safe.directory "${GITHUB_WORKSPACE}"
        shell: bash

      - name: Download data artifacts
        if: inputs.push-templates-data
        uses: actions/download-artifact@v3
        with:
          name: data_templates

      - name: Download doc artifacts
        if: inputs.push-templates-data
        uses: actions/download-artifact@v3
        with:
          name: doc_templates

      # detect diffs om templates data
      - name: Check templates changes
        if: inputs.push-templates-data
        id: changes
        run: |
          head_data_folder=$(git ls-tree -r HEAD data) # check if data folder exists on remote main branch
          unzip -o "${GITHUB_WORKSPACE}/data.zip" -d "${GITHUB_WORKSPACE}/inst/extdata/"
          unzip -o "${GITHUB_WORKSPACE}/doc.zip" -d "${GITHUB_WORKSPACE}/R"
          # move every rda files inside data folder
          mkdir -p data
          for rda_file in inst/extdata/*.rda; do
            filename=$(basename "$rda_file")
            mv "$rda_file" "data/$filename"
          done
          if [ -z "${head_data_folder}" ]
          then {
            echo "data folder not existing on main branch yet - init mode (all templates data files will be pushed)"
            diff_files=$(ls "${GITHUB_WORKSPACE}/data")
          } else {
            git add data R
            diff_files_data=$(git status data | grep 'data/.*\.rda$')
            diff_files_doc=$(git status R | grep 'R/.*\.R$')
            diff_files="${diff_files_data}${diff_files_doc}"
          }
          fi
          if [ -z "$diff_files" ]
          then {
            echo "No changes detected after running templates. Stop here"
            echo "diff=false" >> $GITHUB_OUTPUT
          } else {
            echo "Changes detected, new PR will be created"
            echo "list of files with new updates: $diff_files"
            echo "diff=true" >> $GITHUB_OUTPUT
          }
          fi
        shell: bash {0}

      - name: Generate man pages
        if: inputs.push-templates-data
        run: |
          Rscript -e "roxygen2::roxygenize('.', roclets = c('rd', 'collate', 'namespace'))"
          git add man
        shell: bash {0}

      - name: Update spellchecks list
        if: inputs.push-templates-data
        run: |
          spelling::update_wordlist(pkg=".", confirm=FALSE)
        shell: Rscript {0}

      - name: set source-branch name
        if: inputs.push-templates-data
        id: branch
        run: |
          repo_name=$(basename "${{ github.repository }}")
          echo "source-branch=${{ env.source-branch }}_${repo_name}@devel" >> $GITHUB_OUTPUT
        shell: bash {0}


      # if diff detected, push changes
      - name: Commit and push changes in {{ env.repo }}
        if: ${{ inputs.push-templates-data && steps.changes.outputs.diff == 'true' }}
        uses: stefanzweifel/git-auto-commit-action@v4
        with:
          commit_message: templates data updates from ${{ github.repository }}
          file_pattern: 'inst/extdata/* data/* R/* inst/WORDLIST'
          commit_user_name: pharmaverse-bot
          commit_user_email: 113703390+pharmaverse-bot@users.noreply.github.com
          branch: "${{ steps.branch.outputs.source-branch }}"
          create_branch: true

      - name: Create Pull Request
        if: ${{ inputs.push-templates-data && steps.changes.outputs.diff == 'true' }}
        uses: actions/github-script@v6
        with:
          github-token: ${{ secrets.PHARMAVERSE_BOT }}
          script: |
            // Look for any open PRs
            const repo_owner = "${{ github.repository_owner }}";
            const target_repo_name = "${{ env.repo }}";
            const result = await github.rest.pulls.list({
              owner: repo_owner,
              repo: target_repo_name,
              state: "open"
            })
            let create_new_pr = true;
            for (const pr of result.data) {
              // Look for distinct PR branch name
              if (pr.head.ref === "${{ steps.branch.outputs.source-branch }}") {
                console.log("PR with head ref " + pr.head.ref + " already exists");
                create_new_pr = false;
                break;
              }
            }
            // If no PR with distinguished branch name has been found
            // create a new PR to track changes to renv.lock.
            if (create_new_pr) {
              console.log("Creating a new PR");
              const result2 = await github.rest.pulls.create({
                title: 'Templates data updates from ${{ github.repository }}',
                owner: repo_owner,
                repo: target_repo_name,
                head: '${{ steps.branch.outputs.source-branch }}',
                base: '${{ env.target-branch }}',
                body: [
                  'This PR has been automatically generated by ',
                  'check-templates workflow from ${{ github.repository }}.',
                  '\n\nPlease review the changes.'
                ].join('')
              });
              // Assign reviewers to the PR
              let rev = process.env.reviewers;
              let reviewers_dict = JSON.parse(rev);
              let current_repo_name_full = "${{ github.repository }}";
              let current_repo_name = current_repo_name_full.split('/').pop();
              const result3 = await github.rest.pulls.requestReviewers({
                owner: repo_owner,
                repo: target_repo_name,
                pull_number: result2.data.number,
                reviewers: reviewers_dict[current_repo_name]["reviewers"]
              });
            }