Add get-weaviate-branch and pull-weaviate-code actions

.github/actions/get-weaviate-branch/action.yml

@@ -0,0 +1,164 @@
+name: get-weaviate-branch
+description: 'Resolves a Weaviate Docker tag to the git ref (commit/branch) from weaviate/weaviate that matches the image code'
+inputs:
+  docker_tag:
+    description: 'The Docker tag to resolve (e.g. 1.36.6, preview-my-pr-abc1234.amd64, main-abc1234, nightly, latest)'
+    required: true
+  gh_token:
+    description: 'GitHub token for API calls to resolve short commit hashes (avoids rate limits). Optional but recommended.'
+    required: false
+    default: ''
+outputs:
+  ref:
+    description: 'Best git ref for checkout (always a full 40-char commit hash)'
+    value: ${{ steps.resolve.outputs.ref }}
+  commit:
+    description: 'Full 40-char commit hash'
+    value: ${{ steps.resolve.outputs.commit }}
+  branch:
+    description: 'Branch name best-guess (empty if not determinable, e.g. for preview/PR tags)'
+    value: ${{ steps.resolve.outputs.branch }}
+runs:
+  using: 'composite'
+  steps:
+    - name: Resolve Docker tag to git ref
+      id: resolve
+      shell: bash
+      env:
+        GH_TOKEN: ${{ inputs.gh_token }}
+      run: |
+        set -euo pipefail
+
+        # weaviate/weaviate is a public repo — git operations (ls-remote, fetch)
+        # work without authentication. The token is only needed for the GitHub
+        # API call that resolves short commit hashes (step 6).
+        WEAVIATE_REPO="https://github.com/weaviate/weaviate.git"
+        WEAVIATE_API="https://api.github.com/repos/weaviate/weaviate"
+        tag="${{ inputs.docker_tag }}"
+        commit=""
+        branch=""
+
+        echo "Resolving Docker tag: ${tag}"
+
+        # Step 1: Strip architecture suffix
+        tag_stripped="${tag}"
+        tag_stripped="${tag_stripped%.amd64}"
+        tag_stripped="${tag_stripped%.arm64}"
+
+        # Step 2: Handle special Docker-only tags
+        if [ "${tag_stripped}" = "nightly" ]; then
+          echo "Tag is 'nightly' — resolving main branch HEAD"
+          commit=$(git ls-remote "${WEAVIATE_REPO}" refs/heads/main | awk '{print $1}')
+          branch="main"
+        elif [ "${tag_stripped}" = "latest" ]; then
+          echo "Tag is 'latest' — finding highest release tag"
+          # Find the latest release tag name — capture full output to avoid
+          # broken pipe from head -1 under pipefail.
+          all_tags=$(git -c 'versionsort.suffix=-' ls-remote --tags --refs --sort='-v:refname' "${WEAVIATE_REPO}" \
+            | grep -oE 'refs/tags/v[0-9]+\.[0-9]+\.[0-9]+$' || true)
+          latest_tag=$(echo "${all_tags}" | head -1 | sed 's|refs/tags/||')
+          if [ -z "${latest_tag}" ]; then
+            echo "::error::Failed to find any release tag for 'latest' Docker tag"
+            exit 1
+          fi
+
+          # Resolve the annotated tag to its underlying commit (use ^{} dereferenced entry)
+          commit=$(git ls-remote --tags "${WEAVIATE_REPO}" "refs/tags/${latest_tag}" "refs/tags/${latest_tag}^{}" \
+            | tail -1 | awk '{print $1}')
+          if [ -z "${commit}" ]; then
+            echo "::error::Failed to resolve commit for tag '${latest_tag}'"
+            exit 1
+          fi
+
+          # Extract major.minor for branch
+          version="${latest_tag#v}"
+          major_minor=$(echo "${version}" | grep -oE '^[0-9]+\.[0-9]+')
+          branch="stable/v${major_minor}"
+          echo "Resolved 'latest' to tag ${latest_tag} (commit ${commit})"
+        else
+          # Step 3: Try to extract 7-char hex hash from end (after last '-')
+          suffix="${tag_stripped##*-}"
+          if [[ "${suffix}" =~ ^[0-9a-f]{7}$ ]]; then
+            commit="${suffix}"
+            prefix="${tag_stripped%-*}"
+
+            # Step 4: Determine branch from prefix
+            if [[ "${prefix}" =~ ^preview- ]]; then
+              # Preview/PR build — branch is not recoverable from tag
+              branch=""
+              echo "Preview tag detected, short commit: ${commit}"
+            elif [[ "${prefix}" =~ ^[0-9]+\.[0-9]+\.[0-9]+ ]]; then
+              # Semver prefix (e.g. 1.37.0-dev, 1.36.6) — branch not directly in tag
+              branch=""
+              echo "Semver+hash tag detected, short commit: ${commit}"
+            else
+              # Branch name prefix (e.g. main, stable-v1.34)
+              # In push_docker.sh, '/' in branch names is replaced with '-'
+              # Try common patterns to restore the original branch name
+              if [[ "${prefix}" =~ ^stable-v[0-9]+\.[0-9]+$ ]]; then
+                branch="${prefix/stable-/stable/}"
+              else
+                # For simple branch names (main, feature-x), keep as-is
+                # Cannot reliably restore '/' for arbitrary branches
+                branch="${prefix}"
+              fi
+              echo "Branch tag detected, branch: ${branch}, short commit: ${commit}"
+            fi
+          else
+            # Step 5: No hash found — must be a release version tag (X.Y.Z or X.Y.Z-rc.N)
+            version="${tag_stripped}"
+            git_tag="v${version}"
+            echo "Release tag detected: ${git_tag} — resolving commit from git"
+
+            # Resolve the annotated tag to its commit (use ^{} dereferenced entry if present)
+            tag_commit=$(git ls-remote --tags "${WEAVIATE_REPO}" "refs/tags/${git_tag}" "refs/tags/${git_tag}^{}" \
+              | tail -1 | awk '{print $1}')
+
+            if [ -z "${tag_commit}" ]; then
+              echo "::error::Could not find git tag ${git_tag} in weaviate/weaviate"
+              exit 1
+            fi
+
+            commit="${tag_commit}"
+            # Extract major.minor for stable branch (grep handles RC versions like 1.36.0-rc.0 → 1.36)
+            major_minor=$(echo "${version}" | grep -oE '^[0-9]+\.[0-9]+')
+            branch="stable/v${major_minor}"
+            echo "Resolved ${git_tag} to commit ${commit}"
+          fi
+        fi
+
+        # Step 6: Resolve short hashes (7-char) to full 40-char SHA via GitHub API
+        if [ -z "${commit}" ]; then
+          echo "::error::No commit hash could be extracted from Docker tag '${tag}'"
+          exit 1
+        fi
+        if [ "${#commit}" -lt 40 ]; then
+          echo "Resolving short hash ${commit} to full SHA..."
+          curl_args=(-sf -H "Accept: application/vnd.github.v3+json")
+          if [ -n "${GH_TOKEN}" ]; then
+            curl_args+=(-H "Authorization: Bearer ${GH_TOKEN}")
+          fi
+          if ! response=$(curl "${curl_args[@]}" "${WEAVIATE_API}/commits/${commit}"); then
+            echo "::error::GitHub API call failed when resolving short hash '${commit}'"
+            exit 1
+          fi
+          full_sha=$(printf '%s\n' "${response}" | jq -r '.sha // empty')
+
+          if [ -z "${full_sha}" ]; then
+            echo "::error::Could not resolve short hash ${commit} to a full SHA"
+            exit 1
+          fi
+          echo "Resolved ${commit} -> ${full_sha}"
+          commit="${full_sha}"
+        fi
+
+        ref="${commit}"
+
+        echo "ref=${ref}" >> "${GITHUB_OUTPUT}"
+        echo "commit=${commit}" >> "${GITHUB_OUTPUT}"
+        echo "branch=${branch}" >> "${GITHUB_OUTPUT}"
+
+        echo "--- Results ---"
+        echo "ref:    ${ref}"
+        echo "commit: ${commit}"
+        echo "branch: ${branch}"

.github/actions/pull-weaviate-code/action.yml

@@ -0,0 +1,68 @@
+name: pull-weaviate-code
+description: 'Clones the weaviate/weaviate repository at the commit matching a given Docker tag'
+inputs:
+  docker_tag:
+    description: 'Weaviate Docker tag to resolve (e.g. 1.36.6, preview-my-pr-abc1234.amd64, main-abc1234, nightly, latest)'
+    required: true
+  directory:
+    description: 'Directory in which to clone the weaviate repository (a "weaviate" folder will be created inside it)'
+    required: false
+    default: '.'
+  gh_token:
+    description: 'GitHub token passed to get-weaviate-branch for API calls (avoids rate limits). Optional but recommended.'
+    required: false
+    default: ''
+outputs:
+  ref:
+    description: 'The git ref that was checked out (full 40-char commit hash)'
+    value: ${{ steps.resolve-tag.outputs.ref }}
+  commit:
+    description: 'Full 40-char commit hash that was checked out'
+    value: ${{ steps.resolve-tag.outputs.commit }}
+  branch:
+    description: 'The branch name (if determinable)'
+    value: ${{ steps.resolve-tag.outputs.branch }}
+  weaviate_path:
+    description: 'Full path to the cloned weaviate directory'
+    value: ${{ steps.clone.outputs.weaviate_path }}
+runs:
+  using: 'composite'
+  steps:
+    - name: Resolve Docker tag to git ref
+      id: resolve-tag
+      uses: ./.github/actions/get-weaviate-branch
+      with:
+        docker_tag: ${{ inputs.docker_tag }}
+        gh_token: ${{ inputs.gh_token }}
+
+    - name: Clone weaviate at resolved ref
+      id: clone
+      shell: bash
+      run: |
+        set -euo pipefail
+
+        # weaviate/weaviate is a public repo — no auth needed for git fetch.
+        WEAVIATE_REPO="https://github.com/weaviate/weaviate.git"
+        TARGET_DIR="${{ inputs.directory }}"
+        WEAVIATE_DIR="${TARGET_DIR}/weaviate"
+        COMMIT="${{ steps.resolve-tag.outputs.commit }}"
+
+        # Check that the weaviate path doesn't already exist (file, directory, or symlink)
+        if [ -e "${WEAVIATE_DIR}" ]; then
+          echo "::error::Path '${WEAVIATE_DIR}' already exists. Remove it before running this action to avoid conflicts."
+          exit 1
+        fi
+
+        # Ensure parent directory exists
+        mkdir -p "${TARGET_DIR}"
+
+        echo "Cloning weaviate at commit ${COMMIT} into ${WEAVIATE_DIR}..."
+        git init "${WEAVIATE_DIR}"
+        cd "${WEAVIATE_DIR}"
+        git remote add origin "${WEAVIATE_REPO}"
+        git fetch --depth=1 origin "${COMMIT}"
+        git checkout FETCH_HEAD
+
+        FULL_PATH="$(pwd)"
+        echo "weaviate_path=${FULL_PATH}" >> "${GITHUB_OUTPUT}"
+        echo "Successfully cloned weaviate at ${COMMIT} to ${FULL_PATH}"