chore: enhance monorepo migration scripts with generic exclusions and BOM substitution guards

Author: meltsufin

monorepo-migration/migrate-bigtable.sh

@@ -29,6 +29,8 @@ export SOURCE_REPO_URL="https://github.com/googleapis/java-bigtable"
 export MIGRATION_HEAD_BRANCH="main"
 export SQUASH_COMMITS="false"
 export CODEOWNER="@googleapis/bigtable-team" 
+export BOM_SUBSTITUTIONS="gapic-libraries-bom:google-cloud-monitoring-bom" 
+export PRE_INSTALL_DEPS="java-monitoring/google-cloud-monitoring-bom" 
 
 # 2. Execute the central migration script
 "${SCRIPT_DIR}/migrate.sh"

monorepo-migration/migrate.sh

@@ -443,6 +443,14 @@ sed -i.bak "s/'java-storage-nio'/'java-storage-nio'\n  '${SOURCE_REPO_NAME}'/" "
 python3 "$UPDATE_CI_FILTERS_SCRIPT" ".github/workflows/ci.yaml" "$SOURCE_REPO_NAME"
 python3 "$UPDATE_CHANGES_FILTERS_SCRIPT" ".github/workflows/ci.yaml" "$SOURCE_REPO_NAME"
 
+if [ -n "${PRE_INSTALL_DEPS}" ]; then
+    echo "Injecting explicit dependencies into always_install_deps list inside .kokoro/common.sh..."
+    for dep in $(echo "${PRE_INSTALL_DEPS}" | tr ',' ' '); do
+        sed -i.bak "s|always_install_deps_list=(|always_install_deps_list=(\n      '${dep}'|" ".kokoro/common.sh"
+    done
+    rm -f .kokoro/common.sh.bak
+fi
+
 echo "Committing common.sh and ci.yaml updates..."
 git add .kokoro/common.sh .github/workflows/ci.yaml
 git commit -n --no-gpg-sign -m "chore($SOURCE_REPO_NAME): exempt from global integration testing matrix"
@@ -465,7 +473,6 @@ fi
 # 7.11 Verify compilation
 echo "Verifying compilation..."
 BUILD_SUBDIR="${SOURCE_REPO_NAME}" JOB_TYPE=test .kokoro/build.sh
-# (cd "$SOURCE_REPO_NAME" && mvn compile -DskipTests -T 1C)
 
 # 7.13 Squash commits
 if [ "${SQUASH_COMMITS:-false}" = "true" ]; then

monorepo-migration/modernize_pom.py

@@ -85,7 +85,7 @@ def get_monorepo_versions(monorepo_root='.'):
     versions = {}
     for root, dirs, files in os.walk(monorepo_root):
         # Skip common directories to improve performance and avoid noise
-        dirs[:] = [d for d in dirs if d not in ['samples', 'test', 'target', '.git', '.cloud', 'verification', 'test_data']]
+        dirs[:] = [d for d in dirs if d not in ['samples', 'test', 'target', '.git', '.cloud', 'verification', 'test_data', 'generation']]
 
         if 'pom.xml' in files:
             pom_path = os.path.join(root, 'pom.xml')
@@ -95,6 +95,14 @@ def get_monorepo_versions(monorepo_root='.'):
     return versions
 
 def modernize_pom(file_path, parent_version, source_repo_name=None, parent_artifactId='google-cloud-jar-parent', relative_path='../google-cloud-jar-parent/pom.xml', monorepo_versions=None):
+    bom_substitutions_env = os.environ.get('BOM_SUBSTITUTIONS')
+    bom_substitutions = {}
+    if bom_substitutions_env:
+        for pair in bom_substitutions_env.split(','):
+            if ':' in pair:
+                old, new = pair.split(':', 1)
+                bom_substitutions[old.strip()] = new.strip()
+
     with open(file_path, 'r') as f:
         lines = f.readlines()
 
@@ -182,12 +190,30 @@ def modernize_pom(file_path, parent_version, source_repo_name=None, parent_artif
                     current_group_id = None
                     current_artifact_id = None
                     has_version = False
+                    in_exclusions = False
                     continue
                 if '</dependency>' in line:
                     in_dependency = False
                     current_dependency_lines.append(line)
 
-                    if current_artifact_id == 'google-cloud-shared-dependencies':
+                    if current_artifact_id == 'google-cloud-shared-dependencies' and '-deps-bom' not in file_path:
+                        continue
+                        
+                    if current_artifact_id in bom_substitutions:
+                        target_bom = bom_substitutions[current_artifact_id]
+                        target_version = monorepo_versions.get(target_bom, '0.0.1-SNAPSHOT')
+                        target_marker = target_bom.replace('-bom', '')
+                        indent = "      "
+                        current_dependency_lines = [
+                            f"{indent}<dependency>\n",
+                            f"{indent}  <groupId>com.google.cloud</groupId>\n",
+                            f"{indent}  <artifactId>{target_bom}</artifactId>\n",
+                            f"{indent}  <version>{target_version}</version><!-- {{x-version-update:{target_marker}:current}} -->\n",
+                            f"{indent}  <type>pom</type>\n",
+                            f"{indent}  <scope>import</scope>\n",
+                            f"{indent}</dependency>\n"
+                        ]
+                        new_lines.extend(current_dependency_lines)
                         continue
 
                     # Preservation logic:
@@ -196,22 +222,29 @@ def modernize_pom(file_path, parent_version, source_repo_name=None, parent_artif
                     # 3. Is com.google.cloud group AND artifactId starts with google-cloud- AND has a version tag
                     is_external = current_group_id and not current_group_id.startswith('com.google')
                     is_google_cloud_lib = current_group_id == 'com.google.cloud' and current_artifact_id and current_artifact_id.startswith('google-cloud-')
+                    is_truth = current_group_id and current_group_id.startswith('com.google.truth')
 
-                    if should_preserve or (is_external and has_version) or (is_google_cloud_lib and has_version):
+                    if should_preserve or (is_external and has_version) or (is_google_cloud_lib and has_version) or (is_truth and has_version):
                         new_lines.extend(current_dependency_lines)
                     continue
 
                 if in_dependency:
-                    if '<groupId>' in line:
-                        match = re.search(r'<groupId>(.*?)</groupId>', line)
-                        if match:
-                            current_group_id = match.group(1).strip()
-                    if '<artifactId>' in line:
-                        match = re.search(r'<artifactId>(.*?)</artifactId>', line)
-                        if match:
-                            current_artifact_id = match.group(1).strip()
-                    if '<version>' in line:
-                        has_version = True
+                    if '<exclusions>' in line:
+                        in_exclusions = True
+                    if '</exclusions>' in line:
+                        in_exclusions = False
+                        
+                    if not in_exclusions:
+                        if '<groupId>' in line:
+                            match = re.search(r'<groupId>(.*?)</groupId>', line)
+                            if match:
+                                current_group_id = match.group(1).strip()
+                        if '<artifactId>' in line:
+                            match = re.search(r'<artifactId>(.*?)</artifactId>', line)
+                            if match:
+                                current_artifact_id = match.group(1).strip()
+                        if '<version>' in line:
+                            has_version = True
 
                     if monorepo_versions and current_artifact_id and current_artifact_id in monorepo_versions:
                         new_version = monorepo_versions[current_artifact_id]