Convert license check to Invoke

Author: blag

tasks/__init__.py

@@ -18,8 +18,9 @@ def all_(ctx):
     pass
 
 
-@task(check.compile_, lint.flake8, lint.pylint, copy.copy_pack_to_subdirectory,
-      check.configs, check.metadata, tests.packs_resource_register, tests.packs_tests)
+@task(check.compile_, check.license, lint.flake8, lint.pylint,
+      copy.copy_pack_to_subdirectory, check.configs, check.metadata,
+      tests.packs_resource_register, tests.packs_tests)
 def all_ci(ctx):
     pass
 

tasks/check.py

@@ -17,6 +17,48 @@ def compile_(ctx):
         raise Exception("Could not compile all files")
 
 
+@task
+def license(ctx):
+    # Verifies repo contains LICENSE file with ASF 2.0 content
+    print("")
+    print("==================== license-check ====================")
+    print("")
+    apache_license_rgx = re.compile(r'.*Apache License.*')
+    license_version_rgx = re.compile(r'.*Version 2\.0.*')
+    apache_url_rgx = re.compile(r'.*www\.apache\.org/licenses/LICENSE-2\.0.*')
+    files = []
+    top_level_git_dir = ctx.run('git rev-parse --show-toplevel').stdout.splitlines()[0]
+    base_branch = os.environ.get('BASE_BRANCH', 'origin/master')
+    license_file = os.path.join(top_level_git_dir, 'LICENSE')
+    with ctx.cd(top_level_git_dir):
+        files = ctx.run("git diff --relative --diff-filter=ACMRTUXB "
+                        "--name-only {}".format(base_branch)).stdout.splitlines()
+    if os.environ.get('FORCE_CHECK_ALL_FILES', False) == 'true' or files:
+        # Check for the existence of a LICENSE file
+        if not os.path.exist(license_file):
+            raise Exception("Missing LICENSE file in {root_dir}".format(root_dir=top_level_git_dir))
+
+        with open(license_file) as f:
+            lines = f.read().splitlines()
+
+        found_apache_license = False
+        found_license_version = False
+        found_apache_url = False
+        for line in lines:
+            if apache_license_rgx.match(line):
+                found_apache_license = True
+            if license_version_rgx.match(line):
+                found_license_version = True
+            if apache_url_rgx.match(line):
+                found_apache_url = True
+            if found_apache_license and found_license_version and found_apache_url:
+                break
+        else:
+            raise Exception("LICENSE file doesn't contain Apache 2.0 license text")
+    else:
+        print("No files have changed, skipping run...")
+
+
 @task(copy.copy_pack_to_subdirectory)
 def configs(ctx):
     print("")