move all ci tag logic into parse_ci_tags

Author: strtgbb

ci/jobs/scripts/workflow_hooks/filter_job.py

@@ -1,6 +1,3 @@
-import re
-from functools import lru_cache
-
 from ci.defs.defs import JobNames
 from ci.defs.job_configs import JobConfigs
 from ci.jobs.scripts.workflow_hooks.new_tests_check import (
@@ -52,15 +49,6 @@ def only_docs(changed_files):
 _info_cache = None
 
 
-@lru_cache
-def get_ci_exclude_tags(pr_body):
-    pattern = r"(#|- \[x\] +<!---ci_exclude_)([|\w]+)"
-    matches = []
-    for match in re.findall(pattern, pr_body):
-        matches.extend(match[-1].split("|"))
-    return matches
-
-
 def should_skip_job(job_name):
     global _info_cache
     if _info_cache is None:
@@ -179,7 +167,7 @@ def should_skip_job(job_name):
             return False, ""
         return True, "Skipped, not labeled with 'pr-performance'"
 
-    ci_exclude_tags = get_ci_exclude_tags(_info_cache.pr_body)
+    ci_exclude_tags = _info_cache.get_kv_data("ci_exclude_tags") or []
     for tag in ci_exclude_tags:
         if tag in job_name:
             return True, f"Skipped, job name includes excluded tag '{tag}'"

ci/jobs/scripts/workflow_hooks/parse_ci_tags.py

@@ -0,0 +1,18 @@
+import re
+
+from ci.praktika.info import Info
+
+
+def get_ci_tags(pr_body, tag_prefix):
+    pattern = rf"(- \[x\] +<!---{tag_prefix}_)([|\w]+)"
+    matches = []
+    for match in re.findall(pattern, pr_body):
+        matches.extend(match[-1].split("|"))
+    return matches
+
+
+if __name__ == "__main__":
+    info = Info()
+
+    info.store_kv_data("ci_exclude_tags", get_ci_tags(info.pr_body, "ci_exclude"))
+    info.store_kv_data("ci_regression_jobs", get_ci_tags(info.pr_body, "ci_regression"))

ci/jobs/scripts/workflow_hooks/regression_config.py

@@ -1,5 +1,4 @@
 import copy
-import re
 
 from ci.defs.job_configs import JobConfigs
 from ci.jobs.scripts.clickhouse_version import CHVersion
@@ -8,13 +7,6 @@
 from ci.praktika.info import Info
 from ci.praktika.utils import Shell
 
-def get_ci_exclude_tags(pr_body):
-    pattern = r"(#|- \[x\] +<!---ci_exclude_)([|\w]+)"
-    matches = []
-    for match in re.findall(pattern, pr_body):
-        matches.extend(match[-1].split("|"))
-    return matches
-
 if __name__ == "__main__":
     info = Info()
 
@@ -31,9 +23,6 @@ def get_ci_exclude_tags(pr_body):
     digest = Digest().calc_job_digest(some_build_job, {}, {}).split("-")[0]
     info.store_kv_data("build_digest", digest)
 
-    # store ci exclude tags
-    info.store_kv_data("ci_exclude_tags", get_ci_exclude_tags(info.pr_body))
-
     if info.git_branch == "master" and info.repo_name == "ClickHouse/ClickHouse":
         # store previous commits for perf tests
         raw = Shell.get_output(

ci/jobs/scripts/workflow_hooks/store_data.py

@@ -63,7 +63,7 @@
     pre_hooks=[
         "python3 ./ci/jobs/scripts/workflow_hooks/store_data.py",
         "python3 ./ci/jobs/scripts/workflow_hooks/version_log.py",
-        "python3 ./ci/jobs/scripts/workflow_hooks/regression_config.py",
+        "python3 ./ci/jobs/scripts/workflow_hooks/parse_ci_tags.py",
         # "python3 ./ci/jobs/scripts/workflow_hooks/merge_sync_pr.py", # NOTE (strtgbb): we don't do this
     ],
     workflow_filter_hooks=[should_skip_job],

ci/workflows/master.py

@@ -133,7 +133,7 @@
         "python3 ./ci/jobs/scripts/workflow_hooks/version_log.py",
         # "python3 ./ci/jobs/scripts/workflow_hooks/quick_sync.py", # NOTE (strtgbb): we don't do this
         # "python3 ./ci/jobs/scripts/workflow_hooks/team_notifications.py",
-        "python3 ./ci/jobs/scripts/workflow_hooks/regression_config.py",
+        "python3 ./ci/jobs/scripts/workflow_hooks/parse_ci_tags.py",
     ],
     workflow_filter_hooks=[should_skip_job],
     post_hooks=[