Remove upload_test_stat_aggregates script (#139915)

Instead of moving these queries to ClickHouse, we're just going to remove it since it's not really used.  We do want something for test aggregates, but we can make a new script instead
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139915
Approved by: https://github.com/huydhn

.github/workflows/nightly-s3-uploads.yml

@@ -1,4 +1,4 @@
-name: Nightly Upload to rockset
+name: Nightly Upload to s3
 
 on:
   schedule:
@@ -7,8 +7,7 @@ on:
   pull_request:
     paths:
       - 'tools/stats/upload_external_contrib_stats.py'
-      - 'tools/stats/upload_test_stat_aggregates.py'
-      - '.github/workflows/nightly-rockset-uploads.yml'
+      - '.github/workflows/nightly-s3-uploads.yml'
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
@@ -16,7 +15,7 @@ concurrency:
 
 jobs:
 
-  upload-stats-to-rockset:
+  upload-stats-to-s3:
     if: github.repository_owner == 'pytorch'
     runs-on: ubuntu-22.04
     environment: upload-stats
@@ -33,16 +32,14 @@ jobs:
           cache: pip
 
       - run: |
-          pip3 install requests==2.32.2 rockset==1.0.3 boto3==1.35.42
+          pip3 install requests==2.32.2 boto3==1.35.42
 
       - name: Upload external contribution stats
         uses: nick-fields/retry@v3.0.0
         env:
-          ROCKSET_API_KEY: ${{ secrets.ROCKSET_API_KEY }}
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-        if: ${{ env.ROCKSET_API_KEY != '' }}
         with:
           timeout_minutes: 10
           max_attempts: 10
@@ -50,5 +47,3 @@ jobs:
           command: |
             echo "Uploading external contribution stats for 10 days starting on" "$(date -d '10 days ago' '+%Y-%m-%d')"
             python3 -m tools.stats.upload_external_contrib_stats --startDate "$(date -d '10 days ago' '+%Y-%m-%d')" --length 10
-            echo "Uploading testing aggregate data" "$(date -d yesterday '+%Y-%m-%d')"
-            python3 -m tools.stats.upload_test_stat_aggregates --date "$(date -d yesterday '+%Y-%m-%d')"

tools/stats/upload_external_contrib_stats.py

@@ -112,7 +112,7 @@ def get_external_pr_data(
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(
-        description="Upload external contribution stats to Rockset"
+        description="Upload external contribution stats to s3"
     )
     parser.add_argument(
         "--startDate",

tools/stats/upload_stats_lib.py

@@ -13,7 +13,6 @@ from typing import Any, Callable, Dict, List, Optional
 
 import boto3  # type: ignore[import]
 import requests
-import rockset  # type: ignore[import]
 
 
 PYTORCH_REPO = "https://api.github.com/repos/pytorch/pytorch"
@@ -128,6 +127,8 @@ def upload_to_rockset(
     workspace: str = "commons",
     client: Any = None,
 ) -> None:
+    import rockset  # type: ignore[import]
+
     if not client:
         client = rockset.RocksetClient(
             host="api.usw2a1.rockset.com", api_key=os.environ["ROCKSET_API_KEY"]

tools/stats/upload_test_stat_aggregates.py

@@ -1,86 +0,0 @@
-from __future__ import annotations
-
-import argparse
-import ast
-import datetime
-import json
-import os
-import re
-from typing import Any
-
-import rockset  # type: ignore[import]
-
-from tools.stats.upload_stats_lib import upload_to_s3
-
-
-def get_oncall_from_testfile(testfile: str) -> list[str] | None:
-    path = f"test/{testfile}"
-    if not path.endswith(".py"):
-        path += ".py"
-    # get oncall on test file
-    try:
-        with open(path) as f:
-            for line in f:
-                if line.startswith("# Owner(s): "):
-                    possible_lists = re.findall(r"\[.*\]", line)
-                    if len(possible_lists) > 1:
-                        raise Exception("More than one list found")  # noqa: TRY002
-                    elif len(possible_lists) == 0:
-                        raise Exception(  # noqa: TRY002
-                            "No oncalls found or file is badly formatted"
-                        )  # noqa: TRY002
-                    oncalls = ast.literal_eval(possible_lists[0])
-                    return list(oncalls)
-    except Exception as e:
-        if "." in testfile:
-            return [f"module: {testfile.split('.')[0]}"]
-        else:
-            return ["module: unmarked"]
-    return None
-
-
-def get_test_stat_aggregates(date: datetime.date) -> Any:
-    # Initialize the Rockset client with your API key
-    rockset_api_key = os.environ["ROCKSET_API_KEY"]
-    rockset_api_server = "api.rs2.usw2.rockset.com"
-    iso_date = date.isoformat()
-    rs = rockset.RocksetClient(host="api.usw2a1.rockset.com", api_key=rockset_api_key)
-
-    # Define the name of the Rockset collection and lambda function
-    collection_name = "commons"
-    lambda_function_name = "test_insights_per_daily_upload"
-    query_parameters = [
-        rockset.models.QueryParameter(name="startTime", type="string", value=iso_date)
-    ]
-    api_response = rs.QueryLambdas.execute_query_lambda(
-        query_lambda=lambda_function_name,
-        version="692684fa5b37177f",
-        parameters=query_parameters,
-    )
-    for i in range(len(api_response["results"])):
-        oncalls = get_oncall_from_testfile(api_response["results"][i]["test_file"])
-        api_response["results"][i]["oncalls"] = oncalls
-    return json.loads(
-        json.dumps(api_response["results"], indent=4, sort_keys=True, default=str)
-    )
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
-        description="Upload test stat aggregates to Rockset."
-    )
-    parser.add_argument(
-        "--date",
-        type=datetime.date.fromisoformat,
-        help="Date to upload test stat aggregates for (YYYY-MM-DD). Must be in the last 30 days",
-        required=True,
-    )
-    args = parser.parse_args()
-    if args.date < datetime.datetime.now().date() - datetime.timedelta(days=30):
-        raise ValueError("date must be in the last 30 days")
-    data = get_test_stat_aggregates(date=args.date)
-    upload_to_s3(
-        bucket_name="torchci-aggregated-stats",
-        key=f"test_data_aggregates/{str(args.date)}",
-        docs=data,
-    )