[autorevert] Add Autorevert Metrics page with visualizations and data retrieval

torchci/clickhouse_queries/autorevert_events_with_commits/params.json

@@ -0,0 +1,8 @@
+{
+  "params": {
+    "startTime": "DateTime64(3)",
+    "stopTime": "DateTime64(3)",
+    "workflowNames": "Array(String)"
+  },
+  "tests": []
+}

torchci/clickhouse_queries/autorevert_events_with_commits/query.sql

@@ -0,0 +1,139 @@
+-- Autorevert Events with Linked Revert Commits
+-- Links autorevert_events_v2 to the actual revert commits via PR number and timestamp
+-- Used for false positive detection
+
+WITH autorevert_events AS (
+    SELECT
+        toString(a.commit_sha) AS reverted_sha,
+        a.ts AS autorevert_time,
+        a.workflows,
+        a.source_signal_keys
+    FROM misc.autorevert_events_v2 a FINAL
+    WHERE
+        a.repo = 'pytorch/pytorch'
+        AND a.action = 'revert'
+        AND a.dry_run = 0
+        AND a.failed = 0
+        AND a.ts >= toDateTime({startTime: DateTime64(3)}) - INTERVAL 1 DAY
+        AND a.ts < toDateTime({stopTime: DateTime64(3)}) + INTERVAL 1 DAY
+        -- Filter by workflow intersection
+        AND hasAny(a.workflows, {workflowNames: Array(String)})
+),
+
+-- Get PR number from the reverted commit's message
+autorevert_with_pr AS (
+    SELECT
+        a.reverted_sha,
+        a.autorevert_time,
+        a.workflows,
+        a.source_signal_keys,
+        p.head_commit.'message' AS reverted_message,
+        -- Extract PR number from "Pull Request resolved: https://github.com/pytorch/pytorch/pull/XXXXX"
+        arrayElement(
+            extractAll(
+                p.head_commit.'message',
+                'Pull Request resolved: https://github.com/pytorch/pytorch/pull/(\\d+)'
+            ), 1
+        ) AS pr_number
+    FROM autorevert_events a
+    JOIN push p ON p.head_commit.'id' = a.reverted_sha
+    WHERE p.repository.'full_name' = 'pytorch/pytorch'
+),
+
+-- Find revert commits in the time range
+revert_commits AS (
+    SELECT
+        push.head_commit.'id' AS revert_sha,
+        push.head_commit.'timestamp' AS revert_time,
+        push.head_commit.'message' AS revert_message,
+        -- Extract mentioned PR numbers from revert message
+        -- For nested reverts like 'Reapply "Back out "..." (#164939)" (#165910)" (#166812)',
+        -- the actual PR is the LAST one mentioned in the title line (before newline)
+        if(
+            arrayElement(
+                extractAll(
+                    push.head_commit.'message',
+                    'Reverted https://github.com/pytorch/pytorch/pull/(\\d+)'
+                ), 1
+            ) != '',
+            arrayElement(
+                extractAll(
+                    push.head_commit.'message',
+                    'Reverted https://github.com/pytorch/pytorch/pull/(\\d+)'
+                ), 1
+            ),
+            -- Get the LAST PR number from title (use -1 for last element)
+            arrayElement(
+                extractAll(
+                    -- Extract just the first line (title) to get the correct PR
+                    arrayElement(
+                        splitByChar('\n', push.head_commit.'message'), 1
+                    ),
+                    '#(\\d+)'
+                ), -1
+            )
+        ) AS pr_reference
+    FROM push
+    WHERE
+        push.ref IN ('refs/heads/master', 'refs/heads/main')
+        AND push.repository.'full_name' = 'pytorch/pytorch'
+        AND push.head_commit.'timestamp' >= {startTime: DateTime64(3)}
+        AND push.head_commit.'timestamp' < {stopTime: DateTime64(3)}
+        AND (
+            push.head_commit.'message' LIKE 'Revert %'
+            OR push.head_commit.'message' LIKE 'Reapply %'
+            OR push.head_commit.'message' LIKE 'Back out%'
+        )
+),
+
+-- Join on PR number only, then filter by time in WHERE clause
+-- This avoids ClickHouse JOIN ON restrictions
+matched_reverts AS (
+    SELECT
+        a.reverted_sha,
+        a.autorevert_time,
+        a.workflows,
+        a.source_signal_keys,
+        a.pr_number,
+        substring(a.reverted_message, 1, 100) AS reverted_message_snippet,
+        r.revert_sha,
+        r.revert_time,
+        substring(r.revert_message, 1, 100) AS revert_message_snippet
+    FROM autorevert_with_pr a
+    LEFT JOIN revert_commits r ON r.pr_reference = a.pr_number
+    WHERE
+        a.pr_number != ''
+        AND (
+            -- Keep autoreverts even if no linked revert found
+            r.revert_sha IS NULL
+            OR (
+                r.revert_time > a.autorevert_time
+                AND r.revert_time < a.autorevert_time + INTERVAL 1 HOUR
+            )
+        )
+),
+
+-- Take the first revert commit after the autorevert event for each PR
+linked_autoreverts AS (
+    SELECT
+        *,
+        row_number() OVER (
+            PARTITION BY reverted_sha
+            ORDER BY revert_time ASC NULLS LAST
+        ) AS rn
+    FROM matched_reverts
+)
+
+SELECT
+    reverted_sha,
+    autorevert_time,
+    workflows,
+    source_signal_keys,
+    pr_number,
+    reverted_message_snippet,
+    revert_sha,
+    revert_time,
+    revert_message_snippet
+FROM linked_autoreverts
+WHERE rn = 1
+ORDER BY autorevert_time DESC

torchci/clickhouse_queries/autorevert_false_positives/params.json

@@ -0,0 +1,7 @@
+{
+  "params": {
+    "startTime": "DateTime64(3)",
+    "stopTime": "DateTime64(3)"
+  },
+  "tests": []
+}

torchci/clickhouse_queries/autorevert_false_positives/query.sql

@@ -0,0 +1,107 @@
+-- Autorevert False Positive Candidates
+-- Finds autoreverted commits that were later re-landed
+-- NOTE: This is a preliminary list. True false positives require GitHub API verification:
+--   - If PR is still OPEN (not relanded) → revert was LEGIT
+--   - If PR had commits after revert before reland → revert was LEGIT (author fixed something)
+--   - If PR was relanded with NO changes → revert was FALSE POSITIVE
+
+WITH autorevert_events AS (
+    SELECT
+        toString(a.commit_sha) AS reverted_sha,
+        min(a.ts) AS revert_time
+    FROM misc.autorevert_events_v2 a FINAL
+    WHERE
+        a.repo = 'pytorch/pytorch'
+        AND a.action = 'revert'
+        AND a.dry_run = 0
+        AND a.failed = 0
+        AND a.ts >= toDateTime({startTime: DateTime64(3)}) - INTERVAL 7 DAY
+        AND a.ts < toDateTime({stopTime: DateTime64(3)}) + INTERVAL 7 DAY
+    GROUP BY reverted_sha
+),
+
+-- Get original commit details and extract PR number
+autorevert_with_pr AS (
+    SELECT
+        a.reverted_sha,
+        a.revert_time,
+        p.head_commit.'message' AS original_message,
+        arrayElement(
+            extractAll(
+                p.head_commit.'message',
+                'Pull Request resolved: https://github.com/pytorch/pytorch/pull/(\\d+)'
+            ), 1
+        ) AS pr_number
+    FROM autorevert_events a
+    JOIN push p ON p.head_commit.'id' = a.reverted_sha
+    WHERE p.repository.'full_name' = 'pytorch/pytorch'
+),
+
+-- Find all commits in the time range
+all_commits AS (
+    SELECT
+        push.head_commit.'id' AS sha,
+        push.head_commit.'timestamp' AS time,
+        push.head_commit.'message' AS message
+    FROM push
+    WHERE
+        push.ref IN ('refs/heads/master', 'refs/heads/main')
+        AND push.repository.'full_name' = 'pytorch/pytorch'
+        AND push.head_commit.'timestamp' >= {startTime: DateTime64(3)}
+        AND push.head_commit.'timestamp' < {stopTime: DateTime64(3)}
+),
+
+-- Find Reland commits
+reland_commits AS (
+    SELECT
+        sha AS reland_sha,
+        time AS reland_time,
+        message AS reland_message,
+        -- Extract the original PR being relanded (first PR number mentioned)
+        arrayElement(extractAll(message, '#(\\d+)'), 1) AS primary_mentioned_pr
+    FROM all_commits
+    WHERE
+        message LIKE 'Reland%'
+        OR message LIKE '[Reland]%'
+        OR message LIKE 'Re-land%'
+),
+
+-- Match autoreverts to relands, deduplicated by original PR
+-- Take the first reland for each autoreverted PR
+matched_relands AS (
+    SELECT
+        a.reverted_sha,
+        a.revert_time,
+        a.pr_number AS original_pr,
+        substring(a.original_message, 1, 100) AS original_message_snippet,
+        r.reland_sha,
+        r.reland_time,
+        substring(r.reland_message, 1, 100) AS reland_message_snippet,
+        dateDiff('hour', a.revert_time, r.reland_time) AS hours_to_reland,
+        row_number()
+            OVER (
+                PARTITION BY a.pr_number
+                ORDER BY r.reland_time ASC
+            )
+            AS rn
+    FROM autorevert_with_pr a
+    JOIN reland_commits r ON r.primary_mentioned_pr = a.pr_number
+    WHERE
+        a.pr_number != ''
+        AND r.reland_time > a.revert_time
+        AND r.reland_time < a.revert_time + INTERVAL 30 DAY
+)
+
+SELECT
+    original_pr,
+    reverted_sha,
+    revert_time,
+    original_message_snippet,
+    reland_sha,
+    reland_time,
+    reland_message_snippet,
+    hours_to_reland,
+    'needs_verification' AS status
+FROM matched_relands
+WHERE rn = 1  -- Only first reland per PR
+ORDER BY revert_time DESC

torchci/clickhouse_queries/autorevert_signal_recovery/params.json

@@ -0,0 +1,29 @@
+{
+  "params": {
+    "startTime": "DateTime64(3)",
+    "stopTime": "DateTime64(3)",
+    "workflowNames": "Array(String)",
+    "minRedCommits": "UInt8",
+    "minGreenCommits": "UInt8"
+  },
+  "tests": [
+    {
+      "startTime": "2024-12-01 00:00:00.000",
+      "stopTime": "2024-12-29 00:00:00.000",
+      "workflowNames": ["Lint", "pull", "trunk", "linux-aarch64"],
+      "minRedCommits": 2,
+      "minGreenCommits": 2
+    },
+    {
+      "startTime": {
+        "from_now": -14
+      },
+      "stopTime": {
+        "from_now": 0
+      },
+      "workflowNames": ["Lint", "pull", "trunk", "linux-aarch64"],
+      "minRedCommits": 2,
+      "minGreenCommits": 2
+    }
+  ]
+}