Add method to analyze reverts: analyze_reverts_missing_from_branch (#6834)

Usage:
```
python github_analyze.py --analyze-missing-reverts-from-branch --repo-path ~/local/pytorch --remote upstream --branch release/2.8
Analyzing reverts in main branch not present in release/2.8 branch
Total commits in main but not in release/2.8: 62
Total commits in release/2.8 but not in main: 1

Found 8 revert(s) in main branch not present in release/2.8 branch:
================================================================================
Commit Hash: 029e2b05c225588098d3eba445fd04189691f77d
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-25 06:05:40
Title: Revert "[Quant][CPU] fix fake_quantize_per_tensor_affine of inf values (#155109)"
Reverted GitHub Commit: 19ffb5e6f7606436249742b0f3efc0bab244dc55
Body Preview:     
    This reverts commit 19ffb5e6f7606436249742b0f3efc0bab244dc55.
    
    Reverted pytorch/pytorch#155109 on behalf of https://github.com/albanD due to The corresponding ...
--------------------------------------------------------------------------------
Commit Hash: 3dd872e6d53560933d8d7fc11357617746d37168
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 17:11:35
Title: Revert "Add DeviceAllocator as the base device allocator (#138222)"
Reverted GitHub Commit: 92409b6c89fbfbd3caa79c81b1e3d9e7917d3bc7
Body Preview:     
    This reverts commit 92409b6c89fbfbd3caa79c81b1e3d9e7917d3bc7.
    
    Reverted pytorch/pytorch#138222 on behalf of https://github.com/Camyll due to internal build fai...
--------------------------------------------------------------------------------
Commit Hash: 6459a5c7a92e7a38db374780b91116cc958b6af9
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 17:11:35
Title: Revert "Add unified memory APIs for torch.accelerator (#152932)"
Reverted GitHub Commit: 35e44067c4d9cc9be2652c0b9098885c5a321029
Body Preview:     
    This reverts commit 35e44067c4d9cc9be2652c0b9098885c5a321029.
    
    Reverted pytorch/pytorch#152932 on behalf of https://github.com/Camyll due to internal build fai...
--------------------------------------------------------------------------------
Commit Hash: fd4bb29410c035b31ca55262c3012cadb1194aae
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 16:45:13
Title: Revert "[logging] dynamo_timed for CachingAutotuner.coordinate_descent_tuning (#156517)"
Reverted GitHub Commit: fb75dea2c1b93c78dccf08d5fd5e20b362ecd405
Body Preview:     
    This reverts commit fb75dea2c1b93c78dccf08d5fd5e20b362ecd405.
    
    Reverted pytorch/pytorch#156517 on behalf of https://github.com/Camyll due to internal reverted ...
--------------------------------------------------------------------------------
Commit Hash: 4bd18e31e5a38d0e84ce915b1fa124058c6373fa
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 16:34:21
Title: Revert "Add fx_graph_runnable tests boilerplate (#156552)"
Reverted GitHub Commit: 0a2ec7681d2af973d8daaf7905431a088739dc90
Body Preview:     
    This reverts commit 0a2ec7681d2af973d8daaf7905431a088739dc90.
    
    Reverted pytorch/pytorch#156552 on behalf of https://github.com/Camyll due to breaking internal ...
--------------------------------------------------------------------------------
Commit Hash: 1dc1eedd4369f6e6bb79d5315e3ffc1bdc59b709
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 13:44:42
Title: Revert "[dynamo] Graph break on `torch.Tensor.data` assignment with mismatched dtype (#156623)"
Reverted GitHub Commit: c1ad4b8e7a16f54c35a3908b56ed7d9f95eef586
Body Preview:     
    This reverts commit c1ad4b8e7a16f54c35a3908b56ed7d9f95eef586.
    
    Reverted pytorch/pytorch#156623 on behalf of https://github.com/albanD due to Breaks Dynamo test...
--------------------------------------------------------------------------------
Commit Hash: aa280ea19fb20923d048909fa98af092e18ca2fb
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 13:05:39
Title: Revert "Remove remaining CUDA 12.4 CI code (#155412)"
Reverted GitHub Commit: 9fed2addedb42da86b657165fe14eadc911232cf
Body Preview:     
    This reverts commit 9fed2addedb42da86b657165fe14eadc911232cf.
    
    Reverted pytorch/pytorch#155412 on behalf of https://github.com/Camyll due to cuda 12.4 still ne...
--------------------------------------------------------------------------------
Commit Hash: 19f851ce10b16f0ed11d18d937ca7b32746153b0
Author: PyTorch MergeBot <[email protected]>
Date: 2025-06-24 13:02:07
Title: Revert "Simplify nvtx3 CMake handling, always use nvtx3 (#153784)"
Reverted GitHub Commit: 099d0d6121125062ebc05771c8330cb7cd8d053a
Body Preview:     
    This reverts commit 099d0d6121125062ebc05771c8330cb7cd8d053a.
    
    Reverted pytorch/pytorch#153784 on behalf of https://github.com/Camyll due to breaking internal ...
--------------------------------------------------------------------------------
```

Author: atalman

tools/analytics/github_analyze.py

@@ -5,6 +5,7 @@
 import enum
 import json
 import os
+import re
 from datetime import datetime, timedelta
 from typing import Any, Dict, Iterable, List, Optional, Union
 from urllib.error import HTTPError
@@ -542,6 +543,106 @@ def analyze_stacks(repo: GitRepo) -> None:
         )
 
 
+def extract_commit_hash_from_revert(text):
+    """
+    Extract commit hash from a revert commit message.
+
+    Args:
+        text (str): The revert commit message
+
+    Returns:
+        str or None: The extracted commit hash, or None if not found
+    """
+    # Pattern to match "This reverts commit <hash>."
+    pattern = r"This reverts commit ([0-9a-f]+)\."
+
+    match = re.search(pattern, text)
+    if match:
+        return match.group(1)
+    return None
+
+
+def analyze_reverts_missing_from_branch(repo: GitRepo, branch: str) -> None:
+    """
+    Analyze reverts applied to main branch but not applied to specified branch.
+    This identifies potential missing revert commits that may need to be cherry-picked
+    to the release branch. Also detects if reverted commits from main were cherry-picked
+    to the branch.
+    """
+    # Get commits from main that are not in the specified branch
+    main_only_commits = build_commit_dict(repo.get_commit_list(branch, "main"))
+
+    # Get commits from the specified branch that are not in main
+    branch_only_commits = build_commit_dict(repo.get_commit_list("main", branch))
+    branch_only_reverts = set()
+
+    print(f"Analyzing reverts in main branch not present in {branch} branch")
+    print(f"Total commits in main but not in {branch}: {len(main_only_commits)}")
+    print(f"Total commits in {branch} but not in main: {len(branch_only_commits)}")
+    print()
+
+    for commit_hash, commit in branch_only_commits.items():
+        revert_hash = extract_commit_hash_from_revert(commit.body)
+        if revert_hash != None:
+            branch_only_reverts.add(revert_hash)
+        if is_revert(commit):
+            branch_only_reverts.add(commit_hash)
+
+    # Find reverts in main that are not in the specified branch
+    reverts_missing_from_branch = []
+
+    for commit_hash, commit in main_only_commits.items():
+        if is_revert(commit):
+            reverts_missing_from_branch.append(commit)
+
+    if not reverts_missing_from_branch:
+        print(f"No reverts found in main branch that are missing from {branch} branch.")
+        return
+
+    print(
+        f"Found {len(reverts_missing_from_branch)} revert(s) in main branch not present in {branch} branch:"
+    )
+    print("=" * 80)
+
+    for commit in reverts_missing_from_branch:
+        # Try to identify what was reverted
+        revert_revision = get_revert_revision(commit)
+        ghf_revert_revision = get_ghf_revert_revision(commit)
+
+        reverted_commit_hash = None
+        if revert_revision:
+            print(f"Reverted Phabricator Diff: {revert_revision}")
+        elif ghf_revert_revision:
+            print(f"Reverted GitHub Commit: {ghf_revert_revision}")
+            reverted_commit_hash = ghf_revert_revision
+
+        # Check if the reverted commit was cherry-picked to the branch
+        cherry_picked_to_branch = False
+        if reverted_commit_hash:
+            if reverted_commit_hash in branch_only_reverts:
+                cherry_picked_to_branch = True
+                print(
+                    f"✅  DETECTED: The reverted commit {reverted_commit_hash} was cherry-picked to {branch}"
+                )
+
+        print(f"Commit Hash: {commit.commit_hash}")
+        print(f"Author: {commit.author}")
+        print(f"Date: {commit.commit_date or commit.author_date}")
+        print(f"Title: {commit.title}")
+        if commit.pr_url:
+            print(f"PR URL: {commit.pr_url}")
+
+        if not cherry_picked_to_branch:
+            print(
+                f"⚠️ STATUS: The reverted commit does not appear to be in {branch}, so this revert may not be needed."
+            )
+
+        print(
+            f"Body Preview: {commit.body[:200]}{'...' if len(commit.body) > 200 else ''}"
+        )
+        print("-" * 80)
+
+
 def parse_arguments():
     from argparse import ArgumentParser
 
@@ -562,6 +663,11 @@ def parse_arguments():
     parser.add_argument("--missing-in-branch", action="store_true")
     parser.add_argument("--missing-in-release", action="store_true")
     parser.add_argument("--analyze-stacks", action="store_true")
+    parser.add_argument(
+        "--analyze-missing-reverts-from-branch",
+        action="store_true",
+        help="Analyze reverts applied to main branch but not applied to specified branch",
+    )
     parser.add_argument("--date", type=lambda d: datetime.strptime(d, "%Y-%m-%d"))
     parser.add_argument("--issue-num", type=int)
     return parser.parse_args()
@@ -586,6 +692,15 @@ def main():
         analyze_stacks(repo)
         return
 
+    if args.analyze_missing_reverts_from_branch:
+        if not args.branch:
+            print(
+                "Error: --branch argument is required for --analyze-missing-reverts-from-branch"
+            )
+            return
+        analyze_reverts_missing_from_branch(repo, args.branch)
+        return
+
     # Use milestone idx or search it along milestone titles
     try:
         milestone_idx = int(args.milestone_id)