Update release notes retrieve PRs script (#2257)

Summary:
as discussed offline w/ nateanl, cherry-picked PRs are currently being included when retrieving PRs between a release branch and newer commits. this PR fixes this by removing duplicates in the commit paths

Pull Request resolved: https://github.com/pytorch/audio/pull/2257

Reviewed By: nateanl

Differential Revision: D34459533

Pulled By: carolineechen

fbshipit-source-id: 3497c1d2dca6f8067e2068146a6e28cce591d3c8

tools/release_notes/retrieve_prs.py

@@ -32,14 +32,14 @@ def commit_title(commit_hash):
     return _run_cmd(cmd)
 
 
-def parse_pr_number(commit_hash, title):
+def parse_pr_number(title):
     regex = r"(#[0-9]+)"
     matches = re.findall(regex, title)
     if len(matches) == 0:
-        print(f"[{commit_hash}: {title}] Could not parse PR number, ignoring PR")
+        print(f"[{title}] Could not parse PR number, ignoring PR")
         return None
     if len(matches) > 1:
-        print(f"[{commit_hash}: {title}] Got two PR numbers, using the last one")
+        print(f"[{title}] Got two PR numbers, using the last one")
         return matches[-1][1:]
     return matches[0][1:]
 
@@ -91,24 +91,33 @@ def gh_labels(pr_number):
 
 def get_features(commit_hash):
     title = commit_title(commit_hash)
-    pr_number = parse_pr_number(commit_hash, title)
+    pr_number = parse_pr_number(title)
     labels = []
     if pr_number is not None:
         labels = gh_labels(pr_number)
     return Features(title, pr_number, labels)
 
 
-def get_commits_between(base_version, new_version):
+def get_merge_base(base_version, new_version):
     cmd = ["git", "merge-base", f"{base_version}", f"{new_version}"]
     merge_base = _run_cmd(cmd)
+    return merge_base
+
+
+def get_commits_between(base_version, new_version):
+    merge_base = get_merge_base(base_version, new_version)
 
     # Returns a list of items in the form
     # a7854f33 Add HuBERT model architectures (#1769)
+    cmd = ["git", "log", "--reverse", "--oneline", f"{merge_base}..{base_version}"]
+    base_commits = _run_cmd(cmd).split("\n")
+    base_prs = [parse_pr_number(commit) for commit in base_commits]
+
     cmd = ["git", "log", "--reverse", "--oneline", f"{merge_base}..{new_version}"]
-    commits = _run_cmd(cmd)
+    new_commits = _run_cmd(cmd).split("\n")
 
-    log_lines = commits.split("\n")
-    hashes, titles = zip(*[log_line.split(" ", 1) for log_line in log_lines])
+    commits = [commit for commit in new_commits if parse_pr_number(commit) not in base_prs]
+    hashes, titles = zip(*[commit.split(" ", 1) for commit in commits])
     return hashes, titles