Configurable

app.py

@@ -1,37 +1,100 @@
+from dataclasses import dataclass
+
 import gradio as gr
 from graphql_calls import get_tag_commit_date, get_commits
 
 
-def get_all_commits(token, repo, tag, branch):
+@dataclass
+class Contributions:
+    additions: int
+    deletions: int
+    descriptions: list[str]
+
+
+def get_release_notes(
+    token: str,
+    repo: str,
+    tag: str,
+    branch: str,
+    contributor_treshold: int,
+    ignore_dependabot: bool,
+    ignore_direct: bool,
+):
     date = get_tag_commit_date(token, repo, tag)
     commits = get_commits(token, repo, date)
 
-    result = ''
+    result = ""
+    contributors = {}
 
     for commit in commits:
-        if '(#' in commit.message:
-            split = commit.message.split('(#')
+        if "Hugging Face" not in commit.user.organizations:
+            if commit.user.name not in contributors:
+                contributors[commit.user.name] = Contributions(
+                    additions=commit.additions,
+                    deletions=commit.deletions,
+                    descriptions=[commit.message],
+                )
+            else:
+                contributors[commit.user.name].additions += commit.additions
+                contributors[commit.user.name].deletions += commit.deletions
+                contributors[commit.user.name].descriptions += [commit.message]
+
+        if "(#" in commit.message:
+            if ignore_dependabot and commit.user.name == 'dependabot[bot]':
+                continue
+
+            split = commit.message.split("(#")
             message = split[0]
-            number = split[1].strip(')')
+            number = split[1].strip(")")
+
+            result += f"* {message} by @{commit.user.name} in #{number}\n"
+        elif not ignore_direct:
+            result += f"* {commit.message} by @{commit.user.name} (direct commit on {branch})\n"
+
+    significant_contributors = {
+        k: v for k, v in contributors.items() if (v.additions + v.deletions) > contributor_treshold
+    }
+
+    if len(significant_contributors):
+        result += (
+            "\n## Significant community contributions\n"
+            "\nThe following contributors have made significant "
+            "changes to the library over the last release:\n\n"
+        )
 
-            result += f"* {message} by @{commit.user} in #{number}\n"
-        else:
-            result += f"* {commit.message} by @{commit.user} (direct commit on {branch})\n"
+        for significant_contributor, contributions in significant_contributors.items():
+            result += f"* @{significant_contributor}\n"
+            for description in contributions.descriptions:
+                result += f"    * {description}\n"
 
     return result
 
 
 demo = gr.Interface(
-    fn=get_all_commits,
+    fn=get_release_notes,
     inputs=[
         gr.inputs.Textbox(lines=1, placeholder="Your GitHub token"),
-        gr.inputs.Textbox(lines=1, placeholder="Repository", default='huggingface/transformers'),
+        gr.inputs.Textbox(
+            lines=1, placeholder="Repository", default="huggingface/transformers"
+        ),
         gr.inputs.Textbox(lines=1, placeholder="The tag from which to get commit"),
-        gr.inputs.Textbox(lines=1, placeholder="The linear branch on which the new version tag will be added", default='main'),
+        gr.inputs.Textbox(
+            lines=1,
+            placeholder="The linear branch on which the new version tag will be added",
+            default="main",
+        ),
+        gr.inputs.Slider(
+            minimum=0,
+            maximum=2000,
+            default=500,
+            label="Threshold for significant contributors",
+        ),
+        gr.inputs.Checkbox(label="Ignore dependabot commits"),
+        gr.inputs.Checkbox(label="Ignore direct commits"),
     ],
-    outputs='text',
+    outputs="text",
 )
 
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()

graphql_calls.py

@@ -3,10 +3,18 @@ from dataclasses import dataclass
 import requests
 
 
+@dataclass
+class User:
+    name: str
+    organizations: list[str]
+
+
 @dataclass
 class Commit:
     message: str
-    user: str
+    user: User
+    additions: int
+    deletions: int
 
 
 def call_with_query(query, token):
@@ -52,9 +60,16 @@ def get_commits(token, repository, since):
                         history(first: 100, since: "{since}"{next_page}) {{
                             nodes {{
                                 message
+                                deletions
+                                additions
                                 author {{
                                     user {{
                                         login
+                                        organizations(first: 100) {{
+                                            nodes {{
+                                                name
+                                            }}
+                                        }}
                                     }}
                                 }}
                             }}
@@ -83,8 +98,18 @@ def get_commits(token, repository, since):
     commits = []
     for node in nodes:
         if node['author']['user'] is None:
-            commits.append(Commit(message=node['message'].split('\n')[0], user='<NOT FOUND>'))
+            commits.append(Commit(
+                message=node['message'].split('\n')[0],
+                user=User(name='<NOT FOUND>', organizations=[]),
+                additions=node.get('additions'),
+                deletions=node.get('deletions')
+            ))
         else:
-            commits.append(Commit(message=node['message'].split('\n')[0], user=node['author']['user']['login']))
+            commits.append(Commit(
+                message=node['message'].split('\n')[0],
+                user=User(name=node['author']['user']['login'], organizations=[n['name'] for n in node['author']['user']['organizations']['nodes']]),
+                additions=node.get('additions'),
+                deletions=node.get('deletions')
+            ))
 
     return commits