Merge branch 'release/v0.1.0' into main

Author: dagpunk

.gitignore

@@ -127,3 +127,7 @@ dmypy.json
 
 # Pyre type checker
 .pyre/
+
+# git-analyzer project settings
+config.json
+data/

README.md

@@ -1,2 +1,34 @@
 # git-analyzer
-Explore the GitHub API
+Analyze the repositories and branches of a GitHub user
+
+## Configure
+### config.json
+```
+{
+    "username": "YOUR_USERNAME_HERE",
+    "github_access_token": "YOUR_GITHUB_ACCESS_TOKEN_HERE",
+    "branch_search_pattern": "BRANCH_NAME_REGEX_PATTERN_HERE",
+    "max_repos_to_download": 300
+}
+```
+### username
+Your GitHub username
+### github_access_token
+The access token from GitHub with access to repositories
+### branch_search_pattern
+The search pattern for repositories containing a branch name matching the 
+pattern
+### max_repos_to_download
+The maximum number of repositories to fetch information for (this can prevent 
+exceeding the rate limit too quickly)
+
+## Usage
+### Download repository and branch data
+```
+python main.py
+```
+
+### Search for branches matching a regex pattern
+```
+python find_branches_by_regex.py
+```

config-sample.json

@@ -0,0 +1,6 @@
+{
+    "username": "YOUR_USERNAME_HERE",
+    "github_access_token": "YOUR_GITHUB_ACCESS_TOKEN_HERE",
+    "branch_search_pattern": "BRANCH_NAME_REGEX_PATTERN_HERE",
+    "max_repos_to_download": 300
+}

find_branches_by_regex.py

@@ -0,0 +1,23 @@
+from get_config_value import get_config_value
+import json
+import re
+
+def main():
+    with open('data/user_repo_branches.json', 'r') as input_file:
+        json_data = json.load(input_file)
+
+    print('Data source access time:', json_data['data_source_access_time'])
+
+    for repo_key in json_data['data']['repos']:
+        repo = json_data['data']['repos'][repo_key]
+        
+        for branch_key in repo['branches']:
+            branch = repo['branches'][branch_key]
+            match = re.search(get_config_value('branch_search_pattern'), 
+                branch['name'])
+            
+            if match:
+                print('Matching branch found in ' + repo['name'] + ': ' + \
+                    branch['name'])
+
+main()

get_config_value.py

@@ -0,0 +1,7 @@
+import json
+
+def get_config_value(key):
+    with open('config.json', 'r') as config_file:
+        config_json = json.load(config_file)
+
+    return config_json[key]

main.py

@@ -0,0 +1,81 @@
+from get_config_value import get_config_value
+from github import Github
+import json
+from ratelimiter import RateLimiter
+from time import gmtime, strftime
+
+def get_user_repo_branches():
+    extract_json = {
+        'data_source_access_time': strftime("%Y%m%dT%H%M%SZ", gmtime()),
+        'data_source_name': 'GitHub API'
+    }
+    g = Github(get_config_value('github_access_token'))
+    user = g.get_user()
+    user_json = {
+        'login': user.login,
+        'public_repos': user.public_repos,
+        'repos': {}
+    }
+    repos = user.get_repos()
+    rate_limiter = RateLimiter(max_calls=5000, period=3600)
+
+    for i, repo in enumerate(repos):
+        with rate_limiter:
+            if i >= get_config_value('max_repos_to_download'):
+                pass
+            else:
+                repo_name = repo.name
+                repo_json = {
+                    'name': repo_name,
+                    'private': repo.private,
+                    'forks_count': repo.forks_count,
+                    'stargazers_count': repo.stargazers_count,
+                    'watchers_count': repo.watchers_count,
+                    'size': repo.size,
+                    'open_issues_count': repo.open_issues_count,
+                    'has_issues': repo.has_issues,
+                    'has_projects': repo.has_projects,
+                    'has_wiki': repo.has_wiki,
+                    'has_downloads': repo.has_downloads,
+                    'pushed_at': repo.pushed_at,
+                    'created_at': repo.created_at,
+                    'updated_at': repo.updated_at,
+                    'branches': {}
+                }
+                branches = repo.get_branches()
+
+                for branch in branches:
+                    branch_name = branch.name
+                    branch_commit = branch.commit
+                    branch_json = {
+                        'name': branch_name,
+                        'commit': {
+                            'sha': branch_commit.sha,
+                            'url': branch_commit.url
+                        },
+                        'protected': branch.protected,
+                    }
+                    repo_json['branches'][branch_name] = branch_json
+
+            user_json['repos'][repo_name] = repo_json
+
+    extract_json['data'] = user_json
+
+    with open('data/user_repo_branches.json', 'w') as output_file:
+        json.dump(extract_json, output_file, indent=4, default=str)
+
+def display_rate_limit():
+    g = Github(get_config_value('github_access_token'))
+    print('Rate limit stats:\nCore limit: ', g.rate_limiting[0])
+
+def prompt_user_to_continue():
+    user_input = input('Do you wish to proceed? (y/n): ')
+    return user_input.upper() == 'Y'
+
+def main():
+    display_rate_limit()
+
+    if prompt_user_to_continue():
+        get_user_repo_branches()
+
+main()