gitlab code searcher based on https://github.com/tuimm/gitlab-search, but added enum for all branches and regex string matches. need to add multithreading and dumping data to disk upon receipt.

gitlab-search.py

import gitlab
import re

def search(gitlab_server, token, file_filter, regextofind, group=None, project_filter=None):
  return_value = []
  gl = gitlab.Gitlab(gitlab_server, private_token=token)
  if (project_filter == '') and (group == ''):
    projects = gl.projects.list(all=True)
  else:
    group_object = gl.groups.get(group)
    group_projects = group_object.projects.list(search=project_filter)
    projects = []
    for group_project in group_projects:
      projects.append(gl.projects.get(group_project.id))
  for project in projects:
    print("enuming project: ", project.name)
    files = []
    try:
      files = project.repository_tree(recursive=True, all=True)
    except Exception as e:
      print(str(e), "Error getting tree in project:", project.name)
    for file in files:
      print("enuming file: ", file.get("name"))
      matches=re.findall(file_filter, file['name'])
      filename_matches = len(matches)>0
      if filename_matches:
        for branch in project.branches.list(get_all=True):
          try: 
            file_content = project.files.raw(file_path=file['path'], ref=branch.name)
            filecontentmatches=re.findall(regextofind, str(file_content))
            filecontent_matches = len(filecontentmatches)>0
            if filecontent_matches:
              return_value.append({
                "project": project.name,
                "branch": branch.name,
                "file": file['path']
              })
          except Exception as e:
            #if a file doesn't exist in the targeted branch, then an error exists...
            # print(str(e), "Error getting file contents for", file['path'], " in project:", project.name, " in branch: ",  branch.name)
            continue
  return return_value


#testing with project specified, to check multiple files found, and check regex
search('https://gitlabinstance', 'access_token', '.*', '(?i)NaMe','infra','vpc')


#testing without project specified
search('https://gitlabinstance', 'access_token', '.*', '(?i)NaMe','infra','')


#searching for joinpath for CVE-2022-32190
thelist = search('gitlabinstance', 'access_token', '.*', '(?i)joinpath','','')
thejsonlist = json.dumps(thelist)
thejsonfile = open("thelist.json","w")
thejsonfile.write(thejsonlist)
thejsonfile.close()