backing up pr data work

pr_data_get.py

@@ -1,15 +1,28 @@
 from perceval.backends.core.git import Git
 import os
 import datetime as dt
+import time
 import shutil 
+import pandas as pd
+import dateutil
+from tqdm import tqdm
 
 key = os.environ.get('KKEXKEY')
 
 early_cutoff = dt.datetime(2008,2, 8)
 temp_dir = "/data/users/mgaughan/tmp"
 
+'''
+- what we really want is the 6 weeks before and after these events 
+- lets start by getting 6 weeks (42 days) before and after the first appearance of a given document
+- rate of change, rate of PRs/day
+'''
 def file_get_pr(upstream_vcs_link):
+    #this is the window of days on either side of the event that we're looking at
+    window = 42
     #print(upstream_vcs_link.split('/')[4])
+    project_dict = {}
+    project_dict['upstream_vcs_link'] = upstream_vcs_link
     full_temp_path = temp_dir + upstream_vcs_link.split('/')[4] + ".git"
     repo = Git(uri=upstream_vcs_link, gitpath=full_temp_path)
     try:
@@ -20,27 +33,59 @@ def file_get_pr(upstream_vcs_link):
     has_readme = False
     has_contributing = False
     merge_pre_rm, merge_post_rm, merge_pre_cont, merge_post_cont = 0, 0, 0, 0
+    #list of tuples which has date and whether it was a merge
+    commit_list = []
     for commit in commits:
         if "Merge" in commit['data'].keys():
+            commit_list.append([commit['data']['CommitDate'], True])
             if has_contributing:
                 merge_post_cont += 1
             else:
                 merge_pre_cont += 1
-            print('merge')
+        else: 
-        else:
+            commit_list.append([commit['data']['CommitDate'], False])
-            print('not')
         files = commit['data']['files']
         #print(commit['data']['CommitDate'])
+        #print(type(dateutil.parser.parse(commit['data']['CommitDate'])))
         for file in files:
-            if "CONTRIBUTING.md" == file['file']:
+            if "CONTRIBUTING.md" == file['file'] and has_contributing == False:
                 has_contributing = True
-            if "README.md" == file['file']:
+                first_date_contributing = dateutil.parser.parse(commit['data']['CommitDate'])
+            if "README.md" == file['file'] and has_readme == False:
                 has_readme = True
+                first_date_readme = dateutil.parser.parse(commit['data']['CommitDate'])
     shutil.rmtree(full_temp_path, ignore_errors=True)
-    print("merge pre cont: " + str(merge_pre_cont))
+    project_dict['first_contributing'] = first_date_contributing
-    print('merge post cont: ' + str(merge_post_cont))
+    project_dict['first_readme'] = first_date_readme
+    before_cont = pr_count(first_date_contributing + dt.timedelta(days=-window, hours=0), first_date_contributing, commit_list)
+    project_dict['b6w_prs_cont'] = before_cont[0]
+    project_dict['b6w_mrg_cont'] = before_cont[1]
+    after_cont = pr_count(first_date_contributing, first_date_contributing + dt.timedelta(days=window, hours=0), commit_list)
+    project_dict['a6w_prs_cont'] = after_cont[0]
+    project_dict['a6w_mrg_cont'] = after_cont[1]
+    before_read = pr_count(first_date_readme+ dt.timedelta(days=-window, hours=0), first_date_readme, commit_list)
+    project_dict['b6w_prs_read'] = before_read[0]
+    project_dict['b6w_mrg_read'] = before_read[1]
+    after_read = pr_count(first_date_readme, first_date_readme + dt.timedelta(days=window, hours=0), commit_list)
+    project_dict['a6w_prs_read'] = after_read[0]
+    project_dict['a6w_mrg_read'] = after_read[1]
+    print(project_dict)
+    return project_dict
+
+def pr_count(start, end, commits):
+    count = 0
+    merge_count = 0
+    for commit in tqdm(commits):
+        if dateutil.parser.parse(commit[0]) > start:
+            count += 1
+            if commit[1]:
+                merge_count += 1
+        if dateutil.parser.parse(commit[0]) > end:
+            return [count, merge_count]
+
 
 
 if __name__ == "__main__":
     file_get_pr("https://github.com/tqdm/tqdm")
+    file_get_pr("https://github.com/GameServerManagers/LinuxGSM")