Using Python to parse git log logs sample _python

Using Git to manage the development of the project, git log is a very useful ' historical ' material that needs to come from here and we want to have a custom strong filter for git log. This script is doing this type of task. For a commit in all branch of a repo, the script will extract a class of commits with the bug ID in the message and provide additional Search_key for custom filtering.

Copy Code code as follows:

#-*-Coding:utf-8-*-
# Created by Vince67 feb.2014
# nuovince@gmail.com

Import re
Import OS
Import subprocess

def run (Project_dir, Date_from, date_to, Search_key, filename):
Bug_dic = {}
Bug_branch_dic = {}
Try
Os.chdir (Project_dir)
Except Exception, E:
Raise E
Branches_list = []
Branches_list = Get_branches ()
For branch in Branches_list:
Bug_branch_dic = Deal_branch (Date_from,
Date_to,
Branch
Search_key)
For item in Bug_branch_dic:
If Item not in Bug_dic:
Bug_dic[item] = Bug_branch_dic[item]
Else
Bug_dic[item] + = Bug_branch_dic[item]
Log_output (filename, bug_dic)

# Abstract log of one branch
def deal_branch (Date_from, date_to, Branch, Search_key):
Try
Os.system (' git checkout ' + branch)
Os.system (' git pull ')
Except Exception, error:
Print error
Cmd_git_log = ["Git",
"Log",
"--stat",
"--no-merges",
"-M",
"--after=" +date_from,
"--before=" +date_to]
proc = subprocess. Popen (Cmd_git_log,
Stdout=subprocess. PIPE,
Stderr=subprocess. PIPE)
stdout, stderr = Proc.communicate ()
Bug_branch_dic = Deal_lines (Date_from,
Date_to,
Search_key,
StdOut
Return Bug_branch_dic

# Write commits log to file
def log_output (filename, bug_dic):
fi = open (filename, ' W ')
For item in Bug_dic:
M1 = '--' *5 + ' BUG: ' + Item + '--' *20 + ' \ n '
Fi.write (M1)
For commit in Bug_dic[item]:
Fi.write (Commit)
Fi.close ()

# Analyze Log
def deal_lines (Date_from, date_to, Search_key, stdout):
Bug_dic = {}
For line in Stdout.split (' Commit '):
If Re.search (' Bug: \d+ ', line ') are not none and Re.search (Search_key, line) are not none:
bug_id = Line.split (' Bug: ') [1].split (' \ n ') [0]
If bug_id not in Bug_dic:
BUG_DIC[BUG_ID] = [line]
Else
BUG_DIC[BUG_ID] + = [line]
Return Bug_dic

# get all branches of a project
Def get_branches ():
Branch_list = []
Branches = []
Tmp_str = ' '
Try
Cmd_git_remote = ' git Remote show origin '
proc = subprocess. Popen (Cmd_git_remote.split (),
Stdout=subprocess. PIPE,
Stderr=subprocess. PIPE)
stdout, stderr = Proc.communicate ()
Tmp_str = Stdout.split (' Local branches configured ') [0]
Try
Tmp_str = Tmp_str.split (' Remote branches:\n ') [1]
Except
Tmp_str = Tmp_str.split (' Remote branch:\n ') [1]
Branches = Tmp_str.split (' \ n ')
For branch in Branches[0:-1]:
If Re.search (' tracked ', branch) is not None:
Branch = Branch.replace (' tracked ', '). Strip (')
Branch_list.append (Branch)
Except Exception, error:
if branch_list = = []:
Print "Can not get any branch!"
Return branch_list

if __name__ = = ' __main__ ':
# Path of the. Git project. Example: "/home/username/projects/jekyll_vincent"
Project_dir = ""
Date_from = "2014-01-25"
date_to = "2014-02-26"
# only Search ' bugs: \d+ ' for default
Search_key = ""
# Name of output file. Example: "/home/username/jekyll_0125_0226.log"
filename = ""
Run (Project_dir, Date_from, date_to, search_key, filename)