Commit 066d6dbe authored by Daniel Brown's avatar Daniel Brown
Browse files

adding redmine tools

parent 15650200
This file contains various functions for processing the Redmine download data for Finesse.
import fileinput
from dateutil.parser import parse
import json
import urllib
def processRedmineProductionLog(log, output):
This function parses the Redmine production log for all Finesse
downloads. The output is a text file that contains the file
downloaded, the IP address and the date.
Production log file is typically big (e.g. 10GB), this takes awhile
to run.
log: path and filename to log file
output: path and filename of output text file to create
found = -1
with open(output, "w") as ofile:
for line in fileinput.input(log_filename):
if found > 0 and found+1 == fileinput.lineno():
file = line.lstrip().split()[3]
file = file.split("=>")[1][1:-2]
found = -2
print("Couldn't parse download, probably a traceback")
found = -1
elif line.startswith("Processing AttachmentsController#download"):
found = fileinput.lineno()
ip = line.split()[3]
date = line.split()[5]
if found == -2:
ext = file.split(".")[-1]
if ext == "zip" or ext == "tar" or ext == "pdf":
ofile.write("%s %s %s\n" %(file, ip, date))
print file, ip, date
found = -1
def readFinesseDownloadData(data_filename, unique=True, geo_filename=None):
Reads the data files generated by processRedmineProductionLog
and getFinesseDownloadIPGeoData.
File download data is returned in a dictionary with filenames as the
key. This then contains another dictionary with the date of downloads
and number of downloads that day, and a list of IP addresses that
downloaded it.
If Geo data filename is provided another dictionary will be returned.
The keys of which are the IP addresses and the latitude and longitude
data_filename: filename and path of processRedmineProductionLog output file
unique: If true only unique IP addresses for each file download are read
geo_filename: filename and path of getFinesseDownloadIPGeoData output file
files = {}
IPGeo = {}
with open(data_filename, "r") as data:
for line in data:
split = line.split()
file = split[0].lower()
ip = split[1]
date = parse(split[2])
if split[0] not in files:
files[split[0]] = ([], {})
if unique and ip in files[split[0]][0]:
if date not in files[split[0]][1]:
files[split[0]][1][date] = 0
files[split[0]][1][date] += 1
if geo_filename != None:
with open(geo_filename, "r") as data:
for line in data:
split = line.split()
IPGeo[split[0]] = (split[1], split[2])
return files, IPGeo
return files
def getFinesseDownloadIPGeoData(input_file, output_file):
Using the information outputted to the file by the function processRedmineProductionLog
this will function will call various free IP to Latitude/Longitude web services.
The file outputted will contain a list of unique IP addresses and it's coordinates.
Note: This takes a long time to run! Free webservice may kick you off too if too many
requests are made per hour.
input_file: filename and path to processRedmineProductionLog output file
output_file: Output filename and path
IPGeo = {}
with open(input_file, "r") as data:
for line in data:
split = line.split()
file = split[0].lower()
ip = split[1]
date = parse(split[2])
if ip not in IPGeo:
IPGeo[ip] = ([], [])
response = urllib.urlopen('' % ip).read()
lat = response.split('\n')[3].split()
lon = response.split('\n')[4].split()
if len(lat) == 2 and len(lon) == 2:
response = json.loads(urllib.urlopen('' % ip).read())
if 'latitude' in response:
print response
with open(output_file, "w") as w:
for ip in IPGeo.keys():
w.write("%s %g %g\n" % (ip, IPGeo[ip][0], IPGeo[ip][1]))
\ No newline at end of file
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment