batch_redirects

#!/usr/bin/env python
"""
Update all_libraries.json with tracking for original URLs
"""

import json

redirects = []
try:
    jsonf = open('all_libraries.json', 'r')
    with jsonf:
        libj = json.loads(jsonf.read())
except(Exception) as inst:
    print(inst)

try:
    logf = open('crawl.log', 'r')
    with logf:
        for line in logf:
           if 'INFO:Redirect: ' in line:
                (oldu, newu) = line[line.find(': ')+1:].strip().split('\t')
                if '?utm_source' in newu:
                    newu = newu[0:newu.find('?utm_source')]
                print("Old: %s, New: %s" % (oldu, newu))
                redirects.append((oldu, newu))
except(Exception) as inst:
    print(inst)

newlibj = []
for lib in libj:
    for oldu, newu in redirects:
        if lib['url'] == oldu:
            lib['url'] = newu
            lib['urlPrevious'] = (oldu)
    if lib['url']:
        lib['url'] = lib['url'].strip().strip('\u200e').strip('\u200f')
    newlibj.append(lib)

try:
    newjsonf = open('new_libraries.json', 'w')
    with newjsonf:
        json.dump(newlibj, newjsonf, indent=4, sort_keys=True)
except(Exception) as inst:
    print(inst)