Tabs Outliner Deduplication

dedupe.py

import json
import os

export_file = 'tree-exported-Sat-Apr-09-2016.tree'
backup_export_file = '{}.bak'.format(export_file)
new_tree = []
seen_urls = set()

os.rename(export_file, backup_export_file)

with open(backup_export_file) as f:
    tree = json.load(f)    
    for node in reversed(tree):
        if isinstance(node, list):
            item = node[1]
            if ('type' not in item) or (item['type'] == 'tab'):
                url = item['data']['url']
                if url in seen_urls:
                    print 'skipping duplicate: {}'.format(url)
                    continue
                seen_urls.add(url)
        new_tree.append(node)
new_tree = list(reversed(new_tree))

window_count = -1
tab_count = -1
for node in new_tree:
    if isinstance(node, list):
        item = node[1]
        pos = []

        if item.get('type') in ('win', 'savedwin'):
            window_count += 1
            tab_count = -1
            pos = [window_count]
        else:
            tab_count += 1
            pos = [window_count, tab_count]

        node[2] = pos

with open(export_file, 'w') as f:
    json.dump(new_tree, f)