# Version Reconstruction

#### Setup Analysis

1. Checkout the `master` branch of chrome-auditor repo.
2. `cd analysis`
3. Set `uri = bolt://guangliang.gtisc.gatech.edu` in `config.cfg`
4. Install the requirements in `requirements.txt`.

#### Recovering Versions from a domain:

In the analysis stage, you can run the following command to reconstruct the versions for a domain. For example, in order to reconstruct versions of `https://www.americandancefestival.org`, the following command would be ran:

python3 analysis.py domain-profiler https://www.americandancefestival.org  label-0 --use-sld  --verbose

#### Output

Security Origin: https://www.americandancefestival.org
Iteration 0:
 Domains-Used 10
 Frames-Filtered 33 (0.75){'privy.com', 'addthis.com', 'googleapis.com', 'moatads.com', 'google-analytics.com', 'facebook.net', 'sharethis.com', 'instagram.com', 'facebook.com', 'addthisedge.com'}
Iteration 1:
 Domains-Used 1
 Frames-Filtered 11 (0.25){'gstatic.com'}
Total Frames: 44
Filtered Frames: 44
Total Frames Filtered: 1.00
Domains Used: 11/11 1.00
================================================================================
**NOTE**: Domains that show up in one day only are excluded.
**NOTE**: Domains that have daily average request count less than 0 are excluded.
================================================================================
Generating domain profile chronology...
               domain    min_date    max_date  daily_avg_req_cnt
          gstatic.com  2020-02-23  2020-03-12                  2
        instagram.com  2020-02-23  2020-03-13                  1
                 None  2020-02-22  2020-03-13                  6
            privy.com  2020-02-22  2020-03-13                  5
          addthis.com  2020-02-22  2020-03-13                  3
         facebook.net  2020-02-22  2020-03-13                  3
        sharethis.com  2020-02-22  2020-03-13                  3
 google-analytics.com  2020-02-22  2020-03-13                  2
      addthisedge.com  2020-02-22  2020-03-13                  1
          moatads.com  2020-02-22  2020-03-13                  1
         facebook.com  2020-02-22  2020-03-13                  1
       googleapis.com  2020-02-22  2020-03-13                  1
================================================================================
Generating domain profile versions...
 Version	Domain Set
       0	[{'privy.com', 'facebook.net', 'addthis.com', 'googleapis.com', 'moatads.com', 'google-analytics.com', None, 'sharethis.com', 'facebook.com', 'addthisedge.com'}, neotime.Date(2020, 2, 22)]
       1	[{'gstatic.com', 'instagram.com'}, neotime.Date(2020, 2, 23)]```

In this scenario, the relevant information for completing the version reconstruction evaluation is the following:

Generating domain profile versions...
 Version	Domain Set
       0	[{'privy.com', 'facebook.net', 'addthis.com', 'googleapis.com', 'moatads.com', 'google-analytics.com', None, 'sharethis.com', 'facebook.com', 'addthisedge.com'}, neotime.Date(2020, 2, 22)]
       1	[{'gstatic.com', 'instagram.com'}, neotime.Date(2020, 2, 23)]

In the above example, the number of versions would be two.

Tasks
====

1. For each securityOrigin in `security_origin.csv` run the `domain-profiler` with the parameters shown in the above example.
2. Save the __entire__ output to a file. The name of the file can be the name of the security origin.
4. In [1] there is a mapping from URL to category. The purpose of the evaluation is to determine the average version count for each category.

[1] `https://github.com/jallen89/chrome-auditor/blob/master/evaluation/weblinks-collector/weblinkCategoryMapping.txt`