Skip to content

Latest commit

 

History

History
118 lines (88 loc) · 4.8 KB

README.md

File metadata and controls

118 lines (88 loc) · 4.8 KB
Raw

browsertrix-yaml-examples

YAML files for including and excluding things

See the browsertrix project for more general documentation.

Full sample configurations that were helpful for crawling certain websites are available in the subdirectiories, such as biph-kiev-ua/crawl-config.yaml.

Default and handy starter file

This includes all sub domains like abc.collection.litme.com.ua, handles http:// to https:// conversions & subdomains.

collection: "collection-litme-com-ua"
workers: 16
saveState: always

# A few examples that exclude parts of a page from being fetched and recorded.
blockRules:

  # Unnecessary trackers
  - url: google-analytics.com
  - url: googletagmanager.com
  - url: googlesyndication.com
  - url: yandex.ru
  - url: liveinternet.ru
  - url: hotlog.ru
  - url: openstat.net
  - url: mycounter.ua
  - url: facebook.(com|net)

  # Malware
  - url: www.acint.net       # spam-seo.sape malware
  - url: news.2xclick.ru     # mwblacklisted35 malware
  - url: culturaltracking.ru # culturaltracking malware
  - url: sape.ru             # sape backlinks SEO malware

  # Non-threatened resources that are bandwidth-intensive and perhaps not
  # a current priority. Uncomment these rules to bypass recording.
  #- url: youtube.com/embed/ # Embedded YouTube videos
  #- url: w.soundcloud.com   # Embedded SoundCloud tracks

seeds:
  - url: http://collection.litme.com.ua/
    scopeType: "domain"

Excluding trouble

If you notice that a crawl is collecting duplicate links due to parameters, like

        - '{"url":"https://nmiu.org/index.php/novyny-museum?iccaldate=2022-4-1","seedId":0,"depth":2}'
        - '{"url":"https://nmiu.org/index.php/novyny-museum?iccaldate=2023-03-1","seedId":0,"depth":2}'
        - '{"url":"https://nmiu.org/index.php/novyny-museum?iccaldate=2022-2-1","seedId":0,"depth":2}'

where each is the same base page when loaded in the browser, then first check that excluding it results in the same webpage. in this case, as https://nmiu.org/index.php/novyny-museum goes to the same webpage, we can exclude all cases where iccaldate= is used as a query parameter / link to follow by adding the following to our yaml file:

exclude:
  - .*iccaldate=.*

This will prevent the same page taking up multiple workers / space in the final file / time. So the whole file will read

collection: "collection-nmiu-org"
workers: 16
saveState: always
seeds:
    - url: https://nmiu.org
      include: .*nmiu\.org.*
      scopeType: "host"
      exclude:
        - .*iccaldate=.*

Making Sure You Don't Go Out Of Depth

Some websites have recursive links. They can look like this

'{"url":"https://nmiu.org/lektoriy/181-ekskursiji-kontent/vysta/3/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/index.php/exponat-tyzhnya","seedId":0,"depth":48}'

If you notice this kind of pattern in the URLs of a site that never seems to end, avoid getting stuck in such a trap by adding the "depth" flag to the yaml file:

collection: "collection-litme-com-ua"
workers: 16
saveState: always
seeds:
    - url: http://collection.litme.com.ua/
      include: 
        - .*collection\.litme\.com.ua.*
      scopeType: "host"
      depth: 25

This will tell the tool to only follow a given path to a depth of 25 different URL clicks, so use it with care!

In the example given at the start of this section, exclude can also be used with

exclude: 
  - .*index.php/index.php.*

as this was what was causing the recursion

Handling an Enormous Site like WikiMedia

This is an example crawl of a WikiMedia site, reducing an enormous, unfinishable crawl to a manageable one, by ignoring all wiki paraphernalia (history pages, etc.).

collection: "wiki-library-kr-ua"
workers: 16
saveState: always
seeds:
- url: https://wiki.library.kr.ua/
include: .*\.wiki\.library\.kr\.ua/
exclude: 
  - .*action\=.*
  - .*page\=.*
  - .*limit\=.*
  - .*oldid\=.*
  - .*title=%D0%A1%D0%BF%D0%B5%D1%86%D1%96%D0%B0%D0%BB%D1%8C%D0%BD%D0%B0\:.*
  - .*returnto\=.*
scopeType: "host"

Line 12 tells it to ignore links with titles beginning with the Ukrainian word for "special" followed by a colon: that might need to be reconfigured for each site.