Remove Google Analytics from Maven Website

[niallkp]

Bug description

Hi Maven Team!

Its great that you've migrated from Google Analytics (GA) to Matomo - thanks for doing that. However there seems to be a large number of pages (52,000+)in old docs that still have GA. The table below shows which components/versions are affected & the number of files containing GA.

Location	File Count
doxia components	6,271
maven components	46,247
TOTAL	52,518

I'm not sure what the best way is to resolve this and I guess there might be different solutions for different components. I'm prepared to help with the work, but perhaps someone could do a triage what the best course of action is for each component?

• Perhaps some of the old versions of the docs for components could be removed?
• Is there an straightforward way to re-generate some of these component versions?
• if these are never updated in svn, then perhaps we could patch some versions to remove GA

Regards

Niall

Location	Component	File Count
doxia	doxia-archives/	5181 => done
doxia	doxia-sitetools-archives/	1015 => done
doxia	doxia-tools/doxia-integration-tools	25 => done
doxia	doxia-tools-archives/	50 => done
maven	ant-tasks-archives/	85 => done
maven	archetype-archives/	1639 => done
maven	archetypes-archives/	436 => done
maven	core-its	1719 => done
maven	enforcer-archives/	264 => done
maven	jxr-archives/	393 => done
maven	maven-indexer/indexer-examples	20 => done
maven	maven-indexer-archives/	511 => done
maven	maven-release-archives/	219 => done
maven	plugins/	127 => done
maven	plugins-archives/	11781 => done
maven	plugin-testing-archives/	386 => done
maven	plugin-tools-archives/	2259 => done
maven	pom-archives/	1239 => done
maven	ref/	1019 => done
maven	resolver-archives/	2237 => done
maven	sandbox/plugins	27 => done
maven	scm-archives/	3156 => done
maven	shared-archives/	2518 => done
maven	skins-archives/	2981 => done
maven	studies/extension-demo	21 => done
maven	surefire-archives/	6127 => done
maven	wagon/	352 => done
maven	wagon-archives/	6758 => done

Maven site URL where bug exists

x

[hboutemy]

Hi @niallkp
good list, thanks
regenerating is not an option: patching is (I did it in the past, just need to remember the search/replace magic formula)
and while at it, we should probably do some cleanup, like keeping only latest patch version for each minor => I'll get consensus on that approach quickly I suppose

[niallkp]

Hi @hboutemy , thanks for the response!

I've added a column to the table above indicating a later minor version (theres 27 of them) - so if you could get agreement to delete those, that would be a start.

There are also six maven plugins where the LATEST version has Google Analytics and I was wondering whether they need patches to ensure the next release doesn't contain GA or whether thats already been fixed?

• maven-antrun-plugin-LATEST
• maven-docck-plugin-LATEST
• maven-ejb-plugin-LATEST
• maven-pdf-plugin-LATEST
• maven-scripting-plugin-LATEST
• maven-wrapper-plugin-LATEST

I've become quite adept at doing search/replace to remove GA, so happy to help generate patches if you want - let me know.

[hboutemy]

@niallkp most of the cleanup done
the only probably missing part is when there are some subdirectories

can you check and report, please, what issues remain? It should be much much less now

[niallkp]

Sorry @hboutemy I messed up. When I checked the maven website out of SVN, I didn't notice it had timed out - so I had only a part of the website locally. So instead of there being 6,000+ files with GA, there are actually 52,000+ files.

Thanks for doing the cleanup of minor versions, it has removed some GA. I have updated the list in the description to reflect the current situation.

There are a few maven components with later minor versions in the current list which could also be removed, but it still leaves a daunting list of docs to fix

• ref/3.8.6
• ref/3.8.7
• ref/4.0.0-alpha-2
• ref/4.0.0-alpha-3
• resolver-archives/resolver-ant-tasks-1.2.0
• resolver-archives/resolver-ant-tasks-1.3.0

[kwin]

Isn't it enough to rely on the ASF CSP to disable Google Analytics for the archived sites?

[slawekjaranowski]

@hboutemy can you share somewhere your commands or script used to cleanups?

[hboutemy]

VS Code search replace...

[hboutemy]

for sparse checkout, ignoring big content like javadoc and jxr, I did a small svn-sparse-co.sh script:

#!/bin/bash

function coDir() {
    local url=$1
    echo "$url"
    svn co --depth immediates $url

    cd `basename $url`

    for d in `find * -depth -maxdepth 0 -type d`
    do
      case $d in
        xref | xref-test | apidocs | testapidocs | css | img | images | fonts | js | cobertura)
        echo "ignore $d in `pwd`"
         ;;
        *)
          coDir $url/$d
          ;;
      esac
    done
    cd ..
}

coDir $1

find `basename $1` -type f -name '*.html' -exec grep google-analytics {} \; | wc -l

then in the directory, I search replace with VS Code and svn commit the result

like what I just did:

./svn-sparse-co.sh https://svn.apache.org/repos/asf/maven/website/components/wagon
(VS Code: pick one .html, search for google-analytics, select one GA snippet, replace with snippet taken from view-source:https://maven.apache.org/ )
svn ci -m "replace Google Analytics with Matomo" wagon

human judgement is useful for searching for google-analytics and iteratively extracting all misc corresponding snippets

[hboutemy]

@niallkp I think it is all done: can you check please and confirm?

[niallkp]

Wow @hboutemy, thats brilliant - thanks for doing that!

I re-scanned the Doxia website which was all clear. For the Maven website I only found 16 pages still with Google Analytics - in the ref/4.0.0-alpha-3/maven-bom component. I'm attaching a patch to fix that,

relative to the following path:

• https://svn.apache.org/repos/asf/maven/website/components/

The patch also changes the docs in the maven-site-plugin to use a different js example other than Google Analytics - but that isn't an issue, except it will prevent a false positive on the report to the Privacy Committee - so feel free to not apply that part.

maven-components-ga.patch