Email Obfuscation (EMO)

[Roope]

ProcessWire Email Obfuscation (EMO)

Download | GitHub
 

Email Obfuscation module for email addresses with 64 base crypting.

This module finds all plaintext emails and email links from the document and replaces them with span elements including configurable replace text. All the addresses are encoded to 64 base strings and stored in spans data attributes. Then on client side we decode these strings back to their original state.

Install

1. Create new 'EmailObfuscation' folder into /site/modules/ and place the content of this repository into the directory.
2. Login to processwire and go to Modules page and click 'Check for new modules'. Find 'EmailObfuscation' and click install.
3. You can make optional configuration changes in the module admin page.

Thanks

This is a ProcessWire module fork from MODX Evolution plugin emo E-Mail Obfuscation.
http://modx.com/extras/package/emoemailobfuscation
 

EDITED ON:
2013-03-03 - Added description.
2020-04-16 - Fixed GitHub links and updated description.

---

Hello all. Just found PW few days ago and it's already looking awesome! Here comes first contribute and some questions for developing it further.

There was one existing email obfuscator on reposity that didn't use any crypting for addresses so I decided to do a little test run and port the one that we currenly use with MODX Evo to ProcessWire module.

I'd like to make PageAutocomplete like gonfigure option to admin so that one could select and set templates to exclude module action. It looks like autocomplete is tied to pages and since templates are not set to system as pages this is option is no go, am I right?

[Pete]

Welcome!

I think ASMSelect would be a good fit instead of autocomplete in module config - I know that works.

[Roope]

OK, ASMSelect did the job perfectly. Thanks!

Another question:

In Hello world example module admin pages are restricted by: if($page->template == 'admin') return;

$template array also holds another admin pages like role and user so is there some easy way skip these in loop without naming all of them?

foreach ($this->templates as $t) {
    if ($t != 'admin') { // !in_array($t, array(admin,role,user)); is this proposed way?
        $f->addOption($t->id, $t->name);
    }
}

I just need same template names that are visible on Setup > Templates.

[teppo]

What I've been doing in another module is removing all templates flagged as system. That's probably what you want to do here too.

foreach ($this->templates as $t) {
    // skip system templates
    if (!($t->flags & Template::flagSystem)) {
       $f->addOption($t->id, $t->name);
    }
} 

[Roope]

Yep, perfect. Thanks!

[Roope]

Is there a reason why $field->collapsed = Inputfield::collapsedBlank; works as expected with InputfieldAsmSelect but does nothing with InputfieldCheckbox in module config?

[ryan]

Is there a reason why $field->collapsed = Inputfield::collapsedBlank; works as expected with InputfieldAsmSelect but does nothing with InputfieldCheckbox in module config?

I think we had a bug related to this awhile back. It should be fixed in 2.3. But for now this might work:

$field->collapsed = $field->attr('checked') ? Inputfield::collapsedYes : Inputfield::collapsedNo; 

Thanks for your work with this module. When you get a chance, please add to the modules.processwire.com directory. 

[Roope]

Thanks Ryan! This got work done perfectly.

There's still one thing I can't get figured out. With InputfieldPageAutocomplete in module config I can set parent_id and template_id to limit results. How could I configure it so that all admin pages plus the ones limited in $this->exclude_tpls config would be left out from results?

I get what I want by using InputfieldAsmSelect but I think InputfieldPageAutocomplete would fit much better here:

$f = $this->modules->get('InputfieldAsmSelect');
$f->label = 'Exclude in pages';
$f->attr('name', 'exclude_pages');
$f->attr('value', $this->exclude_pages);
$home = $this->pages->get("/");
$pages = $home->children;
$pages->prepend($home);
foreach ($pages as $p) {
    if (!in_array($p->template->id, $this->exclude_tpls)) {
        $f->addOption($p->id, $p->name);
    }
}
$inputfields->add($f);

[ryan]

The parent_id and template_id are one way to go, but you could instead use 'findPagesSelector'. You can set this to a Field object, or you can set it to an InputfieldPageAutocomplete instance, like this:

$f = $this->modules->get('InputfieldPageAutocomplete');
$f->label = 'Exclude in pages';
$f->attr('name', 'exclude_pages');
$f->attr('value', $this->exclude_pages);
 
$selector = "parent_id=1, template!=admin";
foreach($this->exclude_tpls as $name) {
  $selector .= "|$name";
}
// example: parent_id=1, template!=admin|this|that|whatever
$f->findPagesSelector = $selector; 

[Roope]

Almost one year later I'm happy to say that I finally got this autocomplete feature implemented to exclude pages option. Also forced mailto links can now be disabled from module config. Version got bumped to 1.0.2.

This also means that our love and relationship with PW is soon having it's first birthday. What a year! Still having butterflies in stomach.

[adrian]

I think we had a bug related to this awhile back. It should be fixed in 2.3. But for now this might work:

$field->collapsed = $field->attr('checked') ? Inputfield::collapsedYes : Inputfield::collapsedNo; 

Thanks for your work with this module. When you get a chance, please add to the modules.processwire.com directory. 

Hey Ryan,

I just came across this issue on the latest dev version - presumably with stable as well. I ended up going with your fix for now, although for anyone stumbling across this, the Yes and No need to be reversed in the ternary operator.

[Joss]

Just installed this module.

It seems to be adding the classes to the email and adding mailto, but in the source, the actual email address is as clear as day...

[Roope]

It seems to be adding the classes to the email and adding mailto, but in the source, the actual email address is as clear as day...

Hi Joss!

If classes are added and it turns email strings to mailto links then it sounds like it's working like it should be. My first guess is that you are looking at generated source code, e.g. using inspect element tool from the browser? Decryption is done on client side so in this case, all is fine.

Document source code from server should have predefined text string wrapped to span elements (with emo class and unique id) in place of email addresses.

[Joss]

Oh, Goody!

Thanks, Roope.

[Macrura]

@Roope -

great module! Used on a recent site; And now trying to use it on a new site, but having 2 issues:

1.) There was a point where there was accidentally an email address in the meta description of the site, and that triggered the emo, which broke the page; any way to disable it in the head section?

2.) the emo is triggering on a retina image (@2x in the image name...)

Thanks!

[Roope]

Thanks Macrura!

I just pushed fix for these, so if you can please see how it does. Thanks!

[Macrura]

@Roope - ok awesome - many thanks ! I will be back on the project soon and will report back..

[Macrura]

@roope - i downloaded the latest version, and replaced it, but it still picks up the retina images, which are marked up like this:

<a href="/" class="retina-logo" data-dark-logo="/site/templates/images/lk-logo-dark@2x.png">

i guess there is no way to prevent this sort of thing, because i see in the code that the module will exclude head, images and forms.., but this is a link with a data attribute...

but maybe there could be an option to allow the module to only work on parts of the page in a setting, for example, "container classes to apply the obfuscation to", then it would be possible for designers to limit to only specific places in the page, which would reduce the chance of it picking up images with the @ symbol..

thanks..!

[Roope]

OK, this one is bit tricky..

 

Currently emo uses regex to leave head, form and img tags untouched. Link element can't be left untouched since we're parsing emails.

 

Parsing unregular language like html with regular expressions is always on risky side. For that reason adding support for emo container class like you suggested would need proper DOM parser and IMO that introduces too much heavy lifting for the small need.

 

So, at least for now, you have the following options:

 

1. Use @2x names on a img tag

2. Disable current page from module settings

3. Disable current template from module settings

 

From these I would personally just go with number one:

 

<a href="/" class="retina-logo">
   <img src="../lk-logo-dark.png" data-dark-logo="../lk-logo-dark@2x.png" />
</a>

[Macrura]

@Roope - thanks - really appreciate the reply!

Will take a look at the options and come up with a workaround for this site...

thanks again for making this module..

[SteveB]

This worked great on my local server but when I installed it on the real server pages with email addresses fail. I get ERR_EMPTY_RESPONSE. I wasn't able to find any more clues but if I ever do I'll let you know. Both setups are running 2.5.10 dev.

[joey102030]

This module works great, but does it have to completely overwrite the classes of the containing element? Can it not just append emo_email to any existing classes?

[kixe]

While testing your module I found out that it doesn't work properly in some cases:

//no need to obfuscate, since there is no E-Mail Adress
<a href="mailto:?subject=Lorem%20Ipsum&body=Lorem%20Ipsum!">Foo</a>
<a href="mailto:?subject=Lorem Ipsum!">Foo</a>

//doesn't work with the following examples
<a href="mailto:mail@example.org?subject=Lorem Ipsum!">Foo</a>
<a href="mailto:mail@example.org?subject=Lorem%20Ipsum&body=Lorem%20Ipsum!">Foo</a>

For the 1st Example the module should leave the code untouched. Replacement only necessary if '@' is found.

For 2nd Example I can offer a solution

REPLACE

// urlencode a possible subject
$matches[1] = preg_replace_callback("!(.*\?(subject|body)=)([^\?]*)!i",
function ($m) { return rawurlencode($m[3]); }, $matches[1]);

// the function removes the questionmark, the words 'subject' and 'body' and the equalsign from the querystring 

WITH

// urlencode a possible subject and body, no matter if already urlencoded
$_matches = preg_replace_callback("!(^\?*)?((\??subject|\??body)=([^&]+)){1,2}!i",
function ($m) { return $m[1].$m[3].'='.rawurlencode(rawurldecode($m[4])); }, $matches[1]);

I tested a little bit and it worked well with my solution.
 

[Roope]

Thanks for your input kixe!

Sorry I'm little late to the party but I finally got some time to look this through on summer vacation. There was also some other issues with the module that I wanted to fix while I'm there so here it goes:

I updated the mailto link regex to match only cases where actual email address is also present so your 1st example doesn't get obfuscated no more.

While your fix for the second example worked really well, I decided to leave whole url encoding part off since it didn't seem to be vital for this module to work.

I also changed the way emo avoids @2x named images to be triggered as false positives so now markup like what @Macrura had in the example here works. Before it was only possible with image tags which were skipped from obfuscation. Old method also failed when there was image inside mailto link.

And last but not least; matched mailto link markup doesn't get replaced anymore by new element. We'll use the original captured html and just add required emo_email class to the link - like what @joey102030 just suggested here one post earlier.

I've pushed a new commit to Github that contains all these modifications. Please test it out and report any issues you may find. Thanks!

https://github.com/BlowbackDesign/emo

[bck2sqr1]

I've doubled up on javascripts and apparently it's causing an issue.  I recently started working on an interactive map on my site, the problem is that since I'm rendering the HTML on an event (click of the state loads .html(<div>...john@email.com...) ), it's loading the replacement text instead of my email addresses.  Are there any steps I can take to make it work properly without blocking the template altogether?