Adapters preferring site.content.id for ecid

[patmmccann]

Type of issue

Widespread bad practice

Description

When a bid adapter prefers site.content.id or an RTD module fills it in, that crowds out other designations of the content id. For example jw player boost module and iris may each have their own content identifier, but that isn't supported in this field as there is no room for collisions. If Prebid lets RTD modules fill in this field, it would result in a race and modules or entities affecting each others' monetization. Configuration may be getting blown away or modified in unpredictable ways.

Known issues: Index exchange ( https://prebid.slack.com/archives/C02C5BW8LGN/p1691768324940859 ) and Microsoft ( https://support.iris.tv/xandr-monetize-passing-contextual-data ) and Magnite ( https://support.iris.tv/magnitectv-passing-contextual-data ) and Pubmatic https://support.iris.tv/pubmatic-passing-contextual-data . The incorrect documentation for pubmatic there which passes ids on dctr param might have been fixed already in #9142 but remains in the wild.

Proposal, if prebid detects an rtd module or publisher filling in site.content.id with a pattern that looks like an iris identifier, it moves it to the correct place site.content.data.ext.cids. Also, we ask iris to please stop distributing documentation with content identifiers in the wrong place. Iris appears to be intentionally claiming this field to the exclusion of others in the community, as they sponsored the below proposal but continue to guide publishers to the legacy field.

https://github.com/InteractiveAdvertisingBureau/openrtb/blob/master/extensions/community_extensions/extended-content-ids.md

It seems Iris is comfortable with the correct location for Gumgum server side connections https://support.iris.tv/gumgum-ssp-passing-contextual-data but client side they recommend it as a parameter.

This mess is a great example of the most frequent publisher complaint we receive as a library, enormous variation in configuration to pass the same information to each partner. As such this param is a rules violation:

Prebid.js/modules/gumgumBidAdapter.js

Line 371 in cd787eb

if (params.irisid && typeof params.irisid === 'string') {

, as are potentially the Microsoft perids as a documented workaround.

Related: #9142

[patmmccann]

Summary of Prebid 9 changes:

Change jw player as documented

Prebid.js/modules/jwplayerRtdProvider.js

Line 87 in ef42add

// TODO: Prebid 9 - replace with ENRICH_WHEN_EMPTY

Gumgum shouldn't need to take iris id as a param, it is in the ecid object, they can support both.

[patmmccann]

@MartinGumGum this might be helpful.... If not we'll be deleting iris support in your 9.0 adapter.

function getCids(site) {
// Check if site.content and site.content.data are defined and are arrays
if (site.content && Array.isArray(site.content.data)) {
// Iterate over each data object
for (const dataItem of site.content.data) {
// Check if dataItem.name contains 'iris.com' or 'iris.tv'
if (dataItem.name.includes('iris.com') || dataItem.name.includes('iris.tv')) {
// Return the cids if the condition is met
return dataItem.ext.cids;
}
}
}
// Return null if the condition is not met or the structure is incorrect
return null;
}

// Example usage:
const site = {
content: {
data: [{
name: 'www.iris.com',
ext: {
segtax: 500,
cids: ['iris_c73g5jq96mwso4d8']
}
}]
},
page: 'http://pub.com/news',
ref: 'http://google.com',
publisher: {
id: 'p10000',
domain: 'pub.com'
}
};

const cids = getCids(site);
console.log(cids); // Output: ['iris_c73g5jq96mwso4d8']

[karimMourra]

Proposal to update the ORTB spec to have content exist in the video impression as well, to accommodate for multiple contents being rendered simultaneously on a page. site.content just doesn't scale well
InteractiveAdvertisingBureau/openrtb2.x#102

[patmmccann]

Removing 9 flag as jw and gumgum changes are now pr'd on the 9.0 branch

[MartinGumGum]

@patmmccann I'm preparing the suggestion provided above and fixing the Unit test. Please let me know how we will be affected and how critical this is.

[patmmccann]

currently i've deleted your iris functionality from the 9.0 branch, i can revert that pr on 9 if you have a fix. We're looking to release this week or early next.

[dgirardi]

@patmmccann what's the next step on this? are we waiting for someone to take action?