Weird css in some rows of ACA search results

[sergiobaiao]

Describe the bug
I'm having some weird ACA search results within my environment, like the picture bellow:

Some rows in the results are showing "n class='aca-highlight'>" within the title, description and/or content. Aparently this only happens to my own Alfresco data, since @aborroy was able to replicate this with his own local data, but did replicate with my data from my Alfresco server.

Environment

• ACA version: 5.2.0
• Alfresco 23.2 from alfresco-docker-installer with this .env

ALFRESCO_CE_TAG=23.2.1
SEARCH_CE_TAG=2.0.9.1
SHARE_TAG=23.2.1
ACA_TAG=4.4.1
POSTGRES_TAG=15.6
#POSTGRES_TAG=14.4
MARIADB_TAG=11.3.2
TRANSFORM_ENGINE_TAG=5.1.0
ACTIVEMQ_TAG=5.18-jre17-rockylinux8

To Reproduce
Steps to reproduce the behavior:

1. Go to ACA
2. Click on search icon and perform a search
3. Scroll down to search results
4. See error

I don't think this will happen to everyone, maybe it's related to wrong character encoding from metadata on the search results, that maybe wrongly escaping quotes or any other chars. Also, the last line from each row is showing a lot of weird chars also.

Screenshots
ACA Search Results:

Share Search Results: (same user and same search term)

P.S - I can provide a Consumer only user with restricted access to anyone who wants to test this against my Alfresco server.

[aborroy]

@DenysVuika my guessing is that there is some error in this method:

https://github.com/Alfresco/alfresco-content-app/blob/5.2.0/projects/aca-content/src/lib/components/search/search-results-row/search-results-row.component.ts#L157C1-L161C4

Some inputs, like a nested "aca-highlight" input would create this issue.

A better approach could be to use HTML elements for this replacement. Something similar to:

private stripHighlighting(highlightedContent: string): string {
  if (!highlightedContent) return '';
  
  const parser = new DOMParser();
  const doc = parser.parseFromString(highlightedContent, 'text/html');
  const highlights = doc.querySelectorAll('.aca-highlight');
  
  highlights.forEach(highlight => {
    const parent = highlight.parentNode;
    if (parent) {
      while (highlight.firstChild) {
        parent.insertBefore(highlight.firstChild, highlight);
      }
      parent.removeChild(highlight);
    }
  });
  
  return doc.body.innerHTML;
}

[sergiobaiao]

while researching today for a probable cause, i've done this to log what i was receiving from the search results:
const highlights: SearchEntryHighlight[] = this.node.entry['search']?.['highlight'];
let name = this.node.entry.name;
console.log("name", name);
const properties = this.node.entry.properties;
console.log("properties", properties);
let title = properties?.['cm:title'] || '';
console.log("title", title);
let description = properties?.['cm:description'] || '';
console.log("description", description);
let content = '';
console.log("content", content);

and this is the search result:

If you look what i've circled in red in the right side, the description field has a lot of back/forward slashes, and this entire piece is being stripped from the search result on the left site. I'm not sure if the slashes are being correctly escaped and this may be one of the causes.

[sergiobaiao]

@DenysVuika my guessing is that there is some error in this method:

https://github.com/Alfresco/alfresco-content-app/blob/5.2.0/projects/aca-content/src/lib/components/search/search-results-row/search-results-row.component.ts#L157C1-L161C4

Some inputs, like a nested "aca-highlight" input would create this issue.

A better approach could be to use HTML elements for this replacement. Something similar to:

private stripHighlighting(highlightedContent: string): string {
  if (!highlightedContent) return '';
  
  const parser = new DOMParser();
  const doc = parser.parseFromString(highlightedContent, 'text/html');
  const highlights = doc.querySelectorAll('.aca-highlight');
  
  highlights.forEach(highlight => {
    const parent = highlight.parentNode;
    if (parent) {
      while (highlight.firstChild) {
        parent.insertBefore(highlight.firstChild, highlight);
      }
      parent.removeChild(highlight);
    }
  });
  
  return doc.body.innerHTML;
}

I tried this and it still shows the same error.

[aborroy]

In that case, the fix should be something like:

private stripHighlighting(highlightedContent: string): string {
  if (!highlightedContent) return '';
  
  // Escape special regex characters in prefix and postfix
  const escapedPrefix = this.highlightPrefix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
  const escapedPostfix = this.highlightPostfix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
  
  return highlightedContent
    .replace(new RegExp(escapedPrefix, 'g'), '')
    .replace(new RegExp(escapedPostfix, 'g'), '');
}

[sergiobaiao]

In that case, the fix should be something like:

private stripHighlighting(highlightedContent: string): string {
  if (!highlightedContent) return '';
  
  // Escape special regex characters in prefix and postfix
  const escapedPrefix = this.highlightPrefix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
  const escapedPostfix = this.highlightPostfix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
  
  return highlightedContent
    .replace(new RegExp(escapedPrefix, 'g'), '')
    .replace(new RegExp(escapedPostfix, 'g'), '');
}

tried that also, didn't work. I think we need to search for wrongly escaped chars not only on the prefix/postfix, but also on the entire title/name/description/content

[sergiobaiao]

my dev just found out that the function :

highlights?.forEach((highlight) => {
switch (highlight.field) {
case 'cm:name':
name = highlight.snippets[0];
break;
case 'cm:title':
title = highlight.snippets[0];
break;
case 'cm:description':
description = highlight.snippets[0];
break;
case 'cm:content':
content = ...${highlight.snippets[0]}...;
break;
default:
break;
}
});

is inserting the wrong code (n class="aca-highlight">). The highlight is imported from @alfresco/js-api.

[sergiobaiao]

any workaround on this?

[MichalKinas]

@sergiobaiao @aborroy I will investigate this issue, can you please provide a pdf and a search phrase that I could use to reproduce this issue on my local env? This would be very helpful as I've never seen this issue before

[sergiobaiao]

i can do better...is there a way i can contact you directly so i can send you an username/password to my alfresco server?

[MichalKinas]

Yes you can message me on MS Teams, Slack or send me an email -> [email protected]

[sergiobaiao]

Hi Michal, did you receive my email?

[MichalKinas]

Hello Sergio, yes I did, thank you! I was able to access the server and replicate the issue, I'm investigating the code now

[MichalKinas]

Okay I analyzed what's going on there and it seems that this is an issue with search engine, instead of providing proper prefix <span class='aca-highlight'> for every highlight it seems to apply n class='aca-highlight'> instead which break the rendering in ACA in that particular case, I will forward this issue to the BE team to analyze it

[MichalKinas]

@sergiobaiao could I use one of the PDFs for which the issue is observed and attach it to the bug report for the BE team? So that they can reproduce it locally in their ACS instance? Also please let me know which search engine you're using in your instance, is it SOLR or ES?

[sergiobaiao]

Yes you can, these are public domain documents. We're using SOLR. You can forward that login info too if needed.

[MichalKinas]

Okay so I was able to reproduce this issue locally with Solr, I created the bug for the BE team and passed the info for them

[sergiobaiao]

is there a link for me to follow the bug report?

[MichalKinas]

Yes -> https://hyland.atlassian.net/browse/ACS-9073 sorry I forgot to put it here

[sergiobaiao]

it seems i cannot access it: [email protected] doesn't have access to Jira on hyland.atlassian.net.

…

On Mon, Dec 16, 2024 at 11:46 AM MichalKinas ***@***.***> wrote: Yes -> https://hyland.atlassian.net/browse/ACS-9073 sorry I forgot to put it here — Reply to this email directly, view it on GitHub <#4254 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGV4PGTGVUE47FL5VIWLGIT2F3RTZAVCNFSM6AAAAABSJW5ULSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNBVHAYTSNRTGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Atenciosamente,

--- ============ --- Sérgio Baião --- ============ ---

[MichalKinas]

I'm sorry but we cannot grant you access there, this bug will be picked up by the BE team as soon as this will be possible, I will update you on the progress and planned release date here