Hi
Share excludes commments from the search.
Comments are modeled as fm
ost - which is extended from content.
If you go to the node browser - you can find them with TEXT:"<Comment text>"
You would have to do a "join" with several steps to get back to the document to which the comment applies.
Andy
Share excludes commments from the search.
Comments are modeled as fm
ost - which is extended from content.If you go to the node browser - you can find them with TEXT:"<Comment text>"
You would have to do a "join" with several steps to get back to the document to which the comment applies.
Andy
Hello,
as Andy has pointed out, comments are separate documents / content items and you need to combine results from several layers to achieve a search result as "documents with comment xy".
We did a proof-of-concept last year based on Alfresco 3.4 where we performed a really low-level modification of Alfresco search & indexing. Comments would be included in Lucene documents (Solr not yet functionally ready on 4.0 HEAD at the time) and you could do a search query like +TYPE:"cm:content" +@cm\:name:"Test*" +COMMENT:"Sample Comment".
Such an approach could be possible in 4.0 with Solr as well, but is neither supported nor recommended - you'd have to change / touch major parts of the search subsystem and index server.
The simplest way to go is writing / adapting a web script to perform a two-phase query, as Andy indicated.
Phase one for fmost => collect the lowest level cm:content ancestor nodes from results
Phase two for OR-ed list of ancestor nodes and any additional query fragment => display results
Note that the first phase can be rather expensive - not from the Solr query side, but from navigating to ancestors, as this will usually access all the metadata on the nodes traversed. Depending on the amount of comments matched, you may need a blazingly fast DB or well-sized caches with not-too-short TTL values…
Come to think of it, a more scalable but bit more complex approach could be the following (Note: only works properly, if the same read privileges apply to comments as on their content anchor, and you are only interested in the document irregardless of which specific comment matches):
1) Define an aspect (e.g. "comment scope cache") with a property of type d:content
2) Define a rule / policy for new / updated / deleted fmost nodes that aggregates all the comments text (not just the modified one) into the aspect property, which should be applied to the cm:content anchor node
3) Perform a regular search and include the aspect property as a selector for "documents with comment X"
Regards
Axel
as Andy has pointed out, comments are separate documents / content items and you need to combine results from several layers to achieve a search result as "documents with comment xy".
We did a proof-of-concept last year based on Alfresco 3.4 where we performed a really low-level modification of Alfresco search & indexing. Comments would be included in Lucene documents (Solr not yet functionally ready on 4.0 HEAD at the time) and you could do a search query like +TYPE:"cm:content" +@cm\:name:"Test*" +COMMENT:"Sample Comment".
Such an approach could be possible in 4.0 with Solr as well, but is neither supported nor recommended - you'd have to change / touch major parts of the search subsystem and index server.
The simplest way to go is writing / adapting a web script to perform a two-phase query, as Andy indicated.
Phase one for fm
ost => collect the lowest level cm:content ancestor nodes from resultsPhase two for OR-ed list of ancestor nodes and any additional query fragment => display results
Note that the first phase can be rather expensive - not from the Solr query side, but from navigating to ancestors, as this will usually access all the metadata on the nodes traversed. Depending on the amount of comments matched, you may need a blazingly fast DB or well-sized caches with not-too-short TTL values…
Come to think of it, a more scalable but bit more complex approach could be the following (Note: only works properly, if the same read privileges apply to comments as on their content anchor, and you are only interested in the document irregardless of which specific comment matches):
1) Define an aspect (e.g. "comment scope cache") with a property of type d:content
2) Define a rule / policy for new / updated / deleted fm
ost nodes that aggregates all the comments text (not just the modified one) into the aspect property, which should be applied to the cm:content anchor node3) Perform a regular search and include the aspect property as a selector for "documents with comment X"
Regards
Axel
