Multi Language Search for WCM

• Type: Story
• Status: Closed
• Priority: Minor
• Resolution: Completed
• Affects Version/s: None
• Fix Version/s: 7.2.0 GA1
• Component/s: Search Infrastructure, WCM
• Labels:
  • no_predic
  • search-feb19
• Epic Link: Multi Language Search in 7.2
• Sprint: Search | S02 Sprint 1, Search | S02 Sprint 3, Search | S02 Sprint 4, Search | S02 Sprint 5, Search | S02 Sprint 6, Search | S02 Sprint 7, Search | S02 Sprint 8, Search | S02 Sprint 9, Search | S02 Sprint 10, Search | S02 Sprint 11, Search | S02 Sprint 12, Search | S02 Sprint 13
• Git Pull Request: https://github.com/brianchandotcom/liferay-portal/pull/66691

Back-end changes

In this story, Journal Article's indexer behavior change this way:

• Given translated fields (title, content, description), each translation of one of them is still indexed in the correspondent document field. I.e., if a web content is e.g. titled "On soccer teams" and this title is translated to Portuguese as "Sobre times de futebol," then its corresponding document will have the field title_pt_BR with the given content.
• However, we do not index the default field anymore (that is, web content documents will not have the fields title, description and content anymore.)
• Also, highlighted snippets are sent to all languages found in the web content.

Journal Article's searcher behavior changes this way:

• Now every translated field from the web content is queried, no more only the current language's field.

Results for users

Search is done over all languages

A word a user searches for will be looked for in every language of the web contents, not only the display language or the user's language.

For example: if you have

• a web content titled "On clocks and time" in en_US;
• and you have another  web content titled "On soccer teams" (en_US) and "Sobre times de futebol" (pt_BR); and
• a user searches for "time";

then both the first and second web contents will be returned because both have the word "time" (even they meaning different things.)

Previously, only the first document would be returned.

Search returns highlights for translated fields

Previously, highlighted snippets were returned only for default fields, now they are returned to all localized fields (and not for default fields, that exist no more).

For example,

if you have

• a web content titled "On clocks and time" in en_US;
• and you have another  web content titled "On soccer teams" (en_US) and "Sobre times de futebol" (pt_BR); and
• a user searches for "time" using Portuguese as the display language;

then both the first and second web contents will be returned. Previously, only the second document would be returned.

Also, the second document will display a snippet highlighting the word "times." Previously, this snippet was not returned and so the displayed results did not show what matched with emphasis.

============================

Notes for QA: integration tests only.