Skip to content

Fix position increment in (Reverse)PathHierarchyTokenizer#12875

Merged
mikemccand merged 2 commits intoapache:mainfrom
lukas-vlcek:FixPositionIncrement
Dec 18, 2023
Merged

Fix position increment in (Reverse)PathHierarchyTokenizer#12875
mikemccand merged 2 commits intoapache:mainfrom
lukas-vlcek:FixPositionIncrement

Conversation

@lukas-vlcek
Copy link
Copy Markdown
Contributor

@lukas-vlcek lukas-vlcek commented Dec 4, 2023 •
edited
Loading

Description

Incrementing position attribute for each token in both PathHieararchyTokenizer and ReversePathHieararchyTokenizer.
This change makes it possible to use both tokenizers in BaseTokenStreamTestCase.assertAnalyzesTo() method (test cases extended to demonstrate this).

This PR solves and surpasses #12750.

@msfroh I included your original commit from your Lucene fork to keep proper commit attribution.

Shall we squash all commits?

@lukas-vlcek lukas-vlcek mentioned this pull request Dec 4, 2023
Closed
mikemccand
mikemccand reviewed Dec 7, 2023
View reviewed changes
lucene/analysis/common/src/java/org/apache/lucene/analysis/path/PathHierarchyTokenizer.java Outdated
posAtt.setPositionIncrement(0);
}
posIncAtt.setPositionIncrement(1);
posLenAtt.setPositionLength(1);
Copy link
Copy Markdown
Member

@mikemccand mikemccand Dec 7, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note that 1 is already the default for PositionLengthAttribute so you could skip pulling/setting it.

mikemccand
mikemccand reviewed Dec 7, 2023
View reviewed changes
.../analysis/common/src/java/org/apache/lucene/analysis/path/ReversePathHierarchyTokenizer.java Outdated
posAtt.setPositionIncrement(0);
}
posIncAtt.setPositionIncrement(1);
posLenAtt.setPositionLength(1);
Copy link
Copy Markdown
Member

@mikemccand mikemccand Dec 7, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same here.

@mikemccand
Copy link
Copy Markdown
Member

mikemccand commented Dec 7, 2023

Thanks for tackling this @lukas-vlcek and @msfroh! I left a couple small comments, but otherwise it looks great.

Given that this alters the indexed tokens (makes them non-overlapping), I think this should be a 10.0 only change? It's highly unlikely any users are relying on how PhraseQuery or synonyms, etc. (any positional tokenfilters / queries) behave on fieleds analyzed with these tokenizers, but, still, could you add an entry in MIGRATE.txt explaining this change? Note that this should make highlighting based on postings offsets (e.g. UnifiedHighlighter, in certain modes) work on such fields when it does not today.

Also please add a CHANGES.txt entry under 10.0 as well. Thanks!

@lukas-vlcek
Copy link
Copy Markdown
Contributor Author

lukas-vlcek commented Dec 7, 2023

Note that this should make highlighting based on postings offsets (e.g. UnifiedHighlighter, in certain modes) work on such fields when it does not today.

Ture... so it sounds like we shall add a test for this as well if we want to claim it in the migration notes!

@lukas-vlcek
Copy link
Copy Markdown
Contributor Author

lukas-vlcek commented Dec 8, 2023

@mikemccand Do you think you can give me some hint about?

(e.g. UnifiedHighlighter, in certain modes)

I am looking at TestUnifiedHighlighter* tests. Does it mean that I need to use specific fieldType? Can I use any fieldType(s) from existing UHTestHelper.parametersFactoryList()? Such as postingsType and postingsWithTvType?

@mikemccand
Copy link
Copy Markdown
Member

mikemccand commented Dec 14, 2023

I am looking at TestUnifiedHighlighter* tests. Does it mean that I need to use specific fieldType? Can I use any fieldType(s) from existing UHTestHelper.parametersFactoryList()? Such as postingsType and postingsWithTvType?

Hi @lukas-vlcek -- I don't think we need to add a specific unit test to UnifiedHighlighter tests for this. I think it's fine to say something general in the MIGRATE.txt like "(Reverse)PathHieararchyTokenizer now produces sequential (instead of overlapping) tokens with accurate offsets, making positional queries and highlighters possible for fields tokenized with this tokenizer" or so?

msfroh and others added 2 commits December 18, 2023 11:56
@msfroh @lukas-vlcek
Fix PathHierarchyTokenizer positions
2e3ebf1 
PathHierarchyTokenizer was emitting multiple tokens in the same position
with changing offsets. To be consistent with EdgeNGramTokenizer (which
is conceptually similar -- it's emitting multiple prefixes/suffixes off
the input string), we can output every token with length 1 with
positions incrementing by 1.
@lukas-vlcek
Fix ReversePathHierarchyTokenizer positions
476d524 
Making ReversePathHierarchyTokenizer consistent with recent changes in PathHierarchyTokenizer.
@lukas-vlcek
Copy link
Copy Markdown
Contributor Author

lukas-vlcek commented Dec 18, 2023

@mikemccand Thanks for review. Done!
Shall I squash all commits are will you do it when merging in GH?

@mikemccand
Copy link
Copy Markdown
Member

mikemccand commented Dec 18, 2023

Thanks @lukas-vlcek -- looks great -- no need to squash, I'll take care during merge. Exciting improvement, finally!

@mikemccand mikemccand merged commit 5d6086e into apache:main Dec 18, 2023
@lukas-vlcek lukas-vlcek deleted the FixPositionIncrement branch December 18, 2023 20:05
@hossman hossman mentioned this pull request Feb 25, 2026
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mikemccand mikemccand mikemccand approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@lukas-vlcek @mikemccand @msfroh