[ https://issues.apache.org/jira/browse/LUCENE-6914?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Hoss Man updated LUCENE-6914: ----------------------------- Attachment: LUCENE-6914.patch updated patch with beefed up randomized testing to reproduce the problem that way (either that or some other problem that looks similar to my naked eye) Then I took a shot in the dark at a fix to the call to StemmerUtil.delete and that seems to make the beast happy. Since i'm way out of my depth here i don't intend on commiting w/o explicit feedback from someone who understands this code. (i'm mainly worried i may have introduced some other equally bad bug w/o realizing it). Anybody who understands this and thinks my patch looks good is welcome to run with it for 5.4, no need to wait for me. > DecimalDigitFilter skips characters in some cases (supplemental?) > ----------------------------------------------------------------- > > Key: LUCENE-6914 > URL: https://issues.apache.org/jira/browse/LUCENE-6914 > Project: Lucene - Core > Issue Type: Bug > Affects Versions: 5.4 > Reporter: Hoss Man > Attachments: LUCENE-6914.patch, LUCENE-6914.patch > > > Found this while writing up the solr ref guide for DecimalDigitFilter. > With input like "𝟙𝟡𝟠𝟜" ("Double Struck" 1984) the filter produces "1𝟡8𝟜" (1, > double struck 9, 8, double struck 4) add some non-decimal characters in > between the digits (ie: "𝟙x𝟡x𝟠x𝟜") and you get the expected output > ("1x9x8x4"). This doesn't affect all decimal characters though, as evident > by the existing test cases. > Perhaps this is an off by one bug in the "if the original was supplementary, > shrink the string" code path? -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org