Hi Alex
That sorted it! Thank you so much.
On Solr 8.9.0 I had to set update.autoCreateFields to be false in the
add-unknown-fields-to-the-schema below:
<updateRequestProcessorChain name="add-unknown-fields-to-the-schema"
default="${update.autoCreateFields:false}"
processor="uuid,remove-blank,field-name-mutating,parse-boolean,parse-long,parse-double,parse-date">
<processor class="solr.LogUpdateProcessorFactory"/>
<processor class="solr.DistributedUpdateProcessorFactory"/>
<processor class="solr.RunUpdateProcessorFactory"/>
</updateRequestProcessorChain>
And that sorted it.
Many thanks again for your help!
Matthew
Matthew Flowerday | Consultant | ULEAF
Unisys | 01908 774830| matthew.flower...@unisys.com
<mailto:matthew.flower...@unisys.com>
Address Enigma | Wavendon Business Park | Wavendon | Milton Keynes | MK17 8LX
<http://www.unisys.com/>
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY
MATERIAL and is for use only by the intended recipient. If you received this in
error, please contact the sender and delete the e-mail and its attachments from
all devices.
<http://www.linkedin.com/company/unisys> <http://twitter.com/unisyscorp>
<http://www.youtube.com/theunisyschannel> <http://www.facebook.com/unisyscorp>
<https://vimeo.com/unisys> <http://blogs.unisys.com/>
From: Alexandre Rafalovitch <arafa...@gmail.com>
Sent: 08 July 2021 00:13
To: users@solr.apache.org
Subject: Re: Solr Phonetic Search funny
EXTERNAL EMAIL - Be cautious of all links and attachments.
Check if you have "schemaless" processing enabled.
If yes, then it is most likely because schemaless pipeline converted your date
to real date during preprocessong, then when the schema said "it is a
text/string", the actual Date got serialized back into a string, using a local
formatting (probably a "date long").
If you check indexed tokens in Admin UI for that field, it may give you a
confirmation.
See https://www.outerthoughts.com/2016/04/solr-5-puzzle-magic-date-answer/
Regards,
Alex
On Wed., Jul. 7, 2021, 3:17 p.m. Flowerday, Matthew J,
<matthew.flower...@gb.unisys.com <mailto:matthew.flower...@gb.unisys.com> >
wrote:
Hi There
I have just noticed a funny with DoubleMetaphone phonetic searching involving
records with certain values in date fields causing a phonetic match to be
returned for that record.
I was searching for the word ‘main’ and I found results being returned with
nothing being highlighted via the unified highlighter. I tracked the issue down
to these date fields holding this values
"statementDate_dtr":"2019-10-28T00:00:00Z",
"statementDate_dt":"28/10/2019",
Now the word ‘main’ has a phonetic value of ‘MN’ (according to the Solr Admin
Tool analysis feature). Searching for either ‘main’ or ‘mn’ in the Admin tool
returns a match on the record with these date values.
The fields are configured as
<dynamicField name="*_dt" type="text_en_uleaf" indexed="true"
stored="true"/>
<dynamicField name="*_dtr" type="pdate" indexed="true" stored="true"/>
I suspect the issue is down to the _dtr field as the _dt field is basically a
string field.
If I place the text 2019-10-28T00:00:00Z in a standard string field on another
record the phonetic search does not match for ‘main’ on that record which does
seem to point to the issue being related to a date field.
If I then update the record and change the date to be say
"takenDate_dtr":"2021-07-07T00:00:00Z",
"takenDate_dt":"07/07/2021",
Then a search for ‘main’ does not find a phonetic match.
I am using Solr 8.8.1 and Solr 8.9.0 and the issue appears in both versions.
I was wondering if anyone has seen this before?
Many Thanks
Matthew
Matthew Flowerday | Consultant | ULEAF
Unisys | 01908 774830| matthew.flower...@unisys.com
<mailto:matthew.flower...@unisys.com>
Address Enigma | Wavendon Business Park | Wavendon | Milton Keynes | MK17 8LX
<http://www.unisys.com/>
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY
MATERIAL and is for use only by the intended recipient. If you received this in
error, please contact the sender and delete the e-mail and its attachments from
all devices.
<http://www.linkedin.com/company/unisys> <http://twitter.com/unisyscorp>
<http://www.youtube.com/theunisyschannel> <http://www.facebook.com/unisyscorp>
<https://vimeo.com/unisys> <http://blogs.unisys.com/>
smime.p7s
Description: S/MIME cryptographic signature
