1620773 - HTML partially indexed by Gloda, some words from the message body not found

Status: NEW

(MailNews Core :: Search, defect)

Description

[Fabrice Coudert]

Attachment: Votre commande a bien été prise en compte.eml

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:73.0) Gecko/20100101 Firefox/73.0

Steps to reproduce:

This are instructions given to reproduce the defect from a fresh new profile. It use a sample message attached here.

• Start profile manager : thunderbird -ProfileManager --no-remote
• Create a fresh new profile and start it
• Create a folder into "Local Folder" and drop the sample message into it
• Take a look into the message. It has several parts. The word "80695218" is found on the first part (under "Numéro de commande"), and "Surimi" is on the second part.
• Try searching each word using quick search or global search : "80695218" is found by both, "Surimi" is found only by quick search (select "body").

This is reproduced on 68.4.1 (64 bits) Ubuntu 18.04, and on 65.5.0 (32 bits) Windows 10.

Actual results:

The words on the second part of the message is not found using the global search.
When digging into the global-messages-db.sqlite file, all the words after the first part have been skipped from indexing.

Expected results:

All the words should be indexed.

Comment 1

[Fabrice Coudert]

This is also reproduced using latest 74.0b2 for win64

Comment 2

[Jorg K (CEST = GMT+2)]

I can confirm this. There are no multiple parts in the message, it's one big HTML part. Any yes, some terms aren't found in Gloda search, but they are found in regular body search. Sadly Gloda isn't actively maintained.