3.5.5 Release Notes#

Features#

Frontend and other UI related enhancements:

  • Restore current page after PDF zoom.

  • New stacked bar chart mode for Bar Chart Widget.

  • Added view all button for react items widget headers.

  • Horizontal mode of Communities widget.

  • Custom widget API architecture for Cards widget.

  • Added endpoint to fetch the last N items read by the current user.

  • Added new share GT labeling config option.

  • Added Carousel for the favorites widget in horizontal mode and adjusted styling to match Items widget.

  • Finishing Tabs React Widget. Added Active Tab URL param that will keep same tab selected after page reload or link share.

  • Added share labeling feature for non-admin users.

  • Add “view all” button and config for favorites widget.

  • Added new modes to the items widget “starred items”, “last read items” and “all items”.

  • The Tabs widget in React gets the active tab from the URL if available.

  • Limit entities header to 2 row and implement hidden entities drop-down.

  • Adds Similar Searches widget to suggest queries in the global search.

  • Added show/hide config option for number of filters in favorites widget.

  • [Link Widget] Extend Link widget with optional Save Favorite Query functionality.

  • [Link Widget] Reorder and rework config options.

  • [DashEdit / Widget Browser] Search now works also for widget keywords, e.g. search for “link” will show “Actions” widget.

  • [Item Detail / Reference Documents] Add Search bar to Reference Documents selector.

Backend related enhancements with tags which represent their area of work:

  • [Query Processor] Create a background task that will keep default query workflow up to date. It makes that all changes and improvements to the query processing will be always reflected in the already created workflows.

  • [Query Processor] Don’t import default query processing workflows because they are created automatically during the project creation. It prevents having multiple default workflows in the same project.

  • [Query Processor] Create query processing workflows in all projects. It assures that all projects that are missing default query processing workflow, will have it from now on.

  • [Search] preserve_scroll_order: Parameterize scan API to choose if index-scrolling should apply provided sort-order (expensive operation) or no sort at all(return documents from index order, fast & lightweight operation).

  • [Search] Add preserve_order parameter to squirro_client.scan() method (perform deep pagination with items considering sorting order).

  • [Search] Refactored/simplified handling of synonym search to support synonyms on managed ES (no usage of synchronized synonym files across the cluster, but usage of inline ES settings).

  • [Typeahead] Merge similar suggestions returned from typeahead.

  • [Typeahead] Remove punctuation from content based typeahead.

  • [Database] Extend possible package size sending to MariaDB server. It resolves problems when big amount of data (like file with synonyms) breaks the SQL server. For more information, please refer to this page.

  • [Pipeline Steps] The Change Pipeline offers the option to configure the batch priority of the generated batches. The default priority is low for the batches generated from this step.

  • [Pipeline Steps] A PDF Conversion step was added. Now incoming Office documents can be converted into PDF files, thus enabling better rendering in frontend and some more sophisticated processing in the backend.

  • [Platform] A new PDF Conversion service was added. It behaves and can be controlled similarly to the existing Squirro services that comprise the Squirro platform. The purpose of this service is to handle the conversion of documents to PDFs.

  • [Platform] The PDF Conversion service removes unused PDF renditions after some time to prevent exceeding storage capacity.

  • [Platform] FileContainer now uses hardlinks for duplicated files to prevent excessive disk space usage.

  • [Data Ingestion] Add retry logic in the dequeue operation of inputstream batches. The maximum number of those retries is controlled by the server config option ingester.stream.max-dequeue-retries (default value is 3).

  • [Dataloaders] Support deletions in OneDrive & SharePoint dataloaders.

Bug Fixes#

  • Fixes a bug where tabs widget closes the community 360 page.

  • Handling database exceptions during project and dashboard modification.

  • Perform correct escaping of all reserved characters of Elastic’s query-string clause.

  • Fix error message for groundtruth creation not being shown in the frontend.

  • [Sales Insights] When selecting from the Sentiment Widget, the filter control is not appearing.

  • Catch exceptions when ES is unreachable (for GT creation/deletion).

  • Fix invalid number of indexed items on Data tab.

  • Fixed bug on items collection.

  • Fixes labels that have boolean values are not sent as strings.

  • The QueryLexer used for query-processing broke on token sequences that contained an invalid COMPARISON syntax.

  • Handling database exceptions during project and dashboard modification

  • Fixed a bug that prevented widget header from being hidden.

  • Fixes feedbacks not being processed correctly or data type being wrong for some fields

  • Fix typeahead suggestions don’t match the input query.

  • Fix sort selector in Result List widget, when used in Global Search dashboards

  • fixed sizing issues with items widget when sidebar panel is open.

  • Fix parsing texts without boxes.

  • Fixes issues with react widgets not reacting to the dashboard search. Fixed issues with react widget not loading when duplicated.

  • Fixed URL widget property input which was not working because of incorrect validation.

  • [data ingestion] Fix bug where some times batches were incorrectly assigned to a processor without being processed by it.

  • AI Studio ML Templates: Activation of quantize option by default for the FastText based templates. This reduces the storage footprint by a magnitude but comes with the cost of longer training duration. More info to it can be found here (https://fasttext.cc/blog/2017/10/02/blog-post.html#model-compression).

  • [Sources List] Ensure right value is displayed in indexed items count field.

  • fix webshot generation for pdfconversion files.

  • Items widget size adjustments to match figma.

  • Fix a bug where Squirro becomes unresponsive after switching dashboards multiple times.

  • [Sources List] Deleting a source will wait for success before refreshing the list.

  • Improved react Carousel component that is used inside Items widget, scrolling will be better now.

  • Add conversion to str in the flatenning dict function.

  • Fixed custom query not working for items widget. Fixed items widget breaking when no results.

  • Fix missing abstract in the communities widget.

Breaking Changes#

  • Content Augmentation and Content Extraction pipeline steps now only process the first file found in the list of files of an item. This change does not require any change to existing Pipeline Workflows, but it is important enough to note it.

Installation and Upgrade#

You will have to resolve at least the following config files when upgrading from Squirro 3.3.0

  • /etc/nginx/conf.d/ssl.inc

  • /etc/squirro/common.ini

For new installations, please follow the Installing Squirro on Linux instructions.

To upgrade an existing installation, please consult Upgrading Squirro.