You no need restart Gramps, but you need turn on the new csv file in settings
Found the right directory to place the csv file on MacOS, it’s ~/Library/Application Support/gramps/WebSearch/csv
Thanks
WebSearch v1.3.2 on MacOS
Every search definition is duplicated - maybe this is a MacOS only problem
I have only one line (except for the header line) in the CSV file.
Same problem with the default CSV files.
1 Like
Did you see the directory here? Is it correct here? Is it opened by click? I did not test it on MAC OS
Yes, its reproduced for me. I will fix it
@Urchello
I was reading the documentation here: GitHub - jurchello/WebSearch at websearch-gramps52
The online-help is almost correct, as it shows /Users/sam/Library/Application Support/gramps/WebSearch. It’s shown as a link, but nothing happens, when clicking.
Maybe there should be a hint, saying CSV-files should be in a subdirectory named ‘csv’.
1 Like
This is a fresh Tab, so it tested on Ubuntu only. I will try to fix it a bit later
yes, I will update readme and the Tab info.
Fixed. It will be available in v.1.5.3+
2 Likes
@csam Two detailed sections about user-data directories, custom csv files and custom json file are added to README.md:
-
JSON instructions: GitHub - jurchello/WebSearch: Gramps addon
-
CSV instructions: GitHub - jurchello/WebSearch: Gramps addon
Also the WebSearch infopanel is updated. Now it contains more user-data:
It will be available in WebSearch 1.5.4+
2 Likes
AI suggestions are optimized in versions 1.7.4+
I’ve made some improvements to how AI-based website suggestions work in WebSearch:
Separated into two distinct categories:
- Resources: Archives, forums, and databases
- Communities: Telegram groups, Facebook groups, Reddit threads, YouTube channels, etc.
Visual distinction by color:
- Green → for Resource-type suggestions
- Blue → for Community-type suggestions
Separate AI prompts/requests:
- Each category now uses its own prompt and request logic
- This ensures more relevant and targeted results
Performance optimization:
- AI is no longer called if there are already 10 or more non-skipped suggestions in the file-system
- This reduces unnecessary API usage and speeds up loading
Code cleanup:
- Imports and logic have been reorganized for better maintainability
One of the major benefits of the UID attributes is that users can cultivate collections of hotlinked tools on the parent database connected to the UID stored in Gramps. Allowing a new/changed hotlinked feature to be managed in a single entry. And have that added/corrected capability cascade to all Persons having the associated UID.
So, while your primary link to WikiTree might be the Person’s WikiTree profile, you could also build links to the most interesting WikiTree addon tools. Such as the RootsSearch or DNA Descendants.
And in FamilySearch, you might link to the Sources or Fan chart.
But this raises another intriguing possibility… some of these site offer relationship tools too. But they need a second person reference.
For some of the tools, the 2nd Person is always the logged in person. (This includes tools like WeRelate, WikiTree’s MyCousins, or FamilySearch’s find my relationship.)
All of the WebSearch substitutions are keyed off the Active Person. Could a 2nd set of substitutions key off the Home Person?
So that would allow to creating links for dual-UID tools like WikiTree’s Relationship Finder.
1 Like
Can you show some real urls/examples pls?
Sure. I started with the Roosevelt Presidents. It seems that they are good examples because FDR married a cousin. And the connection of President Teddy Roosevelt was closer to FDRs wife than him
So…
Assume that the Home Person was President FDR (Roosevelt-1) and cousin Active Person was President Teddy Roosevelt (Roosevelt-18) in a Gramps tree:
The default (shortest) connection:
https://www.wikitree.com/index.php?title=Special:Connection&action=connect&person1Name=Roosevelt-1&person2Name=Roosevelt-18
The common ancestor connection:
https://www.wikitree.com/index.php?title=Special:Connection&action=connect&person1Name=Roosevelt-1&person2Name=Roosevelt-18&relation=2
The common paternal ancestor connection
https://www.wikitree.com/index.php?title=Special:Connection&action=connect&person1Name=Roosevelt-1&person2Name=Roosevelt-18&relation=4
1 Like
Got it. OK, I will add this feature. Thanks for the idea )
1 Like
@emyoulation what is your current WebSearch version?
I’m away from the computer and using the phone. So that answer will have to be later. But it was last updated a couple weeks ago.
1 Like
So, its 1.0.0+ probably, with major updates. I would like to ask you and other users who use this version if it working stable, no new issues occur?
1 Like
I’ve had some difficulty with some UIDs not working. But need to troubleshoot my custom CSV and JSON … it is likely that I created my own problem.
I am hoping that @Nick-Hall will ammend the Gramps model for storing UIDs for People, Places, Sources, Citations and Repositories. Where there is an UID structure instead of just a value.
My Attributes custom type lists have grown far too cluttered for a 1 dimensional list. So it would be nice to have UID records winnowed out.
The mix of Phenotype, permanent tags and UID Attributes makes it difficult to find any of them. (And that they don’t have a sort order or filtering makes it that much more difficult.)
1 Like
So, looks like we need add one more json key "home_person_key_name" as optional. @emyoulation could you check please is it what you are expectiong? Both people will have attributes like WikiTreeID. When click any person the link below will extract WikiTreeID values from current person and home person and then insert both values to the link keys: %(WikiTree.ID)s and %(HomePersonWikiTree.ID)s. Correct?
{
"nav_type": "Person",
"attribute_name": "WikiTreeID",
"url_regex": ".*gedbas\\.de/uid/.*",
"key_name": "WikiTree.ID",
"home_person_key_name": "HomePersonWikiTree.ID",
}
https://www.wikitree.com/index.php?title=Special:Connection&action=connect&person1Name=%(WikiTree.ID)s&person2Name=%(HomePersonWikiTree.ID)s&relation=4