DeepBlueCLtd / Fi3ldMan

Field Service Manual with advanced data exploitation
https://deepbluecltd.github.io/Fi3ldMan/
Apache License 2.0
1 stars 1 forks source link

Promote country or class page in WebHelp search #63

Open IanMayo opened 1 year ago

IanMayo commented 1 year ago

The term spain will appear a lot in the content. But, when the user types Spain we would like spain.dita to be the first match.

Check if this works as expected. If no, see if/how we can add weighting to matches where the title matches the search term.

IanMayo commented 1 year ago

This does work as expected. A page with "Spain" in the title or filename is promoted over one with Spain in the content somewhere.

IanMayo commented 3 months ago

Re-open, this isn't working as well as hoped for class pages.

For a class such as anchors, anchor pics gets shown higher up the results listing. In the target content, the supporting pages don't have pics in the title, so it's even less clear which is the higher priority.

I guess we need to investigate the webhelp indexer, and see how we persuade it that the class file should have highest priority in the results.

Note: we can inform the indexer that a specific page is the main source for a class using keywords: image

We should probably do some custom python dev to parse the DITA docs, find the class files (presence of <tgroup cols="6"> or <p outputclass="h1">PROPULSION</p>), get the document title, and generate a keyword for it. While we do that, we should probably also add a keyword for class-file to aid future parsing.