slacgismo / docs-browser Goto Github PK

The CSS style is not interesting. We need to come up with a design that is distinctively SLAC GISMo, including things like a nice banner, our vision, and a link to our homepage.

Clicking on a link to another page doesn’t load the target page correctly.

If a user triggers a 404 from the github api, e.g. typing in an input that doesn't resolve to something valid from github's perspective, we need to let the user know.

There are multiple options, we can show a toast message that says the requested doc could not be found. We can simply highlight the input box red with a tooltip. Or we can change the whole main-content to show there was a 404 error.

Clicking "edit document" takes you to GitHub log in but after log Github generates "WHAT!?" error.

The installation process needs a Python script to deploy on cloud platforms like AWS, e.g., simple script that works as follows:

bash$ git clone https://github.com/slacgismo/docs-browser
bash$ cd docs-browser
bash$ nano src/config.js
bash$ ./deploy.sh src www.mysite.org

Users are not able to login to the host when using the browser. As a result, they are limited to about 50 queries (or about 15 page refreshes) per hour.

We need to add user authentication.

Once user authentication is done, the user/org and project inputs should be changed to selects based on what the user is authorized to see.

We need to add support for embedded latex code, e.g., MathJax, xymatrix, and tikz.

Some project docs folders are more complex and have a tree, not just a flat list. Support for unfolding folders in the tree is needed. See dchassin/gridlabd/develop for a good example.

GitHub supports searching using https://api.github.org/search/code. However it has some important limitations, chief among which is that it will only search the master branch of the original project and not search the forks. We also will need to restrict the search to the docs folder .md files, and separate exact file name matches from content matches. When content matches are found, some context should be given, and if possible when the user clicks on it, the page should be display at the location of the first match with the browser's search bar configured to find the matching contents further down the page. When the browser navigates back to the search page, the page should be redisplayed as it was so that the user does not lose their place in the search list and can tell where to continue. A "Next" button would be really nice alongside the "Go" button.

When the document browser is loaded, we should attempt to parse the query string params, if any. If they are given, then we populate the input fields with those values and perform the initial page load request with them.

We should support something like 👇
?owner=slacgismo&project=gridlabd&branch=develop&folder=/Module&doc=/README.md

The inputs are not read by the Refresh button and the action does not reload the Table of Contents.