jjfiv / bdl-p2-bookz

Sr. Software Engineer, Fortanix

• https://jjfoley.me
• [email protected]
• he/his/him

Previously:

• Assistant Professor of Computer Science at Middlebury College (2020-2022)
• Visiting Assistant Professor at Smith College (2018-2020)
• Research Assistant, PhD Student at UMass Amherst (2013-2019)

bdl-p2-bookz's Issues

Author pages would be nice

Can we extract data from all the creator fields to make author pages?

Not all fields presented in book html

Extract Author Date Range if available to field

By using the birthdate or deathdate of authors, we could sort books (roughly at least) by year. It would be interesting to see both the oldest and newest books.

Some encoding issues still present

http://libread.org/book/etext2116
Maybe the html encodings need to be specified.

Create a system for flagging bad entries

• It should load up previous flags when the system starts, and keep them both in memory and on disk
• There should be some sort of visual indication when a book has been flagged as having a messy metadata entry.
• There should be a way to indicate which book is broken.
• [Bonus/Extra] Allow users to indicate the field that is broken (title, creator, etc.)

Long pages currently only show 20 items; need paging system

Don't let users click on letter links that don't have any results!

And if they do edit the URL or something, have some kind of "No results found" base case in your paging system.

Showing only a few page links at a time.

If we had 10 items per page, we might end up with 3700 or more pages if the user searches for "the" or something like that. As a default, using 50 items per page or 100 will help (and require less clicking for the user).

However, it would be nice to have a system that (like Amazon's Search) recommends a small number of pages, even if there are 400 results (try searching for iphone on amazon) -- it generates clickable links for the first 3 pages and a next page button. When you go to page 3, it shows you a few more, etc.

Simple search system would be nice

Can just linearly match metadata -- turns out even O(37000) is pretty small.

No links generated for numeric title pages

But there are results for, e.g. http://localhost:1234/title/2

Titles often have the author listed in them

This is kind of silly, since we usually have the author in the creator field:

https://libread.org/book/etext44568

A Comprehensive History of Norwich by A. D. Bayne
Bayne, A. D.
On Project Gutenberg

Seems like we could automagically detect this and come up with a title without the "by" part.