aelzenaar/bucephalus

LaTeX document store

Go to my website https://aelzenaar.github.io/

bucephalus's Issues

Clean up HP/bucvac to allow file inclusions and ability for templates to fill Buc_* tags

Currently the two prototypes (asst.mustache and note.mustache) have a lot of shared TeX; it should be possible to shift this out into an include file somehow.

This may require a shift from mustache to jinja for bucvac so we can do this in Python - TeX does not make it easy to change search paths for \input commands!

Combine bucadd and bucfup

Can a -u command be added to bucadd (similar to that for bucvac) so that bucfup can be removed?

More sane file storage

Instead of storing file.tex and file.pdf like YYYY/MM/DD/src/file.tex and YYYY/MM/DD/file.pdf, store them like YYYY/MM/DD/file/file.tex and YYYY/MM/DD/file/file.pdf. This allows us to serve support files (e.g. images) along with the files themselves, allowing us to properly implement #9 .

When removing files, disallow reuse of IDs

Change remove_record_by_id so that when removing files, the ID is marked as 'deleted'.

The server can be changed to return a 410 error when such a file is accessed.

Date timestamps not zero filled on web

timestamp_for_item in web/server.py does not zero fill dates to two chars, only times.

Centralise database open

Currently, every subroutine which wants to access database.db will open its own database instance. This should be moved into its own function in dbops, so that we can globally change database opening parameters like indent.

asst.mustache should have commands for question and part

Indeed, incorrect usage of subsectioning commands in the amsart document class can cause loss of marks (a cautionary tale)!

Add support for HTML output into bucvac

(Depends on #3.)

Add support for outputting an HTML render rather than a PDF render from bucvac upon adding. This will also require testing to make sure that HTML outputs are rendered correctly by the browser in the iframe, otherwise we may need to do some mimetype fiddling like the small piece of code already in server.py that handles markdown.

Implement chains and next/prev links

On /v/ pages, the next/prev links should point to (for example) the next/previous day with articles. (This will require us to pass the list of items in each db entry around.)

Inside the database, there will be a `chains' table which points to the first article in each chain; then each item in the chain will be linked (1) back to the chains table entry, and (2) to both the next and previous items as a doubly-linked list.

Caching in dbops

Code like this:

  if year == None:
    years = []
    for item in db:
      if not(item['ts_year'] in years):
        years.append(item['ts_year'])
    return sorted(years)

Belongs on tdwtf (does anyone still read that? :o). Iterating through every item in the database just to enumerate the possible values of 'ts_year' tags? Really? We need to cache this, and every place where this kind of pattern occurs, before my database gets too big.

Possible implementation: add an additional table in the database to store [years/months/whatever], add a command-line to update it manually (based on iterating the full db), and update it whenever we add/update an article (i.e. check whether the year attached to the article is in the table, if not add it).

`bucvac` and `bucadd` should print ID of new files

Nice easy one: for convenience, every way of adding files should print out the docid of the new database entry.

Hp2 is not adding source files correctly

When adding files via bucvac, the source .tex file is not actually added to the database.

Searching will only return first match if multiple results have same filename

If two articles with the same name exist, any search which matches both will only return one of the two.

Better tag management

This feature would consist of two parts:

1. Expose multi-tag views in the web UI.
2. Provide a nice command-line interface, like buctask, for tag management. (This tool could also handle chain management, for #2).