Terms of servers

Intro

Terms of servers is a terminal based publication that runs on a temporary, communal server called Cerealbox. Working with the server for almost a year now evoked questions around its materiality and reliability. We discuss these questions among us, the users, but also direct them towards Cerealbox itself. Terms of servers' form is malleable, it can be read in a computer shell or as printed text.

Images

• 

  Terms of Servers Title Screen (wip at the studio)

• 

  Terms of Servers installed at the Telecommueum next to the Archive

• 

  close-up of the installation at the museum

• 

  Cover of the print publication

Process

links to pads and wiki pages

pink postit scan

https://pad.xpub.nl/p/si27-claudio-kim

https://pad.xpub.nl/p/si27-claudio-kim-text

process documentation on kims wiki: User:Kim/Special Issue 27/uptime

Code

code tour ...

from rich.console import Console 

from rich import print

from rich.markdown import Markdown

from rich.align import Align

from time import sleep

import os 

import sys

...

#variable declaration
time_lines=n
title=""

...

console = Console()

...

with open("path/to/file.md") as name:
    md_name = Markdown(name.read())

with console.capture() as capture:
    console.print(md_name)

...

all_lines_name = capture.get().splitlines()

...

def exaust(file, time):
    while file:
        page="\n".join(file[:5])
        sys.stdout.write(page)
        sleep(time)
        file=file[5:]

...

while True:
    copy_name = all_lines_name
    
    console.clear()

    print("")
    exaust(copy_name, time_lines)
    sleep()

    print(Align.center("\n$ script.sh"))
    print(Align.center(os.popen(script).read()))

    sleep()
sleep()

Things we didn't end up using

On our way towards the practical realization of the project we ran into multiple possible ways we could achieve presenting formatted text.

Initially we had though of using various text manipulating tools from the linux core tools:

• tr
• pr
• cut
• col
• sed
• printf
• ANSI escape codes (which make the file unreadable)

Afterwards, we ran into tpp (text presentation program), a ruby-ncurses presentation tool that works with files written with some special description synthax (ex. -title, -newpage). Tpp looked like a solution for a while, it even has the possibility to run commands! But we were having issues to make it work on Kim's computer and when you actually ran commands inside of it the output was displayed at the bottom of the file. Working with tpp looks fun but it also does not seem to be the best for large quantities of text.(as presentation files become unreadable)

At the same time of this discovery, Joseph mentioned urwid, a python console user interface library for unix-like OSses. It allows for many possiblities like widgets for building custom TUI, animations, style support, but we also were struggling with making sense of the documentation.

Xdotool is an automation tool that allows for keyboard automation and more, at some point we were thinking of making a script that would open tpp presentations and send keyboard input every x seconds to cycle through them, the files would then be intermitted by linux commands. We did not settle for something too far from this in terms of logic, but we dropped this idea since xdotool works only with a desktop environment.

After some time thinking of how we might solve this, Joseph came up with another recommendation: a python library called pyboxen. Pyboxen is built on top of another library called rich (the one that we ended up using) and it is a python port of another popular npm library called boxen. It allows to make boxes, titles, and controlled formatting. We found out we did not need that much and so we started to look into rich (that also allows for less costumizable boxed frames btw).

Michael helped us writing what is now the final code, he was suggesting to look into textualize, another library from the same creator of rich, but in the end we had what we need and textualize was adding another (interesting) layer of abstraction we needed to wrap our head around in a very limited amount of time.

At some point it really helped sitting down and listing our essentials for this project, we needed a way to format text, but to also run both linux commands and custom bash scripts. We then decided to settle for python and both the rich and os library.

ncurses, tpp, urwid, pyboxen, xdotool, textualize

Print

Steps

the following 3 steps enabled us to turn a folder of markdown files into a ready to print booklet, only using terminal commands

pandoc *.md -o book.html

Pandoc is a document converter. We grab all .md files from the current directory and output them in one .html file. To ensure the right order of files, filenames were numbered.

weasyprint -s stylesheet.css book.html book.pdf

Weasyprint allows us to attach a stylesheet with the -s option while converting the html into a pdf

pdfbook2 --paper=a4paper --short-edge --no-crop book.pdf

To make the pdf ready for print we need imposition. pdfbook2 takes care of that and enabled us to define some other specifics like paper format and size.

Resources

The following (wiki) pages were super helpful to come to the set-up we used:

• Pandoc especially this section: Pandoc#PAD_to_MARKDOWN_to_HTML_to_PDF_to_BOOKLET_PDF
• Weasyprint
• CSS Print (for step 2 defining the stylesheet for the pdf)

What's next

We are interested in understanding how can we distribute this publication inside of the terminal, independently from when we first made it public at the Telecom Museum Rotterdam, the following is a list of open possibilities:

• uv, pip

we could package the project and then have people install it via pip install

• telnet

• ssh guest user

we could host a server on our own and set a guest user with no sudo rights open a process on connection that would read the publication. (eg. arthur @ beubeu)