H&D website v2: Difference between revisions
No edit summary |
No edit summary |
||
Line 100: | Line 100: | ||
|- | |- | ||
! March 2022 | ! March 2022 | ||
| backend work continues, front-end work starts | | backend work continues, front-end, design work starts | ||
|- | |- | ||
! May 2022 | ! May 2022 |
Revision as of 11:19, 19 February 2022
H&D website v2 | |
---|---|
Name | H&D website v2 |
Owner | André Fincato |
Hours | [[]] |
Budget | [[]] |
Categories | [[]] |
Period | [[]] |
Intro
The following is a broad outline for the plans we have to re-build the current website of H&D.
The current setup is composed of a Python program that reads from the MediaWiki instance of H&D (that we use as a CMS of sort), and build up a series of pages:
- a main page showing upcoming and past events
- a section page, listing pages under a specific Category
- an article page, displaying a single wiki article
While this structure holds well, there’s been some evolution in the type of content and the way we want to organize it. An explicit example of these problems can been seen in the ever-growing number of items forming the main navigation bar. There are also still some ongoing difficulties when using some functionalities of MediaWiki (eg when using Concepts, or when making articles with particular titles), that should either be better documented, or re-thought.
Moreover, the current set-up doesn't take into consideration different points of access: the website is not screen-reader-friendly, takes time to load (as pages are rendered when requested), and it only supports creating content in the English language. This has significantly slowed down the processes of creating and editing pages, and limited our viewership to an audience that is able-bodied, english-speaking, and well-connected to the internet. There is an imminent need to address these issues.
Proposal
Taking into account the above points, we propose a project that is in two parts: (1) general updates for accessibility and (2) internationalization. These are both inter-connected but because of the way budget has been allocated and shifted in the last two years (see section: spreadsheet tetris), this project has to be dealt with in two separate sections:
General Updates for Accessibility
This section details the task of completely re-writing the website's front- and back-end in order to improve reader experience, create more points of access, employ less server-intensive and more ecologically sustainable rendering processes, and introduce a new stylesheet and navigation logic.
To do:
- re-write the current Python program (currently using Werkzeug) into a more approachable codebase (eg by converting it to Flask — yes, Flask is built on top of Werkzeug), that can be understood and extended by more people; this includes:
- improve routing
- split functions and make them composable
- make it easier to add new features (when necessary)
- server side rendering: HTML pages are rendered when they are created or edited in the wiki jnstead of on every request
- better error handling when something goes wrong in the wiki (eg it does not bring the frontend website completely down, as it happens now)
- text-only version / with git integration
- Design and layout
- add a search function to the frontend, to query the wiki-backend
- add edit button
- re-think how top navigation works, both on wiki level as well as on frontend side (see Concepts problem)
- integrate more the wiki bits (metadata) of MediaWiki into the frontend, eg show article’s last change timestamp, article’s history, if the article you are reading has been referenced in another article (within the wiki), etc; this can be put under “transparency”
- new design and stylesheet
- text-only version
- Install the mediawiki visual editor and code editor
- Install a more accessible wiki theme
- Creating and Improving Points of Access
- Implement keyboard navigation (tab-indexing)
- Comply with ARIA standards for screen-reader accessibility (aria-labels / aria-roles)
- remove unnecessary javascript; (navigation and for adding an image slideshow)
- convert image captions from mediawiki image uploads into alt text on front-end
- machine learning => suggestions?
- Do a web audit with a professional (Eric?)
- Document Mediawiki (how to use) and www-v2 code (how to develop)
- We should explore more ideas but we are very limited in our thinking here; so i am leaving an empty point:
- -
- GDPR & Web Analytics
- We use web analytics but our readers don't know => this is illegal
- "This website collects user data about page visits so that H&D is able to make reports to its funders"
To discuss (not yet budgeted for):
- explore a new solution for hosting/servinng audio and video
- better integration to the rest of the software that H&D setup and or built in these years: etherpad, livestream, mailing-list, etc
- Install ethercalc and migrate database from karls.computer
- revisit the <tool/> implementation
References:
- Flask: https://flask.palletsprojects.com/en/1.1.x/
- ARIA: https://developer.mozilla.org/en-US/docs/Web/Accessibility/ARIA
Internationalzation
To do:
- add multi-language support to MediaWiki
- implement multi-language option in the frontend and routing
- translating meta texts such as navigation, buttons, error messages, etc...
- prepare message for untranslated texts (link to email suggested changnes / translatios)
To discuss (not yet budgeted for):
- do we consider the possibility of implementing more languages than english and dutch? this significantly affects the front end part.
References:
- https://www.mediawiki.org/wiki/Manual:Language
- https://stackoverflow.com/questions/12144303/how-to-make-a-mediawiki-site-multilingual
- https://www.mediawiki.org/wiki/MediaWiki_Language_Extension_Bundle
- https://www.mediawiki.org/wiki/Extension:Translate
Timeline
November 2020 | drafted the project wiki and started on the backend (see article history) |
---|---|
April 2021 | research and experimentation around accessibility practices (obfuscation) |
May 2021 | work halted (and hours were not invoiced) |
February 2022 | re-write project plan to account for 2022 language and translation budget |
March 2022 | backend work continues, front-end, design work starts |
May 2022 | accessibility audit of working prototype, translations of essential texts and meta-texts |
June 2022 | www-v2 goes live, including documentation and license (soft deadline) |
July 2022 | pre-summer academy (hard deadline) |
Budget / Spreadsheet Tetris
In 2021, Initially, the project was budgeted 6.700,- under the theme of "digital accessibility". Not much of this work was done and no hours were invoiced for it. Although, notably, 800 euros were spent from the digital accessibility budget (by karl and andré) to cofinance research on standard web-accessibility practices for the obfuscation workshop platframe. What remains is 5.900,- (see row 17).
In 2022, there has been a budget of 5.200,- allocated for "Translation". This project should only have access to part of this budget, as a large part of it would go the actual work of translation of wiki pages, open calls, application texts, newsletters, etc... This work of translation is not accounted for in this project and should be given a dedicated project page, although they both feed from the same budget.
Budget availability, plan, and tracking are on this spreadsheet: https://ethercalc.hackersanddesigners.nl/www-v2
People Involved
- H&D:
- André (back, front, design)
- Karl (front, design, accessibility, mediawiki i18n)
- (anyone else wanna join?)
- One dutch-speaking member of H&D for basic meta-text translations
- External:
- Eric, Loes' colleague (for the audit)
Should someone else also be involved from the start in brainstorming points of access together? Maybe there are things we are not considering here since we are able-bodied? A friendly consultant of sorts would be nice to be able to have during this process.
Links
- budget: https://ethercalc.hackersanddesigners.nl/www-v2
- process: https://wiki.hackersanddesigners.nl/index.php?title=H%26D_website_v2_working_notes
- pad: https://etherpad.hackersanddesigners.nl/p/2022-infrastructures
- pad accessibility: https://etherpad.hackersanddesigners.nl/p/website-accessibility-check
- repo: https://github.com/hackersanddesigners/www-v2
hackersanddesigners.nl v2
setup
we're using python 3.10 at the time of writing (check .python-version
). we're also using a nix-based program (devenv) to create a complete dev environment for the project. you don't have to use it. we provide a classic requirements.txt
to install packages with pip
.
we also use rigprep (rg
) to search across the WIKI_DIR
(the folder with the static HTML articles fetched from the MediaWiki). please install ripgrep if it's missing from your system — if missing, you'll get an error in the terminal fro the website telling the rg
binary was not found.
devenv
if you're using devenv
, do the following:
- make sure to install it first, see instructions
devenv shell
virtual environment
if you want to stick to Python's standard tools, then:
- make a new virtual environment:
python3 -m venv env
- activate virtual environment:
source env/bin/activate
packages
to install all packages do:
- try:
python3 -m pip install -r requirements.txt
- else, make sure to upgrade pip:
python3 -m pip install --upgrade pip
and try again with the above command
whenever you install a new package with pip
, update the requirements list with:
pip freeze > requirements.txt
systemd config
the software is controlled through some systemd unit files (you can inspect them under ./systemd
):
hd-www-frontend.service
: runs the server (handling routes, redirects, etc)hd-www-udp.service
: runs the UDP server which listen to the MediaWiki instance sending messages whenever an article is created / modified / deleted
the above two systemd services are crucial for the functioning of the website.
there are two more systemd files that acts as a cronjob-like service. these are used to keep the frontpage up-to-date on a daily basis — necessary given that the frontpage is automatically updated only when any of the articles that it displays has received an update; so in cases when nothing changes for days, the upcoming events section might get out of date.
these two systemd services are:
hd-www-bg-task.service
: to setup the necessary command to run on a given period of timehd-www-bg-task.timer
: to set the actual timer
everyday usage
there are two ways to use this software:
- run the local server that listens to MediaWiki UDP messages
- (re-) build part of, or the entire wiki, at once
first of all run the local MediaWiki instance (from this repo):
- open a terminal,
cd
to the MW repo and rundevenv up
then, if wanting to do 1:
- open another terminal and
cd
into this repo - run
source env/bin/activate
(ordevenv shell
if you use devenv) - then run
python cli.py server
to listen to the MW changes
otherwise, if interested in 2:
cd
into this repo and activate the environment- run
python cli.py build-article
or any other command to build the entire wiki, a specific category page, the frontpage, etc.
type python cli.py --help
for a list of all the options.
scripts
there are two Python helper scripts to lint and format code:
- the former runs flake8 and reports you a list of suggestions;
- the latter runs isort and black to update the codebase by (1) re-sorting the list of imports at the top of a file, and (2) rewriting code in a specific style (eg using single quote, keeping correct whitespace between lines of code, etc).
you can run them manually by simple shell invocation:
./scripts/lint
./scripts/format
intro
this program acts as a static site builder for a MediaWiki instance. we define in settings.toml
the list of categories we use to fetch articles from the wiki, and output a folder of static HTML files.
a backend server, besides serving HTML files also run a search feature by interfacing with the MediaWiki APIs.
originally the idea was to export the MediaWiki's wikitext beside the HTML, but the extra complexity added to handle this while building a static website grew way bigger than imagined. so we took a step back and now rely on MediaWiki's APIs to retrieve data (article HTML, article metadata, working with images, etc.).
overall we still achieve the plan to generate a set of HTML files that can be easily backed up, parsed, re-generated, and so on.
mediawiki
a local instance of MediaWiki must be run, in order to have this software working correctly.
check this repo to set up a MW instance: https://github.com/hackersanddesigners/hd-mw
then make sure to add the following to the .env.php
file (RCFeeds
example):
$wgRCFeeds['exampleirc'] = array(
'formatter' => 'JSONRCFeedFormatter',
'uri' => 'udp://localhost:1338',
'add_interwiki_prefix' => false,
'omit_bots' => false,
'omit_anon' => false, # to detect wiki changes from scripted API requests
'omit_minor' => false,
'omit_patrolled' => false
);
local dev setup and TLS certificates
.env, settings.toml
this project needs two settings files to function:
.env
settings.toml
.env
rename env.sample
to .env
and fill out the file using this reference:
ENV
: set eitherdev
orprod
; this is mostly used to decide if using a local certificate when doing HTTP operation or notSERVER_IP
: set a host for theserver.py
function => eg.localhost
SERVER_PORT
: set a port number for theserver.py
function (ergo, opening a port to listen to UDP messages from the MediaWiki instance) => eg.1331
WIKI_DIR
: path to static HTML output folder. choose a name for it (eg.wiki
), create it, and set its name hereASSETS_DIR
: path to static folder: eg. CSS, JS, imagesMEDIA_DIR
: path for the media directory of WIKI_DIR => eg =><WIKI_DIR>/assets/media
LOG_DIR
: set the directory where to write log filesBASE_URL
: base API URL path => eg. for local setup:http://localhost/api.php?
; for an online wikihttps://wikixyz.tld/api.php?
we create a bot user to help programmatically creating, editing or deleting a wiki article. get credentials by visiting the Special:BotPasswords
page of your wiki. then:
BOT_USR
: use lgnameBOT_PWD
: use lgpasswordLOCAL_CA
: see below under local certificateSEMAPHORE
: number of max operations happening at the same time when doing async HTTP call. above this number the Python interpreter will throw an error. a good number is between 150-175, try and see what works.
local certificate
install mkcert or similar to create a local certificate.
for mkcert:
# if it is the first time, install it
mkcert -install
# then make local CA for current website, for instance
mkcert hd-v2.nl
(you can use a different name for the certificate, eg mkcert <name>
)
this will create two files in the current folder:
hd-v2.nl-key.pem
_hd-v2.nl.pem
add an entry to .env
with:
LOCAL_CA=hd-v2.nl.pem
after this you can use https also in the dev environment while using this codebase!
settings.toml
this file mainly set website preferences, eg general wiki options:
wiki.stylespage
: name of stylesheet pagewiki.langs
: list of wiki languages, including defaultwiki.frontpage.article
: which wiki article do we want to use for the website frontpage?wiki.frontpage.category
: which category do we want to fetch for the articles displayed in the frontpage?wiki.categories.<cat>
: sets a list of categories to define which wiki articles we want to display on the website. the<cat>
is the actual MediaWiki category we want to use. then, each category has some more options:parse
: should we parse it (eg download every article of that category) or not; useful when working on the codebase to speed up parsing process if usingpython cli.py build-wiki
, for instancenav
: should the category be displayed in the navigationindex
: should we build a category index page for itfallback
: using this category as fallback, in case the wiki article has no matching category with the given list of categorieslabel
: some categories in the wiki might be called something, and we might want to display them in a different way in the navigation, in the URL path, etc.
wiki.footer_links.<page>
(Conduct, Accessibility, Privacy Policy):nav
: add Conduct page to footer list of linkslabel
: set label for the above link
domain.canonical_url
: set default website URL
domain.mw_url
: set URL to the running MediaWiki instance we want to work with
and specific plugin options:
tool-plugin.host_default
: which git hosting service fallback is the H&D MW Tool plugin using?tool-plugin.host_default
: a dictionary of git hosting services the H&D MW Tool wants to usetool-plugin.branch_default
: a list of branches to use when parsing information from the git repo set in the plugin; this works as a progressive list of fallback branch names
commands
there is a CLI program at cli.py
to run common operations. currently available commands are:
build-article
<edit/delete>: helper function to trigger a change in the MediaWiki instance, instead of manually logging in to the MW editor and commit a change. the command takes two arguments:PageTitle
and type of operation (edit
,delete
); theedit
operation creates a new article if it does not exist yet, beside making a change to an existing wiki article.build-frontpage
: builds the frontpagebuild-category-index
: builds the given category index pagebuild-wiki
: builds the entire wiki, where by entire it's meant the list of articles with specific categories defined insettings.toml
server
: starts a local server and listen to specified port at UDP messages from the MediaWiki instancesetup
: creates the necessary folders to run the website
start off by running python cli.py --help
to see the available options.
to run a local web-server in order to browse the wiki
folder, you can do:
uvicorn app.main:app --reload
or uselocal-server.sh
.
License
This repository is published under the CC4r* license.