Damast
This repository contains code developed for the digital humanities project "Dhimmis & Muslims – Analysing Multireligious Spaces in the Medieval Muslim World". The project was funded by the VolkswagenFoundation within the scope of the "Mixed Methods" initiative. The project was a collaboration between the Institute for Medieval History II of the Goethe University in Frankfurt/Main, Germany, and the Institute for Visualization and Interactive Systems at the University of Stuttgart, and took place there from 2018 to 2021.
The objective of this joint project was to develop a novel visualization approach in order to gain new insights on the multi-religious landscapes of the Middle East under Muslim rule during the Middle Ages (7th to 14th century). In particular, information on multi-religious communities were researched and made available in a database accessible through interactive visualization as well as through a pilot web-based geo-temporal multi-view system to analyze and compare information from multiple sources. A publicly explorable version of the research will soon be available, and will be linked here. An export of the data collected in the project can be found in the data repository of the University of Stuttgart (DaRUS) (draft, not yet public).
Database
The historical data is collected in a relational PostgreSQL database. For the project, we have used PostgreSQL version 10. Since the project also deals with geographical data, we additionally use the PostGIS extension. A suitable database setup is to use the postgis/postgis:10-3.1 Docker image. An SQL script for creating the database schema is located in util/postgres/schema.sql, and in DaRUS. An overview of the interplay between tables of the database, and a general explanation, can be found in the docs/ directory.
Software
The server is programmed in Python using Flask. Functionalities are split up into a hierarchy of Flask blueprints; for example, there is a blueprints for the landing page, one for the visualization, and a nested hierarchy of blueprints for the REST API. The server provides multiple pages, as well as a HTTP interface for reading from and writing to the PostgreSQL database. The server is built and deployed as a Docker container that contains all necessary dependencies.
An overview and explanation of the different pages and functionalities is located in the docs/ directory. The web pages consist of HTML, CSS and JavaScript. The HTML content is in most cases served via Jinja2 templates that are processed by Flask. The JavaScript code is compiled from TypeScript source, and the CSS is compiled from SCSS.
Getting Started
Basic knowledge with build tools and Docker are required. The instructions below assume a Linux machine with the Bash shell are used.
Installing Dependencies
On the build system, Docker and NodeJS need to be installed. If the Makefile is used, the build-essentials package is required as well. In the root of the repository, run the following code to install the build dependencies:
$ npm install
Building the Frontend
To build all required files for the frontend (JavaScript, CSS, documentation), the Makefile can be used, or consulted for the appropriate commands:
$ make prod
Building the Docker Image
All frontend content and backend Flask code and contents are bundled within a Docker image. In that image, the required software dependencies are also installed in the correct versions. A few configuration options need to be baked into the Docker image on creation, which are dependent on the setup in which the Docker container will later run. Please refer to the deploy.sh shell script for details and examples, as well as to the section on running the server. The Dockerfile is constructed from parts from util/docker/, and then enriched with runtime information to ensure that certain steps are repeated when data changes. An exemplary creation of a Docker image (fictional values, please refer to deploy.sh before copying) could look as follows:
# calculate hash of server files to determine if the COPY instruction should be repeated
$ fs_hash=$(find dhimmis -type f \
| xargs sha1sum \
| awk '{print $1}' \
| sha1sum - \
| awk '{print $1}')
# assemble Dockerfile
$ cat util/docker/{base,prod}.in \
| sed "s/@REBUILD_HASH@/$fs_hash/g" \
> Dockerfile
# build Dockerfile (warning: dummy parameters!)
$ sudo docker build -t damast:latest \
--build-arg=USER_ID=50 \
--build-arg=GROUP_ID=50 \
--build-arg=DHIMMIS_ENVIRONMENT=PRODUCTION \
--build-arg=DHIMMIS_VERSION=v1.0.0 \
--build-arg=DHIMMIS_PORT=8000 \
.
The resulting Docker image can then be transferred to the host machine, for example, using docker save and docker load. Of course, the image can be built directly on the host machine as well.
Running the Server
The server infrastructure consists of three components:
- The Flask server in its Docker container,
- the PostgreSQL database, for example in the form of a
postgis/postgis:10-3.1Docker container, and - a reverse HTTP proxy on the host machine that handles traffic from the outside and SSL.
The util/ directory contains configuration templates for an NGINX reverse proxy, cron, the start script, the systemd configuration, and the user authentication file. The documentation also goes into more details about the setup. A directory on the host machine is mapped as a volume to the /data directory in the docker container. The /data directory contains runtime configuration files (users.db, reports.db, as well as log files). The main Docker container requires some additional runtime configuration, for example for the PostgreSQL password, which can be passed as environment variables to Docker using the --env and --env-file flags. The following configuration environment variables exist, although most have a sensible default:
| Environment Variable | Default Value | Description |
|---|---|---|
DHIMMIS_ENVIRONMENT |
Server environment (PRODUCTION, TESTING, or PYTEST). This decides with which PostgreSQL database to connect (ocn, testing, and pytest (on Docker container) respectively. This is usually set via the Docker image. |
|
DHIMMIS_VERSION |
Software version. This is usually set via the Docker image. | |
DHIMMIS_USER_FILE |
/data/users.db |
Path to SQLite3 file with users, passwords, roles. |
DHIMMIS_REPORT_FILE |
/data/reports.db |
File to which reports are stored during generation. |
DHIMMIS_SECRET_FILE |
/dev/null |
File with JWT and app secret keys. These are randomly generated if not passed, but that is impractical for testing with hot reload (user sessions do not persist). For a production server, this should be empty. |
DHIMMIS_PROXYCOUNT |
1 |
How many reverse proxies the server is behind. This is necessary for proper HTTP redirection and cookie paths. |
DHIMMIS_PROXYPREFIX |
/ |
Reverse proxy prefix. |
FLASK_ACCESS_LOG |
/data/access_log |
Path to access_log (for logging). |
FLASK_ERROR_LOG |
/data/error_log |
Path to error_log (for logging). |
DHIMMIS_PORT |
8000 |
Port at which gunicorn serves the content. Note: This is set via the Dockerfile, and also only used in the Dockerfile. |
PGHOST |
localhost |
PostgreSQL hostname. |
PGPASSWORD |
PostgreSQL password. This is important to set and depends on how the database is set up. | |
PGPORT |
5432 |
PostgreSQL port |
PGUSER |
api |
PostgreSQL user |
1 Dec 14, 2021
502 Jan 03, 2023
3 Dec 16, 2022
302 Dec 29, 2022
3 Apr 27, 2022
1 Jan 27, 2022
1.7k Jan 06, 2023
1 Feb 13, 2022
711 Dec 26, 2022
24 Nov 29, 2022
1 Jan 04, 2022
1.2k Dec 31, 2022
1 Jan 09, 2022
11 Dec 13, 2022
359 Dec 22, 2022
13 Dec 15, 2022
2 Jan 07, 2022
4 Jun 05, 2022
1 Jan 26, 2022
37 Dec 15, 2022