A web scraper for chess coaches.
 
 
 
Go to file
Joshua Potter 1c0dc05b42 Separate initialization from loading. Prefer upserts. 2023-12-04 08:14:33 -07:00
.githooks Use more robust pre-commit. 2023-12-03 14:31:36 -07:00
app Apply pyls-isort. 2023-12-01 16:37:05 -07:00
sql Separate initialization from loading. Prefer upserts. 2023-12-04 08:14:33 -07:00
.envrc Initial commit. 2023-11-27 13:09:40 -07:00
.gitignore Rewrite export as NDJSON and include script to load result into postgres. (#3) 2023-12-01 10:30:44 -07:00
README.md Separate initialization from loading. Prefer upserts. 2023-12-04 08:14:33 -07:00
default.nix Initial commit. 2023-11-27 13:09:40 -07:00
flake.lock Initial commit. 2023-11-27 13:09:40 -07:00
flake.nix Apply pyls-isort. 2023-12-01 16:37:05 -07:00
poetry.lock Use lxml to speed up parsing. 2023-12-01 07:12:40 -07:00
pyproject.toml Use lxml to speed up parsing. 2023-12-01 07:12:40 -07:00

README.md

coach-scraper

Caution! Be careful running this script.

We intentionally delay each batch of requests. Make sure any adjustments to this script appropriately rate-limit.

Overview

This is a simple web scraper for coaches listed on:

The program searches for coach usernames as well as specific information about each of them (their profile, recent activity, and stats). The result will be found in a newly created data directory with the following structure:

data
└── <site>
│   ├── coaches
│   │   ├── <username>
│   │   │   ├── <username>.html
│   │   │   ├── export.json
│   │   │   └── ...
│   │   ├── ...
└── pages
    ├── <n>.txt
    ├── ...

Usage

If you have nix available, run:

$ nix run . -- --user-agent <your-email> -s <site>

If not, ensure you have poetry on your machine and instead run the following:

$ poetry run python3 -m app -u <your-email> -s <site>

Database

Included in the development shell of this flake is a Postgres client (version 15.5). Generate an empty Postgres cluster at /db by running

$ pg_ctl -D db init

To start the database, run the following:

$ pg_ctl -D db -l db/logfile -o --unix_socket_directories=@scraper start

In the above command, @scraper refers to an abstract socket name. Rename to whatever is appropriate for your use case. To then connect to this database instance, run:

$ psql -h @scraper

To later shut the database down, run:

$ pg_ctl -D db stop

Loading Data

To load all exported coach data into a local Postgres instance, use the provided sql/*.sql files. First initialize the export schema/table:

$ psql -h @scraper -f sql/init.sql

Next, concatenate all exported content and dump into the newly created table:

$ cat data/{chesscom,lichess}/export.json > data/export.json
$ psql -h @scraper -f sql/export.sql -v export="'$PWD/data/export.json'"

Re-running will automatically create backups and replace the coach data found in coach_scraper.export.

E2E

With the above section on loading files, we now have the individual components necessary to scrape coach data from our chess website and dump the results into the database in one fell swoop. Assuming our database is open with a socket connection available at @scraper:

nix run . -- --user-agent <your-email> -s chesscom -s lichess
cat data/{chesscom,lichess}/export.json > data/export.json
psql -h @scraper -f sql/init.sql
psql -h @scraper -f sql/export.sql -v export="'$PWD/data/export.json'"

Development

nix is used for development. The included flakes.nix file automatically loads in Python (version 3.11.6) with packaging and dependency management handled by poetry (version 1.7.0). direnv can be used to a launch a dev shell upon entering this directory (refer to .envrc). Otherwise run via:

$ nix develop

Language Server

The python-lsp-server (version v1.9.0) is included in this flake, along with the python-lsp-black and pyls-isort plugins. Additionally, pylsp is expected to be configured to use:

Refer to your editor for configuration details.

Formatting

Formatting depends on the black (version 23.9.1) tool. A pre-commit hook is included in .githooks that can be used to format all *.py files prior to commit. Install via:

$ git config --local core.hooksPath .githooks/

If running direnv, this hook is installed automatically when entering the directory.