[](https://travis-ci.com/eyeseast/feed-to-sqlite)
[](https://github.com/eyeseast/feed-to-sqlite/actions?query=workflow%3ATest)
[](https://pypi.org/project/feed-to-sqlite/)
[](https://github.com/eyeseast/feed-to-sqlite/blob/master/LICENSE)
# feed-to-sqlite
Download an RSS or Atom feed and save it to a SQLite database. This is meant to work well with [datasette](https://github.com/simonw/datasette).
## Installation
```sh
pip install feed-to-sqlite
```
## CLI Usage
Let's grab the ATOM feeds for [items I've shared on NewsBlur](http://chrisamico.newsblur.com/social/rss/35501/chrisamico) and [my instapaper favorites](https://www.instapaper.com/starred/rss/13475/qUh7yaOUGOSQeANThMyxXdYnho) save each its own table.
```sh
feed-to-sqlite feeds.db http://chrisamico.newsblur.com/social/rss/35501/chrisamico https://www.instapaper.com/starred/rss/13475/qUh7yaOUGOSQeANThMyxXdYnho
```
This will use a SQLite database called `feeds.db`, creating it if necessary. By default, each feed gets its own table, named based on a slugified version of the feed's title.
To load all items from multiple feeds into a common (or pre-existing) table, pass a `--table` argument:
```sh
feed-to-sqlite feeds.db --table links <url> <url>
```
That will put all items in a table called `links`.
Each feed also creates an entry in a `feeds` table containing top-level metadata for each feed. Each item will have a foreign key to the originating feed. This is especially useful if combining feeds into a shared table.
## Python API
One function, `ingest_feed`, does most of the work here. The following will create a database called `feeds.db` and download my NewsBlur shared items into a new table called `links`.
```python
from feed_to_sqlite import ingest_feed
url = "http://chrisamico.newsblur.com/social/rss/35501/chrisamico"
ingest_feed("feeds.db", url=url, table_name="links")
```
### Transforming data on ingest
When working in Python directly, it's possible to pass in a function to transform rows before they're saved to the database.
The `normalize` argument to `ingest_feed` is a function that will be called on each feed item, useful for fixing links or doing additional work.
It's signature is `normalize(table, entry, feed_details, client)`:
- `table` is a SQLite table ([from sqlite-utils](https://sqlite-utils.datasette.io/en/stable/python-api.html#accessing-tables))
- `entry` is one feed item, as a dictionary
- `feed_details` is a dictionary of top-level feed information, as a dictionary
- `client` is an instance of `httpx.Client`, which can be used for outgoing HTTP requests during normalization
That function should return a dictionary representing the row to be saved. Returning a falsey value for a given row will cause that row to be skipped.
## Development
Tests use [pytest](https://docs.pytest.org/). Run `pytest tests/` to run the test suite.
