# JSON

??? abstract "TLDR"

    ```{ .python .no-check }

    import edsnlp

    stream = edsnlp.data.read_json(path, converter="omop")

    stream = stream.map_pipeline(nlp)

    res = stream.to_json(path, converter="omop")

    # or equivalently

    edsnlp.data.to_json(stream, path, converter="omop")

    ```

We provide methods to read and write documents (raw or annotated) from and to json files.

As an example, imagine that we have the following document that uses the OMOP schema

```{ title="data.jsonl" }

{ "note_id": 0, "note_text": "Le patient ...", "note_datetime": "2021-10-23", "entities": [...] }

{ "note_id": 1, "note_text": "Autre doc ...", "note_datetime": "2022-12-24", "entities": [] }

...

```

You could also have multiple `.json` files in a directory, the reader will read them all.

## Reading JSON files {: #edsnlp.data.json.read_json }

::: edsnlp.data.json.read_json

    options:

        heading_level: 3

        show_source: false

        show_toc: false

        show_bases: false

## Writing JSON files {: #edsnlp.data.json.write_json }

::: edsnlp.data.json.write_json

    options:

        heading_level: 3

        show_source: false

        show_toc: false

        show_bases: false