# `_source.json` Sidecar Schema

Every downloaded file (or group of files from the same source) in CatholicCorpus has a companion `_source.json` sidecar that documents where it came from, who made it available, and how it may be used. This is the formal schema for those sidecars.

The machine-readable JSON Schema (draft 2020-12) is at [`schema.json`](schema.json).

## Fields

### Top-level

| Field | Type | Required | Description |
|-------|------|----------|-------------|
| `corpus` | string | Yes | Always `"CatholicCorpus.org"`. Identifies this file as part of the project. |
| `source_name` | string | Yes | Human-readable name of the source. |
| `source_url` | string (URI) | Yes | URL from which the content was downloaded. |
| `upstream_url` | string (URI) | Yes | Canonical URL of the original project. May equal `source_url`. |
| `description` | string | Yes | What this source contains and why it matters. |
| `language` | string | Yes | Primary language(s). See controlled values below. |
| `date_downloaded` | string (date) | Yes | ISO 8601 date of download (YYYY-MM-DD). |
| `license` | string | Yes | Short license identifier. See controlled values below. |
| `license_url` | string or null | Yes | URL to full license text. Null for public domain. |
| `license_details` | string | Yes | Extended explanation of the licensing situation. |
| `attribution_required` | boolean | Yes | Whether attribution must be given when using this content. |
| `attribution_text` | string or null | Yes | Copy-paste-ready attribution text. Null if not required. |
| `credit` | object | Yes | Structured credit information (see below). |
| `format` | string | Yes | File format(s) of the content. |
| `estimated_words` | string or number | Yes | Approximate word count. |
| `notes` | string or null | Yes | Additional context, or null. |
| `article_count` | string | No | Number of discrete articles (encyclopedia-type sources only). |

### The `credit` object

| Field | Type | Required | Description |
|-------|------|----------|-------------|
| `digitized_by` | string | Yes | Institution or person who created the digital version. |
| `digitizer_url` | string (URI) | Yes | URL of the digitization project. |
| `original_author` | string | Yes | Original author(s) of the text. |
| `translator` | string or null | No | Translator, if applicable. Null for original-language texts. |
| `editor` | string or null | No | Editor of the edition used. |
| `publisher` | string or null | No | Historical print publisher. |
| `publication_year` | string or null | No | Year or range of the print edition. |
| `thank_you` | string | Yes | Human-readable acknowledgment. |

## Controlled Vocabularies

### `license` values

| Value | Meaning |
|-------|---------|
| `Public Domain` | Work is in the public domain (pre-1928 U.S. publication or author deceased 70+ years). No restrictions on use. |
| `CC BY 4.0` | Creative Commons Attribution 4.0. Free to use with attribution. |
| `CC BY-NC-SA 4.0` | Creative Commons Attribution-NonCommercial-ShareAlike 4.0. |
| `CC BY 4.0 with additional EULA` | CC BY 4.0 with an additional end-user license for specific uses (e.g., SBLGNT commercial use restriction). |
| `Non-commercial use permitted` | May be used for non-commercial purposes. Check `license_details` for specifics. |
| `Copyrighted (not redistributable)` | Content is under copyright. Scripts to download it are included but the content itself is not in the corpus. |
| `Unclear` | License status could not be determined. Check `license_details` for what is known. |

### `language` values

The `language` field uses standard language names, not ISO codes. Common values in the corpus:

- `Latin`
- `Ancient Greek`
- `English`
- `Latin (primary), English (translations)`
- `Latin, Greek`
- `Multilingual`

For multilingual sources, the primary language is listed first.

## How Edition and Citation References Work

### Patrologia Latina references

For texts from the Migne Patrologia Latina (Task 02), the standard scholarly citation format is PL volume:column (e.g., PL 32:659 for Augustine's Confessions, Book 1, Chapter 1). The Corpus Corporum download preserves the internal structure of each text but does not embed column numbers in the TEI XML. Researchers needing precise PL column references should cross-reference with the original Migne volumes (available in Task 16 for the Graeca, or via the PDF volumes on archive.org for the Latina).

### Archive.org item IDs

For texts downloaded from archive.org, the `source_url` contains the archive.org identifier (e.g., `https://archive.org/details/summatheologica01thom`). This identifier can be used to access the original metadata, alternative formats, and lending status via the archive.org API.

## Three-Tier License Model

The corpus uses a three-tier approach:

1. **Tier 1 (Public Domain)**: The vast majority. No restrictions. Use freely.
2. **Tier 2 (Creative Commons)**: A small number of texts. Follow the CC license terms (attribute, and for NC licenses, do not use commercially).
3. **Tier 3 (Copyrighted)**: Task 17 only. Scripts are MIT-licensed; content is not redistributed. Users who run the scripts are responsible for their own compliance.

The `license` field in each sidecar tells you which tier a given source belongs to.

## Validation

To validate your sidecars against this schema:

```bash
pip install jsonschema
python3 -c "
import json, jsonschema, glob
schema = json.load(open('docs/schema.json'))
files = glob.glob('**/_source.json', recursive=True)
errors = 0
for f in files:
    try:
        jsonschema.validate(json.load(open(f)), schema)
    except jsonschema.ValidationError as e:
        print(f'{f}: {e.message}')
        errors += 1
print(f'Validated {len(files)} files, {errors} errors')
"
```