Update export structure

content/project/export/index.md

@@ -7,6 +7,8 @@ Projects can be exported as SQLite databases. This page describes its structure
 
 To learn how to export a project, see [Export a project](@/howto/export/index.md).
 
+A [Python library](https://pypi.org/project/arkindex-export/) is available to help in using these exports within Python code.
+
 ## Database structure
 
 <figure>
@@ -20,7 +22,7 @@ To learn how to export a project, see [Export a project](@/howto/export/index.md
 
 Name | Type | Description
 ---- | ---- | -----------
-version | `INTEGER` | Version number of the database structure. Currently set to `3`.
+version | `INTEGER` | Version number of the database structure. Currently set to `8`.
 
 ### classification
 
@@ -36,6 +38,24 @@ high_confidence | `INTEGER` | Whether the worker states this class is the correc
 worker_version_id | `VARCHAR(37)` | UUID of a worker version that created this classification. Nullable.
 worker_run_id | `VARCHAR(37)` | UUID of a worker run that created this classification. Nullable.
 
+### dataset
+
+Name | Type | Description
+---- | ---- | -----------
+id | `VARCHAR(37)` | UUID of the dataset.
+name | `VARCHAR(100)` | Name of the dataset.
+state | `VARCHAR(50)` | State of the dataset. Either `open`, `building`, `complete` or `error`.
+sets | `TEXT` | Comma-separated list of allowed set names.
+
+### dataset_element
+
+Name | Type | Description
+---- | ---- | -----------
+id | `VARCHAR(37)` | UUID of the dataset-element relationship.
+dataset_id | `VARCHAR(37)` | UUID of the dataset the element belongs to.
+element_id | `VARCHAR(37)` | UUID of the element contained in the dataset.
+set_name | `VARCHAR(50)` | Name of the set within the dataset where the element is.
+
 ### element
 
 Name | Type | Description
@@ -168,8 +188,11 @@ id | `VARCHAR(37)` | UUID of the worker version.
 name | `VARCHAR(100)` | Human-readable name of the worker.
 slug | `VARCHAR(100)` | Slug of the worker.
 type | `VARCHAR(50)` | Type of the worker (recognizer, classifier, etc.).
-revision | `VARCHAR(50)` | Hash of the Git commit that created this worker version.
-repository_url | `TEXT` | URL of the Git repository that hosts this worker version.
+version | `INTEGER` | Version number for this worker version, when it does not have a Git commit. Nullable.
+revision | `VARCHAR(50)` | Hash of the Git commit that created this worker version. Nullable.
+repository_url | `TEXT` | URL of the Git repository that hosts this worker version. Nullable.
+
+Either `version`, or both `revision` and `repository_url` must be set. All three fields cannot be set at once.
 
 ### worker_run
 

content/project/export/structure.txt

@@ -26,3 +26,5 @@ graph LR
     metadata --> worker_run
     worker_run --> worker_version
     image --> image_server
+    dataset_element --> element
+    dataset_element --> dataset