Storage

from aryn_sdk.client.client import Client

docset = client.create_docset(name="My DocSet")
docset_id = docset.docset_id

docset = client.get_docset(docset_id="your-docset-id")

docsets = client.list_docsets().get_all()
for docset in docsets:
    print(f"DocSet: {docset.name}")

client.delete_docset(docset_id="your-docset-id")

doc = client.add_doc(file="/path/to/myfile.pdf", docset_id="your-docset-id")

docs = client.list_docs(docset_id="your-docset-id")
for doc in docs:
    print(f"Document: {doc.name}")

doc = client.get_doc(docset_id="your-docset-id", doc_id="your-doc-id")

client.delete_doc(docset_id="your-docset-id", doc_id="your-doc-id")

output = "output.pdf"
client.get_doc_binary(docset_id="your-docset-id", doc_id="your-doc-id", file=output)

with open("my-favorite-pdf.pdf", "rb") as f:
    res = client.add_doc_async(file=f, docset_id="your-docset-id")
    print(f"Task ID: {res.task_id}")
    doc_metadata = res.result().value
    print(f"Document ID: {doc_metadata.doc_id}")

{
    "task_id": "aryn:t-47gpd3604e5tz79z1jro5fc"
}

import time

with open("/path/to/pdf", "rb") as f:
    res = client.add_doc_async(file=f, docset_id=docset_id)
    task_id = res.task_id

while True:
    result = client.get_async_result(task_id)
    if result.status_code == 202:
        print("Task still pending...")
    else:
        print(f"Task completed with status code: {result.status_code}")
        print(f"Result: {result.value}")
        break
    time.sleep(1)

{
    "task_status":"done",
    "result": ...
}

with open("/path/to/pdf", "rb") as f:
    task = client.add_doc_async(file=f, docset_id=docset_id)
    task_id = task.task_id
    print(f"Cancelling task {task_id}")
    res = client.cancel_async_task(task)
    if res.status_code == 200:
        print("Cancel successful.")

tasks = client.list_async_tasks().value
print(tasks.tasks)

{
    "aryn:t-sc0v0lglkauo774pioflp4l": {
    "task_status": "pending"
},
    "aryn:t-b9xp7ny0eejvqvbazjhg8rn": {
    "task_status": "pending"
}
}

from aryn_sdk.types import ReplaceOperation

updates = [
    ReplaceOperation(
        path="/properties/status",
        value="reviewed"
    )
]
client.update_doc_properties(
    docset_id="your-docset-id",
    doc_id="your-doc-id",
    operations=updates
)

from aryn_sdk.types.schema import Schema, SchemaField

schema = Schema(fields=[
    SchemaField(name="category", field_type="string")
])
client.extract_properties(docset_id="your-docset-id", schema=schema)

client.delete_properties(docset_id="your-docset-id", schema=schema)

from aryn_sdk.client import Client
client = Client(region="US", timeout=240.0)