doc-agent / docagent /__init__.py
Sanidhya Ravi
Turn the TrOCR demo into an agentic document pipeline
5eb40dc
Raw History Blame Contribute Delete
916 Bytes
"""An agentic document-processing pipeline.
Drop in anything -- image, PDF, Office file, ZIP -- and the agent works out
what it is, reads it with the right model, and produces the structured output
that suits it: a spreadsheet for tabular data, JSON for a form, a summary PDF
for prose.
Typical use::
from docagent import process
doc = process("invoice.pdf", "pull the line items into a spreadsheet")
for artifact in doc.artifacts:
print(artifact.kind, artifact.path)
"""
from .agent import process, run, run_stream
from .ingest import load, supported_extensions
from .state import Artifact, Block, Box, DocKind, Document, Field, Page, Table
__version__ = "0.2.0"
__all__ = [
"process",
"run",
"run_stream",
"load",
"supported_extensions",
"Artifact",
"Block",
"Box",
"DocKind",
"Document",
"Field",
"Page",
"Table",
"__version__",
]