Old Assyrian / Old Babylonian with Text‑Fabric — Colab Notebook (README)

Audience: Assyriology PhD students (new to programming).
Notebook: Text_Fabric_OA_OB_Colab.ipynb (Colab‑ready) / Text_Fabric_OA_OB_notebook.ipynb (local Jupyter).

Purpose

This notebook introduces the Text‑Fabric (TF) data model and APIs for exploring Old Assyrian (OA) and Old Babylonian (OB) corpora. You will learn how to load a TF “app,” inspect node types and features, traverse text structure (documents → lines → words), run searches, and export results (e.g., node IDs and line text) to CSV for downstream analysis.

Learning goals

• Understand TF’s core objects A, F, L, T, TF and the idea of annotated text as a graph (nodes + features).
• Navigate hierarchy: e.g., L.u(node, otype="line") (up to line), L.d(line, otype="w") (down to words).
• Interpret node types used in OA/OB: doc/text (document level), line, w (word), and wdiv (word divider).
• Use features such as docnumber, pnumber, lnno, ARK, and render text with T.text(...) / display with A.plain(...).
• Export structured data to CSV (e.g., one ID per line, including document and line references).

Quick start

1) Environment

• Colab: open the Colab notebook and run the first cell to install packages (internet required).
• Local Jupyter (Python ≥ 3.10):

  pip install text-fabric

2) Load an app

from tf.app import use
A = use("oldassyrian", hoist=globals())  # or "oldbabylonian"

hoist=globals() gives you the standard handles: A (app), F (features), L (locality/up/down), T (text), TF (metadata).

3) Inspect corpus structure

F.otype.all             # available node types
list(TF.features)[:10]  # some feature names

4) Basic traversal & export (example: wdiv → line)

import csv

ids = sorted(F.type.s("wdiv"))          # all word-divider nodes
rows = []
for n in ids:
    ln = L.u(n, otype="line")[0]        # parent line
    doc = (L.u(ln, otype="doc") or L.u(ln, otype="text") or [None])[0]
    rows.append({
        "wdiv_node_id": n,
        "line_node_id": ln,
        "docnumber": F.docnumber.v(doc) if doc else None,
        "lnno": F.lnno.v(ln),
        "ARK": F.ARK.v(doc) if doc else None,
    })

with open("wdiv_index.csv", "w", newline="", encoding="utf-8") as f:
    w = csv.DictWriter(f, fieldnames=rows[0].keys())
    w.writeheader(); w.writerows(rows)

Notes for newcomers to Python

• Run a cell with Shift+Enter. If variables “disappear,” restart the kernel and run cells from the top.
• Many TF queries return sets; sort them for reproducibility: sorted(...).
• CSV writing requires import csv. Pandas alternative:

  import pandas as pd
  pd.DataFrame(rows).to_csv("wdiv_index.csv", index=False)

Common pitfalls & quick fixes

• NameError: name 'csv' is not defined → add import csv (or use Pandas).
• %load_ext autoreload fails on Python 3.12 (missing imp) → either upgrade IPython (>= 8.25) or inject a one‑cell shim:

  import types, importlib, sys
  imp = types.ModuleType("imp"); imp.reload = importlib.reload; sys.modules["imp"] = imp
  %load_ext autoreload
  %autoreload 2

Citing & attribution

Please cite Text‑Fabric and the Old Assyrian TF app when you publish results based on this notebook.

• Dirk Roorda, Text‑Fabric v8.3.4: Text‑Fabric with a new display algorithm, Zenodo (2020). DOI: 10.5281/zenodo.3909582.
• Nino‑cunei/oldassyrian: First public version, Zenodo (2020).
• Text‑Fabric documentation: https://annotation.github.io/text-fabric/tf/
• Tutorial foundation: Text‑Fabric Old Assyrian tutorial (nbviewer link in the notebook comments).

Additional attribution: This notebook was adapted and extended by the UC Berkeley team — Micaela Montes and Adam Anderson — building on the Text‑Fabric OA/OB environment by Dirk Roorda and collaborators.

Acknowledgments

We gratefully acknowledge the TF ecosystem and the OA/OB corpora providers that made this computational exploration possible. This teaching material is intended as an on‑ramp for Assyriologists to engage directly with annotated cuneiform corpora in Python.

License

Data, text, and documentation: Dedicated to the public domain under CC0 1.0 Universal.

Code (Python scripts, code cells, utilities): Licensed under the MIT License.

SPDX identifiers:

• Data/text/docs: CC0-1.0
• Code: MIT