Metadata-Version: 2.4
Name: arispec
Version: 0.0.1
Summary: Anchor-relative identification: declarative extraction from irregular text and grids
Author-email: "Matthew C. Tedder" <matthewct@gmail.com>
License-Expression: Apache-2.0
Project-URL: Homepage, https://github.com/Solifugus/etl-tools
Project-URL: Roadmap, https://github.com/Solifugus/etl-tools/blob/master/docs/roadmap.md
Keywords: parsing,text,extraction,report,fixed-width,legacy,etl
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Financial and Insurance Industry
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
License-File: NOTICE
Dynamic: license-file

# arispec

Anchor-relative identification: declarative extraction from irregular text and grids

**Status: name reserved. Nothing is implemented yet.**

Declarative extraction of fields from irregular, semi-structured text: legacy
print reports, mainframe spool, teller totals, and the 2D grids that arrive as
spreadsheets.

ARI locates a value by anchoring on nearby landmarks and **delimiting it by
type** -- "the last money-shaped token on this row" -- rather than by a fixed
column or a column heading. Neither of those survives contact with real
reports: headings drift out of alignment with their own data over decades of
separate edits, and a trailing-minus negative is one column wider than the
positive above it because the sign is appended after right-justification.

The design and schedule live in [the roadmap](https://github.com/Solifugus/etl-tools/blob/master/docs/roadmap.md).

Apache-2.0. Copyright 2026 Matthew C. Tedder.
