AFRP-History

The One Line — The Ramallah Family

David Saah, PhD · John Mogannam

The whole history of the Ramallah family as a single unbroken timeline — from Adam and Eve to today, with the evidence attached at every date. Compiled for the American Federation of Ramallah, Palestine.

This repository builds the book. It is not a document with a build script bolted on; the book is the program's output, and every figure, map, chart, citation and correction in it is generated from source at build time.

python build.py          # → out/AFRP_The_One_Line_v5.html

The build is deterministic: the same sources produce byte-identical HTML.

---

What is in the book

Sixty-six generations from Adam to Rāshid al-Ḥaddādīn, thirty-six of them named, thirty with no name at all — then about eighteen more to this morning. It runs as one chronological line through fourteen eras of Palestinian history, and the family's own chain of nine links is threaded through it, era by era, saying at each point which link we have reached and what evidence there is for it. Often the honest answer is none, and the book says so.

• 38 figures, all generated: Lambert conformal conic maps drawn from Natural Earth, the generational ladder, the register comparison, the evidence chain.
• 223 numbered notes, renumbered into document order at build time.
• A test harness: thirty-six checkable claims from the family's own presentation, each with a verdict, so the audit can be re-run by anyone.
• Corrections printed in the open, at the same size as the claims they replace, with the superseded reading quoted.

---

Layout

build.py              the entry point
src/
  paths.py            where everything lives
  build_v5.py         the book: entries, eras, assembly, CSS
  build_recon.py      the base figure library and GENS (the chart's 36 rows)
  build_figs2.py      the void, the diaspora, the population plates
  build_figs3.py      the chain of nine links, the town plan, the Azd dispersal
  build_figs4.py      the deep past, the branch at Eber, the ladder,
                      the Karak wheel, the register comparison
  build_atlas.py      the cartography — Sheet, fit_box, Sutherland–Hodgman clip
  build_timeline_test.py  the test harness
  build_defters.py    the 1553 and 1562 register names, encoded
assets/               inputs the build reads (the bibliography, the OCR index)
data/                 machine-readable outputs (the register names as CSV)
out/                  the built book
qa/                   the QA suite
docs/                 method, changelog, open questions

On a telephone

The book is one page, and on a phone it is about four hundred thousand pixels tall. Three things make that work:

• The plates keep a floor width and scroll sideways. An 860-unit sheet poured into a 390px screen is a scale of 0.36, and a 9pt label becomes 3pt — present and unreadable. On a narrow screen each plate holds 720px and the reader swipes it, which is what one does with a large map on a small table. The figure number says so: Figure 18 · swipe the plate →.
• content-visibility:auto on entries and figures, so the browser lays out and paints only what is near the viewport. First load drops from 1.4s to 0.4s. Find-in-page still reaches inside; contain-intrinsic-size keeps the scrollbar honest to within about three per cent.
• Footnote jumps are instantaneous on mobile. A note can be three hundred thousand pixels away, and animating that is several seconds of the page flying past.

GitHub Pages serves the file gzipped: about 360 KB over the wire.

The QA suite

python qa/qa_final.py    # overflow at 1280/900/420, page errors, broken anchors
python qa/qa_ov.py       # text-overlap detection inside every figure
python qa/qa_cover.py    # text painted over by a later opaque shape

qa_ov.py walks every <text> node in all 38 figures, takes its getBBox(), and reports any two that overlap. It should print [].

---

Method

Four standing rules, set out in full in docs/METHOD.md:

1. The co-authors' own chart is a guide, not a source. A chart cannot be its own evidence, least of all when the people printing it drew it. Every correction to it in this book is a self-correction.
2. Palestinian and Arab scholarship first. Eastern sources are cited as the source; Western editors are cited as the route.
3. Print uncertainty rather than smooth it. Where the record corrects the family, the correction appears in a marked box at the date where it happens.
4. Cite everything — author, work, year, page, DOI or URL.

---

Sources

The library this book is built on — some ninety PDFs of primary and secondary sources, the Ottoman register facsimiles, and the OCR of ʿAzīz Shāhīn's Kashf al-Niqāb (1982) — is not in this repository. Most of it is not ours to redistribute. docs/SOURCES.md lists what the build expects and where it came from.

Contributing

Corrections are the point. If you can show that something here is wrong, open an issue with the source — author, work, year, page. The book is designed to absorb that: it prints its own retractions.

Open questions are in docs/OPEN_QUESTIONS.md. The two that matter most both need somebody in a Turkish archive.

---

Licence

The code in this repository is MIT — see LICENSE.

The book is the authors'. The sources quoted in its notes — published scholarship, the Ottoman register facsimiles, the Karak wheel — belong to their own rights-holders and are cited, not redistributed. See docs/SOURCES.md.