lekhaka

lekhaka (लेखक, a scribe) turns a printed book of verses into its chapters. Each page is read by a vision model through vishalakshi, local or hosted, as blocks: verses with their printed numbers, headings, colophons. lekhaka gathers the verses into chapters, mends the numbering slips a reader makes, repairs malformed Devanagari akṣaras, and lists what a proofreader should check.

Install

pip install lekhaka

How to use

from lekhaka import read_book, book_md
from vishalakshi import Vault

b = read_book('gita.pdf', Vault('books.db'), model='claude/sonnet', title='Bhagavad Gita',
              hint='Each verse ends in ॥ n ॥.', counts=[47, 72])
b.chapters[0]['verses'][:2]     # [{n, text, page, heading}, ...]
b.probs                         # numbers mended, counts that differ, for a proofreader
print(book_md(b.chapters, 'Bhagavad Gita'))

With a vault the pages are kept as versions of one document: vault.proofread(b.doc_id, readers) has other models read them again, and vault.accept applies the changes the owner chooses. Without a vault nothing is stored.

The pieces work on their own, on the pages vishalakshi.read_doc returns or any {page, blocks} list:

On real books

nbs/samples holds a few pages each of three books vedicreader has read, as a model returned them. Sixteen pages of Varāhamihira’s Bṛhajjātaka assemble into its third and fourth chapters, each opened by its heading and closed by its colophon:

import json
bj = json.load(open('samples/brihat_jataka.json'))
chs, probs = assemble(bj)
[(c['heading'], len(c['verses']), (c['colophon'] or '')[:40]) for c in chs]

[('वियोनिजन्माध्यायः', 8, 'इति श्रीवराहमिहिरकृते बृहज्जातके वियोनिज'),
 ('निषेकाध्यायः', 22, 'इति श्रीवराहमिहिरकृते बृहज्जातके निषेकाध')]

probs

[]

Its colophons number the chapters, so assemble_colophon can cut there instead, reading each chapter’s number and name from its colophon; cut lays the chapters into titles and numbering reports a run’s numbers:

cc, _ = assemble_colophon(bj)
[(c['part'], c['no'], c['name'], len(c['verses'])) for c in cc]

[('पूर्व', 3, 'वियोनिजन्म', 8), ('पूर्व', 4, 'निषेक', 22)]

[t['title'] for t in cut(cc, 'chapters', book='बृहज्जातकम्')], numbering(cc[1]['verses'])

(['बृहज्जातकम् १ वियोनिजन्म', 'बृहज्जातकम् २ निषेक'],
 {'numbered': 22,
  'unnumbered': 0,
  'skipped': [],
  'repeated': [],
  'first': 1,
  'last': 22})

The Vidhānamālai prints each verse’s number at the end of its urai, the prose gloss after it. The urai goes into its verse’s notes, and gives the verse its number:

vm = json.load(open('samples/vidhanamalai.json'))
chs, probs = assemble(vm, notes=('urai',), number_from_note=True)
v = chs[0]['verses'][0]
len(chs[0]['verses']), v['n'], v['text'].split('\n')[0], v['notes']['urai'][0][:60]

(11,
 1,
 'பூமகள் கேள்வன்பொன் னுடையன் புட்கொடி யோன் புனித,',
 '(என்றது.) ஸ்ரீமகாலக்ஷ்மிக்கு நாயகரும், பீதாம்பரத்தையுடையவரும')

Uromarishi’s Pañcapakṣi (1917) numbers its verses in Tamil numerals with the old signs for ten and a hundred:

uro = json.load(open('samples/uromarishi.json'))
[(b['n'], vnum(b['n'])) for p in uro for b in p['blocks'] if '௱' in str(b.get('n') or '')][:4]

[('௨௱௫', 205), ('௨௱௬', 206), ('௨௱௭', 207), ('௨௱௮', 208)]

The pieces

• vnum, mend, renumber, assemble, assemble_colophon, check_counts: numbers and chapters. A page’s blocks may sit under blocks or items; a head is a heading (KINDS); join_cont rejoins a block a page break cut.
• apply_fixes, number_runs, borrow_numbers, cut_by_plan, drop_pages: what a reading gets wrong, put right before assemble. fix_chapters, sutras: after it.
• fix_aksharas, bad_aksharas, norm, virama: malformed akṣaras and typing slips in Devanagari.
• cut, numbering: verses laid into titles. witness_verses, text_changes, ref_check, move_lines: a reading checked against another edition. verse_text, disagree, compare: two readings of a page compared.
• read_book, book_md: a book end to end, through vishalakshi.

fix_aksharas('ध्ार्म'), vnum('२६॥२७'), vnum('२६॥ १', pick='first')

('धार्म', 27, 26)