Ty_Tech/old-books
6
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Structure: fuzzy CHAPTER keyword for OCR, cap run-on titles, prefer contents-page title when it has fewer unknown words; metadata: multi-resolution OCR of title pages, OCR name correction from repeated words; ingest: qpdf optional

Browse Source
This commit is contained in:
Audiobook Producer
2026-10-18 13:51:40 -05:00
parent 2e33a24ea3
commit a1138b016a
4 changed files with 113 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
_pipeline/tests/test_structure.py
+24
View File
@@ -48,6 +48,30 @@ class HeadingTests(unittest.TestCase):
self.assertTrue(got[0]["synthetic"])
self.assertEqual((got[0]["unit"], got[0]["number"]), (2, 1))
def test_ocr_damaged_chapter_word_is_recognised(self):
units = [unit("CHAPTER I.\nOne\n\n" + BODY), unit("OUAPTER II.\nTwo\n\n" + BODY),
unit("CUAPTER TX.\nNine?\n\n" + BODY), unit("CIEAPTER NEL,\nTwelve?\n\n" + BODY)]
got = structure.find_headings(units)
self.assertEqual([h["number"] for h in got], [1, 2, 3, 4]) # unreadable numerals count up in sequence
self.assertFalse(structure.is_chapter_word("CHAPTERS"))
self.assertFalse(structure.is_chapter_word("CAPTAIN"))
self.assertFalse(structure.is_chapter_word("SCIENCE"))
def test_long_title_without_blank_line_keeps_only_the_upper_case_lines(self):
body = "the Yogis not only bring about desired mental qualities and properties by will-power coupled with rhythmic breathing, but they also develop"
units = [unit("CHAPTER XVI.\nYOGI SPIRITUAL BREATHING.\n" + body + "\n" + body + "\n\n" + BODY)]
h = structure.find_headings(units)[0]
self.assertEqual(h["title"], "Yogi Spiritual Breathing.")
self.assertEqual(h["title_end"], 2)
def test_contents_in_chapter_line_form(self):
units = [unit("CONTENTS\nCHAPTER I.\nBreath Is Life........ 8\nCHAPTER II.\nThe Four Methods of Respiration ..... 27\n"), unit("CHAPTER I.\nX\n\n" + BODY)]
toc = structure.toc_chapter_lines(units, 1)
self.assertEqual(toc[2][0], "The Four Methods of Respiration")
self.assertEqual(toc[1][0], "Breath Is Life")
self.assertEqual(structure.unknown_words("The Four Metiiods of Respiration"), 1)
self.assertEqual(structure.unknown_words("The Four Methods of Respiration"), 0)
def test_nothing_readable_gives_no_headings(self):
self.assertEqual(structure.find_headings([unit(""), unit("x")]), [])