import tempfile import unittest from pathlib import Path from snippets.fichiers import Extrait, decouper_markdown, decouper_pdf, ecrire_textes, lire_textes TOC = ("Contents\nInstall . . . . . . 1\n Does it start? . . . . . 1\n How long does the\n battery last? . . . . 2\n" "Data . . . . 3 Which formats? . . . . 3\n") CORPS = ("Install\nDoes it start?\nCheck power.\nRevision 1.5\n2\nAIMsight FAQ\nCheck fuse.\n" "How long does the battery last?\nAbout 3 days: � sun; � mains.\nData\nWhich formats?\nCSV.\n") PARASITES = ["Revision 1.5", "AIMsight FAQ"] class TestDecouperPdf(unittest.TestCase): def test_un_extrait_par_question_avec_corps(self): x = decouper_pdf(TOC + CORPS, PARASITES) self.assertEqual([e.titre for e in x], ["Does it start?", "How long does the battery last?", "Which formats?"]) def test_parasites_et_numeros_de_page_retires(self): x = decouper_pdf(TOC + CORPS, PARASITES) self.assertEqual(x[0].contenu, "Check power.\nCheck fuse.") def test_page_precedee_d_un_joint_invisible(self): # pdftotext rend « . . . ⁠1 » (U+2060) dans les vrais PDF x = decouper_pdf(TOC.replace(" 1\n", " ⁠1\n") + CORPS, PARASITES) self.assertEqual(x[0].titre, "Does it start?") def test_titre_du_corps_sur_deux_lignes(self): # pdftotext coupe les longues questions dans le corps corps = CORPS.replace("How long does the battery last?\n", "How long does the battery\nlast?\n") x = decouper_pdf(TOC + corps, PARASITES) self.assertEqual([e.titre for e in x], ["Does it start?", "How long does the battery last?", "Which formats?"]) self.assertEqual(x[0].contenu, "Check power.\nCheck fuse.") def test_saut_de_page_dans_la_table(self): # « Revision 1.5 / 1 / AIMsight FAQ » entre deux entrées de la table toc = TOC.replace("Data . . . . 3 ", "Data . . . . 3\nRevision 1.5\n1\nAIMsight FAQ\n") self.assertIn("Which formats?", [e.titre for e in decouper_pdf(toc + CORPS, PARASITES)]) def test_puces(self): self.assertIn("\n- sun;\n- mains.", decouper_pdf(TOC + CORPS, [])[1].contenu) class TestDecouperMarkdown(unittest.TestCase): def test_markdown_court_un_extrait(self): x = decouper_markdown("# Le cycle thermique\n\nLe jour, ...\n", "labeling/connaissances/cycle-thermique.md") self.assertEqual([(e.titre, e.source) for e in x], [("Le cycle thermique", "labeling/connaissances/cycle-thermique.md")]) def test_markdown_long_par_section(self): texte = "# Doc\n\n## A\n\n" + "a" * 4000 + "\n\n## B\n\n" + "b" * 4000 + "\n" self.assertEqual([e.titre for e in decouper_markdown(texte, "x.md")], ["Doc — A", "Doc — B"]) class TestCorpus(unittest.TestCase): def test_aller_retour_des_textes(self): with tempfile.TemporaryDirectory() as t: d = Path(t) x = [Extrait("Which formats?", "CSV.", "AIMsight FAQ, rév. 1.5")] ecrire_textes(x, d) self.assertEqual(lire_textes(d), x) self.assertEqual((d / "which-formats.md").read_text().splitlines()[-1], "Source : AIMsight FAQ, rév. 1.5") if __name__ == "__main__": unittest.main()