"""Tests for text and HTML cleaning."""

import unittest

from converter.extractors import clean_html, clean_text


class CleanTextTests(unittest.TestCase):
    def test_empty_input(self):
        self.assertEqual(clean_text(""), "")
        self.assertEqual(clean_text(None), "")

    def test_collapses_whitespace_inside_paragraphs(self):
        self.assertEqual(clean_text("a   b \t c"), "a b c")

    def test_paragraph_breaks_survive(self):
        # Blank lines are the smart chunker's paragraph boundaries (and
        # reset its quotation state), so they survive cleaning while a
        # run of blank lines collapses to a single break.
        self.assertEqual(clean_text("a\n\n   b \t c"), "a\n\nb c")
        self.assertEqual(clean_text("a\n\n\n\nb"), "a\n\nb")
        self.assertEqual(clean_text("a\n  \nb"), "a\n\nb")
        self.assertEqual(clean_text("line one\nline two"), "line one line two")
        self.assertEqual(clean_text("a \n\n b"), "a\n\nb")

    def test_paragraph_breaks_not_glued_to_text(self):
        self.assertEqual(
            clean_text("End of chapter.\n\n\n  New chapter.  \n\nStarts here."),
            "End of chapter.\n\nNew chapter.\n\nStarts here.",
        )

    def test_preserves_inline_numbers(self):
        self.assertEqual(clean_text("He was 42 years old."), "He was 42 years old.")

    def test_preserves_grouped_and_decimal_numbers(self):
        self.assertEqual(
            clean_text("Over 1,000 pages and 3.5 stars."),
            "Over 1,000 pages and 3.5 stars.",
        )

    def test_removes_standalone_page_numbers(self):
        # The page number's own line becomes a paragraph break (a safe
        # chunk boundary), not a glued sentence.
        self.assertEqual(
            clean_text("End of page.\n7\nNext page text."),
            "End of page.\n\nNext page text.",
        )

    def test_page_number_removal_leaves_paragraph_break(self):
        result = clean_text("Chapter one\n\n12\n\nChapter two")
        self.assertEqual(result, "Chapter one\n\nChapter two")
        self.assertNotIn("  ", result)


class CleanHtmlTests(unittest.TestCase):
    def test_strips_tags(self):
        self.assertEqual(clean_html("<p>Hello <b>world</b></p>"), "Hello world")

    def test_removes_script_and_style(self):
        html = "<style>.x{color:red}</style><p>Text</p><script>var a=1;</script>"
        self.assertEqual(clean_html(html), "Text")

    def test_unescapes_entities(self):
        self.assertEqual(clean_html("Tom &amp; Jerry"), "Tom & Jerry")

    def test_empty(self):
        self.assertEqual(clean_html(""), "")
        self.assertEqual(clean_html(None), "")

    def test_inline_markup_never_splits_words(self):
        # Inline tags must not inject spaces mid-word (they become chunk
        # boundaries and corrupt pronunciation).
        self.assertEqual(
            clean_html("<p>He was un<em>believ</em>able and didn<i>'</i>t stop.</p>"),
            "He was unbelievable and didn't stop.",
        )

    def test_block_tags_become_paragraph_breaks(self):
        self.assertEqual(
            clean_html("<p>First para.</p><p>Second para.</p><h2>Head</h2>"),
            "First para.\n\nSecond para.\n\nHead",
        )

    def test_div_sections_keep_paragraph_boundaries(self):
        html = ('<div>“An unfinished quotation.</div>'
                '<div>A new paragraph outside the quotation.</div>')
        self.assertEqual(
            clean_html(html),
            "“An unfinished quotation.\n\nA new paragraph outside the quotation.",
        )

    def test_table_cells_do_not_glue(self):
        self.assertEqual(
            clean_html("<table><tr><td>A</td><td>B</td></tr>"
                       "<tr><td>C</td><td>D</td></tr></table>"),
            "A\n\nB\n\nC\n\nD",
        )

    def test_regex_fallback_matches_bs4_behavior(self):
        html = "<p>un<em>believ</em>able</p><div>After a block.</div>"
        self.assertEqual(clean_html(html), "unbelievable\n\nAfter a block.")


if __name__ == "__main__":
    unittest.main()