import contextlib import inspect import io import os import sys import tempfile import unittest from pathlib import Path from types import SimpleNamespace from unittest import mock SCRIPT_DIR = Path(__file__).resolve().parents[1] / "scripts" if str(SCRIPT_DIR) not in sys.path: sys.path.insert(0, str(SCRIPT_DIR)) import merge_and_build # noqa: E402 class GenerateFormatTests(unittest.TestCase): def _write_file(self, path, content="data"): Path(path).write_text(content, encoding="utf-8") def _set_mtime(self, path, timestamp): os.utime(path, (timestamp, timestamp)) def test_skips_when_output_is_up_to_date(self): with tempfile.TemporaryDirectory() as temp_dir: html_file = os.path.join(temp_dir, "book_doc.html") output_file = os.path.join(temp_dir, "book.epub") self._write_file(html_file, "") self._write_file(output_file, "epub") self._set_mtime(html_file, 100) self._set_mtime(output_file, 200) with mock.patch.object(merge_and_build.subprocess, "run") as run_mock: result = merge_and_build.generate_format( html_file, temp_dir, ".epub", "zh-CN" ) self.assertEqual(result, output_file) run_mock.assert_not_called() def test_rebuilds_when_image_assets_are_newer(self): with tempfile.TemporaryDirectory() as temp_dir: html_file = os.path.join(temp_dir, "book_doc.html") output_file = os.path.join(temp_dir, "book.epub") images_dir = os.path.join(temp_dir, "images") image_file = os.path.join(images_dir, "cover.jpg") os.makedirs(images_dir, exist_ok=True) self._write_file(html_file, "") self._write_file(output_file, "epub") self._write_file(image_file, "image") self._set_mtime(html_file, 100) self._set_mtime(output_file, 200) self._set_mtime(image_file, 300) with mock.patch.object( merge_and_build.subprocess, "run", return_value=SimpleNamespace(stdout="", stderr=""), ) as run_mock: result = merge_and_build.generate_format( html_file, temp_dir, ".epub", "zh-CN" ) self.assertEqual(result, output_file) run_mock.assert_called_once() cmd = run_mock.call_args.args[0] self.assertEqual(cmd[0], "python3") self.assertEqual(cmd[2], html_file) self.assertEqual(cmd[4], output_file) @unittest.skipUnless( "cover" in inspect.signature(merge_and_build.generate_format).parameters, "cover parameter unavailable", ) def test_rebuilds_epub_when_cover_is_explicitly_requested(self): with tempfile.TemporaryDirectory() as temp_dir: html_file = os.path.join(temp_dir, "book_doc.html") output_file = os.path.join(temp_dir, "book.epub") cover_file = os.path.join(temp_dir, "cover.jpg") self._write_file(html_file, "") self._write_file(output_file, "epub") self._write_file(cover_file, "image") self._set_mtime(html_file, 100) self._set_mtime(output_file, 200) self._set_mtime(cover_file, 300) with mock.patch.object( merge_and_build.subprocess, "run", return_value=SimpleNamespace(stdout="", stderr=""), ) as run_mock: result = merge_and_build.generate_format( html_file, temp_dir, ".epub", "zh-CN", cover=cover_file ) self.assertEqual(result, output_file) run_mock.assert_called_once() cmd = run_mock.call_args.args[0] self.assertIn("--cover", cmd) self.assertIn(cover_file, cmd) class MissingCoverPathTests(unittest.TestCase): @unittest.skipUnless( "cover" in inspect.signature(merge_and_build.generate_format).parameters, "cover parameter unavailable", ) def test_main_rejects_missing_cover_path(self): with tempfile.TemporaryDirectory() as temp_dir: missing_cover = os.path.join(temp_dir, "missing-cover.jpg") with mock.patch.object( merge_and_build, "load_config", return_value={} ), mock.patch.object( merge_and_build, "get_lang_config", return_value={"lang_attr": "zh-CN"} ), mock.patch.object( merge_and_build, "merge_markdown_files", return_value=True ), mock.patch.object( merge_and_build, "convert_md_to_html", return_value=True ), mock.patch.object( merge_and_build, "add_toc", return_value=True ), mock.patch.object( merge_and_build, "generate_formats" ) as generate_formats_mock, mock.patch.object( sys, "argv", ["merge_and_build.py", "--temp-dir", temp_dir, "--cover", missing_cover] ): with self.assertRaises(SystemExit) as exc: merge_and_build.main() self.assertNotEqual(exc.exception.code, 0) generate_formats_mock.assert_not_called() class ExportAliasTests(unittest.TestCase): def _write(self, path, content="x"): Path(path).write_text(content, encoding="utf-8") def test_export_named_aliases_copies_canonical_outputs(self): with tempfile.TemporaryDirectory() as temp_dir: for name in ["book.html", "book_doc.html", "book.docx", "book.epub", "book.pdf"]: self._write(Path(temp_dir) / name, name) copied = merge_and_build.export_named_aliases(temp_dir, "Translated Book") self.assertEqual( set(copied), { "Translated Book.html", "Translated Book_doc.html", "Translated Book.docx", "Translated Book.epub", "Translated Book.pdf", }, ) self.assertEqual( (Path(temp_dir) / "Translated Book.epub").read_text(encoding="utf-8"), "book.epub", ) self.assertTrue((Path(temp_dir) / "book.epub").exists()) def test_export_name_rejects_paths(self): with self.assertRaises(ValueError): merge_and_build.export_named_aliases("/tmp", "../bad") class MergeBlankOutputTests(unittest.TestCase): """A whitespace-only output chunk must abort the merge instead of being silently dropped from the final book.""" def _write(self, path, content): Path(path).write_text(content, encoding="utf-8") def _workspace(self, tmp): from manifest import create_manifest temp_dir = Path(tmp) self._write(temp_dir / "input.md", "One.\n\nTwo.\n") self._write(temp_dir / "chunk0001.md", "One.\n") self._write(temp_dir / "chunk0002.md", "Two.\n") self._write(temp_dir / "output_chunk0001.md", "一。\n") self._write(temp_dir / "output_chunk0002.md", "二。\n") create_manifest( str(temp_dir), ["chunk0001.md", "chunk0002.md"], str(temp_dir / "input.md"), ) return temp_dir def _merge(self, temp_dir): buf = io.StringIO() with contextlib.redirect_stdout(buf): ok = merge_and_build.merge_markdown_files(str(temp_dir)) return ok, buf.getvalue() def test_manifest_merge_fails_on_blank_output(self): with tempfile.TemporaryDirectory() as tmp: temp_dir = self._workspace(tmp) self._write(temp_dir / "output_chunk0002.md", "\n \n") ok, out = self._merge(temp_dir) self.assertFalse(ok) self.assertFalse((temp_dir / "output.md").exists()) self.assertIn("Blank output", out) def test_legacy_merge_fails_on_blank_output(self): with tempfile.TemporaryDirectory() as tmp: temp_dir = self._workspace(tmp) (temp_dir / "manifest.json").unlink() self._write(temp_dir / "output_chunk0002.md", "\n \n") ok, out = self._merge(temp_dir) self.assertFalse(ok) self.assertFalse((temp_dir / "output.md").exists()) self.assertIn("Blank output", out) def test_merge_succeeds_with_substantive_outputs(self): with tempfile.TemporaryDirectory() as tmp: temp_dir = self._workspace(tmp) ok, _ = self._merge(temp_dir) self.assertTrue(ok) merged = (temp_dir / "output.md").read_text(encoding="utf-8") self.assertIn("一。", merged) self.assertIn("二。", merged) class ImageValidationTests(unittest.TestCase): """Validates _validate_chunk_images, _check_generated_html_sanity, and the basic-regex alt-escape fix. Together these guard against subagent-produced malformed tags surviving into the final HTML.""" def _write(self, path, content): Path(path).write_text(content, encoding="utf-8") def _run_validator(self, temp_dir): buf = io.StringIO() with contextlib.redirect_stdout(buf): ok = merge_and_build._validate_chunk_images(temp_dir) return ok, buf.getvalue() def _run_html_sanity(self, html_path): buf = io.StringIO() with contextlib.redirect_stdout(buf): ok = merge_and_build._check_generated_html_sanity(html_path) return ok, buf.getvalue() def test_passes_for_clean_chunks(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Hello A and ![fig](images/b.png).', ) self._write( Path(temp_dir) / "output_chunk0001.md", 'Translated 译 and ![图](images/b.png).', ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_fails_on_unescaped_double_quote_in_alt(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Bottle Drink Me bottle', ) self._write( Path(temp_dir) / "output_chunk0001.md", '瓶子 标着', ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("output_chunk0001.md", out) self.assertIn("malformed ", out) def test_passes_for_curly_quote_alt(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Drink Me bottle', ) self._write( Path(temp_dir) / "output_chunk0001.md", '标着“喝我”的瓶子', ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_passes_for_html_entity_alt(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Drink Me', ) self._write( Path(temp_dir) / "output_chunk0001.md", '标着"喝我"的瓶子', ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_fails_on_missing_image_src(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'A some text B', ) self._write( Path(temp_dir) / "output_chunk0001.md", '译 译文', ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("images/b.png", out) self.assertIn("missing ", out) def test_fails_on_changed_src_path(self): with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'orig', ) self._write( Path(temp_dir) / "output_chunk0001.md", '译', ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("images/000034.png", out) self.assertIn("images/000035.png", out) def test_fails_on_repeated_image_dropped_to_one(self): # Counter-based regression: source has same src twice, translated has it once. # A set-based comparison would miss this. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'first middle second', ) self._write( Path(temp_dir) / "output_chunk0001.md", '译一 中间译文', ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("images/a.png", out) def test_fails_when_output_md_uptodate_but_chunks_bad(self): # Regression test for cache-bypass: even if output.md exists and is "up to date", # bad chunks must still cause merge_markdown_files() to fail, and the stale # output.md must be removed. with tempfile.TemporaryDirectory() as temp_dir: chunk = Path(temp_dir) / "chunk0001.md" out_chunk = Path(temp_dir) / "output_chunk0001.md" output_md = Path(temp_dir) / "output.md" self._write(chunk, 'A') self._write(out_chunk, '标') self._write(output_md, "stale merged content") os.utime(chunk, (100, 100)) os.utime(out_chunk, (150, 150)) os.utime(output_md, (200, 200)) # newer than chunks buf = io.StringIO() with contextlib.redirect_stdout(buf): result = merge_and_build.merge_markdown_files(temp_dir) out = buf.getvalue() self.assertFalse(result) self.assertFalse(output_md.exists(), msg=f"stale output.md should be deleted; stdout=\n{out}") self.assertIn("output_chunk0001.md", out) def test_passes_when_code_block_preserves_broken_img_example(self): # Regression: a tech book may legitimately ship a fenced code block that # demonstrates a deliberately-broken tag. Both source and output # carry the same example, so the per-chunk delta is empty — must pass. broken_example = ( "Here is a buggy tag to demonstrate the parser:\n\n" "```html\n" 'he said \n' "```\n" ) with tempfile.TemporaryDirectory() as temp_dir: self._write(Path(temp_dir) / "chunk0001.md", broken_example) self._write( Path(temp_dir) / "output_chunk0001.md", broken_example.replace("Here is a buggy tag to demonstrate the parser:", "下面这个有 bug 的标签是用来演示解析器的："), ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_fails_when_output_introduces_new_broken_img(self): # If source had one broken example and output adds a *second* broken tag # (not present in source), the new bad attr shows up in the delta and # we must flag it — even when there's a baseline of broken attrs. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", "Demo block:\n\n" "```html\n" 'he said \n' "```\n\n" 'And a real image: A\n', ) self._write( Path(temp_dir) / "output_chunk0001.md", "演示代码块：\n\n" "```html\n" 'he said \n' "```\n\n" # New corruption: subagent broke the real image alt with unescaped quote '真实图片：标着\n', ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("introduced malformed ", out) self.assertIn("output_chunk0001.md", out) def test_fails_when_real_image_replaced_with_escaped_markdown(self): # Regression: a regex that didn't honor `\!` would count `\![](path)` as # an image, masking real loss when the subagent escaped it accidentally. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", "See ![Fig 1](images/a.png) for details.", ) self._write( Path(temp_dir) / "output_chunk0001.md", "见 \\![图 1](images/a.png) 了解详情。", ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("missing ![](path)", out) self.assertIn("images/a.png", out) def test_passes_when_both_chunks_have_escaped_markdown_image(self): # Symmetric case: both chunks intentionally use `\![...]` as literal text. # Neither counts as a real image; counts match. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", "Use \\![alt](path) syntax to write the example as text.", ) self._write( Path(temp_dir) / "output_chunk0001.md", "用 \\![alt](path) 语法把示例写成文本。", ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_fails_when_markdown_image_missing_closing_paren(self): # Regression: the regex must require the closing `)` — a fragment like # `![图](images/a.png` does NOT render as an image, so counting it as a # preserved reference would mask real image loss. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", "See ![Fig 1](images/a.png) for details.", ) self._write( Path(temp_dir) / "output_chunk0001.md", "见 ![图 1](images/a.png 了解详情。", # missing closing ) ) ok, out = self._run_validator(temp_dir) self.assertFalse(ok) self.assertIn("missing ![](path)", out) self.assertIn("images/a.png", out) def test_passes_for_markdown_image_with_title(self): # Forward-compatibility: standard `![alt](url "title")` syntax must keep # parsing as a single image when both chunks preserve the title. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Look at ![Fig 1](images/a.png "Diagram of the system") below.', ) self._write( Path(temp_dir) / "output_chunk0001.md", '看下方 ![图 1](images/a.png "系统示意图")。', ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_passes_with_gt_in_quoted_alt(self): # Regression: a regex like ]*> would truncate at the first `>` # inside a quoted attribute value, producing a false-positive malformed # report for legitimate math/comparison content. HTMLParser handles this. with tempfile.TemporaryDirectory() as temp_dir: self._write( Path(temp_dir) / "chunk0001.md", 'Compare x > y and a < b here.', ) self._write( Path(temp_dir) / "output_chunk0001.md", '比较 x > y 且 a < b 这里。', ) ok, out = self._run_validator(temp_dir) self.assertTrue(ok, msg=out) def test_html_canary_passes_when_prose_mentions_img_tag(self): # Regression: a book that legitimately discusses HTML will render `` # mentions as `<img>` in prose or inside 
. That is not
        # corruption and must not block the build.
        with tempfile.TemporaryDirectory() as temp_dir:
            html_file = Path(temp_dir) / "book.html"
            self._write(
                html_file,
                ""
                "
The <img> tag is used for inline images.
"
                "
<img src="example.png">
"
                "
Here is a real image: \"real\"
"
                "",
            )
            ok, out = self._run_html_sanity(str(html_file))
            self.assertTrue(ok, msg=out)

    def test_basic_regex_escapes_alt_with_quote(self):
        # The basic-regex fallback used to inline alt verbatim, so a literal " in
        # markdown alt text would produce malformed raw . Now it html.escapes.
        with tempfile.TemporaryDirectory() as temp_dir:
            md_file = Path(temp_dir) / "in.md"
            html_file = Path(temp_dir) / "out.html"
            self._write(md_file, '![Title with "quote" inside](images/x.png)\n')

            ok = merge_and_build.convert_with_basic_regex(str(md_file), str(html_file), "t")
            self.assertTrue(ok)

            html_text = html_file.read_text(encoding="utf-8")
            self.assertIn(""", html_text)
            self.assertNotIn('alt="Title with "quote"', html_text)

            sanity_ok, sanity_out = self._run_html_sanity(str(html_file))
            self.assertTrue(sanity_ok, msg=sanity_out)


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