jobyssey/tests/test_generate_cv.py

"""Contract tests for tools/generate_cv.py.

The generator turns documents/sources/ YAML plus a role config into a
LaTeX CV. What must hold:

- Source description blocks split into the right bullets, including the
  leading-prose case (projects/ecosystem-devrel.yml).
- Every escape from 05-cv-templates.md's special-character table is
  applied; `%` is the silent killer (truncates the rest of the line).
- Date ranges use a single ASCII hyphen, never `--` (en-dash ligature
  breaks ATS date parsing - confirmed Workday import failure).
- `use: N` bullet refs resolve 1-based and fail loudly out of range.
- Skill labels absent from documents/sources/ warn unless allowlisted
  via `extra-skills`; the shipped role configs generate warning-free.
- Generated documents never contain `\\item [` (F9: clipped label trap).
"""
import datetime
import re
import sys
import tempfile
import unittest
from pathlib import Path

REPO = Path(__file__).resolve().parent.parent
sys.path.insert(0, str(REPO / "tools"))

import generate_cv  # noqa: E402

UNBRACED_BRACKET_ITEM = re.compile(r"\\item\s*\[")


class TestParseBullets(unittest.TestCase):
    def test_splits_bullets_and_joins_continuations(self):
        text = "- first bullet\n  continues here\n- second bullet\n"
        self.assertEqual(
            generate_cv.parse_bullets(text),
            ["first bullet continues here", "second bullet"],
        )

    def test_leading_prose_becomes_first_bullet(self):
        text = "Intro paragraph.\n\n- real bullet\n"
        self.assertEqual(
            generate_cv.parse_bullets(text),
            ["Intro paragraph.", "real bullet"],
        )

    def test_empty_description(self):
        self.assertEqual(generate_cv.parse_bullets(None), [])


class TestLatexEscape(unittest.TestCase):
    def test_special_characters(self):
        self.assertEqual(
            generate_cv.latex_escape("99% uptime & $4.7M for C# devs_v1"),
            "99\\% uptime \\& \\$4.7M for C\\# devs\\_v1",
        )

    def test_unicode_punctuation(self):
        self.assertEqual(generate_cv.latex_escape("a — b – c"), "a --- b -- c")
        self.assertEqual(generate_cv.latex_escape("\u201cMotr\u201d"), "``Motr''")
        self.assertEqual(generate_cv.latex_escape("Yu\u0113"), "Yu\\=e")
        self.assertEqual(
            generate_cv.latex_escape("A \u00b7 B"), "A \\textperiodcentered{} B"
        )


class TestFormatRange(unittest.TestCase):
    def test_open_range_uses_ascii_hyphen_and_present(self):
        start = datetime.date(2021, 3, 1)
        self.assertEqual(generate_cv.format_range(start, None), "Mar 2021 - Present")

    def test_year_only_range(self):
        start = datetime.date(2022, 1, 1)
        end = datetime.date(2023, 8, 1)
        self.assertEqual(
            generate_cv.format_range(start, end, month=False), "2022 - 2023"
        )

    def test_never_emits_double_hyphen(self):
        start = datetime.date(2016, 7, 1)
        end = datetime.date(2019, 5, 1)
        self.assertNotIn("--", generate_cv.format_range(start, end))


class TestResolveBullets(unittest.TestCase):
    SOURCE = {"description": "- alpha\n- beta\n- gamma\n"}
    PATH = Path("fake.yml")

    def test_use_index_is_one_based(self):
        entry = {"bullets": [{"use": 2}, "literal text"]}
        self.assertEqual(
            generate_cv.resolve_bullets(entry, self.SOURCE, self.PATH),
            ["beta", "literal text"],
        )

    def test_default_is_all_bullets(self):
        self.assertEqual(
            generate_cv.resolve_bullets({}, self.SOURCE, self.PATH),
            ["alpha", "beta", "gamma"],
        )

    def test_out_of_range_index_fails(self):
        with self.assertRaises(generate_cv.ConfigError):
            generate_cv.resolve_bullets(
                {"bullets": [{"use": 4}]}, self.SOURCE, self.PATH
            )


class TestSkillValidation(unittest.TestCase):
    def test_unknown_skill_warns(self):
        warnings = []
        generate_cv.validate_skills(
            [{"group": "X", "items": ["Definitely Not A Skill"]}], [], warnings.append
        )
        self.assertEqual(len(warnings), 1)

    def test_extra_skills_allowlist_silences(self):
        warnings = []
        generate_cv.validate_skills(
            [{"group": "X", "items": ["Definitely Not A Skill"]}],
            ["Definitely Not A Skill"],
            warnings.append,
        )
        self.assertEqual(warnings, [])

    def test_parenthetical_alias_grounds_label(self):
        warnings = []
        # "Go (Golang)" grounds via "Go"; "DIDs" via the sources'
        # "Decentralized Identifiers (DIDs)".
        generate_cv.validate_skills(
            [{"group": "X", "items": ["Go (Golang)", "DIDs"]}], [], warnings.append
        )
        self.assertEqual(warnings, [])


class TestShippedRoleConfigs(unittest.TestCase):
    """Every role config in documents/cv/roles/ must generate cleanly."""

    def generate_all(self):
        results = {}
        with tempfile.TemporaryDirectory() as tmp:
            for role in generate_cv.list_roles():
                warnings = []
                target = generate_cv.generate(
                    role, output=Path(tmp) / f"{role}.tex", warn=warnings.append
                )
                results[role] = (target.read_text(encoding="utf-8"), warnings)
        return results

    def test_expected_roles_exist(self):
        self.assertLessEqual(
            {
                "example",
                "staff-software-engineer",
                "senior-software-engineer",
                "founding-software-engineer",
            },
            set(generate_cv.list_roles()),
        )

    def test_configs_generate_without_grounding_warnings(self):
        for role, (_, warnings) in self.generate_all().items():
            self.assertEqual(warnings, [], f"{role}: ungrounded skills: {warnings}")

    def test_output_is_wellformed_latex(self):
        for role, (text, _) in self.generate_all().items():
            body = "\n".join(
                line for line in text.splitlines() if not line.startswith("%%")
            )
            self.assertIn("\\begin{document}", body, role)
            self.assertIn("\\end{document}", body, role)
            self.assertIn("prnk28@gmail.com", body, role)
            # F9 guard: no bullet parsed as an optional label.
            self.assertIsNone(UNBRACED_BRACKET_ITEM.search(body), role)
            # No unescaped % (silent truncation) or & (compile error).
            self.assertIsNone(re.search(r"(?<!\\)%", body), role)
            self.assertIsNone(re.search(r"(?<!\\)&", body), role)
            # No en-dash-producing "--" inside a date \textit{...}.
            for match in re.finditer(r"\\textit\{([^}]*\d{4}[^}]*)\}", body):
                self.assertNotIn("--", match.group(1), f"{role}: {match.group(0)}")


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