"""Tests for cj-scan.py — org-file cj-annotation scanner.

The script parses an org file and emits JSON describing:
- cj_blocks: every cj annotation found (source-block or legacy-inline form)
- verify_tasks: every VERIFY heading + placement validity (top-level or first-level child only)
- unclosed_blocks: any source-block fence that opened but never closed
"""

import json
import subprocess
from pathlib import Path

import pytest

SCRIPT = Path(__file__).parent.parent / "cj-scan.py"


@pytest.fixture
def run_scan(tmp_path):
    """Write content to a temp org file and run cj-scan; return parsed JSON output."""
    def _run(content: str) -> dict:
        f = tmp_path / "test.org"
        f.write_text(content)
        result = subprocess.run(
            ["python3", str(SCRIPT), str(f)],
            capture_output=True,
            text=True,
            check=True,
        )
        return json.loads(result.stdout)
    return _run


# ----------------------------------------------------------------------
# cj-block detection
# ----------------------------------------------------------------------

class TestCjScanCjBlockDetection:
    """Detection of cj annotations — source-block and legacy-inline forms."""

    def test_cj_scan_source_block_single_detected(self, run_scan):
        """Normal: a single source-block cj is detected with correct line range and body."""
        content = "* Section\n#+begin_src cj: comment\nplease check this\n#+end_src\n"
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 1
        b = result["cj_blocks"][0]
        assert b["form"] == "source-block"
        assert b["body"] == "please check this"
        assert b["start_line"] == 2
        assert b["end_line"] == 4

    def test_cj_scan_source_block_multiline_body_preserved(self, run_scan):
        """Normal: multi-line body is preserved with embedded newlines."""
        content = "* S\n#+begin_src cj: comment\nline 1\nline 2\nline 3\n#+end_src\n"
        result = run_scan(content)
        assert result["cj_blocks"][0]["body"] == "line 1\nline 2\nline 3"

    def test_cj_scan_multiple_source_blocks_each_detected(self, run_scan):
        """Normal: multiple source-blocks in a file are detected as separate items."""
        content = (
            "* A\n#+begin_src cj: comment\nfirst\n#+end_src\n"
            "* B\n#+begin_src cj: comment\nsecond\n#+end_src\n"
        )
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 2
        bodies = [b["body"] for b in result["cj_blocks"]]
        assert bodies == ["first", "second"]

    def test_cj_scan_legacy_inline_single_line_detected(self, run_scan):
        """Normal: a legacy inline cj line is detected with form=legacy-inline."""
        content = "* Section\ncj: please check this\n"
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 1
        b = result["cj_blocks"][0]
        assert b["form"] == "legacy-inline"
        assert b["body"] == "please check this"
        assert b["start_line"] == 2
        assert b["end_line"] == 2

    def test_cj_scan_mixed_forms_in_same_file(self, run_scan):
        """Normal: source-block + legacy inline coexist; both detected as separate items."""
        content = (
            "* A\ncj: legacy form\n"
            "* B\n#+begin_src cj: comment\nnew form\n#+end_src\n"
        )
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 2
        forms = sorted(b["form"] for b in result["cj_blocks"])
        assert forms == ["legacy-inline", "source-block"]

    def test_cj_scan_empty_file_returns_empty_lists(self, run_scan):
        """Boundary: empty file → empty cj_blocks and verify_tasks lists."""
        result = run_scan("")
        assert result["cj_blocks"] == []
        assert result["verify_tasks"] == []
        assert result["unclosed_blocks"] == []

    def test_cj_scan_no_cj_content_returns_empty_blocks(self, run_scan):
        """Boundary: org file with no cj content → empty cj_blocks."""
        content = "* Section\n** TODO Task\nbody text\n** TODO Another\n"
        result = run_scan(content)
        assert result["cj_blocks"] == []

    def test_cj_scan_block_before_any_heading_empty_chain(self, run_scan):
        """Boundary: cj block at top of file (before any heading) → empty parent chain."""
        content = "#+begin_src cj: comment\ntop-level note\n#+end_src\n"
        result = run_scan(content)
        assert result["cj_blocks"][0]["parent_heading_chain"] == []
        assert result["cj_blocks"][0]["parent_depth"] == 0

    @pytest.mark.parametrize("fence", [
        "#+begin_src cj: comment",
        "#+begin_src cj:",
        "#+begin_src cj: anything",
        "#+BEGIN_SRC cj: comment",  # case-insensitive
    ])
    def test_cj_scan_source_block_fence_variants_all_recognized(self, run_scan, fence):
        """Boundary: fence label and case variants are all valid forms."""
        content = f"* S\n{fence}\nbody\n#+end_src\n"
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 1
        assert result["cj_blocks"][0]["body"] == "body"

    def test_cj_scan_unclosed_source_block_reported(self, run_scan):
        """Error: a source-block that opens but never closes → reported in unclosed_blocks."""
        content = "* S\n#+begin_src cj: comment\nbody that never ends\n"
        result = run_scan(content)
        assert result["cj_blocks"] == []
        assert len(result["unclosed_blocks"]) == 1
        assert result["unclosed_blocks"][0]["start_line"] == 2


# ----------------------------------------------------------------------
# Parent heading chain reconstruction
# ----------------------------------------------------------------------

class TestCjScanParentChain:
    """Parent heading chain construction — walking the org tree backward."""

    def test_cj_scan_nested_parent_chain_three_levels(self, run_scan):
        """Normal: cj block inside three nested headings → chain reflects all three."""
        content = (
            "* Work\n"
            "** DOING [#A] Kostya's contract\n"
            "*** VERIFY Question?\n"
            "#+begin_src cj: comment\nanswer\n#+end_src\n"
        )
        result = run_scan(content)
        chain = result["cj_blocks"][0]["parent_heading_chain"]
        assert len(chain) == 3
        assert chain[0] == {"depth": 1, "heading": "Work"}
        assert chain[1] == {"depth": 2, "heading": "DOING [#A] Kostya's contract"}
        assert chain[2] == {"depth": 3, "heading": "VERIFY Question?"}
        assert result["cj_blocks"][0]["parent_depth"] == 3

    def test_cj_scan_depth_skip_only_actual_ancestors(self, run_scan):
        """Normal: heading depth skip (e.g., * then ***) → chain captures only present headings."""
        content = "* Section\n*** Deep child\n#+begin_src cj: comment\nbody\n#+end_src\n"
        result = run_scan(content)
        chain = result["cj_blocks"][0]["parent_heading_chain"]
        assert [h["depth"] for h in chain] == [1, 3]

    def test_cj_scan_shallower_sibling_pops_deeper_frames(self, run_scan):
        """Normal: when a shallower heading appears, deeper frames pop off the stack."""
        content = (
            "* A\n** A.1\n*** A.1.1\n"
            "** B\n"
            "#+begin_src cj: comment\nunder B\n#+end_src\n"
        )
        result = run_scan(content)
        chain = result["cj_blocks"][0]["parent_heading_chain"]
        assert len(chain) == 2
        assert chain[0]["heading"] == "A"
        assert chain[1]["heading"] == "B"


# ----------------------------------------------------------------------
# VERIFY task detection + placement audit
# ----------------------------------------------------------------------

class TestCjScanVerifyPlacement:
    """VERIFY task detection and placement audit per the canonical rule."""

    def test_cj_scan_verify_at_depth_2_is_valid(self, run_scan):
        """Normal: ** VERIFY (top-level) is valid placement."""
        content = "* Work\n** VERIFY [#C] Hayk's Farearth Evaluation :research:hayk:\n"
        result = run_scan(content)
        assert len(result["verify_tasks"]) == 1
        v = result["verify_tasks"][0]
        assert v["depth"] == 2
        assert v["valid_depth"] is True
        assert v["promotion_target"] is None

    def test_cj_scan_verify_at_depth_3_is_valid(self, run_scan):
        """Normal: *** VERIFY (first-level child) is valid placement."""
        content = "* Work\n** TODO Parent\n*** VERIFY Question?\n"
        result = run_scan(content)
        v = result["verify_tasks"][0]
        assert v["depth"] == 3
        assert v["valid_depth"] is True

    def test_cj_scan_verify_at_depth_4_invalid_promote_to_3(self, run_scan):
        """Normal: **** VERIFY is buried; suggests promotion to depth 3."""
        content = "* W\n** P\n*** Q\n**** VERIFY Buried?\n"
        result = run_scan(content)
        v = result["verify_tasks"][0]
        assert v["depth"] == 4
        assert v["valid_depth"] is False
        assert v["promotion_target"] == 3

    def test_cj_scan_verify_at_depth_6_invalid_promote_to_3(self, run_scan):
        """Normal: ****** VERIFY at any deep level → promotion target is still 3."""
        content = "* W\n** P\n*** Q\n**** Q2\n***** Q3\n****** VERIFY Very buried?\n"
        result = run_scan(content)
        v = result["verify_tasks"][0]
        assert v["depth"] == 6
        assert v["promotion_target"] == 3

    def test_cj_scan_verify_at_depth_1_invalid_promote_to_2(self, run_scan):
        """Boundary: * VERIFY at top-section depth → promotion target is 2 (top-level under section)."""
        content = "* VERIFY Should-be-deeper\n"
        result = run_scan(content)
        v = result["verify_tasks"][0]
        assert v["depth"] == 1
        assert v["valid_depth"] is False
        assert v["promotion_target"] == 2

    def test_cj_scan_verify_heading_with_priority_and_tags(self, run_scan):
        """Boundary: VERIFY heading with priority cookie + tags → heading text captured fully."""
        content = "* W\n** VERIFY [#C] Hayk's Farearth Evaluation :research:hayk:\n"
        result = run_scan(content)
        v = result["verify_tasks"][0]
        assert "Hayk's Farearth Evaluation" in v["heading"]
        assert ":research:" in v["heading"]

    def test_cj_scan_no_verify_tasks_empty_list(self, run_scan):
        """Boundary: file with only TODO/DOING headings → empty verify_tasks list."""
        content = "* W\n** TODO X\n*** DOING Y\n"
        result = run_scan(content)
        assert result["verify_tasks"] == []

    def test_cj_scan_verify_word_in_body_is_not_a_task(self, run_scan):
        """Error: the word VERIFY appearing in body prose is not detected as a task."""
        content = (
            "* Work\n"
            "** TODO Important task\n"
            "Body line mentioning VERIFY in prose.\n"
        )
        result = run_scan(content)
        assert result["verify_tasks"] == []


# ----------------------------------------------------------------------
# Nested-fence suppression (no false positives inside wrapper blocks)
# ----------------------------------------------------------------------

class TestCjScanNestedFencesIgnored:
    """A cj fence nested inside another #+begin_<type> block is content, not an annotation.

    Documentation patterns frequently embed the cj marker inside `#+begin_example`
    blocks or inside `#+begin_src snippet` yasnippet definitions to *show* what
    the marker looks like.  A naive line-by-line scanner that only watches for
    `#+begin_src cj:` and `#+end_src` matches those literally and misclassifies
    the documentation as live annotations.
    """

    def test_cj_scan_fence_inside_begin_example_ignored(self, run_scan):
        """Normal: cj fence inside #+begin_example ... #+end_example is documentation, not annotation."""
        content = (
            "* Section\n"
            "Here is what the cj marker looks like:\n"
            "#+begin_example\n"
            "#+begin_src cj: comment\n"
            "I am documentation, not a real annotation\n"
            "#+end_src\n"
            "#+end_example\n"
        )
        result = run_scan(content)
        assert result["cj_blocks"] == []
        assert result["unclosed_blocks"] == []

    def test_cj_scan_fence_inside_begin_src_other_lang_ignored(self, run_scan):
        """Normal: cj fence inside #+begin_src snippet (or any non-cj src block) is content.

        The outer #+begin_src snippet block claims everything until the FIRST
        matching #+end_src, so the inner `#+begin_src cj:` line is literal text
        inside that block — not a nested cj annotation.
        """
        content = (
            "* Section\n"
            "#+begin_src snippet\n"
            "# name: cj-comment-block\n"
            "# --\n"
            "#+begin_src cj: comment\n"
            "$0\n"
            "#+end_src\n"
            "#+end_src\n"
        )
        result = run_scan(content)
        assert result["cj_blocks"] == []

    def test_cj_scan_fence_inside_begin_quote_ignored(self, run_scan):
        """Boundary: cj fence inside #+begin_quote ... #+end_quote is quoted prose, not annotation."""
        content = (
            "* Section\n"
            "#+begin_quote\n"
            "#+begin_src cj: comment\n"
            "quoted, not active\n"
            "#+end_src\n"
            "#+end_quote\n"
        )
        result = run_scan(content)
        assert result["cj_blocks"] == []

    def test_cj_scan_real_cj_after_example_block_still_detected(self, run_scan):
        """Normal: after a wrapper block closes, a subsequent real cj fence is detected.

        Regression guard: the wrapper-tracking state must reset when the
        wrapper closes; otherwise everything after the first example block
        would be silently swallowed.
        """
        content = (
            "* Section\n"
            "#+begin_example\n"
            "#+begin_src cj: comment\n"
            "doc, ignored\n"
            "#+end_src\n"
            "#+end_example\n"
            "#+begin_src cj: comment\n"
            "this one is real\n"
            "#+end_src\n"
        )
        result = run_scan(content)
        assert len(result["cj_blocks"]) == 1
        assert result["cj_blocks"][0]["body"] == "this one is real"

    def test_cj_scan_legacy_inline_inside_wrapper_ignored(self, run_scan):
        """Boundary: legacy `cj: ...` line inside a wrapper block is content, not annotation."""
        content = (
            "* Section\n"
            "#+begin_example\n"
            "cj: this is a documentation example, not a real annotation\n"
            "#+end_example\n"
        )
        result = run_scan(content)
        assert result["cj_blocks"] == []

    def test_cj_scan_unclosed_wrapper_does_not_swallow_rest_of_file_silently(self, run_scan):
        """Error: an unclosed wrapper consumes everything after it — cj_blocks empty, not unclosed_blocks.

        Pinning current behaviour: an unclosed `#+begin_example` is a separate
        org-level malformation that this scanner doesn't currently report.  The
        guard here is that the unclosed-wrapper case doesn't produce
        false-positive cj_blocks downstream.
        """
        content = (
            "* Section\n"
            "#+begin_example\n"
            "#+begin_src cj: comment\n"
            "should be content of the unclosed example\n"
            "#+end_src\n"
            "more content here, no end_example\n"
        )
        result = run_scan(content)
        assert result["cj_blocks"] == []