tests/test_translate.py

# tests/test_translate.py
"""Tests for agent/translate.py — pure functions only, no LLM calls."""
import copy
import pytest
from agent.translate import _collect_segments, _set_at_path
from agent.prompts import PROSE_FIELDS, PROSE_LIST_FIELDS


# ── _collect_segments ──────────────────────────────────────────────────────────

def test_collect_segments_prose_field():
    brief = {"what_matters_most": "Revenue grew strongly."}
    segs = _collect_segments(brief)
    assert len(segs) == 1
    path, text = segs[0]
    assert path == ("what_matters_most",)
    assert text == "Revenue grew strongly."


def test_collect_segments_excludes_evidence_snippet():
    brief = {"evidence_snippet": "verbatim quote here", "headline": "Q3 beat"}
    segs = _collect_segments(brief)
    paths = [p for p, _ in segs]
    assert ("evidence_snippet",) not in paths
    assert ("headline",) in paths


def test_collect_segments_skips_quarter_deltas_subtree():
    brief = {
        "what_matters_most": "Growth accelerated.",
        "quarter_deltas": {"headline": "Should NOT be collected", "text": "Also skip"},
    }
    segs = _collect_segments(brief)
    texts = [t for _, t in segs]
    assert "Growth accelerated." in texts
    assert "Should NOT be collected" not in texts
    assert "Also skip" not in texts


def test_collect_segments_prose_list_field():
    brief = {"what_to_watch": ["Watch margins closely.", "Monitor capex spend."]}
    segs = _collect_segments(brief)
    assert len(segs) == 2
    paths = [p for p, _ in segs]
    assert ("what_to_watch", 0) in paths
    assert ("what_to_watch", 1) in paths


def test_collect_segments_nested_dict():
    brief = {
        "bull_points": [
            {"text": "Revenue beat expectations.", "source": "10-K", "reliability": "HIGH"}
        ]
    }
    segs = _collect_segments(brief)
    paths = [p for p, _ in segs]
    assert ("bull_points", 0, "text") in paths
    # source and reliability are enum fields, should not be collected
    assert ("bull_points", 0, "source") not in paths
    assert ("bull_points", 0, "reliability") not in paths


def test_collect_segments_skips_empty_strings():
    brief = {"what_matters_most": "", "headline": "Has content"}
    segs = _collect_segments(brief)
    paths = [p for p, _ in segs]
    assert ("what_matters_most",) not in paths
    assert ("headline",) in paths


def test_collect_segments_nested_key_quote():
    brief = {"key_quote": {"text": "This is the key quote.", "source": "transcript"}}
    segs = _collect_segments(brief)
    paths = [p for p, _ in segs]
    assert ("key_quote", "text") in paths
    assert ("key_quote", "source") not in paths


# ── _set_at_path ──────────────────────────────────────────────────────────────

def test_set_at_path_top_level():
    obj = {"what_matters_most": "English text"}
    _set_at_path(obj, ("what_matters_most",), "Texte français")
    assert obj["what_matters_most"] == "Texte français"


def test_set_at_path_nested():
    obj = {"bull_points": [{"text": "English", "reliability": "HIGH"}]}
    _set_at_path(obj, ("bull_points", 0, "text"), "Français")
    assert obj["bull_points"][0]["text"] == "Français"
    assert obj["bull_points"][0]["reliability"] == "HIGH"  # untouched


def test_set_at_path_missing_key_is_silent():
    obj = {"bull_points": []}
    _set_at_path(obj, ("bull_points", 0, "text"), "Français")
    # Should not raise; list is empty so index 0 doesn't exist


def test_set_at_path_empty_path_is_silent():
    obj = {"x": 1}
    _set_at_path(obj, (), "value")
    assert obj == {"x": 1}  # unchanged


# ── drift guard ──────────────────────────────────────────────────────────────

def test_prose_fields_in_language_directive():
    """Every name in PROSE_FIELDS | PROSE_LIST_FIELDS appears in language_directive output."""
    from agent.prompts import language_directive
    directive = language_directive("French")
    for field in PROSE_FIELDS | PROSE_LIST_FIELDS:
        assert field in directive, f"Field '{field}' missing from language_directive"