README.md

@@ -104,9 +104,19 @@ Follow these steps before running the compaction test suite:
    ```bash
    gcloud emulators firestore start --host-port=localhost:8153
    ```
+   In the therminal where execute the test:
+   ```bash
+   export FIRESTORE_EMULATOR_HOST=localhost:8153
+   ```
 3. Execute the tests with `pytest` through `uv`:
    ```bash
    uv run pytest tests/test_compaction.py -v
    ```
 
 If any step fails, double-check that the tools are installed and available on your `PATH` before trying again.
+
+### Filter emojis
+Execute the tests with `pytest` command:
+```bash
+uv run pytest tests/test_governance_emojis.py
+```

tests/test_governance_emojis.py

@@ -0,0 +1,69 @@
+"""Unit tests for the emoji filtering regex."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+import pytest
+
+
+os.environ.setdefault("CONFIG_YAML", str(Path(__file__).resolve().parents[1] / "config.yaml"))
+
+from va_agent.governance import GovernancePlugin
+
+
+def _make_plugin() -> GovernancePlugin:
+    plugin = object.__new__(GovernancePlugin)
+    plugin._combined_pattern = plugin._get_combined_pattern()
+    return plugin
+
+
+@pytest.fixture()
+def plugin() -> GovernancePlugin:
+    return _make_plugin()
+
+
+@pytest.mark.parametrize(
+    ("original", "expected_clean", "expected_removed"),
+    [
+        ("Hola 🔪 mundo", "Hola  mundo", ["🔪"]),
+        ("No 🔪💀🚬 permitidos", "No  permitidos", ["🔪", "💀", "🚬"]),
+        ("Dedo 🖕 grosero", "Dedo  grosero", ["🖕"]),
+        ("Dedo 🖕🏾 grosero", "Dedo  grosero", ["🖕🏾"]),
+        ("Todo Amor: 👩‍❤️‍👨 | 👩‍❤️‍👩 | 🧑‍❤️‍🧑 | 👨‍❤️‍👨 | 👩‍❤️‍💋‍👨 | 👩‍❤️‍💋‍👩 | 🧑‍❤️‍💋‍🧑 | 👨‍❤️‍💋‍👨", "Todo Amor:  |  |  |  |  |  |  |", ["👩‍❤️‍👨", "👩‍❤️‍👩", "🧑‍❤️‍🧑", "👨‍❤️‍👨", "👩‍❤️‍💋‍👨", "👩‍❤️‍💋‍👩", "🧑‍❤️‍💋‍🧑", "👨‍❤️‍💋‍👨"]),
+        ("Amor 👩🏽‍❤️‍👨🏻 bicolor", "Amor  bicolor", ["👩🏽‍❤️‍👨🏻"]),
+        ("Beso 👩🏻‍❤️‍💋‍👩🏿 bicolor gay", "Beso  bicolor gay", ["👩🏻‍❤️‍💋‍👩🏿"]),
+        ("Emoji compuesto permitido 👨🏽‍💻", "Emoji compuesto permitido 👨🏽‍💻", []),
+    ],
+)
+def test_remove_emojis_blocks_forbidden_sequences(
+    plugin: GovernancePlugin,
+    original: str,
+    expected_clean: str,
+    expected_removed: list[str],
+) -> None:
+    cleaned, removed = plugin._remove_emojis(original)
+
+    assert cleaned == expected_clean
+    assert removed == expected_removed
+
+
+def test_remove_emojis_preserves_allowed_people_with_skin_tones(
+    plugin: GovernancePlugin,
+) -> None:
+    original = "Persona 👩🏽 hola"
+
+    cleaned, removed = plugin._remove_emojis(original)
+
+    assert cleaned == original
+    assert removed == []
+
+
+def test_remove_emojis_trims_whitespace_after_removal(
+    plugin: GovernancePlugin,
+) -> None:
+    cleaned, removed = plugin._remove_emojis("   🔪Hola🔪   ")
+
+    assert cleaned == "Hola"
+    assert removed == ["🔪", "🔪"]