Upgrade to hassil 2.0 (#130544)

* Working on hassil 2.0 * Bump to hassil 2.0 * Update snapshots * Remove debug logging
2024-11-13 15:50:08 -06:00 · 2024-11-13 15:50:08 -06:00 · 51c6ee97b1
commit 51c6ee97b1
parent 4002bc3c25
12 changed files with 53 additions and 102 deletions
--- a/homeassistant/components/conversation/default_agent.py
+++ b/homeassistant/components/conversation/default_agent.py
@ -16,11 +16,11 @@ from hassil.expression import Expression, ListReference, Sequence
 from hassil.intents import Intents, SlotList, TextSlotList, WildcardSlotList
 from hassil.recognize import (
    MISSING_ENTITY,
-    MatchEntity,
    RecognizeResult,
-    UnmatchedTextEntity,
    recognize_all,
+    recognize_best,
 )
+from hassil.string_matcher import UnmatchedRangeEntity, UnmatchedTextEntity
 from hassil.util import merge_dict
 from home_assistant_intents import ErrorKey, get_intents, get_languages
 import yaml
@ -499,6 +499,7 @@ class DefaultAgent(ConversationEntity):
        maybe_result: RecognizeResult | None = None
        best_num_matched_entities = 0
        best_num_unmatched_entities = 0
+        best_num_unmatched_ranges = 0
        for result in recognize_all(
            user_input.text,
            lang_intents.intents,
@ -517,10 +518,14 @@ class DefaultAgent(ConversationEntity):
                    num_matched_entities += 1

            num_unmatched_entities = 0
+            num_unmatched_ranges = 0
            for unmatched_entity in result.unmatched_entities_list:
                if isinstance(unmatched_entity, UnmatchedTextEntity):
                    if unmatched_entity.text != MISSING_ENTITY:
                        num_unmatched_entities += 1
+                elif isinstance(unmatched_entity, UnmatchedRangeEntity):
+                    num_unmatched_ranges += 1
+                    num_unmatched_entities += 1
                else:
                    num_unmatched_entities += 1

@ -532,15 +537,24 @@ class DefaultAgent(ConversationEntity):
                    (num_matched_entities == best_num_matched_entities)
                    and (num_unmatched_entities < best_num_unmatched_entities)
                )
+                or (
+                    # Prefer unmatched ranges
+                    (num_matched_entities == best_num_matched_entities)
+                    and (num_unmatched_entities == best_num_unmatched_entities)
+                    and (num_unmatched_ranges > best_num_unmatched_ranges)
+                )
                or (
                    # More literal text matched
                    (num_matched_entities == best_num_matched_entities)
                    and (num_unmatched_entities == best_num_unmatched_entities)
+                    and (num_unmatched_ranges == best_num_unmatched_ranges)
                    and (result.text_chunks_matched > maybe_result.text_chunks_matched)
                )
                or (
                    # Prefer match failures with entities
                    (result.text_chunks_matched == maybe_result.text_chunks_matched)
+                    and (num_unmatched_entities == best_num_unmatched_entities)
+                    and (num_unmatched_ranges == best_num_unmatched_ranges)
                    and (
                        ("name" in result.entities)
                        or ("name" in result.unmatched_entities)
@ -550,6 +564,7 @@ class DefaultAgent(ConversationEntity):
                maybe_result = result
                best_num_matched_entities = num_matched_entities
                best_num_unmatched_entities = num_unmatched_entities
+                best_num_unmatched_ranges = num_unmatched_ranges

        return maybe_result

@ -562,76 +577,15 @@ class DefaultAgent(ConversationEntity):
        language: str,
    ) -> RecognizeResult | None:
        """Search intents for a strict match to user input."""
-        custom_found = False
-        name_found = False
-        best_results: list[RecognizeResult] = []
-        best_name_quality: int | None = None
-        best_text_chunks_matched: int | None = None
-        for result in recognize_all(
+        return recognize_best(
            user_input.text,
            lang_intents.intents,
            slot_lists=slot_lists,
            intent_context=intent_context,
            language=language,
-        ):
-            # Prioritize user intents
-            is_custom = (
-                result.intent_metadata is not None
-                and result.intent_metadata.get(METADATA_CUSTOM_SENTENCE)
-            )
-
-            if custom_found and not is_custom:
-                continue
-
-            if not custom_found and is_custom:
-                custom_found = True
-                # Clear builtin results
-                name_found = False
-                best_results = []
-                best_name_quality = None
-                best_text_chunks_matched = None
-
-            # Prioritize results with a "name" slot
-            name = result.entities.get("name")
-            is_name = name and not name.is_wildcard
-
-            if name_found and not is_name:
-                continue
-
-            if not name_found and is_name:
-                name_found = True
-                # Clear non-name results
-                best_results = []
-                best_text_chunks_matched = None
-
-            if is_name:
-                # Prioritize results with a better "name" slot
-                name_quality = len(cast(MatchEntity, name).value.split())
-                if (best_name_quality is None) or (name_quality > best_name_quality):
-                    best_name_quality = name_quality
-                    # Clear worse name results
-                    best_results = []
-                    best_text_chunks_matched = None
-                elif name_quality < best_name_quality:
-                    continue
-
-            # Prioritize results with more literal text
-            # This causes wildcards to match last.
-            if (best_text_chunks_matched is None) or (
-                result.text_chunks_matched > best_text_chunks_matched
-            ):
-                best_results = [result]
-                best_text_chunks_matched = result.text_chunks_matched
-            elif result.text_chunks_matched == best_text_chunks_matched:
-                # Accumulate results with the same number of literal text matched.
-                # We will resolve the ambiguity below.
-                best_results.append(result)
-
-        if best_results:
-            # Successful strict match
-            return best_results[0]
-
-        return None
+            best_metadata_key=METADATA_CUSTOM_SENTENCE,
+            best_slot_name="name",
+        )

    async def _build_speech(
        self,
--- a/homeassistant/components/conversation/http.py
+++ b/homeassistant/components/conversation/http.py
@ -6,12 +6,8 @@ from collections.abc import Iterable
 from typing import Any

 from aiohttp import web
-from hassil.recognize import (
-    MISSING_ENTITY,
-    RecognizeResult,
-    UnmatchedRangeEntity,
-    UnmatchedTextEntity,
-)
+from hassil.recognize import MISSING_ENTITY, RecognizeResult
+from hassil.string_matcher import UnmatchedRangeEntity, UnmatchedTextEntity
 import voluptuous as vol

 from homeassistant.components import http, websocket_api
--- a/homeassistant/components/conversation/manifest.json
+++ b/homeassistant/components/conversation/manifest.json
@ -6,5 +6,5 @@
  "documentation": "https://www.home-assistant.io/integrations/conversation",
  "integration_type": "system",
  "quality_scale": "internal",
-  "requirements": ["hassil==1.7.4", "home-assistant-intents==2024.11.6"]
+  "requirements": ["hassil==2.0.1", "home-assistant-intents==2024.11.13"]
 }
--- a/homeassistant/components/conversation/trigger.py
+++ b/homeassistant/components/conversation/trigger.py
@ -4,7 +4,8 @@ from __future__ import annotations

 from typing import Any

-from hassil.recognize import PUNCTUATION, RecognizeResult
+from hassil.recognize import RecognizeResult
+from hassil.util import PUNCTUATION_ALL
 import voluptuous as vol

 from homeassistant.const import CONF_COMMAND, CONF_PLATFORM
@ -20,7 +21,7 @@ from .const import DATA_DEFAULT_ENTITY, DOMAIN
 def has_no_punctuation(value: list[str]) -> list[str]:
    """Validate result does not contain punctuation."""
    for sentence in value:
-        if PUNCTUATION.search(sentence):
+        if PUNCTUATION_ALL.search(sentence):
            raise vol.Invalid("sentence should not contain punctuation")

    return value
--- a/homeassistant/package_constraints.txt
+++ b/homeassistant/package_constraints.txt
@ -32,10 +32,10 @@ go2rtc-client==0.1.1
 ha-ffmpeg==3.2.2
 habluetooth==3.6.0
 hass-nabucasa==0.84.0
-hassil==1.7.4
+hassil==2.0.1
 home-assistant-bluetooth==1.13.0
 home-assistant-frontend==20241106.2
-home-assistant-intents==2024.11.6
+home-assistant-intents==2024.11.13
 httpx==0.27.2
 ifaddr==0.2.0
 Jinja2==3.1.4
--- a/requirements_all.txt
+++ b/requirements_all.txt
@ -1093,7 +1093,7 @@ hass-nabucasa==0.84.0
 hass-splunk==0.1.1

 # homeassistant.components.conversation
-hassil==1.7.4
+hassil==2.0.1

 # homeassistant.components.jewish_calendar
 hdate==0.10.9
@ -1130,7 +1130,7 @@ holidays==0.60
 home-assistant-frontend==20241106.2

 # homeassistant.components.conversation
-home-assistant-intents==2024.11.6
+home-assistant-intents==2024.11.13

 # homeassistant.components.home_connect
 homeconnect==0.8.0
--- a/requirements_test_all.txt
+++ b/requirements_test_all.txt
@ -928,7 +928,7 @@ habluetooth==3.6.0
 hass-nabucasa==0.84.0

 # homeassistant.components.conversation
-hassil==1.7.4
+hassil==2.0.1

 # homeassistant.components.jewish_calendar
 hdate==0.10.9
@ -956,7 +956,7 @@ holidays==0.60
 home-assistant-frontend==20241106.2

 # homeassistant.components.conversation
-home-assistant-intents==2024.11.6
+home-assistant-intents==2024.11.13

 # homeassistant.components.home_connect
 homeconnect==0.8.0
--- a/script/hassfest/docker/Dockerfile
+++ b/script/hassfest/docker/Dockerfile
@ -23,7 +23,7 @@ RUN --mount=from=ghcr.io/astral-sh/uv:0.5.0,source=/uv,target=/bin/uv \
        -c /usr/src/homeassistant/homeassistant/package_constraints.txt \
        -r /usr/src/homeassistant/requirements.txt \
        stdlib-list==0.10.0 pipdeptree==2.23.4 tqdm==4.66.5 ruff==0.7.3 \
-        PyTurboJPEG==1.7.5 ha-ffmpeg==3.2.2 hassil==1.7.4 home-assistant-intents==2024.11.6 mutagen==1.47.0 pymicro-vad==1.0.1 pyspeex-noise==1.0.2
+        PyTurboJPEG==1.7.5 ha-ffmpeg==3.2.2 hassil==2.0.1 home-assistant-intents==2024.11.13 mutagen==1.47.0 pymicro-vad==1.0.1 pyspeex-noise==1.0.2

 LABEL "name"="hassfest"
 LABEL "maintainer"="Home Assistant <hello@home-assistant.io>"
--- a/tests/components/assist_pipeline/snapshots/test_websocket.ambr
+++ b/tests/components/assist_pipeline/snapshots/test_websocket.ambr
@ -697,7 +697,7 @@
        'speech': dict({
          'plain': dict({
            'extra_data': None,
-            'speech': 'Sorry, I am not aware of any area called are',
+            'speech': 'Sorry, I am not aware of any area called Are',
          }),
        }),
      }),
@ -741,7 +741,7 @@
        'speech': dict({
          'plain': dict({
            'extra_data': None,
-            'speech': 'Sorry, I am not aware of any area called are',
+            'speech': 'Sorry, I am not aware of any area called Are',
          }),
        }),
      }),
--- a/tests/components/conversation/snapshots/test_http.ambr
+++ b/tests/components/conversation/snapshots/test_http.ambr
@ -639,7 +639,7 @@
        'details': dict({
          'brightness': dict({
            'name': 'brightness',
-            'text': '100%',
+            'text': '100',
            'value': 100,
          }),
          'name': dict({
@ -654,7 +654,7 @@
        'match': True,
        'sentence_template': '[<numeric_value_set>] <name> brightness [to] <brightness>',
        'slots': dict({
-          'brightness': '100%',
+          'brightness': '100',
          'name': 'test light',
        }),
        'source': 'builtin',
--- a/tests/components/conversation/test_default_agent.py
+++ b/tests/components/conversation/test_default_agent.py
@ -770,8 +770,8 @@ async def test_error_no_device_on_floor_exposed(
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "turn on test light on the ground floor", None, Context(), None
@ -838,8 +838,8 @@ async def test_error_no_domain(hass: HomeAssistant) -> None:
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "turn on the fans", None, Context(), None
@ -873,8 +873,8 @@ async def test_error_no_domain_exposed(hass: HomeAssistant) -> None:
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "turn on the fans", None, Context(), None
@ -1047,8 +1047,8 @@ async def test_error_no_device_class(hass: HomeAssistant) -> None:
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "open the windows", None, Context(), None
@ -1096,8 +1096,8 @@ async def test_error_no_device_class_exposed(hass: HomeAssistant) -> None:
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "open all the windows", None, Context(), None
@ -1207,8 +1207,8 @@ async def test_error_no_device_class_on_floor_exposed(
    )

    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[recognize_result],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=recognize_result,
    ):
        result = await conversation.async_converse(
            hass, "open ground floor windows", None, Context(), None
@ -1229,8 +1229,8 @@ async def test_error_no_device_class_on_floor_exposed(
 async def test_error_no_intent(hass: HomeAssistant) -> None:
    """Test response with an intent match failure."""
    with patch(
-        "homeassistant.components.conversation.default_agent.recognize_all",
-        return_value=[],
+        "homeassistant.components.conversation.default_agent.recognize_best",
+        return_value=None,
    ):
        result = await conversation.async_converse(
            hass, "do something", None, Context(), None
--- a/tests/components/conversation/test_trace.py
+++ b/tests/components/conversation/test_trace.py
@ -56,7 +56,7 @@ async def test_converation_trace(
        "intent_name": "HassListAddItem",
        "slots": {
            "name": "Shopping List",
-            "item": "apples ",
+            "item": "apples",
        },
    }