Add Pydantic-based config validation and type correction#310
Add Pydantic-based config validation and type correction#310
Conversation
MaStr
changed the title
MaStr
added
enhancement
There was a problem hiding this comment.
Pull request overview
This PR introduces centralized configuration validation/coercion via new Pydantic models, aiming to handle Home Assistant “numbers as strings” issues at config load time and remove scattered runtime conversions across Batcontrol.
Changes:
- Added
src/batcontrol/config_model.pywith Pydantic models and avalidate_config()entrypoint. - Integrated Pydantic validation into
load_config()and removed several downstream coercions/parsers (core time resolution, tariff float casting, HA consumption list parsing, inverter legacy rename). - Added unit/integration tests covering validation, coercion, backward compatibility, and
load_config()behavior; addedpydanticdependency.
Reviewed changes
Copilot reviewed 10 out of 10 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
| tests/batcontrol/test_core.py | Updates time-resolution tests to reflect validation happening before Batcontrol init. |
| tests/batcontrol/test_config_model.py | Adds unit tests for Pydantic config models and coercion/validation behavior. |
| tests/batcontrol/test_config_load_integration.py | Adds integration tests for YAML load_config() + validation. |
| src/batcontrol/setup.py | Calls validate_config() during config loading. |
| src/batcontrol/inverter/inverter.py | Removes legacy field/default handling; adds cache_ttl wiring for MQTT inverter. |
| src/batcontrol/forecastconsumption/consumption.py | Removes semicolon-list parsing and int coercion (now expected from validation). |
| src/batcontrol/dynamictariff/dynamictariff.py | Removes float() casting; assumes validated numeric types. |
| src/batcontrol/core.py | Removes time-resolution string coercion/validation; assumes validated config. |
| src/batcontrol/config_model.py | New Pydantic models + validate_config() for centralized validation/coercion. |
| pyproject.toml | Adds pydantic as a dependency. |
There was a problem hiding this comment.
Pull request overview
Adds a centralized, Pydantic v2–based configuration validation layer so config type coercion/validation happens once at load time (helping address HA addon “numbers as strings” issues) and downstream modules can rely on properly typed config values.
Changes:
- Introduces
src/batcontrol/config_model.pywith Pydantic models plusvalidate_config()and wires it intoload_config(). - Removes scattered runtime type conversions in
core.py,dynamictariff.py,forecastconsumption/consumption.py, and legacy mapping ininverter.py. - Adds extensive unit + integration tests for config validation/coercion and updates existing tests accordingly.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
src/batcontrol/config_model.py |
New Pydantic models + validate_config() entry point to coerce/validate config. |
src/batcontrol/setup.py |
Calls validate_config() during load_config() to centralize validation. |
src/batcontrol/core.py |
Removes time_resolution_minutes coercion/validation, relying on validated config. |
src/batcontrol/dynamictariff/dynamictariff.py |
Removes float() casting for utility numeric fields, relying on validated config. |
src/batcontrol/forecastconsumption/consumption.py |
Removes manual parsing for history_days/history_weights (now intended to be handled by validation). |
src/batcontrol/inverter/inverter.py |
Removes legacy rename/default handling (moved to Pydantic) and passes cache_ttl for MQTT inverter. |
tests/batcontrol/test_config_model.py |
New unit tests for models, coercions, constraints, and validate_config(). |
tests/batcontrol/test_config_load_integration.py |
New integration tests verifying load_config() + YAML loading + validation behavior. |
tests/batcontrol/test_core.py |
Updates time-resolution tests to reflect validation occurring before Batcontrol.__init__(). |
pyproject.toml |
Adds pydantic>=2.0,<3.0 dependency. |
docs/pydantic-config-migration.md |
Migration notes, design decisions, and follow-ups. |
There was a problem hiding this comment.
Pull request overview
Adds centralized configuration validation/coercion using Pydantic so config type fixes happen once at load time (addressing Home Assistant addon string-typed numeric values) and downstream modules can rely on normalized types.
Changes:
- Introduces
BatcontrolConfigand subsystem config models +validate_config()for type coercion and validation. - Integrates validation into
load_config()and removes scattered manual coercions in core/tariff/consumption/inverter code paths. - Adds unit + integration tests and migration documentation.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
src/batcontrol/config_model.py |
New Pydantic models + validate_config() to validate/coerce loaded config. |
src/batcontrol/setup.py |
Calls validate_config() during config load. |
src/batcontrol/core.py |
Removes local time resolution coercion/validation, assumes validated config. |
src/batcontrol/dynamictariff/dynamictariff.py |
Removes float() casts, relies on validated float types. |
src/batcontrol/forecastconsumption/consumption.py |
Simplifies list parsing; keeps fallback parsing for nested HA config. |
src/batcontrol/inverter/inverter.py |
Removes legacy rename/default handling (now in Pydantic); passes cache_ttl. |
tests/batcontrol/test_config_model.py |
New unit tests for models and coercion behavior. |
tests/batcontrol/test_config_load_integration.py |
New integration tests for load_config() + YAML loading. |
tests/batcontrol/test_core.py |
Updates time-resolution test to reflect validated-config expectations. |
pyproject.toml |
Adds pydantic>=2.0,<3.0 runtime dependency. |
docs/pydantic-config-migration.md |
Documents migration status, design decisions, and remaining follow-ups. |
There was a problem hiding this comment.
Pull request overview
Adds centralized configuration validation/coercion using Pydantic to address Home Assistant add-on string-typed options and remove scattered runtime type conversions across Batcontrol.
Changes:
- Introduces
src/batcontrol/config_model.pyPydantic v2 models +validate_config()to validate/coerce config at load time. - Integrates validation into
setup.load_config()and simplifies downstream modules (core.py,dynamictariff.py,inverter.py,forecastconsumption/consumption.py) by removing redundant casts/renames. - Adds extensive unit + integration tests for config validation and YAML load behavior.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
src/batcontrol/config_model.py |
New Pydantic models + validate_config() that returns a coerced dict (model_dump(exclude_none=True)). |
src/batcontrol/setup.py |
Calls validate_config() during config load before further checks. |
src/batcontrol/core.py |
Removes local time_resolution_minutes coercion/validation and relies on validated config. |
src/batcontrol/dynamictariff/dynamictariff.py |
Removes float casts for utility pricing parameters assuming validated types. |
src/batcontrol/inverter/inverter.py |
Removes legacy rename/default handling (moved to Pydantic) and adds cache_ttl passthrough for MQTT inverter config. |
src/batcontrol/forecastconsumption/consumption.py |
Adjusts parsing/coercion for nested HA config cases and ensures numeric types for annual consumption. |
tests/batcontrol/test_config_model.py |
New unit tests covering model defaults, coercions, validators, and validate_config() output shape. |
tests/batcontrol/test_config_load_integration.py |
New integration tests validating YAML load + Pydantic coercion through load_config(). |
tests/batcontrol/test_core.py |
Updates time-resolution test expectations to reflect load-time validation behavior. |
pyproject.toml |
Adds pydantic>=2.0,<3.0 dependency. |
docs/pydantic-config-migration.md |
Adds migration/status documentation for the new validation layer. |
There was a problem hiding this comment.
Pull request overview
Adds centralized, Pydantic v2–based configuration validation/coercion at config load time to address Home Assistant “numbers as strings” issues and reduce scattered runtime type conversions across Batcontrol.
Changes:
- Introduces
BatcontrolConfigand subsystem config models insrc/batcontrol/config_model.py, plusvalidate_config()returning a plain validated dict (exclude_none=True). - Integrates validation into
setup.load_config()and removes/reduces downstream manual coercions (e.g.,core.py,dynamictariff.py,inverter.py,forecastconsumption/consumption.py). - Adds extensive unit + integration tests covering coercion, defaults, validation errors, and YAML load behavior.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| tests/batcontrol/test_core.py | Updates time-resolution tests to reflect validation happening at load time. |
| tests/batcontrol/test_config_model.py | Adds unit tests for new Pydantic config models and coercion/validation behavior. |
| tests/batcontrol/test_config_load_integration.py | Adds integration tests for load_config() validating/coercing YAML configs. |
| src/batcontrol/setup.py | Calls validate_config() after YAML load and before PV installation checks. |
| src/batcontrol/inverter/inverter.py | Removes legacy/default handling now covered by Pydantic; passes cache_ttl for MQTT inverter. |
| src/batcontrol/forecastconsumption/consumption.py | Adjusts parsing/coercion to handle nested HA config edge cases and numeric strings. |
| src/batcontrol/dynamictariff/dynamictariff.py | Removes float casts assuming config has already been coerced by Pydantic. |
| src/batcontrol/core.py | Removes manual time_resolution_minutes coercion/validation and uses validated value. |
| src/batcontrol/config_model.py | New Pydantic models and validate_config() entry point. |
| pyproject.toml | Adds pydantic>=2.0,<3.0 dependency. |
| docs/pydantic-config-migration.md | Documents migration status, design decisions, and remaining follow-ups. |
There was a problem hiding this comment.
Pull request overview
This PR centralizes configuration validation/coercion by introducing Pydantic v2 models and wiring them into config loading, reducing scattered runtime type conversions (notably for Home Assistant add-on stringly-typed options).
Changes:
- Added
BatcontrolConfig+ subsystem config models and avalidate_config()entry point to validate/coerce raw config dicts at load time. - Integrated validation into
setup.load_config()and simplified downstream modules (core, inverter factory, dynamic tariff factory, consumption forecast factory) by removing redundant casts/legacy handling. - Added extensive unit + integration tests for config validation and YAML loading, plus a migration note doc.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
src/batcontrol/config_model.py |
New Pydantic models + validation entry point for centralized config coercion. |
src/batcontrol/setup.py |
Calls validate_config() during config load; adds basic YAML shape check. |
src/batcontrol/core.py |
Removes local time_resolution_minutes coercion/validation and relies on validated config. |
src/batcontrol/inverter/inverter.py |
Drops legacy key mutation/defaulting in factory; passes through cache_ttl. |
src/batcontrol/dynamictariff/dynamictariff.py |
Removes float() casts for already-coerced utility config values. |
src/batcontrol/forecastconsumption/consumption.py |
Keeps parsing for nested HA config dicts; adjusts coercions for numeric/list fields. |
tests/batcontrol/test_config_model.py |
New unit tests for model defaults, coercion, constraints, and backwards-compat rename. |
tests/batcontrol/test_config_load_integration.py |
New integration tests covering YAML load + validation behavior. |
tests/batcontrol/test_core.py |
Updates time resolution tests to reflect validation now happening in load_config(). |
pyproject.toml |
Adds pydantic>=2,<3 dependency. |
docs/pydantic-config-migration.md |
Documents migration status, design decisions, and remaining follow-ups. |
src/batcontrol/config_model.py
Outdated
| battery_control: BatteryControlConfig = BatteryControlConfig() | ||
| battery_control_expert: Optional[BatteryControlExpertConfig] = None | ||
| inverter: InverterConfig = InverterConfig() | ||
| utility: UtilityConfig | ||
| mqtt: Optional[MqttConfig] = None | ||
| evcc: Optional[EvccConfig] = None | ||
| pvinstallations: List[PvInstallationConfig] | ||
| consumption_forecast: ConsumptionForecastConfig = ( | ||
| ConsumptionForecastConfig() | ||
| ) |
| # Coerce numeric fields that may arrive as strings from nested HA config | ||
| cache_ttl_hours = float(ha_config.get('cache_ttl_hours', 48.0)) | ||
| multiplier = float(ha_config.get('multiplier', 1.0)) |
| # in config_model.py (must be int, must be 15 or 60) | ||
| self.time_resolution = config.get('time_resolution_minutes', 60) | ||
|
|
Round 5 Review Feedback - All items addressedAddressed the 3 remaining comments from the latest review round: 1. Mutable defaults in BatcontrolConfig (✅ Fixed)Comment: Action: Switched to 2.
|
There was a problem hiding this comment.
Pull request overview
Introduces centralized, Pydantic v2–based configuration validation/coercion at config-load time to remove scattered runtime type conversions and improve robustness (notably for Home Assistant add-on stringified numeric options).
Changes:
- Added
BatcontrolConfigand subsystem config models plus avalidate_config()entry point to validate/coerce raw config dicts. - Integrated validation into
setup.load_config()and simplified downstream modules by removing redundant casts/parsing. - Added extensive unit + integration test coverage for validation and load-time behavior, plus migration documentation.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
src/batcontrol/config_model.py |
New Pydantic models + validate_config() to validate/coerce config at load time |
src/batcontrol/setup.py |
Calls validate_config() after YAML load and adds basic YAML-shape validation |
src/batcontrol/core.py |
Removes string→int coercion logic for time_resolution_minutes, relies on validated config |
src/batcontrol/dynamictariff/dynamictariff.py |
Removes redundant float() casting for utility price parameters and tariff zones |
src/batcontrol/forecastconsumption/consumption.py |
Keeps targeted parsing/coercion for nested HA config edge-cases; coerces CSV annual consumption |
src/batcontrol/inverter/inverter.py |
Removes legacy rename/default handling (now in Pydantic) and passes cache_ttl to MQTT inverter |
tests/batcontrol/test_config_model.py |
New unit tests covering model defaults, coercion, constraints, legacy rename |
tests/batcontrol/test_config_load_integration.py |
New integration tests for load_config() + dummy config file loading |
tests/batcontrol/test_core.py |
Updates time-resolution tests to reflect validation happening before Batcontrol.__init__() |
pyproject.toml |
Adds pydantic>=2.0,<3.0 dependency |
docs/pydantic-config-migration.md |
Migration/status notes and remaining follow-ups |
| if not isinstance(config, dict): | ||
| raise RuntimeError(f'Configfile {configfile} is empty or not a valid YAML mapping') | ||
|
|
||
| # Validate and coerce types via Pydantic before any other checks | ||
| config = validate_config(config) | ||
|
|
There was a problem hiding this comment.
load_config() now calls validate_config(), which can raise pydantic.ValidationError, but the function docstring (and callers like main.py) still imply RuntimeError-only behavior. Consider either (a) catching ValidationError and re-raising RuntimeError with a user-friendly message, or (b) updating the docstring/type contract and ensuring the CLI handles ValidationError cleanly (no stack trace for config mistakes).
| """Write config dict to a YAML file and return the path.""" | ||
| path = os.path.join(tmpdir, 'test_config.yaml') | ||
| with open(path, 'w', encoding='UTF-8') as f: | ||
| yaml.dump(config_dict, f) |
There was a problem hiding this comment.
Tests write YAML using yaml.dump(), which can emit Python-specific tags for non-primitive objects and is less constrained than safe_dump(). Using yaml.safe_dump() here would better match load_config()'s yaml.safe_load() behavior and avoid surprising serialization output if test fixtures evolve.
| yaml.dump(config_dict, f) | |
| yaml.safe_dump(config_dict, f) |
Round 6 Review Feedback - All items addressed1.
|
src/batcontrol/setup.py
Outdated
| @@ -66,9 +71,17 @@ def load_config(configfile:str) -> dict: | |||
|
|
|||
| config = yaml.safe_load(config_str) | |||
There was a problem hiding this comment.
load_config() calls yaml.safe_load() directly; malformed YAML will raise a yaml.YAMLError, not the RuntimeError promised by the docstring. Consider catching yaml.YAMLError around the load and re-raising as RuntimeError with a clear message (and from exc).
| config = yaml.safe_load(config_str) | |
| try: | |
| config = yaml.safe_load(config_str) | |
| except yaml.YAMLError as exc: | |
| raise RuntimeError(f'Configfile {configfile} is not valid YAML: {exc}') from exc |
src/batcontrol/config_model.py
Outdated
| type: str = 'dummy' | ||
| address: Optional[str] = None | ||
| user: Optional[str] = None | ||
| password: Optional[str] = None | ||
| max_grid_charge_rate: float = 5000 |
There was a problem hiding this comment.
address/user/password are optional here, but Inverter.create_inverter() later subscripts them (e.g. config['address']) for fronius_gen24. Because validate_config() uses model_dump(exclude_none=True), YAML nulls like address: will be dropped and turn into KeyError at runtime. Add a type-dependent validator to require these fields when type='fronius_gen24' (and fail validation with a clear error).
| # MQTT inverter fields | ||
| capacity: Optional[int] = None | ||
| min_soc: Optional[int] = None | ||
| max_soc: Optional[int] = None | ||
| base_topic: Optional[str] = None | ||
| cache_ttl: int = 120 |
There was a problem hiding this comment.
MQTT inverter fields like capacity are optional in the model, but Inverter.create_inverter() uses config['capacity'] for type='mqtt', which will raise KeyError if omitted (and exclude_none=True will also drop nulls). Consider validating required MQTT fields when type='mqtt' so config errors are caught at load time instead of during runtime inverter creation.
src/batcontrol/config_model.py
Outdated
| name: str = '' | ||
| type: Optional[str] = None | ||
| lat: Optional[float] = None | ||
| lon: Optional[float] = None | ||
| declination: Optional[float] = None |
There was a problem hiding this comment.
PvInstallationConfig.name defaults to an empty string, which means a PV installation missing the name key will still pass validation and end up with name=''. Downstream code uses unit['name'] as a cache key; an empty/default name can silently collide across installations and hide config mistakes. Consider making name required (no default) and/or enforcing min_length=1.
Round 7 Review Feedback - All 4 items addressed1.
|
Round 13 Review Feedback - Addressed
|
Introduces Pydantic models for all configuration sections that validate and coerce types at load time, fixing HA addon string-to-int/float issues (e.g. MQTT/EVCC port, time_resolution_minutes, vat/fees/markup). - Add config_model.py with typed models for all config sections - Integrate validate_config() into load_config() in setup.py - Remove manual type coercion from core.py and dynamictariff.py - Add 33 new tests covering model validation and type coercion - Add pydantic>=2.0 dependency https://claude.ai/code/session_015rEfxnRN99qtwpBmEZ3GpP
…el validation - Add field_validators for history_days/history_weights semicolon string parsing (e.g. "-7;-14;-21" -> [-7, -14, -21]) from HA addon - Add model_validator for legacy max_charge_rate -> max_grid_charge_rate rename - Add loglevel validator (must be debug/info/warning/error, normalized to lowercase) - Remove manual type coercion from consumption.py and inverter.py - Add 10 new tests (367 total passing) https://claude.ai/code/session_015rEfxnRN99qtwpBmEZ3GpP
- Add cache_ttl field to InverterConfig Pydantic model (used by MQTT inverter) - Fix pre-existing bug: cache_ttl was not passed through in inverter factory's MQTT path, always falling back to default 120 https://claude.ai/code/session_015rEfxnRN99qtwpBmEZ3GpP
Documents what was done, what remains, config model reference, and HA addon coercion table. https://claude.ai/code/session_015rEfxnRN99qtwpBmEZ3GpP
…default Fixes all 5 review comments from Copilot: 1. Use model_dump(exclude_none=True) so None optional fields don't appear as keys in the output dict. Downstream code checks key presence (e.g. 'csv' in config, required_fields checks) and would break with None values present. 2. Make pvinstallations required (no default) - Pydantic now raises ValidationError if missing, instead of silently defaulting to []. 3. Give cache_ttl a real default (120) instead of Optional[None] to prevent None propagating into TTLCache. 4. Reorder load_config() to run validate_config() before the pvinstallations check, so Pydantic provides clear error messages. 5. tariff_zone_* None issue resolved by exclude_none=True. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Fixes 2 valid review comments (3 others were already fixed): 1. Restore history_days/history_weights parsing in consumption factory for the nested homeassistant_api dict case. Pydantic only validates top-level ConsumptionForecastConfig fields; when values come from the nested homeassistant_api dict they bypass Pydantic validation and may still be semicolon-separated strings. 2. Copy input dict in InverterConfig model_validator before mutating via pop(). Prevents surprising side effects on callers who reuse the same dict. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…cion, lint Fixes 4 new review comments: 1. Make vat/fees/markup Optional[float]=None in UtilityConfig so downstream required_fields checks in dynamictariff.py can detect missing config (previously defaults of 0.0 masked misconfigurations). 2. Add float() coercion for cache_ttl_hours, multiplier, and annual_consumption in consumption factory for the nested homeassistant_api/csv dict paths that bypass Pydantic validation. 3. Remove unused import tempfile from test_config_load_integration.py. 4. Remove unused import BatteryControlExpertConfig from test_config_model.py. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…cription - vat/fees/markup are Optional[float]=None, not float=0.0 - Remove nonexistent 'entsoe' provider, list actual providers - Clarify that semicolon parsing was kept for nested HA config case - Note exclude_none=True in design decisions - Remove nonexistent sungrow/huawei inverter types Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…lass - setup.py: raise RuntimeError when yaml.safe_load() returns None (empty or non-mapping YAML) before calling validate_config(), giving a clear error instead of a TypeError from **None unpacking - test_core.py: rename TestTimeResolutionString -> TestTimeResolutionValidation and update docstring to reflect tests use int values (string coercion is covered in test_config_model.py) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…sive validation - Use Field(default_factory=...) for nested model defaults in BatcontrolConfig to follow Pydantic best practices (even though v2 deep-copies, this is cleaner) - Guard against None values before float() cast in consumption.py HA config (YAML null values would previously raise TypeError) - Add defensive time_resolution_minutes validation in Batcontrol.__init__ so direct instantiation without load_config() gives a clear error Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…n tests - Catch pydantic.ValidationError in load_config() and re-raise as RuntimeError so callers (e.g. __main__.py) continue to see only RuntimeError as documented - Update docstring to reflect that malformed config raises RuntimeError - Replace yaml.dump() with yaml.safe_dump() in integration test helper to match load_config()'s use of yaml.safe_load() and avoid Python-specific YAML tags - Update test expecting ValidationError to expect RuntimeError (new behavior) - Remove now-unused ValidationError import from integration test file Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ion, name required - Wrap yaml.safe_load() in try/except yaml.YAMLError → RuntimeError so malformed YAML gives a clean error consistent with the rest of load_config() - Add type-dependent model_validator to InverterConfig: fronius_gen24 requires address/user/password; mqtt requires capacity — prevents KeyError at runtime - Make PvInstallationConfig.name required (no default) to prevent empty-string cache key collisions across PV installations - Add 5 new tests covering all new validation paths Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…fix comment - Change max_grid_charge_rate/max_pv_charge_rate/min_pv_charge_rate to int in InverterConfig — they represent watts and are used with %d formatting in logs - Sanitize ValidationError in load_config(): use exc.errors() with only field path and message (no input_value) to avoid leaking secrets like inverter passwords - Fix misleading comment in consumption.py: clarify it's an explicit None check (not 'or default') so it's clear why 0 is handled correctly as a valid value - Update test asserting max_grid_charge_rate is float → int Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
max_grid_charge_rate/max_pv_charge_rate/min_pv_charge_rate are int in the model (watts); update the field reference tree and coercion table accordingly. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
After changing InverterConfig power rate fields to int in round 8, three test assertions still compared against float literals (3000.0, 4000.0). Update them to int and add isinstance(_, int) checks to catch regressions. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…nual_consumption None - InverterConfig.validate_type_required_fields: reject empty/whitespace strings for fronius_gen24 address/user/password (not just None), and reject capacity <= 0 for mqtt (0 or negative is invalid for downstream calculations) - consumption.py CSV factory: guard annual_consumption against None value (YAML null) with 'or 0' pattern — consistent with HA config guard added earlier - Add 2 new tests: mqtt capacity=0 rejected; fronius empty-string fields rejected - Update test match string to align with new error message wording Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…equired fields - PvInstallationConfig: add field_validator for name to reject empty/whitespace strings (used as cache keys in forecastsolar — empty would cause collisions); strip leading/trailing whitespace from valid names - UtilityConfig: add model_validator(mode='after') to enforce provider-specific required fields at config load time instead of at DynamicTariff runtime: * awattar_at/awattar_de: require vat, markup, fees * energyforecast/energyforecast_96: require vat, markup, fees, apikey * tibber: require apikey * evcc: require url * tariff_zones: require tariff_zone_1, zone_1_hours, tariff_zone_2, zone_2_hours - Update all affected tests to provide required fields for their provider type - Add 6 new tests covering name whitespace rejection, name stripping, and provider-specific validation errors (awattar, tibber, evcc) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The dummy config documents timezone as "not optional" and core logic uses it for scheduling/price-window alignment. Silently defaulting to 'Europe/Berlin' masks misconfigured setups. Now fails fast at config load with a clear validation error when timezone is absent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
MaStr
force-pushed
the
claude/pydantic-config-conversion-Qc7UW
branch
from
d1b7f58 to
eb9c525
Compare
- UtilityConfig: update comment to reflect that vat/fees/markup are now validated by validate_provider_required_fields, not by downstream code - dynamictariff.py: add note that required_fields checks are secondary safety guards — primary validation is now in Pydantic at load time Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Post-rebase local reviewRebased on latest Local review found two stale comments addressed in a housekeeping commit:
No functional changes. 397 tests pass. |
Summary
Introduces comprehensive configuration validation using Pydantic models to eliminate scattered type conversion code throughout the codebase and fix Home Assistant addon issues where numeric values arrive as strings.
Key Changes
New Pydantic models (
src/batcontrol/config_model.py):BatcontrolConfig: Top-level configuration with validation fortime_resolution_minutes(15 or 60) andloglevelBatteryControlConfig,InverterConfig,UtilityConfig,MqttConfig,EvccConfig: Specialized config models for each subsystemPvInstallationConfig,ConsumptionForecastConfig: Installation and forecast-specific modelsvalidate_config()function: Entry point for validating raw config dicts and returning validated dictsType coercion at config load time:
history_days="-7;-14;-21"→[-7, -14, -21])max_charge_rate→max_grid_charge_rate)Integration with config loading (
src/batcontrol/setup.py):load_config()now callsvalidate_config()to validate and coerce types before returning the config dictSimplified downstream code:
core.py(time resolution validation)dynamictariff.pyconsumption.pyinverter.pyComprehensive test coverage (
tests/batcontrol/test_config_model.py):Integration tests (
tests/batcontrol/test_config_load_integration.py):load_config()with Pydantic validationNotable Implementation Details
extra='allow'to preserve unknown fields for forward compatibilityhttps://claude.ai/code/session_015rEfxnRN99qtwpBmEZ3GpP
Closes #144