truthound-dashboard 1.4.4__py3-none-any.whl → 1.5.1__py3-none-any.whl

truthound_dashboard/api/profile.py

@@ -1,6 +1,10 @@
 """Profile API endpoints.
 
 This module provides endpoints for data profiling and comparison.
+
+Note: truthound's th.profile() only supports (data, source) parameters.
+Advanced options like sampling strategies, pattern detection configuration,
+and correlation analysis are NOT supported by the underlying library.
 """
 
 from __future__ import annotations
@@ -11,12 +15,12 @@ from fastapi import APIRouter, HTTPException, Path, Query
 
 from truthound_dashboard.schemas import (
     LatestComparisonResponse,
+    ProfileAdvancedRequest,
     ProfileComparisonRequest,
     ProfileComparisonResponse,
     ProfileListResponse,
     ProfileRequest,
     ProfileResponse,
-    ProfileTrendRequest,
     ProfileTrendResponse,
 )
 
@@ -29,7 +33,7 @@ router = APIRouter()
     "/sources/{source_id}/profile",
     response_model=ProfileResponse,
     summary="Profile source",
-    description="Run data profiling on a source with optional sampling and pattern detection",
+    description="Run data profiling on a source",
 )
 async def profile_source(
     service: ProfileServiceDep,
@@ -39,19 +43,17 @@ async def profile_source(
 ) -> ProfileResponse:
     """Run data profiling on a source.
 
-    Supports advanced configuration including:
-    - Sampling strategies: none, head, random, systematic, stratified, reservoir, adaptive, hash
-    - Pattern detection: email, phone, uuid, url, ip_address, credit_card, etc.
-    - Statistical analysis options: histograms, correlations, cardinality
+    Note: truthound's th.profile() only supports (data, source) parameters.
+    Advanced configuration options are not supported.
 
     Args:
         service: Injected profile service.
         source_service: Injected source service.
         source_id: Source to profile.
-        request: Optional profiling configuration.
+        request: Optional request body (not used, kept for API compatibility).
 
     Returns:
-        Profiling result with column statistics, detected patterns, and sampling metadata.
+        Profiling result with column statistics.
 
     Raises:
         HTTPException: 404 if source not found, 500 on profiling error.
@@ -61,38 +63,73 @@ async def profile_source(
     if source is None:
         raise HTTPException(status_code=404, detail="Source not found")
 
-    # Build profiling kwargs from request
-    profile_kwargs: dict = {}
-
-    if request:
-        # Handle sampling configuration
-        if request.sampling:
-            # Advanced sampling config takes precedence
-            profile_kwargs["sampling_strategy"] = request.sampling.strategy
-            profile_kwargs["sample_size"] = request.sampling.sample_size
-            profile_kwargs["confidence_level"] = request.sampling.confidence_level
-            profile_kwargs["margin_of_error"] = request.sampling.margin_of_error
-            profile_kwargs["strata_column"] = request.sampling.strata_column
-            profile_kwargs["seed"] = request.sampling.seed
-        elif request.sample_size:
-            # Backward compatible simple sample_size
-            profile_kwargs["sample_size"] = request.sample_size
-
-        # Handle pattern detection configuration
-        if request.pattern_detection:
-            profile_kwargs["enable_pattern_detection"] = request.pattern_detection.enabled
-            profile_kwargs["pattern_sample_size"] = request.pattern_detection.sample_size
-            profile_kwargs["min_pattern_confidence"] = request.pattern_detection.min_confidence
-            profile_kwargs["patterns_to_detect"] = request.pattern_detection.patterns_to_detect
-
-        # Additional profiling options
-        profile_kwargs["include_histograms"] = request.include_histograms
-        profile_kwargs["include_correlations"] = request.include_correlations
-        profile_kwargs["include_cardinality"] = request.include_cardinality
+    try:
+        result = await service.profile_source(source_id)
+        return ProfileResponse.from_result(result)
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@router.post(
+    "/sources/{source_id}/profile/advanced",
+    response_model=ProfileResponse,
+    summary="Profile source with advanced configuration",
+    description="Run data profiling with custom ProfilerConfig options",
+)
+async def profile_source_advanced(
+    service: ProfileServiceDep,
+    source_service: SourceServiceDep,
+    source_id: Annotated[str, Path(description="Source ID to profile")],
+    request: ProfileAdvancedRequest,
+) -> ProfileResponse:
+    """Run advanced data profiling with custom configuration.
+
+    Uses truthound's ProfilerConfig for fine-grained control over:
+    - Sampling: sample_size, random_seed
+    - Features: include_patterns, include_correlations, include_distributions
+    - Pattern detection: pattern_sample_size, min_pattern_match_ratio
+    - Output: top_n_values, correlation_threshold
+    - Performance: n_jobs
+
+    Args:
+        service: Injected profile service.
+        source_service: Injected source service.
+        source_id: Source to profile.
+        request: Advanced profiling configuration.
+
+    Returns:
+        Profiling result with column statistics.
+
+    Raises:
+        HTTPException: 404 if source not found, 501 if not supported, 500 on error.
+    """
+    # Verify source exists
+    source = await source_service.get_by_id(source_id)
+    if source is None:
+        raise HTTPException(status_code=404, detail="Source not found")
 
     try:
-        result = await service.profile_source(source_id, **profile_kwargs)
+        # Convert request to config dict
+        config = {
+            "sample_size": request.sample_size,
+            "random_seed": request.random_seed,
+            "include_patterns": request.include_patterns,
+            "include_correlations": request.include_correlations,
+            "include_distributions": request.include_distributions,
+            "top_n_values": request.top_n_values,
+            "pattern_sample_size": request.pattern_sample_size,
+            "correlation_threshold": request.correlation_threshold,
+            "min_pattern_match_ratio": request.min_pattern_match_ratio,
+            "n_jobs": request.n_jobs,
+        }
+        result = await service.profile_source_advanced(source_id, config=config)
         return ProfileResponse.from_result(result)
+    except ImportError as e:
+        raise HTTPException(
+            status_code=501,
+            detail=f"Advanced profiling not available: {e}. "
+            "Please upgrade truthound to the latest version.",
+        )
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
 
@@ -102,6 +139,41 @@ async def profile_source(
 # =============================================================================
 
 
+@router.get(
+    "/sources/{source_id}/profile/latest",
+    response_model=ProfileResponse | None,
+    summary="Get latest profile",
+    description="Retrieve the most recent profile result for a source",
+)
+async def get_latest_profile(
+    service: ProfileServiceDep,
+    source_service: SourceServiceDep,
+    source_id: Annotated[str, Path(description="Source ID")],
+) -> ProfileResponse | None:
+    """Get the latest profile for a source.
+
+    Args:
+        service: Injected profile service.
+        source_service: Injected source service.
+        source_id: Source ID.
+
+    Returns:
+        Latest profile result or null.
+
+    Raises:
+        HTTPException: 404 if source not found.
+    """
+    source = await source_service.get_by_id(source_id)
+    if source is None:
+        raise HTTPException(status_code=404, detail="Source not found")
+
+    profile = await service.get_latest_profile(source_id)
+    if profile is None:
+        return None
+
+    return ProfileResponse.from_result(profile)
+
+
 @router.get(
     "/sources/{source_id}/profiles",
     response_model=ProfileListResponse,
@@ -255,3 +327,93 @@ async def get_latest_profile_comparison(
         raise HTTPException(status_code=404, detail="Source not found")
 
     return await comparison_service.get_latest_comparison(source)
+
+
+# =============================================================================
+# Rule Generation from Profile
+# =============================================================================
+
+
+@router.post(
+    "/sources/{source_id}/profiles/generate-rules",
+    summary="Generate validation rules from profile",
+    description="Automatically generate validation rules based on profiled data characteristics.",
+)
+async def generate_rules_from_profile(
+    source_id: Annotated[str, Path(description="Source ID")],
+    service: ProfileServiceDep,
+    source_service: SourceServiceDep,
+    strictness: str = Query(
+        default="medium",
+        description="Rule strictness: loose, medium, strict",
+    ),
+    preset: str = Query(
+        default="default",
+        description="Rule preset: default, strict, loose, minimal, comprehensive, ci_cd, schema_only, format_only",
+    ),
+    include_categories: list[str] | None = Query(
+        default=None,
+        description="Rule categories to include (schema, stats, pattern, completeness, uniqueness, distribution)",
+    ),
+    exclude_categories: list[str] | None = Query(
+        default=None,
+        description="Rule categories to exclude",
+    ),
+    profile_if_needed: bool = Query(
+        default=True,
+        description="Profile source if no recent profile exists",
+    ),
+    sample_size: int | None = Query(
+        default=None,
+        description="Sample size for profiling if needed",
+    ),
+) -> dict:
+    """Generate validation rules from source profile.
+
+    Uses truthound's generate_suite() to automatically create validation
+    rules based on the profiled data characteristics.
+
+    The generated rules can be used directly with th.check() or saved
+    as a schema file for repeated validation.
+
+    Args:
+        source_id: Source ID to generate rules for.
+        service: Profile service.
+        source_service: Source service.
+        strictness: Rule strictness level.
+        preset: Rule generation preset.
+        include_categories: Rule categories to include.
+        exclude_categories: Rule categories to exclude.
+        profile_if_needed: Profile source if no recent profile exists.
+        sample_size: Sample size for profiling if needed.
+
+    Returns:
+        Generated rules with YAML content and metadata.
+
+    Raises:
+        HTTPException: 404 if source not found, 400 if no profile available.
+    """
+    # Verify source exists
+    source = await source_service.get_by_id(source_id)
+    if source is None:
+        raise HTTPException(status_code=404, detail="Source not found")
+
+    try:
+        result = await service.generate_rules_from_profile(
+            source_id,
+            strictness=strictness,
+            preset=preset,
+            include_categories=include_categories,
+            exclude_categories=exclude_categories,
+            profile_if_needed=profile_if_needed,
+            sample_size=sample_size,
+        )
+        return result
+    except ValueError as e:
+        raise HTTPException(status_code=400, detail=str(e))
+    except ImportError as e:
+        raise HTTPException(
+            status_code=501,
+            detail=f"Rule generation not available: {e}. "
+            "Please upgrade truthound to the latest version.",
+        )