gaard-api 0.1.3__tar.gz → 0.2.1__tar.gz

{gaard_api-0.1.3/src/gaard_api.egg-info → gaard_api-0.2.1}/PKG-INFO

@@ -1,13 +1,13 @@
 Metadata-Version: 2.4
 Name: gaard-api
-Version: 0.1.3
+Version: 0.2.1
 Summary: GAARD backend web services providing admin interface
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown
-Requires-Dist: gaard-core==0.1.0
-Requires-Dist: gaard-connectors==0.1.1
-Requires-Dist: gaard-llm==0.1.0
-Requires-Dist: gaard-plugin-api<0.2.0,>=0.1.0
+Requires-Dist: gaard-core==0.2.1
+Requires-Dist: gaard-connectors==0.2.1
+Requires-Dist: gaard-llm==0.2.1
+Requires-Dist: gaard-plugin-api<0.3.0,>=0.2.0
 Requires-Dist: fastapi>=0.111.0
 Requires-Dist: uvicorn[standard]>=0.30.0
 Requires-Dist: pydantic>=2.7.0

{gaard_api-0.1.3 → gaard_api-0.2.1}/pyproject.toml

@@ -4,16 +4,16 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "gaard-api"
-version = "0.1.3"
+version = "0.2.1"
 description = "GAARD backend web services providing admin interface"
 readme = "README.md"
 requires-python = ">=3.11"
 
 dependencies = [
-    "gaard-core==0.1.0",
-    "gaard-connectors==0.1.1",
-    "gaard-llm==0.1.0",
-    "gaard-plugin-api>=0.1.0,<0.2.0",
+    "gaard-core==0.2.1",
+    "gaard-connectors==0.2.1",
+    "gaard-llm==0.2.1",
+    "gaard-plugin-api>=0.2.0,<0.3.0",
     "fastapi>=0.111.0",
     "uvicorn[standard]>=0.30.0",
     "pydantic>=2.7.0",

{gaard_api-0.1.3 → gaard_api-0.2.1}/src/gaard_api/admin/database.py

@@ -1,7 +1,7 @@
 from collections.abc import Iterator
 import json
 
-from sqlalchemy import create_engine, select
+from sqlalchemy import create_engine, delete, select
 from sqlalchemy.engine import Engine
 from sqlalchemy import inspect, text
 from sqlalchemy.orm import Session, sessionmaker
@@ -24,6 +24,7 @@ from gaard_api.core.settings import settings
 _engine: Engine | None = None
 _session_factory: sessionmaker[Session] | None = None
 _engine_url: str | None = None
+LEGACY_PROMPT_KEYS = {"investigation_readiness"}
 
 
 def get_engine() -> Engine:
@@ -105,10 +106,12 @@ def seed_settings(session: Session) -> None:
         "gaard_sql_generation_mode": settings.gaard_sql_generation_mode,
         "gaard_result_interpretation_mode": settings.gaard_result_interpretation_mode,
         "gaard_output_classification_mode": settings.gaard_output_classification_mode,
-        "gaard_investigation_mode": settings.gaard_investigation_mode,
-        "gaard_investigation_ambiguity_mode": settings.gaard_investigation_ambiguity_mode,
         "gaard_query_max_rows": str(settings.gaard_query_max_rows),
         "gaard_query_timeout_seconds": str(settings.gaard_query_timeout_seconds),
+        "gaard_analysis_loop_count": str(settings.gaard_analysis_loop_count),
+        "gaard_analysis_auto_enable_business_logic": (
+            "true" if settings.gaard_analysis_auto_enable_business_logic else "false"
+        ),
         "gaard_llm_provider": settings.gaard_llm_provider,
         "gaard_llm_base_url": settings.gaard_llm_base_url,
         "gaard_llm_api_key": settings.gaard_llm_api_key,
@@ -154,6 +157,10 @@ def apply_runtime_settings(session: Session) -> None:
 
 
 def seed_prompts(session: Session) -> None:
+    session.execute(
+        delete(PromptTemplate).where(PromptTemplate.prompt_key.in_(LEGACY_PROMPT_KEYS))
+    )
+
     for prompt in DEFAULT_PROMPTS:
         existing = session.scalar(
             select(PromptTemplate).where(PromptTemplate.prompt_key == prompt["prompt_key"])
@@ -399,6 +406,15 @@ def ensure_data_query_audit_schema(engine: Engine) -> None:
                 )
             )
 
+    if "llm_sql_language" not in columns:
+        with engine.begin() as connection:
+            connection.execute(
+                text(
+                    "ALTER TABLE data_query_audit_logs "
+                    "ADD COLUMN llm_sql_language VARCHAR(50) DEFAULT ''"
+                )
+            )
+
     with engine.begin() as connection:
         for index in DataQueryAuditLog.__table__.indexes:
             index.create(bind=connection, checkfirst=True)

{gaard_api-0.1.3 → gaard_api-0.2.1}/src/gaard_api/admin/defaults.py

@@ -86,66 +86,6 @@ Return one JSON object with:
 - reason: short explanation
 """
 
-DEFAULT_INVESTIGATION_READINESS_SYSTEM_PROMPT = """You are GAARD Investigation Readiness.
-
-Your task is to decide whether GAARD already knows enough to create a correct SQL query for the user's question.
-
-Assume nothing. Verify continuously.
-
-Use only:
-- the user's question,
-- the active datasource schema,
-- the approved or previously saved business logic supplied in the payload.
-
-You do not generate SQL.
-You do not answer the user.
-You decide only whether normal SQL generation may start safely.
-
-Return ready_for_sql=true only when all information needed for correct SQL is explicit in the question, schema, and business logic:
-- requested business entity or metric,
-- relevant tables, views and columns,
-- required filters and dictionary/status values,
-- required joins or relationships,
-- requested output shape such as count, list, detail, or aggregation.
-
-Return ready_for_sql=false when any material element is missing, ambiguous, inferred only from the model, or would require checking data values before SQL can be trusted. In that case route must be analysis.
-
-Output rules:
-- Return only a JSON object.
-- Do not include markdown.
-- Do not include reasoning outside the JSON.
-- Do not include <think> blocks.
-- Use exactly this JSON shape:
-  {"ready_for_sql":false,"route":"analysis","confidence":0.0,"reason":"short reason","missing_information":[],"required_analysis":[],"required_analysis_tasks":[],"assumptions":[]}
-
-Required analysis task shape:
-{"missing_information":"what is missing","required_analysis":"specific read-only data question for SQL analysis","category":"dictionary_value","expected_output":"what kind of result would resolve this"}
-
-Allowed categories:
-- dictionary_value
-- relationship_logic
-- filter_logic
-- aggregation_logic
-- entity_mapping
-- unknown
-"""
-
-DEFAULT_INVESTIGATION_READINESS_USER_PROMPT = """Assess whether normal SQL generation can start.
-
-Input JSON:
-{payload}
-
-Return one JSON object with:
-- ready_for_sql: boolean
-- route: sql or analysis
-- confidence: number from 0 to 1
-- reason: short explanation
-- missing_information: list of missing or ambiguous items
-- required_analysis: list of checks that Analysis mode should perform when ready_for_sql=false
-- required_analysis_tasks: list of structured SQL-analysis tasks with missing_information, required_analysis, category, expected_output
-- assumptions: list of any assumptions that would affect SQL correctness
-"""
-
 DEFAULT_RESULT_INTERPRETATION_SYSTEM_PROMPT = """You are GAARD Data Result Interpreter.
 
 Your task is to explain SQL query results to the user.
@@ -247,13 +187,6 @@ DEFAULT_PROMPTS = [
         "system_prompt": DEFAULT_SQL_GENERATION_SYSTEM_PROMPT,
         "user_prompt_template": DEFAULT_SQL_GENERATION_USER_PROMPT,
     },
-    {
-        "prompt_key": "investigation_readiness",
-        "name": "Investigation: readiness",
-        "description": "Decides whether Investigation can safely delegate to normal SQL generation.",
-        "system_prompt": DEFAULT_INVESTIGATION_READINESS_SYSTEM_PROMPT,
-        "user_prompt_template": DEFAULT_INVESTIGATION_READINESS_USER_PROMPT,
-    },
     {
         "prompt_key": "result_interpretation",
         "name": "Result interpretation",

{gaard_api-0.1.3 → gaard_api-0.2.1}/src/gaard_api/admin/models.py

@@ -89,6 +89,7 @@ class DataQueryAuditLog(Base):
     question: Mapped[str] = mapped_column(Text)
     answer: Mapped[str] = mapped_column(Text)
     sql: Mapped[str] = mapped_column(Text)
+    llm_sql_language: Mapped[str] = mapped_column(String(50), default="")
     output_classification: Mapped[OutputClassification] = mapped_column(
         SAEnum(
             OutputClassification,
@@ -251,3 +252,22 @@ class BusinessKnowledgeClaim(Base):
         onupdate=utc_now,
     )
     updated_by: Mapped[str] = mapped_column(String(255), default="system")
+
+
+class AnalysisSessionRecord(Base):
+    __tablename__ = "analysis_sessions"
+
+    id: Mapped[int] = mapped_column(Integer, primary_key=True)
+    session_id: Mapped[str] = mapped_column(String(64), unique=True, index=True)
+    status: Mapped[str] = mapped_column(String(50), index=True, default="running")
+    user_id: Mapped[str] = mapped_column(String(255), index=True)
+    datasource_id: Mapped[str] = mapped_column(String(255), index=True)
+    question: Mapped[str] = mapped_column(Text)
+    context_json: Mapped[str] = mapped_column(Text, default="{}")
+    events_json: Mapped[str] = mapped_column(Text, default="[]")
+    created_at: Mapped[datetime] = mapped_column(DateTime(timezone=True), default=utc_now)
+    updated_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True),
+        default=utc_now,
+        onupdate=utc_now,
+    )

{gaard_api-0.1.3 → gaard_api-0.2.1}/src/gaard_api/admin/prompt_runtime.py

@@ -1,5 +1,4 @@
 from gaard_core.errors import ConfigurationError
-from gaard_core.investigation.models import InvestigationContext
 from gaard_core.json_utils import json_dumps
 from gaard_core.prompt_compiler.models import CompiledPrompt, SqlGenerationPromptRequest
 from gaard_core.prompt_compiler.schema_formatter import SchemaPromptFormatter
@@ -54,9 +53,7 @@ class MetadataSqlGenerationPromptCompiler:
             return request.formatted_schema
 
         if request.database_schema is None:
-            raise ConfigurationError(
-                "Either database_schema or formatted_schema must be provided."
-            )
+            raise ConfigurationError("Either database_schema or formatted_schema must be provided.")
 
         return self.schema_formatter.format(request.database_schema)
 
@@ -88,37 +85,6 @@ class MetadataIntentClassificationPromptCompiler:
         )
 
 
-class MetadataInvestigationReadinessPromptCompiler:
-    def __init__(self, prompt_template: PromptTemplate) -> None:
-        self.prompt_template = prompt_template
-
-    def compile(self, context: InvestigationContext) -> CompiledPrompt:
-        payload = {
-            "question": context.question,
-            "datasource_id": context.datasource_id,
-            "user_id": context.user_id,
-            "schema": context.formatted_schema,
-            "business_logic": context.business_logic,
-        }
-        payload_json = json_dumps(payload, ensure_ascii=False, indent=2)
-
-        return CompiledPrompt(
-            system_prompt=self.prompt_template.system_prompt,
-            user_prompt=self.prompt_template.user_prompt_template.format(
-                payload=payload_json,
-                question=context.question,
-                datasource_id=context.datasource_id,
-                user_id=context.user_id,
-                schema=context.formatted_schema,
-                business_logic=context.business_logic,
-            ),
-            metadata={
-                "prompt_key": self.prompt_template.prompt_key,
-                "prompt_version": self.prompt_template.version,
-            },
-        )
-
-
 class MetadataResultInterpretationPromptCompiler:
     def __init__(self, prompt_template: PromptTemplate) -> None:
         self.prompt_template = prompt_template
@@ -204,17 +170,6 @@ def get_intent_classification_prompt_compiler() -> (
     return MetadataIntentClassificationPromptCompiler(prompt_template=prompt_template)
 
 
-def get_investigation_readiness_prompt_compiler() -> (
-    MetadataInvestigationReadinessPromptCompiler | None
-):
-    prompt_template = get_active_prompt_template_safe("investigation_readiness")
-
-    if prompt_template is None:
-        return None
-
-    return MetadataInvestigationReadinessPromptCompiler(prompt_template=prompt_template)
-
-
 def get_result_interpretation_prompt_compiler() -> (
     MetadataResultInterpretationPromptCompiler | None
 ):