remdb 0.3.146__py3-none-any.whl → 0.3.181__py3-none-any.whl

rem/settings.py

@@ -77,6 +77,7 @@ class LLMSettings(BaseSettings):
         LLM__ANTHROPIC_API_KEY or ANTHROPIC_API_KEY - Anthropic API key
         LLM__EMBEDDING_PROVIDER or EMBEDDING_PROVIDER - Default embedding provider (openai)
         LLM__EMBEDDING_MODEL or EMBEDDING_MODEL - Default embedding model name
+        LLM__DEFAULT_STRUCTURED_OUTPUT - Default structured output mode (False = streaming text)
     """
 
     model_config = SettingsConfigDict(
@@ -138,6 +139,11 @@ class LLMSettings(BaseSettings):
         description="Default embedding model (provider-specific model name)",
     )
 
+    default_structured_output: bool = Field(
+        default=False,
+        description="Default structured output mode for agents. False = streaming text (easier), True = JSON schema validation",
+    )
+
     @field_validator("openai_api_key", mode="before")
     @classmethod
     def validate_openai_api_key(cls, v):
@@ -1028,7 +1034,7 @@ class ChatSettings(BaseSettings):
     - Prevents context window bloat while maintaining conversation continuity
 
     User Context (on-demand by default):
-    - Agent system prompt includes: "User ID: {user_id}. To load user profile: Use REM LOOKUP users/{user_id}"
+    - Agent system prompt includes: "User: {email}. To load user profile: Use REM LOOKUP \"{email}\""
     - Agent decides whether to load profile based on query
     - More efficient for queries that don't need personalization
 
@@ -1114,6 +1120,14 @@ class APISettings(BaseSettings):
         ),
     )
 
+    rate_limit_enabled: bool = Field(
+        default=True,
+        description=(
+            "Enable rate limiting for API endpoints. "
+            "Set to false to disable rate limiting entirely (useful for development)."
+        ),
+    )
+
 
 class ModelsSettings(BaseSettings):
     """
@@ -1471,6 +1485,172 @@ class DBListenerSettings(BaseSettings):
         return [c.strip() for c in self.channels.split(",") if c.strip()]
 
 
+class EmailSettings(BaseSettings):
+    """
+    Email service settings for SMTP.
+
+    Supports passwordless login via email codes and transactional emails.
+    Uses Gmail SMTP with App Passwords by default.
+
+    Generate app password at: https://myaccount.google.com/apppasswords
+
+    Environment variables:
+        EMAIL__ENABLED - Enable email service (default: false)
+        EMAIL__SMTP_HOST - SMTP server host (default: smtp.gmail.com)
+        EMAIL__SMTP_PORT - SMTP server port (default: 587 for TLS)
+        EMAIL__SENDER_EMAIL - Sender email address
+        EMAIL__SENDER_NAME - Sender display name
+        EMAIL__APP_PASSWORD - Gmail app password (from secrets)
+        EMAIL__USE_TLS - Use TLS encryption (default: true)
+        EMAIL__LOGIN_CODE_EXPIRY_MINUTES - Login code expiry (default: 10)
+
+    Branding environment variables (for email templates):
+        EMAIL__APP_NAME - Application name in emails (default: REM)
+        EMAIL__LOGO_URL - Logo URL for email templates (40x40 recommended)
+        EMAIL__TAGLINE - Tagline shown in email footer
+        EMAIL__WEBSITE_URL - Main website URL for email links
+        EMAIL__PRIVACY_URL - Privacy policy URL for email footer
+        EMAIL__TERMS_URL - Terms of service URL for email footer
+    """
+
+    model_config = SettingsConfigDict(
+        env_prefix="EMAIL__",
+        env_file=".env",
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    enabled: bool = Field(
+        default=False,
+        description="Enable email service (requires app_password to be set)",
+    )
+
+    smtp_host: str = Field(
+        default="smtp.gmail.com",
+        description="SMTP server host",
+    )
+
+    smtp_port: int = Field(
+        default=587,
+        description="SMTP server port (587 for TLS, 465 for SSL)",
+    )
+
+    sender_email: str = Field(
+        default="",
+        description="Sender email address",
+    )
+
+    sender_name: str = Field(
+        default="REM",
+        description="Sender display name",
+    )
+
+    # Branding settings for email templates
+    app_name: str = Field(
+        default="REM",
+        description="Application name shown in email templates",
+    )
+
+    logo_url: str | None = Field(
+        default=None,
+        description="Logo URL for email templates (40x40 recommended)",
+    )
+
+    tagline: str = Field(
+        default="Your AI-powered platform",
+        description="Tagline shown in email footer",
+    )
+
+    website_url: str = Field(
+        default="https://rem.ai",
+        description="Main website URL for email links",
+    )
+
+    privacy_url: str = Field(
+        default="https://rem.ai/privacy",
+        description="Privacy policy URL for email footer",
+    )
+
+    terms_url: str = Field(
+        default="https://rem.ai/terms",
+        description="Terms of service URL for email footer",
+    )
+
+    app_password: str | None = Field(
+        default=None,
+        description="Gmail app password for SMTP authentication",
+    )
+
+    use_tls: bool = Field(
+        default=True,
+        description="Use TLS encryption for SMTP",
+    )
+
+    login_code_expiry_minutes: int = Field(
+        default=10,
+        description="Login code expiry in minutes",
+    )
+
+    trusted_email_domains: str = Field(
+        default="",
+        description=(
+            "Comma-separated list of trusted email domains for new user registration. "
+            "Existing users can always login regardless of domain. "
+            "New users must have an email from a trusted domain. "
+            "Empty string means all domains are allowed. "
+            "Example: 'siggymd.ai,example.com'"
+        ),
+    )
+
+    @property
+    def trusted_domain_list(self) -> list[str]:
+        """Get trusted domains as a list, filtering empty strings."""
+        if not self.trusted_email_domains:
+            return []
+        return [d.strip().lower() for d in self.trusted_email_domains.split(",") if d.strip()]
+
+    def is_domain_trusted(self, email: str) -> bool:
+        """Check if an email's domain is in the trusted list.
+
+        Args:
+            email: Email address to check
+
+        Returns:
+            True if domain is trusted (or if no trusted domains configured)
+        """
+        domains = self.trusted_domain_list
+        if not domains:
+            # No restrictions configured
+            return True
+
+        email_domain = email.lower().split("@")[-1].strip()
+        return email_domain in domains
+
+    @property
+    def is_configured(self) -> bool:
+        """Check if email service is properly configured."""
+        return bool(self.sender_email and self.app_password)
+
+    @property
+    def template_kwargs(self) -> dict:
+        """
+        Get branding kwargs for email templates.
+
+        Returns a dict that can be passed to template functions:
+            login_code_template(..., **settings.email.template_kwargs)
+        """
+        kwargs = {
+            "app_name": self.app_name,
+            "tagline": self.tagline,
+            "website_url": self.website_url,
+            "privacy_url": self.privacy_url,
+            "terms_url": self.terms_url,
+        }
+        if self.logo_url:
+            kwargs["logo_url"] = self.logo_url
+        return kwargs
+
+
 class TestSettings(BaseSettings):
     """
     Test environment settings.
@@ -1585,6 +1765,7 @@ class Settings(BaseSettings):
     chunking: ChunkingSettings = Field(default_factory=ChunkingSettings)
     content: ContentSettings = Field(default_factory=ContentSettings)
     schema_search: SchemaSettings = Field(default_factory=SchemaSettings)
+    email: EmailSettings = Field(default_factory=EmailSettings)
     test: TestSettings = Field(default_factory=TestSettings)
 
 

rem/sql/background_indexes.sql

@@ -21,6 +21,11 @@ CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_embeddings_moments_vector_hnsw
 ON embeddings_moments
 USING hnsw (embedding vector_cosine_ops);
 
+-- HNSW vector index for embeddings_ontologies
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_embeddings_ontologies_vector_hnsw
+ON embeddings_ontologies
+USING hnsw (embedding vector_cosine_ops);
+
 -- HNSW vector index for embeddings_ontology_configs
 CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_embeddings_ontology_configs_vector_hnsw
 ON embeddings_ontology_configs

rem/sql/migrations/001_install.sql

@@ -44,6 +44,33 @@ BEGIN
     RAISE NOTICE '✓ All required extensions installed successfully';
 END $$;
 
+-- ============================================================================
+-- NORMALIZATION HELPER
+-- ============================================================================
+
+-- Normalize entity keys to lower-kebab-case for consistent lookups
+-- "Mood Disorder" -> "mood-disorder"
+-- "mood_disorder" -> "mood-disorder"
+-- "MoodDisorder" -> "mood-disorder"
+CREATE OR REPLACE FUNCTION normalize_key(input TEXT)
+RETURNS TEXT AS $$
+BEGIN
+    RETURN lower(
+        regexp_replace(
+            regexp_replace(
+                regexp_replace(input, '([a-z])([A-Z])', '\1-\2', 'g'),  -- camelCase -> kebab
+                '[_\s]+', '-', 'g'  -- underscores/spaces -> hyphens
+            ),
+            '-+', '-', 'g'  -- collapse multiple hyphens
+        )
+    );
+END;
+$$ LANGUAGE plpgsql IMMUTABLE;
+
+COMMENT ON FUNCTION normalize_key IS
+'Normalizes entity keys to lower-kebab-case for consistent lookups.
+Examples: "Mood Disorder" -> "mood-disorder", "mood_disorder" -> "mood-disorder"';
+
 -- ============================================================================
 -- MIGRATION TRACKING
 -- ============================================================================
@@ -237,10 +264,11 @@ BEGIN
 
     -- First lookup in KV store to get entity_type (table name)
     -- Include user-owned AND public (NULL user_id) entries
+    -- Normalize input key for consistent matching
     SELECT kv.entity_type INTO entity_table
     FROM kv_store kv
     WHERE (kv.user_id = effective_user_id OR kv.user_id IS NULL)
-    AND kv.entity_key = p_entity_key
+    AND kv.entity_key = normalize_key(p_entity_key)
     LIMIT 1;
 
     -- If not found, return empty
@@ -414,6 +442,7 @@ BEGIN
     FOR graph_keys IN
         WITH RECURSIVE graph_traversal AS (
             -- Base case: Find starting entity (user-owned OR public)
+            -- Normalize input key for consistent matching
             SELECT
                 0 AS depth,
                 kv.entity_key,
@@ -424,7 +453,7 @@ BEGIN
                 ARRAY[kv.entity_key]::TEXT[] AS path
             FROM kv_store kv
             WHERE (kv.user_id = effective_user_id OR kv.user_id IS NULL)
-            AND kv.entity_key = p_entity_key
+            AND kv.entity_key = normalize_key(p_entity_key)
 
             UNION ALL
 
@@ -441,7 +470,7 @@ BEGIN
             JOIN kv_store source_kv ON source_kv.entity_key = gt.entity_key
                 AND (source_kv.user_id = effective_user_id OR source_kv.user_id IS NULL)
             CROSS JOIN LATERAL jsonb_array_elements(COALESCE(source_kv.graph_edges, '[]'::jsonb)) AS edge
-            JOIN kv_store target_kv ON target_kv.entity_key = (edge->>'dst')::VARCHAR(255)
+            JOIN kv_store target_kv ON target_kv.entity_key = normalize_key((edge->>'dst')::VARCHAR(255))
                 AND (target_kv.user_id = effective_user_id OR target_kv.user_id IS NULL)
             WHERE gt.depth < p_max_depth
             AND (p_rel_type IS NULL OR (edge->>'rel_type')::VARCHAR(100) = p_rel_type)
@@ -657,7 +686,7 @@ BEGIN
         MIN(msg_counts.first_msg)::TIMESTAMP AS first_message_at,
         MAX(msg_counts.last_msg)::TIMESTAMP AS last_message_at
     FROM shared_sessions ss
-    LEFT JOIN users u ON u.user_id = ss.owner_user_id AND u.tenant_id = ss.tenant_id
+    LEFT JOIN users u ON u.id::text = ss.owner_user_id AND u.tenant_id = ss.tenant_id
     LEFT JOIN (
         SELECT
             m.session_id,