horizon-sdk 16.0.0

//! Domain model types for the Horizon platform.
//!
//! Each struct maps to a table in the `horizon_public` `PostgreSQL` schema.
//! Fields use `Option` where the database column is nullable or has a server-side default.

use chrono::{DateTime, Utc};
use serde::{Deserialize, Serialize};
use sqlx::error::BoxDynError;
use sqlx::postgres::{PgTypeInfo, PgValueRef};
use uuid::Uuid;

use crate::types::error::{HorizonError, Result as HorizonResult};

/// Minimum byte length for a `PostgreSQL` `POINT` binary value.
const POINT_BYTE_LEN: usize = 16;

/// A geographic position as (longitude, latitude).
///
/// Maps to `PostgreSQL` `POINT` type with native sqlx decode support.
/// The binary wire format is two big-endian `f64` values (x, y).
#[derive(Debug, Clone, Copy, PartialEq, Serialize, Deserialize)]
pub struct Position(pub f64, pub f64);

impl sqlx::Type<sqlx::Postgres> for Position {
    fn compatible(ty: &PgTypeInfo) -> bool {
        *ty == Self::type_info()
    }

    fn type_info() -> PgTypeInfo {
        PgTypeInfo::with_name("POINT")
    }
}

#[allow(
    clippy::as_conversions,
    clippy::indexing_slicing,
    clippy::std_instead_of_core,
    clippy::big_endian_bytes,
    clippy::float_arithmetic,
    reason = "byte slicing, Box cast, and BE f64 decode required for PostgreSQL POINT wire format"
)]
impl<'de> sqlx::Decode<'de, sqlx::Postgres> for Position {
    fn decode(value: PgValueRef<'de>) -> Result<Self, BoxDynError> {
        let buf = <&[u8] as sqlx::Decode<sqlx::Postgres>>::decode(value)?;
        if buf.len() < POINT_BYTE_LEN {
            return Err("POINT requires 16 bytes".into());
        }
        let (x_bytes, rest) = buf.split_at(8_usize);
        let x = f64::from_be_bytes(
            x_bytes
                .try_into()
                .map_err(|err| Box::new(err) as BoxDynError)?,
        );
        let y = f64::from_be_bytes(
            rest[..8]
                .try_into()
                .map_err(|err| Box::new(err) as BoxDynError)?,
        );
        Ok(Self(x, y))
    }
}

/// A label/annotation on a platform.
///
/// Time and value coordinates describe the labeled region. `time_coordinates`
/// pairs with `value_coordinates` element-wise (a point, line, or polygon
/// depending on length). `bearing_time_record_specification_id` and
/// `spectrogram_specification_id` are mutually exclusive — only one may be
/// set. `data_stream_id` is `None` for legacy platform-wide rows; inserts
/// require `Some(...)`. User attribution is managed by database triggers.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct Annotation {
    pub bearing_time_record_specification_id: Option<Uuid>,
    pub confidence: Option<f64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub data_stream_id: Option<Uuid>,
    pub duration_seconds: Option<i32>,
    pub feed_context: Option<serde_json::Value>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub notes: Option<String>,
    pub ontology_class_id: Option<Uuid>,
    pub organization_id: Option<Uuid>,
    pub parent_annotation_id: Option<Uuid>,
    pub platform_id: Uuid,
    pub spectrogram_specification_id: Option<Uuid>,
    pub time_coordinates: Vec<DateTime<Utc>>,
    pub value_coordinates: Vec<f64>,
}

/// A specification for audio data processing.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct AudioSpecification {
    pub bit_depth: i64,
    pub channel_count: i64,
    pub channel_index: i64,
    pub created_datetime: Option<DateTime<Utc>>,
    pub encoding: String,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub sample_rate: i64,
}

/// A specification for beamgram data processing.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct BeamgramSpecification {
    pub center_bearing: Option<f64>,
    pub center_bin_width: Option<f64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub elevation_increment: Option<f64>,
    pub fft_sample_count: Option<i32>,
    pub id: Option<Uuid>,
    pub lower_elevation: Option<f64>,
    pub max_frequency: Option<f64>,
    pub min_frequency: Option<f64>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub normalizer: Option<String>,
    pub organization_id: Option<Uuid>,
    pub update_rate_ms: Option<i64>,
    pub upper_elevation: Option<f64>,
}

/// A user-defined alias for a bearing resolution, naming a bearing bin count.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct BearingResolutionAlias {
    pub bearing_bin_count: Option<i32>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// A specification for bearing-time record data processing.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct BearingTimeRecordSpecification {
    pub amplitude_unit_mode: Option<String>,
    pub baffle_bearing_list: Option<Vec<f64>>,
    pub bearing_bin_count: Option<i32>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub elevation_increment: Option<f64>,
    pub fft_sample_count: Option<i32>,
    pub focus_range: Option<String>,
    pub frequency_spacing: Option<f64>,
    pub heading_data_type: Option<String>,
    pub heading_vector_index: Option<i32>,
    pub id: Option<Uuid>,
    pub lower_elevation: Option<f64>,
    pub max_frequency: Option<f64>,
    pub max_pixel: Option<i32>,
    pub min_frequency: Option<f64>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub normalizer: Option<String>,
    pub organization_id: Option<Uuid>,
    pub update_rate_ms: Option<i64>,
    pub upper_elevation: Option<f64>,
}

/// Discriminator for a [`VariantDataRow`] payload arm.
///
/// Stored as text in Postgres/Iceberg. Legacy rows may have `variant = NULL`
/// and still carry a vector payload.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
#[non_exhaustive]
#[serde(rename_all = "snake_case")]
pub enum DataRowVariant {
    Float32,
    Float32Array,
    Float64,
    Float64Array,
    Int,
    IntArray,
    String,
    Struct,
    Vector,
}

impl DataRowVariant {
    #[must_use]
    pub const fn as_str(self) -> &'static str {
        match self {
            Self::Float32 => "float32",
            Self::Float32Array => "float32_array",
            Self::Float64 => "float64",
            Self::Float64Array => "float64_array",
            Self::Int => "int",
            Self::IntArray => "int_array",
            Self::String => "string",
            Self::Struct => "struct",
            Self::Vector => "vector",
        }
    }

    #[must_use]
    pub fn parse(value: &str) -> Option<Self> {
        match value {
            "float32" => Some(Self::Float32),
            "float32_array" => Some(Self::Float32Array),
            "float64" => Some(Self::Float64),
            "float64_array" => Some(Self::Float64Array),
            "int" => Some(Self::Int),
            "int_array" => Some(Self::IntArray),
            "string" => Some(Self::String),
            "struct" => Some(Self::Struct),
            "vector" => Some(Self::Vector),
            _ => None,
        }
    }

    /// Column name for the payload arm this variant selects.
    #[must_use]
    pub const fn payload_arm(self) -> &'static str {
        match self {
            Self::Float32 => "payload_float32",
            Self::Float32Array => "payload_float32_array",
            Self::Float64 => "payload_float64",
            Self::Float64Array => "payload_float64_array",
            Self::Int => "payload_int",
            Self::IntArray => "payload_int_array",
            Self::String => "payload_string",
            Self::Struct => "payload_struct",
            Self::Vector => "vector",
        }
    }
}

/// Typed payload carried by a [`VariantDataRow`].
///
/// Exactly one arm is active. Unstructured data uses [`DataRowPayload::Struct`].
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
#[non_exhaustive]
pub enum DataRowPayload {
    Float32(f32),
    Float32Array(Vec<f32>),
    Float64(f64),
    Float64Array(Vec<f64>),
    Int(i32),
    IntArray(Vec<i32>),
    String(String),
    /// Unstructured JSON. Serialized as `jsonb` in Postgres and JSON text in Iceberg.
    Struct(serde_json::Value),
    Vector {
        vector: Vec<f64>,
        vector_end_bound: f64,
        vector_start_bound: f64,
    },
}

impl DataRowPayload {
    #[must_use]
    pub const fn variant(&self) -> DataRowVariant {
        match self {
            Self::Float32(_) => DataRowVariant::Float32,
            Self::Float32Array(_) => DataRowVariant::Float32Array,
            Self::Float64(_) => DataRowVariant::Float64,
            Self::Float64Array(_) => DataRowVariant::Float64Array,
            Self::Int(_) => DataRowVariant::Int,
            Self::IntArray(_) => DataRowVariant::IntArray,
            Self::String(_) => DataRowVariant::String,
            Self::Struct(_) => DataRowVariant::Struct,
            Self::Vector { .. } => DataRowVariant::Vector,
        }
    }
}

/// A vector row of data in the `data_row` table.
///
/// This preserves the original Rust SDK model used by existing stream
/// producers. Convert it to [`VariantDataRow`] at storage boundaries.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct DataRow {
    pub created_datetime: Option<DateTime<Utc>>,
    pub data_stream_id: Uuid,
    pub data_type: String,
    pub datetime: DateTime<Utc>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub specification_id: Uuid,
    pub vector: Vec<f64>,
    pub vector_end_bound: f64,
    pub vector_start_bound: f64,
}

/// A row with any supported payload variant in the `data_row` table.
///
/// This type is used for both Postgres inserts and Arrow IPC serialization.
/// Payload columns are flat `Option` fields so `sqlx` can map them directly;
/// use [`VariantDataRow::from_payload`] / [`VariantDataRow::payload`] for the typed union.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct VariantDataRow {
    pub created_datetime: Option<DateTime<Utc>>,
    pub data_stream_id: Uuid,
    pub data_type: String,
    pub datetime: DateTime<Utc>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub payload_float32: Option<f32>,
    pub payload_float32_array: Option<Vec<f32>>,
    pub payload_float64: Option<f64>,
    pub payload_float64_array: Option<Vec<f64>>,
    pub payload_int: Option<i32>,
    pub payload_int_array: Option<Vec<i32>>,
    pub payload_string: Option<String>,
    pub payload_struct: Option<serde_json::Value>,
    pub specification_id: Uuid,
    pub variant: Option<String>,
    pub vector: Option<Vec<f64>>,
    pub vector_end_bound: Option<f64>,
    pub vector_start_bound: Option<f64>,
}

impl VariantDataRow {
    /// Build a row from a typed payload, setting `variant` and clearing unused columns.
    #[must_use]
    pub fn from_payload(
        data_stream_id: Uuid,
        datetime: DateTime<Utc>,
        data_type: String,
        specification_id: Uuid,
        payload: DataRowPayload,
    ) -> Self {
        let mut row = Self {
            created_datetime: None,
            data_stream_id,
            data_type,
            datetime,
            modified_datetime: None,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id,
            variant: Some(payload.variant().as_str().to_owned()),
            vector: None,
            vector_end_bound: None,
            vector_start_bound: None,
        };
        match payload {
            DataRowPayload::Float32(value) => row.payload_float32 = Some(value),
            DataRowPayload::Float32Array(value_list) => {
                row.payload_float32_array = Some(value_list);
            }
            DataRowPayload::Float64(value) => row.payload_float64 = Some(value),
            DataRowPayload::Float64Array(value_list) => {
                row.payload_float64_array = Some(value_list);
            }
            DataRowPayload::Int(value) => row.payload_int = Some(value),
            DataRowPayload::IntArray(value_list) => row.payload_int_array = Some(value_list),
            DataRowPayload::String(value) => row.payload_string = Some(value),
            DataRowPayload::Struct(value) => row.payload_struct = Some(value),
            DataRowPayload::Vector {
                vector,
                vector_end_bound,
                vector_start_bound,
            } => {
                row.vector = Some(vector);
                row.vector_end_bound = Some(vector_end_bound);
                row.vector_start_bound = Some(vector_start_bound);
            }
        }
        row
    }

    /// Resolve the active payload arm.
    ///
    /// Legacy rows (`variant` null with a vector present) map to
    /// [`DataRowPayload::Vector`]. An unrecognized non-null `variant` fails
    /// even when `vector` is populated.
    pub fn payload(&self) -> Option<DataRowPayload> {
        let variant = match self.variant.as_deref() {
            None if self.vector.is_some() => DataRowVariant::Vector,
            None => return None,
            Some(variant_text) => DataRowVariant::parse(variant_text)?,
        };

        match variant {
            DataRowVariant::Float32 => self.payload_float32.map(DataRowPayload::Float32),
            DataRowVariant::Float32Array => self
                .payload_float32_array
                .clone()
                .map(DataRowPayload::Float32Array),
            DataRowVariant::Float64 => self.payload_float64.map(DataRowPayload::Float64),
            DataRowVariant::Float64Array => self
                .payload_float64_array
                .clone()
                .map(DataRowPayload::Float64Array),
            DataRowVariant::Int => self.payload_int.map(DataRowPayload::Int),
            DataRowVariant::IntArray => {
                self.payload_int_array.clone().map(DataRowPayload::IntArray)
            }
            DataRowVariant::String => self.payload_string.clone().map(DataRowPayload::String),
            DataRowVariant::Struct => self.payload_struct.clone().map(DataRowPayload::Struct),
            DataRowVariant::Vector => {
                let vector = self.vector.clone()?;
                let vector_end_bound = self.vector_end_bound?;
                let vector_start_bound = self.vector_start_bound?;
                Some(DataRowPayload::Vector {
                    vector,
                    vector_end_bound,
                    vector_start_bound,
                })
            }
        }
    }

    /// Require exactly one of `vector` / `payload_*` and a matching `variant`.
    ///
    /// Legacy rows may omit `variant` when only `vector` is set. A `vector` arm
    /// also requires both `vector_start_bound` and `vector_end_bound` so
    /// [`Self::payload`] can resolve it. Call this before Postgres writes and
    /// Arrow serialization so rows built outside [`Self::from_payload`] cannot
    /// persist mixed, mismatched, or incomplete payloads.
    ///
    /// # Errors
    ///
    /// Returns [`HorizonError::InvalidDataRowPayload`] when the invariant does not hold.
    pub fn validate_payload(&self) -> HorizonResult<()> {
        let populated_arm_list: Vec<&'static str> = [
            ("payload_float32", self.payload_float32.is_some()),
            (
                "payload_float32_array",
                self.payload_float32_array.is_some(),
            ),
            ("payload_float64", self.payload_float64.is_some()),
            (
                "payload_float64_array",
                self.payload_float64_array.is_some(),
            ),
            ("payload_int", self.payload_int.is_some()),
            ("payload_int_array", self.payload_int_array.is_some()),
            ("payload_string", self.payload_string.is_some()),
            ("payload_struct", self.payload_struct.is_some()),
            ("vector", self.vector.is_some()),
        ]
        .into_iter()
        .filter_map(|(field_name, is_set)| is_set.then_some(field_name))
        .collect();

        let active_arm = if let [arm] = populated_arm_list.as_slice() {
            *arm
        } else {
            let message = "exactly one of vector or payload_* must be set".to_owned();
            return Err(HorizonError::InvalidDataRowPayload { message });
        };
        if active_arm == "vector"
            && (self.vector_start_bound.is_none() || self.vector_end_bound.is_none())
        {
            let message =
                "vector payload requires vector_start_bound and vector_end_bound".to_owned();
            return Err(HorizonError::InvalidDataRowPayload { message });
        }

        match self.variant.as_deref() {
            None => {
                if active_arm != "vector" {
                    let message =
                        "variant is required when the payload is not a legacy vector".to_owned();
                    return Err(HorizonError::InvalidDataRowPayload { message });
                }
            }
            Some(variant_text) => {
                let Some(variant) = DataRowVariant::parse(variant_text) else {
                    let message = format!("unknown variant: {variant_text}");
                    return Err(HorizonError::InvalidDataRowPayload { message });
                };
                let expected_arm = variant.payload_arm();
                if expected_arm != active_arm {
                    let message = format!(
                        "variant '{variant_text}' does not match populated payload arm '{active_arm}'"
                    );
                    return Err(HorizonError::InvalidDataRowPayload { message });
                }
            }
        }
        Ok(())
    }
}

impl From<&DataRow> for VariantDataRow {
    fn from(row: &DataRow) -> Self {
        Self {
            created_datetime: row.created_datetime,
            data_stream_id: row.data_stream_id,
            data_type: row.data_type.clone(),
            datetime: row.datetime,
            modified_datetime: row.modified_datetime,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id: row.specification_id,
            variant: Some(DataRowVariant::Vector.as_str().to_owned()),
            vector: Some(row.vector.clone()),
            vector_end_bound: Some(row.vector_end_bound),
            vector_start_bound: Some(row.vector_start_bound),
        }
    }
}

impl TryFrom<VariantDataRow> for DataRow {
    type Error = HorizonError;

    fn try_from(row: VariantDataRow) -> Result<Self, Self::Error> {
        let Some(DataRowPayload::Vector {
            vector,
            vector_end_bound,
            vector_start_bound,
        }) = row.payload()
        else {
            let message = "a non-vector variant cannot be converted to DataRow".to_owned();
            return Err(HorizonError::InvalidDataRowPayload { message });
        };

        Ok(Self {
            created_datetime: row.created_datetime,
            data_stream_id: row.data_stream_id,
            data_type: row.data_type,
            datetime: row.datetime,
            modified_datetime: row.modified_datetime,
            specification_id: row.specification_id,
            vector,
            vector_end_bound,
            vector_start_bound,
        })
    }
}

/// One upload session of sensor data from a platform (mission or sortie).
///
/// `organization_id`, `created_datetime`, and `modified_datetime` are
/// server-assigned and ignored on create.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct DataCollection {
    pub collected_secs: Option<i32>,
    pub collected_size: Option<i64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub ingest_pipeline_version: Option<String>,
    pub ingested_datetime: Option<DateTime<Utc>>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: String,
    pub organization_id: Option<Uuid>,
    pub platform_id: Uuid,
    pub platform_software_version: Option<String>,
    pub s3_prefix: Option<String>,
    pub status: Option<DataCollectionStatus>,
    pub upload_complete_datetime: Option<DateTime<Utc>>,
    pub xml_sensor_config: Option<serde_json::Value>,
}

/// A contiguous block of sensor data time within a data collection.
///
/// `organization_id`, `created_datetime`, and `modified_datetime` are
/// server-assigned and ignored on create.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct DataCollectionInterval {
    pub collected_secs: Option<i32>,
    pub collected_size: Option<i64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub data_collection_id: Uuid,
    pub data_stream_id: Uuid,
    pub end_datetime: DateTime<Utc>,
    pub group_hash_id: Option<String>,
    pub has_classification_override: Option<bool>,
    pub has_detection_override: Option<bool>,
    pub has_track_override: Option<bool>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub processing_end_datetime: Option<DateTime<Utc>>,
    pub processing_start_datetime: Option<DateTime<Utc>>,
    pub start_datetime: DateTime<Utc>,
    pub status: Option<DataCollectionStatus>,
}

/// Lifecycle status of a data collection or one of its intervals.
///
/// Maps to the `horizon_public.data_collection_status` Postgres ENUM.
#[allow(
    clippy::exhaustive_enums,
    reason = "variants mirror a closed Postgres ENUM, so exhaustive matching downstream is desirable"
)]
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
#[serde(rename_all = "lowercase")]
pub enum DataCollectionStatus {
    Failed,
    Ingested,
    Pending,
    Processing,
    Registered,
}

/// Backend that powers tile queries for a data stream.
///
/// Maps to the `horizon_public.data_stream_query_source` Postgres ENUM.
#[allow(
    clippy::exhaustive_enums,
    reason = "variants mirror a closed Postgres ENUM, so exhaustive matching downstream is desirable"
)]
#[derive(Debug, Default, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, sqlx::Type)]
#[sqlx(
    type_name = "horizon_public.data_stream_query_source",
    rename_all = "lowercase"
)]
#[serde(rename_all = "lowercase")]
pub enum DataStreamQuerySource {
    /// Query the row-store `data_row` table directly.
    #[default]
    Postgres,
    /// Query Iceberg through Trino.
    Trino,
}

/// A data stream belonging to a platform.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct DataStream {
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Option<Uuid>,
    #[serde(default)]
    pub query_source: DataStreamQuerySource,
}

impl DataStream {
    /// Create a `DataStream` with a deterministic UUID derived from the platform ID and name.
    ///
    /// Defaults `query_source` to [`DataStreamQuerySource::Postgres`]; override the
    /// field after construction if the stream should be served from Trino.
    #[must_use]
    pub fn from_platform_and_name(platform_id: Uuid, data_stream_name: &str) -> Self {
        let name = capitalize(data_stream_name);
        Self {
            created_datetime: None,
            id: Some(data_stream_uuid(platform_id, data_stream_name)),
            modified_datetime: None,
            name: Some(name),
            organization_id: None,
            platform_id: Some(platform_id),
            query_source: DataStreamQuerySource::Postgres,
        }
    }
}

/// A user-defined alias for an elevation range (lower/upper elevation and increment).
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct ElevationAlias {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub elevation_increment: Option<f64>,
    pub id: Option<Uuid>,
    pub lower_elevation: Option<f64>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub upper_elevation: Option<f64>,
}

/// A user-defined focus range label for categorizing acoustic conditions.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct FocusRange {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// A user-defined alias for a frequency band (lower/upper frequency pair).
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct FrequencyBandAlias {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub lower_frequency: Option<f64>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub upper_frequency: Option<f64>,
}

/// A user-defined alias for a frequency resolution, naming a frequency bin count.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct FrequencyResolutionAlias {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub frequency_bin_count: Option<i32>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// A row of metadata in the Iceberg `metadata_row` table.
///
/// This type is used for both Postgres inserts and Arrow IPC serialization.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct MetadataRow {
    pub altitude: Option<f64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub data_stream_id: Uuid,
    pub datetime: DateTime<Utc>,
    pub heading: Option<f64>,
    pub latitude: Option<f64>,
    pub longitude: Option<f64>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub pitch: Option<f64>,
    pub roll: Option<f64>,
    pub speed: Option<f64>,
    pub speed_over_ground: Option<f64>,
}

/// A group that can group platforms around some bounded time or purpose.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct Mission {
    pub created_datetime: Option<DateTime<Utc>>,
    pub end_datetime: Option<DateTime<Utc>>,
    pub free_text: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub position: Option<Position>,
    pub start_datetime: Option<DateTime<Utc>>,
}

/// A relationship that associates a platform with a mission.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct MissionPlatform {
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub mission_id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Option<Uuid>,
}

/// A named normalization algorithm referenced by feed specifications.
///
/// Uses a soft delete: `is_archived` frees the name for reuse because the
/// unique constraint is partial (`WHERE NOT is_archived`). There is no hard
/// delete; use `archive_normalizer` instead.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct Normalizer {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub is_archived: Option<bool>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// An ontology for organizing platforms.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct Ontology {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// A class within an ontology.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct OntologyClass {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub ontology_id: Uuid,
    pub order: Option<i32>,
    pub organization_id: Option<Uuid>,
    pub parent_id: Option<Uuid>,
    pub relationship_type: Option<String>,
}

/// An instance of a platform.
///
/// The `position` field maps to `PostgreSQL`'s `POINT` type as `(longitude, latitude)`.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct Platform {
    pub created_datetime: Option<DateTime<Utc>>,
    pub end_datetime: Option<DateTime<Utc>>,
    pub free_text: Option<String>,
    pub id: Option<Uuid>,
    pub kind_id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub position: Option<Position>,
    pub start_datetime: Option<DateTime<Utc>>,
}

/// A relationship that associates a platform with an audio specification.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformAudioSpecification {
    pub audio_specification_id: Uuid,
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Uuid,
}

/// A relationship that associates a platform with a beamgram specification.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformBeamgramSpecification {
    pub beamgram_specification_id: Option<Uuid>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Option<Uuid>,
}

/// A relationship that associates a platform with a bearing-time record specification.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformBearingTimeRecordSpecification {
    pub bearing_time_record_specification_id: Option<Uuid>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Option<Uuid>,
}

/// Key-value information stored for a platform (append-only).
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformInformation {
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Uuid,
    pub properties: serde_json::Value,
}

/// A kind of platform (e.g. buoy, vessel, aircraft).
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformKind {
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub image_url: Option<String>,
    pub long_description: Option<String>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub short_description: Option<String>,
}

/// A relationship that associates a platform with a spectrogram specification.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PlatformSpectrogramSpecification {
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub platform_id: Option<Uuid>,
    pub spectrogram_specification_id: Option<Uuid>,
}

/// A directional spectrogram specification: leading/lagging bearing row offsets
/// used to derive directional hue on top of a spectrogram.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct DirectionalSpectrogramSpecification {
    pub bearing_lagging_rows: Option<i64>,
    pub bearing_leading_rows: Option<i64>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub organization_id: Option<Uuid>,
    pub spectrogram_specification_id: Option<Uuid>,
}

/// A specification for spectrogram data processing.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct SpectrogramSpecification {
    pub amplitude_unit_mode: Option<String>,
    pub channel: Option<i32>,
    pub channel_role: Option<String>,
    pub created_datetime: Option<DateTime<Utc>>,
    pub fft_sample_count: Option<i32>,
    pub fft_sample_overlap_count: Option<i32>,
    pub frequency_bin_count: Option<i32>,
    pub frequency_spacing: Option<f64>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub normalizer: Option<String>,
    pub organization_id: Option<Uuid>,
}

/// A user-defined alias for an update rate, naming an interval.
///
/// `update_rate_ms` maps the `PostgreSQL` `interval` column to whole
/// milliseconds, mirroring the beamgram/bearing-time-record specifications.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct UpdateRateAlias {
    pub created_datetime: Option<DateTime<Utc>>,
    pub description: Option<String>,
    pub id: Option<Uuid>,
    pub modified_datetime: Option<DateTime<Utc>>,
    pub name: Option<String>,
    pub organization_id: Option<Uuid>,
    pub update_rate_ms: Option<i64>,
}

/// Capitalize a string: uppercase first character, lowercase rest.
///
/// Matches Python's `str.capitalize()` for cross-SDK parity.
#[must_use]
pub fn capitalize(s: &str) -> String {
    let mut chars = s.chars();
    chars.next().map_or_else(String::new, |first| {
        let upper: String = first.to_uppercase().collect();
        format!("{upper}{}", chars.as_str().to_lowercase())
    })
}

/// Compute the deterministic UUID for a data stream from its platform and name.
///
/// Applies [`capitalize`] to the name for Python SDK parity, then hashes
/// the `"{platform_id}:{capitalized_name}"` string via [`name_to_uuid`].
#[must_use]
pub fn data_stream_uuid(platform_id: Uuid, data_stream_name: &str) -> Uuid {
    let name = capitalize(data_stream_name);
    name_to_uuid(&format!("{platform_id}:{name}"))
}

/// Deterministic UUID generation for data idempotency.
///
/// Uses UUID v5 (SHA-1 namespace) with a fixed namespace to produce
/// the same UUID for the same input string across invocations.
#[must_use]
pub fn name_to_uuid(name: &str) -> Uuid {
    const NAMESPACE: Uuid = Uuid::from_bytes([
        0x13, 0xdb, 0x1b, 0xb5, 0x27, 0xac, 0x4b, 0xf7, 0x81, 0x86, 0x10, 0x43, 0x34, 0x7c, 0x42,
        0x47,
    ]);
    let normalized = name.trim().to_lowercase();
    Uuid::new_v5(&NAMESPACE, normalized.as_bytes())
}

#[cfg(test)]
#[allow(
    clippy::default_numeric_fallback,
    clippy::unwrap_used,
    reason = "tests use numeric literals and unwrap for brevity"
)]
mod tests {
    use super::*;

    #[test]
    fn capitalize_matches_python() {
        assert_eq!(capitalize("hydrophone"), "Hydrophone");
        assert_eq!(capitalize("Hydrophone"), "Hydrophone");
        assert_eq!(capitalize("hYDROPHONE"), "Hydrophone");
        assert_eq!(capitalize("HYDROPHONE"), "Hydrophone");
        assert_eq!(capitalize(""), "");
        assert_eq!(capitalize("a"), "A");
    }

    #[test]
    fn data_stream_from_platform_and_name() {
        let platform_id = Uuid::new_v4();
        let ds = DataStream::from_platform_and_name(platform_id, "audio");
        assert_eq!(ds.platform_id, Some(platform_id));
        assert_eq!(ds.name.as_deref(), Some("Audio"));
        assert!(ds.id.is_some());

        let ds2 = DataStream::from_platform_and_name(platform_id, "audio");
        assert_eq!(ds.id, ds2.id);
    }

    #[test]
    fn data_stream_uuid_is_deterministic() {
        let platform_id = Uuid::new_v4();
        let a = data_stream_uuid(platform_id, "audio");
        let b = data_stream_uuid(platform_id, "audio");
        assert_eq!(a, b);
    }

    #[test]
    fn data_stream_uuid_matches_from_platform_and_name() {
        let platform_id = Uuid::new_v4();
        let ds = DataStream::from_platform_and_name(platform_id, "audio");
        assert_eq!(ds.id, Some(data_stream_uuid(platform_id, "audio")));
    }

    #[test]
    fn mission_serde_round_trip() {
        let mission = Mission {
            created_datetime: None,
            end_datetime: None,
            free_text: None,
            id: Some(Uuid::new_v4()),
            modified_datetime: None,
            name: Some("Test Mission".to_owned()),
            organization_id: None,
            position: Some(Position(1.5_f64, 2.5_f64)),
            start_datetime: Some(Utc::now()),
        };
        let json = serde_json::to_string(&mission).unwrap();
        let deserialized: Mission = serde_json::from_str(&json).unwrap();
        assert_eq!(mission, deserialized);
    }

    #[test]
    fn name_to_uuid_is_deterministic() {
        let a = name_to_uuid("test");
        let b = name_to_uuid("test");
        assert_eq!(a, b);
    }

    #[test]
    fn name_to_uuid_matches_python_sdk() {
        let expected = Uuid::parse_str("e32be881-c656-5b8e-8dd7-889066ac1c83").unwrap();
        assert_eq!(name_to_uuid("Test Platform"), expected);

        let expected2 = Uuid::parse_str("eaeebe45-0eaa-59e4-9b79-7ceda69789e2").unwrap();
        assert_eq!(name_to_uuid("hello world"), expected2);
    }

    #[test]
    fn name_to_uuid_normalizes_whitespace_and_case() {
        let a = name_to_uuid("  Hello World  ");
        let b = name_to_uuid("hello world");
        assert_eq!(a, b);
    }

    #[test]
    fn platform_information_serde_with_json_properties() {
        let info = PlatformInformation {
            created_datetime: None,
            id: None,
            modified_datetime: None,
            organization_id: None,
            platform_id: Uuid::new_v4(),
            properties: serde_json::json!({"key": "value", "nested": {"a": 1_i32}}),
        };
        let json = serde_json::to_string(&info).unwrap();
        let deserialized: PlatformInformation = serde_json::from_str(&json).unwrap();
        assert_eq!(info, deserialized);
    }

    #[test]
    fn data_row_from_payload_passes_validation() {
        let row = VariantDataRow::from_payload(
            Uuid::nil(),
            Utc::now(),
            "cot".to_owned(),
            Uuid::nil(),
            DataRowPayload::Struct(serde_json::json!({"callsign": "ALPHA"})),
        );
        row.validate_payload().unwrap();
    }

    #[test]
    fn data_row_allows_legacy_vector_without_variant() {
        let row = VariantDataRow {
            created_datetime: None,
            data_stream_id: Uuid::nil(),
            data_type: "audio".to_owned(),
            datetime: Utc::now(),
            modified_datetime: None,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id: Uuid::nil(),
            variant: None,
            vector: Some(vec![1.0, 2.0]),
            vector_end_bound: Some(1.0),
            vector_start_bound: Some(0.0),
        };
        row.validate_payload().unwrap();
    }

    #[test]
    fn data_row_rejects_vector_without_bounds() {
        let row = VariantDataRow {
            created_datetime: None,
            data_stream_id: Uuid::nil(),
            data_type: "audio".to_owned(),
            datetime: Utc::now(),
            modified_datetime: None,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id: Uuid::nil(),
            variant: None,
            vector: Some(vec![1.0, 2.0]),
            vector_end_bound: None,
            vector_start_bound: Some(0.0),
        };
        let error = row.validate_payload().unwrap_err();
        assert!(
            error
                .to_string()
                .contains("vector payload requires vector_start_bound and vector_end_bound"),
            "got: {error}"
        );
        assert!(row.payload().is_none());
    }

    #[test]
    fn data_row_rejects_mixed_payload_fields() {
        let mut row = VariantDataRow::from_payload(
            Uuid::nil(),
            Utc::now(),
            "cot".to_owned(),
            Uuid::nil(),
            DataRowPayload::Struct(serde_json::json!({})),
        );
        row.payload_float64 = Some(22.5);
        let error = row.validate_payload().unwrap_err();
        assert!(
            error
                .to_string()
                .contains("exactly one of vector or payload_* must be set"),
            "got: {error}"
        );
    }

    #[test]
    fn data_row_rejects_mismatched_variant() {
        let mut row = VariantDataRow::from_payload(
            Uuid::nil(),
            Utc::now(),
            "cot".to_owned(),
            Uuid::nil(),
            DataRowPayload::Struct(serde_json::json!({})),
        );
        row.variant = Some("float64".to_owned());
        let error = row.validate_payload().unwrap_err();
        assert!(
            error
                .to_string()
                .contains("does not match populated payload arm"),
            "got: {error}"
        );
    }

    #[test]
    fn data_row_payload_resolves_legacy_vector() {
        let row = VariantDataRow {
            created_datetime: None,
            data_stream_id: Uuid::nil(),
            data_type: "audio".to_owned(),
            datetime: Utc::now(),
            modified_datetime: None,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id: Uuid::nil(),
            variant: None,
            vector: Some(vec![1.0, 2.0]),
            vector_end_bound: Some(1.0),
            vector_start_bound: Some(0.0),
        };
        assert!(matches!(row.payload(), Some(DataRowPayload::Vector { .. })));
    }

    #[test]
    fn data_row_payload_rejects_unknown_variant_even_with_vector() {
        let row = VariantDataRow {
            created_datetime: None,
            data_stream_id: Uuid::nil(),
            data_type: "audio".to_owned(),
            datetime: Utc::now(),
            modified_datetime: None,
            payload_float32: None,
            payload_float32_array: None,
            payload_float64: None,
            payload_float64_array: None,
            payload_int: None,
            payload_int_array: None,
            payload_string: None,
            payload_struct: None,
            specification_id: Uuid::nil(),
            variant: Some("not-a-variant".to_owned()),
            vector: Some(vec![1.0, 2.0]),
            vector_end_bound: Some(1.0),
            vector_start_bound: Some(0.0),
        };
        assert!(row.payload().is_none());
    }
}