use std::sync::Arc;
use arrow::array::{ArrayRef, StringBuilder, UInt64Builder};
use arrow::datatypes::{DataType, Field, Schema, SchemaRef};
use arrow::error::ArrowError;
use arrow::record_batch::RecordBatch;
use polyc_eventlog::Event;
use polyc_proto::kinds;
#[derive(Debug, Clone)]
pub(crate) struct ModelCallRow {
pub partition: String,
pub position: u64,
pub turn_id: Option<String>,
pub provider: String,
pub model: String,
pub captured_clock_unix_ms: u64,
}
#[must_use]
pub(crate) fn schema() -> SchemaRef {
Arc::new(Schema::new(vec![
Field::new("partition", DataType::Utf8, false),
Field::new("position", DataType::UInt64, false),
Field::new("turn_id", DataType::Utf8, true),
Field::new("provider", DataType::Utf8, false),
Field::new("model", DataType::Utf8, false),
Field::new("captured_clock_unix_ms", DataType::UInt64, false),
]))
}
pub(crate) fn decode_model_call_batch(rows: &[ModelCallRow]) -> Result<RecordBatch, ArrowError> {
let mut partition_b = StringBuilder::with_capacity(rows.len(), rows.len() * 8);
let mut position_b = UInt64Builder::with_capacity(rows.len());
let mut turn_id_b = StringBuilder::with_capacity(rows.len(), rows.len() * 36);
let mut provider_b = StringBuilder::with_capacity(rows.len(), rows.len() * 12);
let mut model_b = StringBuilder::with_capacity(rows.len(), rows.len() * 16);
let mut captured_clock_unix_ms_b = UInt64Builder::with_capacity(rows.len());
for row in rows {
partition_b.append_value(&row.partition);
position_b.append_value(row.position);
match &row.turn_id {
Some(id) => turn_id_b.append_value(id),
None => turn_id_b.append_null(),
}
provider_b.append_value(&row.provider);
model_b.append_value(&row.model);
captured_clock_unix_ms_b.append_value(row.captured_clock_unix_ms);
}
let columns: Vec<ArrayRef> = vec![
Arc::new(partition_b.finish()),
Arc::new(position_b.finish()),
Arc::new(turn_id_b.finish()),
Arc::new(provider_b.finish()),
Arc::new(model_b.finish()),
Arc::new(captured_clock_unix_ms_b.finish()),
];
RecordBatch::try_new(schema(), columns)
}
#[must_use]
pub(crate) fn decode_model_call_events(
partition: &str,
events: &[(u64, Event)],
) -> Vec<ModelCallRow> {
crate::decode::decode_typed_kind_events(
partition,
events,
&[kinds::MODEL_CALL],
"model_call",
polyc_facts::fold_model_call_event,
|partition, position, turn_id, fact: polyc_facts::ModelCallFact| ModelCallRow {
partition,
position,
turn_id,
provider: fact.provider,
model: fact.model,
captured_clock_unix_ms: fact.captured_clock_unix_ms,
},
)
}
#[cfg(test)]
mod tests {
use arrow::array::Array as _;
use buffa::Message as _;
use polyc_proto::proto::polychrome::events::v1::ModelCallEvent;
use uuid::Uuid;
use super::*;
fn sample_model_call(
provider: &str,
model: &str,
captured_clock_unix_ms: u64,
) -> ModelCallEvent {
ModelCallEvent {
provider: provider.to_string(),
model: model.to_string(),
decode_params: buffa::MessageField::default(),
system_config_ref: String::new(),
captured_clock_unix_ms,
clear_trigger_bytes: 0,
clear_keep_recent: 0,
clear_marker: String::new(),
persona_block_placement: 0,
__buffa_unknown_fields: buffa::UnknownFields::default(),
}
}
#[test]
fn schema_shape() {
let schema = schema();
let names: Vec<&str> = schema.fields().iter().map(|f| f.name().as_str()).collect();
assert_eq!(
names,
vec![
"partition",
"position",
"turn_id",
"provider",
"model",
"captured_clock_unix_ms",
]
);
let expect = [
("partition", DataType::Utf8, false),
("position", DataType::UInt64, false),
("turn_id", DataType::Utf8, true),
("provider", DataType::Utf8, false),
("model", DataType::Utf8, false),
("captured_clock_unix_ms", DataType::UInt64, false),
];
for (field, (name, ty, nullable)) in schema.fields().iter().zip(expect) {
assert_eq!(field.name(), name);
assert_eq!(field.data_type(), &ty);
assert_eq!(field.is_nullable(), nullable);
}
}
#[test]
fn decode_model_call_batch_round_trips() {
let rows = vec![
ModelCallRow {
partition: "conv-a".to_string(),
position: 5,
turn_id: None,
provider: "stub-provider".to_string(),
model: "stub-model-a".to_string(),
captured_clock_unix_ms: 1_700_000_000_000,
},
ModelCallRow {
partition: "conv-a".to_string(),
position: 9,
turn_id: Some("turn-xyz".to_string()),
provider: "alt-provider".to_string(),
model: "stub-model-b".to_string(),
captured_clock_unix_ms: 1_700_000_050_000,
},
];
let batch = decode_model_call_batch(&rows).expect("batch build");
assert_eq!(batch.num_rows(), 2);
assert_eq!(batch.schema(), schema());
let partition = batch
.column(0)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(partition.value(0), "conv-a");
assert_eq!(partition.value(1), "conv-a");
let position = batch
.column(1)
.as_any()
.downcast_ref::<arrow::array::UInt64Array>()
.unwrap();
assert_eq!(position.values(), &[5, 9]);
let turn_id = batch
.column(2)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert!(turn_id.is_null(0));
assert_eq!(turn_id.value(1), "turn-xyz");
let provider = batch
.column(3)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(provider.value(0), "stub-provider");
assert_eq!(provider.value(1), "alt-provider");
let model = batch
.column(4)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(model.value(0), "stub-model-a");
assert_eq!(model.value(1), "stub-model-b");
let captured_clock = batch
.column(5)
.as_any()
.downcast_ref::<arrow::array::UInt64Array>()
.unwrap();
assert_eq!(
captured_clock.values(),
&[1_700_000_000_000, 1_700_000_050_000]
);
}
#[test]
fn decode_model_call_events_filters_and_decodes_real_buffa_bytes() {
let turn = Uuid::from_u128(0x0195_abcd_ef01_2345_6789_abcd_ef01_2345);
let model_call = sample_model_call("stub-provider", "stub-model", 1_700_000_000_123);
let bytes = model_call.encode_to_vec();
let events = vec![
(1, Event::new(kinds::TURN_START, Vec::new())),
(
2,
Event::new(kinds::tagged(kinds::MODEL_CALL, &turn), bytes),
),
(3, Event::new(kinds::USER_MSG, b"not a model call".to_vec())),
];
let decoded = decode_model_call_events("conv-real", &events);
assert_eq!(decoded.len(), 1);
assert_eq!(decoded[0].partition, "conv-real");
assert_eq!(decoded[0].position, 2);
assert_eq!(decoded[0].turn_id, Some(turn.to_string()));
assert_eq!(decoded[0].provider, "stub-provider");
assert_eq!(decoded[0].model, "stub-model");
assert_eq!(decoded[0].captured_clock_unix_ms, 1_700_000_000_123);
}
#[test]
fn decode_model_call_events_bare_kind_has_no_turn_id() {
let events = vec![(7, Event::new(kinds::MODEL_CALL, Vec::new()))];
let decoded = decode_model_call_events("conv-bare", &events);
assert_eq!(decoded.len(), 1);
assert_eq!(decoded[0].partition, "conv-bare");
assert_eq!(decoded[0].position, 7);
assert_eq!(decoded[0].turn_id, None);
}
#[test]
fn empty_payload_model_call_event_decodes_to_defaults() {
let events = vec![(1, Event::new(kinds::MODEL_CALL, Vec::new()))];
let decoded = decode_model_call_events("conv-empty", &events);
assert_eq!(decoded.len(), 1, "an empty payload must decode, not skip");
assert_eq!(decoded[0].provider, "");
assert_eq!(decoded[0].model, "");
assert_eq!(decoded[0].captured_clock_unix_ms, 0);
}
#[test]
fn undecodable_non_empty_payload_is_skipped() {
let events = vec![
(1, Event::new(kinds::MODEL_CALL, vec![0xFF, 0xFE, 0xFD])),
(2, Event::new(kinds::MODEL_CALL, Vec::new())),
];
let decoded = decode_model_call_events("conv-corrupt", &events);
assert_eq!(decoded.len(), 1, "only the empty (valid) payload decodes");
assert_eq!(decoded[0].position, 2);
assert_eq!(decoded[0].provider, "");
assert_eq!(decoded[0].model, "");
}
#[test]
fn model_call_event_round_trip_through_decode_and_batch() {
let turn = Uuid::from_u128(0x0195_abcd_ef01_2345_6789_abcd_ef01_9999);
let model_call = sample_model_call("stub-provider", "stub-model", 1_700_000_099_999);
let bytes = model_call.encode_to_vec();
let events = vec![(
1,
Event::new(kinds::tagged(kinds::MODEL_CALL, &turn), bytes),
)];
let decoded = decode_model_call_events("conv-rt", &events);
let batch = decode_model_call_batch(&decoded).expect("batch build");
let partition = batch
.column(0)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(partition.value(0), "conv-rt");
let position = batch
.column(1)
.as_any()
.downcast_ref::<arrow::array::UInt64Array>()
.unwrap();
assert_eq!(position.value(0), 1);
let turn_id = batch
.column(2)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(turn_id.value(0), turn.to_string());
let provider = batch
.column(3)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(provider.value(0), "stub-provider");
let model = batch
.column(4)
.as_any()
.downcast_ref::<arrow::array::StringArray>()
.unwrap();
assert_eq!(model.value(0), "stub-model");
let captured_clock = batch
.column(5)
.as_any()
.downcast_ref::<arrow::array::UInt64Array>()
.unwrap();
assert_eq!(captured_clock.value(0), 1_700_000_099_999);
}
}