add data validation

This commit is contained in:
2026-08-25 23:33:54 +01:00
parent 3220d1d9a4
commit f9cfb99916
4 changed files with 170 additions and 113 deletions
+25 -4
View File
@@ -1,4 +1,3 @@
// src/main.rs
mod api;
mod cli;
mod db;
@@ -150,8 +149,24 @@ async fn run_ingest(config: &types::Config, full: bool) -> anyhow::Result<()> {
db::update_lookups(&conn, &lookup_maps)?;
let species_codes = ingest::extract_non_total_values(&metadata, ingest::DIM_SPECIES);
let gear_codes = ingest::extract_non_total_values(&metadata, ingest::DIM_GEAR);
let zone_codes = ingest::extract_non_total_values(&metadata, ingest::DIM_ZONE);
tracing::info!(
species_count = species_codes.len(),
gear_count = gear_codes.len(),
zone_count = zone_codes.len(),
"Extracted non-TOTAL dimension values for leaf-only query"
);
let all_months = ingest::extract_available_months(&metadata);
if full {
conn.execute("DELETE FROM landings", [])?;
tracing::info!("Cleared landings table for full re-ingest");
}
let months_to_fetch: Vec<String> = if full {
all_months
} else {
@@ -203,7 +218,7 @@ async fn run_ingest(config: &types::Config, full: bool) -> anyhow::Result<()> {
"fetching batch"
);
let query = ingest::build_query(batch)?;
let query = ingest::build_query(batch, &species_codes, &gear_codes, &zone_codes)?;
let data = match ingest::fetch_data(&client, &config.data_source_url, &query).await {
Ok(d) => d,
@@ -225,6 +240,12 @@ async fn run_ingest(config: &types::Config, full: bool) -> anyhow::Result<()> {
for j in 0..data.dataset.value.len() {
match ingest::parse_row(j, &data.dataset, &lookup_maps) {
Ok(row) => {
if row.species_code == "TOTAL"
|| row.gear_code == "TOTAL"
|| row.zone_code == "TOTAL"
{
continue;
}
landings.push(ingest::data_row_to_landing(row));
}
Err(e) => {
@@ -331,8 +352,8 @@ mod tests {
month: "2024M01".to_string(),
species_code: "COD".to_string(),
species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(),
zone_code: "TOTAL".to_string(),
gear_code: "TR1".to_string(),
zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(),