FLO/hagfish
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add data validation

Browse Source
This commit is contained in:
bartal-lsn
2026-08-25 23:33:54 +01:00
parent 3220d1d9a4
commit f9cfb99916
4 changed files with 170 additions and 113 deletions
Download Patch File Download Diff File Expand all files Collapse all files
justfile
+30 -34
View File
@@ -6,7 +6,19 @@ build:
echo "Building hagfisk binary..." echo "Building hagfisk binary..."
cargo build --release cargo build --release
# Sync binary to server # Stop app service (must stop before DB operations)
stop-app:
ssh {{host}} "systemctl --user stop hagfisk.service"
# Start app service
start-app:
ssh {{host}} "systemctl --user start hagfisk.service"
# Restart app service
restart-app:
ssh {{host}} "systemctl --user restart hagfisk.service"
# Deploy and restart (safe for regular updates)
deploy: build deploy: build
ssh -q {{host}} "rm -f {{hagfisk_dir}}/hagfisk" ssh -q {{host}} "rm -f {{hagfisk_dir}}/hagfisk"
echo "Syncing binary to remote..." echo "Syncing binary to remote..."
@@ -16,50 +28,34 @@ deploy: build
ssh -q {{host}} "systemctl --user daemon-reload" ssh -q {{host}} "systemctl --user daemon-reload"
ssh -q {{host}} "systemctl --user restart hagfisk.service" ssh -q {{host}} "systemctl --user restart hagfisk.service"
# Enable and start service # Deploy with full re-ingest (stops service first)
enable-now-app: deploy-reingest: build stop-app
ssh {{host}} "systemctl --user enable --now hagfisk.service" echo "Syncing binary to remote..."
rsync -avz --progress ./target/release/hagfisk {{host}}:{{hagfisk_dir}}/hagfisk
# Start service on remote echo "Restoring SELinux contexts..."
start-app: ssh -t -q {{host}} "sudo restorecon -RF {{hagfisk_dir}}"
ssh -q {{host}} "systemctl --user daemon-reload"
echo "Running full re-ingest..."
ssh {{host}} "cd {{hagfisk_dir}} && ./hagfisk ingest --full"
echo "Starting service..."
ssh {{host}} "systemctl --user start hagfisk.service" ssh {{host}} "systemctl --user start hagfisk.service"
# Stop service on remote # Manual full ingest (assumes service is stopped)
stop-app: ingest-full:
ssh {{host}} "systemctl --user stop hagfisk.service" ssh {{host}} "cd {{hagfisk_dir}} && ./hagfisk ingest --full"
# Restart service on remote # Stream logs from remote app
restart-app:
ssh {{host}} "systemctl --user restart hagfisk.service"
# Stream logs from remote
logs-app: logs-app:
ssh {{host}} "journalctl --user -u hagfisk -f" ssh {{host}} "journalctl --user -u hagfisk -f"
# Show status of hagfisk # Show status of app
ps-app: ps-app:
ssh {{host}} "systemctl --user status hagfisk.service" ssh {{host}} "systemctl --user status hagfisk.service"
# Enable and start ingest service and timer # Stream logs from remote ingest (manual run)
enable-now-ingest:
ssh {{host}} "systemctl --user enable --now hagfisk-ingest.service hagfisk-ingest.timer"
# Start ingest on remote
start-ingest:
ssh {{host}} "systemctl --user start hagfisk-ingest.service hagfisk-ingest.timer"
# Stop ingest on remote
stop-ingest:
ssh {{host}} "systemctl --user stop hagfisk-ingest.service hagfisk-ingest.timer"
# Restart ingest on remote
restart-ingest:
ssh {{host}} "systemctl --user restart hagfisk-ingest.service hagfisk-ingest.timer"
# Stream ingest logs from remote
logs-ingest: logs-ingest:
ssh {{host}} "journalctl --user -u hagfisk-ingest -f" ssh {{host}} "journalctl --user -u hagfisk-ingest -f"
# Show status of hagfisk ingest # Show status of ingest
ps-ingest: ps-ingest:
ssh {{host}} "systemctl --user status hagfisk-ingest.service" ssh {{host}} "systemctl --user status hagfisk-ingest.service"
src/api.rs
+32 -24
View File
@@ -48,6 +48,9 @@ fn default_limit() -> u32 {
const MAX_LIMIT: u32 = 10000; const MAX_LIMIT: u32 = 10000;
const LEAF_ONLY_FILTER: &str =
"species_code != 'TOTAL' AND gear_code != 'TOTAL' AND zone_code != 'TOTAL'";
#[derive(Debug, Clone, Deserialize, Default)] #[derive(Debug, Clone, Deserialize, Default)]
pub struct SummaryQuery { pub struct SummaryQuery {
pub species: Option<String>, pub species: Option<String>,
@@ -75,6 +78,8 @@ impl SummaryQuery {
let mut conditions: Vec<String> = Vec::new(); let mut conditions: Vec<String> = Vec::new();
let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new(); let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new();
conditions.push(format!("({LEAF_ONLY_FILTER})"));
if let Some(ref month) = self.month { if let Some(ref month) = self.month {
conditions.push("month = ?".to_string()); conditions.push("month = ?".to_string());
args.push(Box::new(month.clone())); args.push(Box::new(month.clone()));
@@ -177,6 +182,8 @@ impl AvailableFiltersQuery {
let mut conditions: Vec<String> = Vec::new(); let mut conditions: Vec<String> = Vec::new();
let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new(); let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new();
conditions.push(format!("({LEAF_ONLY_FILTER})"));
if let Some(ref month_from) = self.month_from { if let Some(ref month_from) = self.month_from {
conditions.push("month >= ?".to_string()); conditions.push("month >= ?".to_string());
args.push(Box::new(month_from.clone())); args.push(Box::new(month_from.clone()));
@@ -323,7 +330,7 @@ async fn get_species(State(state): State<AppState>) -> ApiResult<Json<Vec<Specie
let conn = state.conn.clone(); let conn = state.conn.clone();
let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<SpeciesDto>> { let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<SpeciesDto>> {
let conn = conn.blocking_lock(); let conn = conn.blocking_lock();
let mut stmt = conn.prepare("SELECT code, label FROM species ORDER BY code")?; let mut stmt = conn.prepare("SELECT code, label FROM species WHERE code != 'TOTAL' ORDER BY code")?;
let rows = stmt.query_map([], |row| { let rows = stmt.query_map([], |row| {
Ok(SpeciesDto { Ok(SpeciesDto {
code: row.get(0)?, code: row.get(0)?,
@@ -354,7 +361,8 @@ async fn get_landings(
let mut sql = String::from( let mut sql = String::from(
"SELECT month, species_code, species_label, gear_code, zone_code, \ "SELECT month, species_code, species_label, gear_code, zone_code, \
processing_code, preservation_code, shipsize_code, measure_code, value \ processing_code, preservation_code, shipsize_code, measure_code, value \
FROM landings WHERE 1=1", FROM landings WHERE 1=1 \
AND species_code != 'TOTAL' AND gear_code != 'TOTAL' AND zone_code != 'TOTAL'",
); );
let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new(); let mut args: Vec<Box<dyn duckdb::ToSql>> = Vec::new();
let mut idx = 1; let mut idx = 1;
@@ -468,7 +476,7 @@ async fn get_summary(
COALESCE(MAX(species_label), species_code) AS species_label, \ COALESCE(MAX(species_label), species_code) AS species_label, \
SUM(CASE WHEN measure_code = 'VALUE' THEN value END) AS total_value, \ SUM(CASE WHEN measure_code = 'VALUE' THEN value END) AS total_value, \
SUM(CASE WHEN measure_code = 'MASS' THEN value END) AS total_mass \ SUM(CASE WHEN measure_code = 'MASS' THEN value END) AS total_mass \
FROM landings WHERE 1=1{where_clause} AND species_code != 'TOTAL' \ FROM landings WHERE 1=1{where_clause} \
GROUP BY species_code, species_label \ GROUP BY species_code, species_label \
ORDER BY total_value DESC NULLS LAST \ ORDER BY total_value DESC NULLS LAST \
LIMIT 10" LIMIT 10"
@@ -774,7 +782,7 @@ async fn get_zones(State(state): State<AppState>) -> ApiResult<Json<Vec<LookupDt
let conn = state.conn.clone(); let conn = state.conn.clone();
let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<LookupDto>> { let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<LookupDto>> {
let conn = conn.blocking_lock(); let conn = conn.blocking_lock();
let mut stmt = conn.prepare("SELECT code, label FROM zone ORDER BY code")?; let mut stmt = conn.prepare("SELECT code, label FROM zone WHERE code != 'TOTAL' ORDER BY code")?;
let rows = stmt.query_map([], |row| { let rows = stmt.query_map([], |row| {
Ok(LookupDto { Ok(LookupDto {
code: row.get(0)?, code: row.get(0)?,
@@ -796,7 +804,7 @@ async fn get_gear(State(state): State<AppState>) -> ApiResult<Json<Vec<LookupDto
let conn = state.conn.clone(); let conn = state.conn.clone();
let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<LookupDto>> { let rows = tokio::task::spawn_blocking(move || -> db::Result<Vec<LookupDto>> {
let conn = conn.blocking_lock(); let conn = conn.blocking_lock();
let mut stmt = conn.prepare("SELECT code, label FROM gear ORDER BY code")?; let mut stmt = conn.prepare("SELECT code, label FROM gear WHERE code != 'TOTAL' ORDER BY code")?;
let rows = stmt.query_map([], |row| { let rows = stmt.query_map([], |row| {
Ok(LookupDto { Ok(LookupDto {
code: row.get(0)?, code: row.get(0)?,
@@ -831,8 +839,8 @@ mod tests {
month: "2024M01".to_string(), month: "2024M01".to_string(),
species_code: "COD".to_string(), species_code: "COD".to_string(),
species_label: "Toskur".to_string(), species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -843,8 +851,8 @@ mod tests {
month: "2024M01".to_string(), month: "2024M01".to_string(),
species_code: "COD".to_string(), species_code: "COD".to_string(),
species_label: "Toskur".to_string(), species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -855,8 +863,8 @@ mod tests {
month: "2024M01".to_string(), month: "2024M01".to_string(),
species_code: "HER".to_string(), species_code: "HER".to_string(),
species_label: "Sild".to_string(), species_label: "Sild".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -867,8 +875,8 @@ mod tests {
month: "2024M01".to_string(), month: "2024M01".to_string(),
species_code: "HER".to_string(), species_code: "HER".to_string(),
species_label: "Sild".to_string(), species_label: "Sild".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -879,8 +887,8 @@ mod tests {
month: "2024M02".to_string(), month: "2024M02".to_string(),
species_code: "COD".to_string(), species_code: "COD".to_string(),
species_label: "Toskur".to_string(), species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -891,8 +899,8 @@ mod tests {
month: "2024M02".to_string(), month: "2024M02".to_string(),
species_code: "COD".to_string(), species_code: "COD".to_string(),
species_label: "Toskur".to_string(), species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),
@@ -1153,7 +1161,7 @@ mod tests {
let state = test_state(); let state = test_state();
let base = spawn_test_server(state).await; let base = spawn_test_server(state).await;
let resp = reqwest::get(format!("{base}/api/summary?zone=TOTAL")) let resp = reqwest::get(format!("{base}/api/summary?zone=FO"))
.await .await
.unwrap(); .unwrap();
assert_eq!(resp.status(), StatusCode::OK); assert_eq!(resp.status(), StatusCode::OK);
@@ -1167,7 +1175,7 @@ mod tests {
let state = test_state(); let state = test_state();
let base = spawn_test_server(state).await; let base = spawn_test_server(state).await;
let resp = reqwest::get(format!("{base}/api/summary?gear=TOTAL")) let resp = reqwest::get(format!("{base}/api/summary?gear=TR1"))
.await .await
.unwrap(); .unwrap();
assert_eq!(resp.status(), StatusCode::OK); assert_eq!(resp.status(), StatusCode::OK);
@@ -1448,8 +1456,8 @@ mod tests {
let body: AvailableFiltersDto = resp.json().await.unwrap(); let body: AvailableFiltersDto = resp.json().await.unwrap();
assert!(body.species.contains(&"COD".to_string())); assert!(body.species.contains(&"COD".to_string()));
assert!(body.zones.contains(&"TOTAL".to_string())); assert!(body.zones.contains(&"FO".to_string()));
assert!(body.gear.contains(&"TOTAL".to_string())); assert!(body.gear.contains(&"TR1".to_string()));
} }
#[tokio::test] #[tokio::test]
@@ -1464,8 +1472,8 @@ mod tests {
let body: AvailableFiltersDto = resp.json().await.unwrap(); let body: AvailableFiltersDto = resp.json().await.unwrap();
assert!(body.species.contains(&"COD".to_string())); assert!(body.species.contains(&"COD".to_string()));
assert!(body.zones.contains(&"TOTAL".to_string())); assert!(body.zones.contains(&"FO".to_string()));
assert!(body.gear.contains(&"TOTAL".to_string())); assert!(body.gear.contains(&"TR1".to_string()));
} }
#[tokio::test] #[tokio::test]
@@ -1473,7 +1481,7 @@ mod tests {
let state = test_state(); let state = test_state();
let base = spawn_test_server(state).await; let base = spawn_test_server(state).await;
let resp = reqwest::get(format!("{base}/api/available-filters?zone_multi=TOTAL")) let resp = reqwest::get(format!("{base}/api/available-filters?zone_multi=FO"))
.await .await
.unwrap(); .unwrap();
assert_eq!(resp.status(), StatusCode::OK); assert_eq!(resp.status(), StatusCode::OK);
src/ingest.rs
+53 -21
View File
@@ -4,9 +4,9 @@ use std::collections::HashMap;
use tracing::{debug, info}; use tracing::{debug, info};
const DIM_MONTH: &str = "month"; const DIM_MONTH: &str = "month";
const DIM_SPECIES: &str = "Species (ASFIS2022)"; pub const DIM_SPECIES: &str = "Species (ASFIS2022)";
const DIM_GEAR: &str = "Fishing Gear (ISSCFG2016)"; pub const DIM_GEAR: &str = "Fishing Gear (ISSCFG2016)";
const DIM_ZONE: &str = "Economic Zone (GEONOM2023)"; pub const DIM_ZONE: &str = "Economic Zone (GEONOM2023)";
const DIM_PROCESSING: &str = "Processing (EUMOFAPresentation)"; const DIM_PROCESSING: &str = "Processing (EUMOFAPresentation)";
const DIM_PRESERVATION: &str = "Preservation (EUMOFAPreservation)"; const DIM_PRESERVATION: &str = "Preservation (EUMOFAPreservation)";
const DIM_SHIPSIZE: &str = "Shipsize"; const DIM_SHIPSIZE: &str = "Shipsize";
@@ -70,6 +70,20 @@ pub fn extract_available_months(meta: &MetadataResponse) -> Vec<String> {
.unwrap_or_default() .unwrap_or_default()
} }
pub fn extract_non_total_values(meta: &MetadataResponse, dim_code: &str) -> Vec<String> {
meta.variables
.iter()
.find(|v| v.code == dim_code)
.map(|v| {
v.values
.iter()
.filter(|val| **val != "TOTAL")
.cloned()
.collect()
})
.unwrap_or_default()
}
pub fn chunk_months(months: &[String], batch_size: usize) -> Vec<Vec<String>> { pub fn chunk_months(months: &[String], batch_size: usize) -> Vec<Vec<String>> {
if months.is_empty() { if months.is_empty() {
return vec![]; return vec![];
@@ -81,8 +95,13 @@ pub fn chunk_months(months: &[String], batch_size: usize) -> Vec<Vec<String>> {
.collect() .collect()
} }
pub fn build_query(all_months: &[String]) -> Result<Query> { pub fn build_query(
if all_months.is_empty() { months: &[String],
species_codes: &[String],
gear_codes: &[String],
zone_codes: &[String],
) -> Result<Query> {
if months.is_empty() {
return Err(IngestError::InvalidValueCode( return Err(IngestError::InvalidValueCode(
"build_query requires at least one month".to_string(), "build_query requires at least one month".to_string(),
)); ));
@@ -94,28 +113,28 @@ pub fn build_query(all_months: &[String]) -> Result<Query> {
code: DIM_MONTH.to_string(), code: DIM_MONTH.to_string(),
selection: Selection { selection: Selection {
filter: "item".to_string(), filter: "item".to_string(),
values: all_months.to_vec(), values: months.to_vec(),
}, },
}, },
QueryItem { QueryItem {
code: DIM_SPECIES.to_string(), code: DIM_SPECIES.to_string(),
selection: Selection { selection: Selection {
filter: "all".to_string(), filter: "item".to_string(),
values: vec!["*".to_string()], values: species_codes.to_vec(),
}, },
}, },
QueryItem { QueryItem {
code: DIM_GEAR.to_string(), code: DIM_GEAR.to_string(),
selection: Selection { selection: Selection {
filter: "all".to_string(), filter: "item".to_string(),
values: vec!["*".to_string()], values: gear_codes.to_vec(),
}, },
}, },
QueryItem { QueryItem {
code: DIM_ZONE.to_string(), code: DIM_ZONE.to_string(),
selection: Selection { selection: Selection {
filter: "all".to_string(), filter: "item".to_string(),
values: vec!["*".to_string()], values: zone_codes.to_vec(),
}, },
}, },
QueryItem { QueryItem {
@@ -463,10 +482,14 @@ mod tests {
maps maps
} }
#[test] #[test]
fn test_build_query_uses_correct_dimension_codes() { fn test_build_query_uses_correct_dimension_codes() {
let months = vec!["2015M01".to_string(), "2015M02".to_string()]; let months = vec!["2015M01".to_string(), "2015M02".to_string()];
let query = build_query(&months).unwrap(); let species = vec!["COD".to_string(), "HER".to_string()];
let gear = vec!["TR1".to_string(), "GN".to_string()];
let zones = vec!["FO".to_string(), "INTL".to_string()];
let query = build_query(&months, &species, &gear, &zones).unwrap();
assert_eq!(query.response.format, "json-stat2"); assert_eq!(query.response.format, "json-stat2");
assert_eq!(query.query.len(), 8); assert_eq!(query.query.len(), 8);
@@ -476,7 +499,16 @@ mod tests {
assert_eq!(query.query[0].selection.values.len(), 2); assert_eq!(query.query[0].selection.values.len(), 2);
assert_eq!(query.query[1].code, DIM_SPECIES); assert_eq!(query.query[1].code, DIM_SPECIES);
assert_eq!(query.query[1].selection.filter, "all"); assert_eq!(query.query[1].selection.filter, "item");
assert_eq!(query.query[1].selection.values, vec!["COD", "HER"]);
assert_eq!(query.query[2].code, DIM_GEAR);
assert_eq!(query.query[2].selection.filter, "item");
assert_eq!(query.query[2].selection.values, vec!["TR1", "GN"]);
assert_eq!(query.query[3].code, DIM_ZONE);
assert_eq!(query.query[3].selection.filter, "item");
assert_eq!(query.query[3].selection.values, vec!["FO", "INTL"]);
assert_eq!(query.query[4].code, DIM_PROCESSING); assert_eq!(query.query[4].code, DIM_PROCESSING);
assert_eq!(query.query[4].selection.filter, "item"); assert_eq!(query.query[4].selection.filter, "item");
@@ -503,13 +535,13 @@ mod tests {
.values .values
.contains(&"VALUE".to_string()) .contains(&"VALUE".to_string())
); );
} }
#[test] #[test]
fn test_build_query_returns_error_on_empty() { fn test_build_query_returns_error_on_empty() {
let result = build_query(&[]); let result = build_query(&[], &[], &[], &[]);
assert!(result.is_err()); assert!(result.is_err());
} }
#[test] #[test]
fn test_decode_key_indices_basic() { fn test_decode_key_indices_basic() {
src/main.rs
+25 -4
View File
@@ -1,4 +1,3 @@
// src/main.rs
mod api; mod api;
mod cli; mod cli;
mod db; mod db;
@@ -150,8 +149,24 @@ async fn run_ingest(config: &types::Config, full: bool) -> anyhow::Result<()> {
db::update_lookups(&conn, &lookup_maps)?; 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); 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 { let months_to_fetch: Vec<String> = if full {
all_months all_months
} else { } else {
@@ -203,7 +218,7 @@ async fn run_ingest(config: &types::Config, full: bool) -> anyhow::Result<()> {
"fetching batch" "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 { let data = match ingest::fetch_data(&client, &config.data_source_url, &query).await {
Ok(d) => d, 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() { for j in 0..data.dataset.value.len() {
match ingest::parse_row(j, &data.dataset, &lookup_maps) { match ingest::parse_row(j, &data.dataset, &lookup_maps) {
Ok(row) => { Ok(row) => {
if row.species_code == "TOTAL"
|| row.gear_code == "TOTAL"
|| row.zone_code == "TOTAL"
{
continue;
}
landings.push(ingest::data_row_to_landing(row)); landings.push(ingest::data_row_to_landing(row));
} }
Err(e) => { Err(e) => {
@@ -331,8 +352,8 @@ mod tests {
month: "2024M01".to_string(), month: "2024M01".to_string(),
species_code: "COD".to_string(), species_code: "COD".to_string(),
species_label: "Toskur".to_string(), species_label: "Toskur".to_string(),
gear_code: "TOTAL".to_string(), gear_code: "TR1".to_string(),
zone_code: "TOTAL".to_string(), zone_code: "FO".to_string(),
processing_code: "TOTAL".to_string(), processing_code: "TOTAL".to_string(),
preservation_code: "TOTAL".to_string(), preservation_code: "TOTAL".to_string(),
shipsize_code: "TOTAL".to_string(), shipsize_code: "TOTAL".to_string(),