Promote candidate table row families into export
This commit is contained in:
parent
6bca2d4761
commit
f7a2342a7c
4 changed files with 370 additions and 10 deletions
|
|
@ -61,10 +61,26 @@ pub(crate) struct RuntimeCandidateTableNamedRunScanReport {
|
|||
pub(crate) files_with_any_numbered_port_runs_count: usize,
|
||||
pub(crate) files_with_any_numbered_warehouse_runs_count: usize,
|
||||
pub(crate) files_with_both_numbered_run_families_count: usize,
|
||||
pub(crate) files_with_port01_11_run_at_45_55_count: usize,
|
||||
pub(crate) files_with_warehouse01_11_run_at_56_66_count: usize,
|
||||
pub(crate) port00_warehouse00_row_pair_map_counts: BTreeMap<String, usize>,
|
||||
pub(crate) port00_warehouse00_row_pair_map_paths: BTreeMap<String, Vec<String>>,
|
||||
pub(crate) numbered_port_warehouse_trailer_family_map_counts: BTreeMap<String, usize>,
|
||||
pub(crate) numbered_port_warehouse_trailer_family_map_paths: BTreeMap<String, Vec<String>>,
|
||||
pub(crate) skipped_file_count: usize,
|
||||
pub(crate) samples: Vec<RuntimeCandidateTableNamedRunScanSample>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Default)]
|
||||
struct RuntimeCandidateTableNamedRunAggregates {
|
||||
files_with_port01_11_run_at_45_55_count: usize,
|
||||
files_with_warehouse01_11_run_at_56_66_count: usize,
|
||||
port00_warehouse00_row_pair_map_counts: BTreeMap<String, usize>,
|
||||
port00_warehouse00_row_pair_map_paths: BTreeMap<String, Vec<String>>,
|
||||
numbered_port_warehouse_trailer_family_map_counts: BTreeMap<String, usize>,
|
||||
numbered_port_warehouse_trailer_family_map_paths: BTreeMap<String, Vec<String>>,
|
||||
}
|
||||
|
||||
pub(crate) fn scan_candidate_table_headers(
|
||||
root_path: &Path,
|
||||
) -> Result<(), Box<dyn std::error::Error>> {
|
||||
|
|
@ -193,6 +209,7 @@ pub(crate) fn scan_candidate_table_named_runs(
|
|||
.iter()
|
||||
.filter(|sample| !sample.port_runs.is_empty() && !sample.warehouse_runs.is_empty())
|
||||
.count();
|
||||
let aggregates = build_named_run_aggregates(&samples);
|
||||
|
||||
let report = RuntimeCandidateTableNamedRunScanReport {
|
||||
root_path: root_path.display().to_string(),
|
||||
|
|
@ -201,6 +218,15 @@ pub(crate) fn scan_candidate_table_named_runs(
|
|||
files_with_any_numbered_port_runs_count,
|
||||
files_with_any_numbered_warehouse_runs_count,
|
||||
files_with_both_numbered_run_families_count,
|
||||
files_with_port01_11_run_at_45_55_count: aggregates.files_with_port01_11_run_at_45_55_count,
|
||||
files_with_warehouse01_11_run_at_56_66_count: aggregates
|
||||
.files_with_warehouse01_11_run_at_56_66_count,
|
||||
port00_warehouse00_row_pair_map_counts: aggregates.port00_warehouse00_row_pair_map_counts,
|
||||
port00_warehouse00_row_pair_map_paths: aggregates.port00_warehouse00_row_pair_map_paths,
|
||||
numbered_port_warehouse_trailer_family_map_counts: aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_counts,
|
||||
numbered_port_warehouse_trailer_family_map_paths: aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_paths,
|
||||
skipped_file_count,
|
||||
samples,
|
||||
};
|
||||
|
|
@ -386,3 +412,233 @@ pub(crate) fn load_candidate_table_named_run_scan_sample(
|
|||
warehouse_runs,
|
||||
})
|
||||
}
|
||||
|
||||
fn build_named_run_aggregates(
|
||||
samples: &[RuntimeCandidateTableNamedRunScanSample],
|
||||
) -> RuntimeCandidateTableNamedRunAggregates {
|
||||
let mut aggregates = RuntimeCandidateTableNamedRunAggregates::default();
|
||||
|
||||
for sample in samples {
|
||||
if find_named_run(&sample.port_runs, "Port01", "Port11", 45, 55, 11).is_some() {
|
||||
aggregates.files_with_port01_11_run_at_45_55_count += 1;
|
||||
}
|
||||
if find_named_run(
|
||||
&sample.warehouse_runs,
|
||||
"Warehouse01",
|
||||
"Warehouse11",
|
||||
56,
|
||||
66,
|
||||
11,
|
||||
)
|
||||
.is_some()
|
||||
{
|
||||
aggregates.files_with_warehouse01_11_run_at_56_66_count += 1;
|
||||
}
|
||||
|
||||
if let (Some(port00_run), Some(warehouse00_run)) = (
|
||||
find_named_run_by_names(&sample.port_runs, "Port00", "Port00", 1),
|
||||
find_named_run_by_names(&sample.warehouse_runs, "Warehouse00", "Warehouse00", 1),
|
||||
) {
|
||||
let row_pair_key =
|
||||
format!("{}/{}", port00_run.start_index, warehouse00_run.start_index);
|
||||
*aggregates
|
||||
.port00_warehouse00_row_pair_map_counts
|
||||
.entry(row_pair_key.clone())
|
||||
.or_insert(0) += 1;
|
||||
aggregates
|
||||
.port00_warehouse00_row_pair_map_paths
|
||||
.entry(row_pair_key)
|
||||
.or_default()
|
||||
.push(sample.path.clone());
|
||||
}
|
||||
|
||||
if let (Some(port_run), Some(warehouse_run)) = (
|
||||
find_named_run(&sample.port_runs, "Port01", "Port11", 45, 55, 11),
|
||||
find_named_run(
|
||||
&sample.warehouse_runs,
|
||||
"Warehouse01",
|
||||
"Warehouse11",
|
||||
56,
|
||||
66,
|
||||
11,
|
||||
),
|
||||
) {
|
||||
let port_signature = trailer_signature(port_run);
|
||||
let warehouse_signature = trailer_signature(warehouse_run);
|
||||
let trailer_family_key = if port_signature == warehouse_signature {
|
||||
port_signature
|
||||
} else {
|
||||
format!("port={port_signature} | warehouse={warehouse_signature}")
|
||||
};
|
||||
*aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_counts
|
||||
.entry(trailer_family_key.clone())
|
||||
.or_insert(0) += 1;
|
||||
aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_paths
|
||||
.entry(trailer_family_key)
|
||||
.or_default()
|
||||
.push(sample.path.clone());
|
||||
}
|
||||
}
|
||||
|
||||
for paths in aggregates
|
||||
.port00_warehouse00_row_pair_map_paths
|
||||
.values_mut()
|
||||
{
|
||||
paths.sort();
|
||||
}
|
||||
for paths in aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_paths
|
||||
.values_mut()
|
||||
{
|
||||
paths.sort();
|
||||
}
|
||||
|
||||
aggregates
|
||||
}
|
||||
|
||||
fn find_named_run<'a>(
|
||||
runs: &'a [RuntimeCandidateTableNamedRun],
|
||||
first_name: &str,
|
||||
last_name: &str,
|
||||
start_index: usize,
|
||||
end_index: usize,
|
||||
count: usize,
|
||||
) -> Option<&'a RuntimeCandidateTableNamedRun> {
|
||||
runs.iter().find(|run| {
|
||||
run.first_name == first_name
|
||||
&& run.last_name == last_name
|
||||
&& run.count == count
|
||||
&& run.start_index == start_index
|
||||
&& run.end_index == end_index
|
||||
})
|
||||
}
|
||||
|
||||
fn find_named_run_by_names<'a>(
|
||||
runs: &'a [RuntimeCandidateTableNamedRun],
|
||||
first_name: &str,
|
||||
last_name: &str,
|
||||
count: usize,
|
||||
) -> Option<&'a RuntimeCandidateTableNamedRun> {
|
||||
runs.iter().find(|run| {
|
||||
run.first_name == first_name && run.last_name == last_name && run.count == count
|
||||
})
|
||||
}
|
||||
|
||||
fn trailer_signature(run: &RuntimeCandidateTableNamedRun) -> String {
|
||||
run.distinct_trailer_hex_words.join("|")
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::{
|
||||
RuntimeCandidateTableNamedRun, RuntimeCandidateTableNamedRunScanSample,
|
||||
build_named_run_aggregates,
|
||||
};
|
||||
|
||||
#[test]
|
||||
fn builds_named_run_aggregate_split_fields() {
|
||||
let samples = vec![
|
||||
RuntimeCandidateTableNamedRunScanSample {
|
||||
path: "Louisiana.gmp".to_string(),
|
||||
profile_family: "unknown".to_string(),
|
||||
source_kind: "map-fixed-catalog-range".to_string(),
|
||||
observed_entry_count: 67,
|
||||
port_runs: vec![
|
||||
sample_run("Port00", "Port00", 35, 35, 1, vec!["0x00000001"]),
|
||||
sample_run("Port01", "Port11", 45, 55, 11, vec!["0x00000001"]),
|
||||
],
|
||||
warehouse_runs: vec![
|
||||
sample_run("Warehouse00", "Warehouse00", 43, 43, 1, vec!["0x00000001"]),
|
||||
sample_run("Warehouse01", "Warehouse11", 56, 66, 11, vec!["0x00000001"]),
|
||||
],
|
||||
},
|
||||
RuntimeCandidateTableNamedRunScanSample {
|
||||
path: "State of Germany.gmp".to_string(),
|
||||
profile_family: "unknown".to_string(),
|
||||
source_kind: "map-fixed-catalog-range".to_string(),
|
||||
observed_entry_count: 67,
|
||||
port_runs: vec![
|
||||
sample_run("Port00", "Port00", 10, 10, 1, vec!["0x00000000"]),
|
||||
sample_run("Port01", "Port11", 45, 55, 11, vec!["0x00000000"]),
|
||||
],
|
||||
warehouse_runs: vec![
|
||||
sample_run("Warehouse00", "Warehouse00", 18, 18, 1, vec!["0x00000000"]),
|
||||
sample_run("Warehouse01", "Warehouse11", 56, 66, 11, vec!["0x00000000"]),
|
||||
],
|
||||
},
|
||||
];
|
||||
|
||||
let aggregates = build_named_run_aggregates(&samples);
|
||||
|
||||
assert_eq!(aggregates.files_with_port01_11_run_at_45_55_count, 2);
|
||||
assert_eq!(aggregates.files_with_warehouse01_11_run_at_56_66_count, 2);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.port00_warehouse00_row_pair_map_counts
|
||||
.get("35/43"),
|
||||
Some(&1)
|
||||
);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.port00_warehouse00_row_pair_map_counts
|
||||
.get("10/18"),
|
||||
Some(&1)
|
||||
);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_counts
|
||||
.get("0x00000001"),
|
||||
Some(&1)
|
||||
);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_counts
|
||||
.get("0x00000000"),
|
||||
Some(&1)
|
||||
);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.port00_warehouse00_row_pair_map_paths
|
||||
.get("35/43")
|
||||
.cloned()
|
||||
.unwrap_or_default(),
|
||||
vec!["Louisiana.gmp".to_string()]
|
||||
);
|
||||
assert_eq!(
|
||||
aggregates
|
||||
.numbered_port_warehouse_trailer_family_map_paths
|
||||
.get("0x00000000")
|
||||
.cloned()
|
||||
.unwrap_or_default(),
|
||||
vec!["State of Germany.gmp".to_string()]
|
||||
);
|
||||
}
|
||||
|
||||
fn sample_run(
|
||||
first_name: &str,
|
||||
last_name: &str,
|
||||
start_index: usize,
|
||||
end_index: usize,
|
||||
count: usize,
|
||||
distinct_trailer_hex_words: Vec<&str>,
|
||||
) -> RuntimeCandidateTableNamedRun {
|
||||
RuntimeCandidateTableNamedRun {
|
||||
prefix: first_name
|
||||
.trim_end_matches(|ch: char| ch.is_ascii_digit())
|
||||
.to_string(),
|
||||
start_index,
|
||||
end_index,
|
||||
count,
|
||||
first_name: first_name.to_string(),
|
||||
last_name: last_name.to_string(),
|
||||
start_offset: 0,
|
||||
end_offset: 0,
|
||||
distinct_trailer_hex_words: distinct_trailer_hex_words
|
||||
.into_iter()
|
||||
.map(str::to_string)
|
||||
.collect(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue