feat: extract database operations module and add extensive test suite

This commit is contained in:
2026-01-28 17:32:27 -06:00
parent 992263205c
commit 1733ee5f86
14 changed files with 1539 additions and 80 deletions
+212
View File
@@ -0,0 +1,212 @@
mod helpers;
use banner::data::batch::batch_upsert_courses;
use sqlx::PgPool;
#[sqlx::test]
async fn test_batch_upsert_empty_slice(pool: PgPool) {
batch_upsert_courses(&[], &pool).await.unwrap();
let count: (i64,) = sqlx::query_as("SELECT COUNT(*) FROM courses")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count.0, 0);
}
#[sqlx::test]
async fn test_batch_upsert_inserts_new_courses(pool: PgPool) {
let courses = vec![
helpers::make_course("10001", "202510", "CS", "1083", "Intro to CS", 25, 30, 0, 5),
helpers::make_course(
"10002",
"202510",
"MAT",
"1214",
"Calculus I",
40,
45,
3,
10,
),
];
batch_upsert_courses(&courses, &pool).await.unwrap();
let rows: Vec<(String, String, String, String, i32, i32, i32, i32)> = sqlx::query_as(
"SELECT crn, subject, course_number, title, enrollment, max_enrollment, wait_count, wait_capacity
FROM courses ORDER BY crn",
)
.fetch_all(&pool)
.await
.unwrap();
assert_eq!(rows.len(), 2);
let (crn, subject, course_number, title, enrollment, max_enrollment, wait_count, wait_capacity) =
&rows[0];
assert_eq!(crn, "10001");
assert_eq!(subject, "CS");
assert_eq!(course_number, "1083");
assert_eq!(title, "Intro to CS");
assert_eq!(*enrollment, 25);
assert_eq!(*max_enrollment, 30);
assert_eq!(*wait_count, 0);
assert_eq!(*wait_capacity, 5);
let (crn, subject, ..) = &rows[1];
assert_eq!(crn, "10002");
assert_eq!(subject, "MAT");
}
#[sqlx::test]
async fn test_batch_upsert_updates_existing(pool: PgPool) {
let initial = vec![helpers::make_course(
"20001",
"202510",
"CS",
"3443",
"App Programming",
10,
35,
0,
5,
)];
batch_upsert_courses(&initial, &pool).await.unwrap();
// Upsert the same CRN+term with updated enrollment
let updated = vec![helpers::make_course(
"20001",
"202510",
"CS",
"3443",
"App Programming",
30,
35,
2,
5,
)];
batch_upsert_courses(&updated, &pool).await.unwrap();
let count: (i64,) = sqlx::query_as("SELECT COUNT(*) FROM courses")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count.0, 1, "upsert should not create a duplicate row");
let (enrollment, wait_count): (i32, i32) =
sqlx::query_as("SELECT enrollment, wait_count FROM courses WHERE crn = '20001'")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(enrollment, 30);
assert_eq!(wait_count, 2);
}
#[sqlx::test]
async fn test_batch_upsert_mixed_insert_and_update(pool: PgPool) {
let initial = vec![
helpers::make_course("30001", "202510", "CS", "1083", "Intro to CS", 10, 30, 0, 5),
helpers::make_course(
"30002",
"202510",
"CS",
"2073",
"Computer Architecture",
20,
30,
0,
5,
),
];
batch_upsert_courses(&initial, &pool).await.unwrap();
// Update both existing courses and add a new one
let mixed = vec![
helpers::make_course("30001", "202510", "CS", "1083", "Intro to CS", 15, 30, 1, 5),
helpers::make_course(
"30002",
"202510",
"CS",
"2073",
"Computer Architecture",
25,
30,
0,
5,
),
helpers::make_course(
"30003",
"202510",
"MAT",
"1214",
"Calculus I",
40,
45,
3,
10,
),
];
batch_upsert_courses(&mixed, &pool).await.unwrap();
let count: (i64,) = sqlx::query_as("SELECT COUNT(*) FROM courses")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count.0, 3, "should have 2 updated + 1 new = 3 total rows");
// Verify updated values
let (enrollment,): (i32,) =
sqlx::query_as("SELECT enrollment FROM courses WHERE crn = '30001'")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(enrollment, 15);
let (enrollment,): (i32,) =
sqlx::query_as("SELECT enrollment FROM courses WHERE crn = '30002'")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(enrollment, 25);
// Verify new row
let (subject,): (String,) = sqlx::query_as("SELECT subject FROM courses WHERE crn = '30003'")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(subject, "MAT");
}
#[sqlx::test]
async fn test_batch_upsert_unique_constraint_crn_term(pool: PgPool) {
// Same CRN, different term codes → should produce two separate rows
let courses = vec![
helpers::make_course("40001", "202510", "CS", "1083", "Intro to CS", 25, 30, 0, 5),
helpers::make_course("40001", "202520", "CS", "1083", "Intro to CS", 10, 30, 0, 5),
];
batch_upsert_courses(&courses, &pool).await.unwrap();
let count: (i64,) = sqlx::query_as("SELECT COUNT(*) FROM courses WHERE crn = '40001'")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(
count.0, 2,
"same CRN with different term codes should be separate rows"
);
let rows: Vec<(String, i32)> = sqlx::query_as(
"SELECT term_code, enrollment FROM courses WHERE crn = '40001' ORDER BY term_code",
)
.fetch_all(&pool)
.await
.unwrap();
assert_eq!(rows[0].0, "202510");
assert_eq!(rows[0].1, 25);
assert_eq!(rows[1].0, "202520");
assert_eq!(rows[1].1, 10);
}
+435
View File
@@ -0,0 +1,435 @@
mod helpers;
use banner::data::models::{ScrapePriority, TargetType};
use banner::data::scrape_jobs;
use serde_json::json;
use sqlx::PgPool;
// ── fetch_and_lock_job ──────────────────────────────────────────────
#[sqlx::test]
async fn fetch_and_lock_empty_queue(pool: PgPool) {
let result = scrape_jobs::fetch_and_lock_job(&pool).await.unwrap();
assert!(result.is_none());
}
#[sqlx::test]
async fn fetch_and_lock_returns_job_and_sets_locked_at(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
false,
0,
3,
)
.await;
let job = scrape_jobs::fetch_and_lock_job(&pool)
.await
.unwrap()
.expect("should return a job");
assert_eq!(job.id, id);
assert!(matches!(job.target_type, TargetType::Subject));
assert_eq!(job.target_payload, json!({"subject": "CS"}));
// Verify locked_at was set in the database
let (locked_at,): (Option<chrono::DateTime<chrono::Utc>>,) =
sqlx::query_as("SELECT locked_at FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert!(locked_at.is_some(), "locked_at should be set after fetch");
}
#[sqlx::test]
async fn fetch_and_lock_skips_locked_jobs(pool: PgPool) {
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
true, // locked
0,
3,
)
.await;
let result = scrape_jobs::fetch_and_lock_job(&pool).await.unwrap();
assert!(result.is_none(), "locked jobs should be skipped");
}
#[sqlx::test]
async fn fetch_and_lock_skips_future_execute_at(pool: PgPool) {
// Insert a job with execute_at in the future via raw SQL
sqlx::query(
"INSERT INTO scrape_jobs (target_type, target_payload, priority, execute_at)
VALUES ('Subject', '{\"subject\": \"CS\"}', 'Medium', NOW() + INTERVAL '1 hour')",
)
.execute(&pool)
.await
.unwrap();
let result = scrape_jobs::fetch_and_lock_job(&pool).await.unwrap();
assert!(result.is_none(), "future execute_at jobs should be skipped");
}
#[sqlx::test]
async fn fetch_and_lock_priority_desc_ordering(pool: PgPool) {
// Insert low priority first, then critical
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "LOW"}),
ScrapePriority::Low,
false,
0,
3,
)
.await;
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CRIT"}),
ScrapePriority::Critical,
false,
0,
3,
)
.await;
let job = scrape_jobs::fetch_and_lock_job(&pool)
.await
.unwrap()
.expect("should return a job");
assert_eq!(
job.target_payload,
json!({"subject": "CRIT"}),
"Critical priority should be fetched before Low"
);
}
#[sqlx::test]
async fn fetch_and_lock_execute_at_asc_ordering(pool: PgPool) {
// Insert an older job and a newer job, both same priority
sqlx::query(
"INSERT INTO scrape_jobs (target_type, target_payload, priority, execute_at)
VALUES ('Subject', '{\"subject\": \"OLDER\"}', 'Medium', NOW() - INTERVAL '2 hours')",
)
.execute(&pool)
.await
.unwrap();
sqlx::query(
"INSERT INTO scrape_jobs (target_type, target_payload, priority, execute_at)
VALUES ('Subject', '{\"subject\": \"NEWER\"}', 'Medium', NOW() - INTERVAL '1 hour')",
)
.execute(&pool)
.await
.unwrap();
let job = scrape_jobs::fetch_and_lock_job(&pool)
.await
.unwrap()
.expect("should return a job");
assert_eq!(
job.target_payload,
json!({"subject": "OLDER"}),
"Older execute_at should be fetched first"
);
}
// ── delete_job ──────────────────────────────────────────────────────
#[sqlx::test]
async fn delete_job_removes_row(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::SingleCrn,
json!({"crn": "12345"}),
ScrapePriority::High,
false,
0,
3,
)
.await;
scrape_jobs::delete_job(id, &pool).await.unwrap();
let (count,): (i64,) = sqlx::query_as("SELECT COUNT(*) FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count, 0, "row should be deleted");
}
#[sqlx::test]
async fn delete_job_nonexistent_id_no_error(pool: PgPool) {
// Deleting a non-existent ID should not error
scrape_jobs::delete_job(999_999, &pool).await.unwrap();
}
// ── unlock_job ──────────────────────────────────────────────────────
#[sqlx::test]
async fn unlock_job_clears_locked_at(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::CrnList,
json!({"crns": [1, 2, 3]}),
ScrapePriority::Medium,
true, // locked
0,
3,
)
.await;
scrape_jobs::unlock_job(id, &pool).await.unwrap();
let (locked_at,): (Option<chrono::DateTime<chrono::Utc>>,) =
sqlx::query_as("SELECT locked_at FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert!(locked_at.is_none(), "locked_at should be cleared");
}
// ── unlock_and_increment_retry ──────────────────────────────────────
#[sqlx::test]
async fn unlock_and_increment_retry_has_retries_remaining(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
true,
0, // retry_count
3, // max_retries
)
.await;
let has_retries = scrape_jobs::unlock_and_increment_retry(id, 3, &pool)
.await
.unwrap();
assert!(has_retries, "should have retries remaining (0→1, max=3)");
// Verify state in DB
let (retry_count, locked_at): (i32, Option<chrono::DateTime<chrono::Utc>>) =
sqlx::query_as("SELECT retry_count, locked_at FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(retry_count, 1);
assert!(locked_at.is_none(), "should be unlocked");
}
#[sqlx::test]
async fn unlock_and_increment_retry_exhausted(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
true,
2, // retry_count
3, // max_retries
)
.await;
let has_retries = scrape_jobs::unlock_and_increment_retry(id, 3, &pool)
.await
.unwrap();
assert!(
!has_retries,
"should NOT have retries remaining (2→3, max=3)"
);
let (retry_count,): (i32,) =
sqlx::query_as("SELECT retry_count FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(retry_count, 3);
}
#[sqlx::test]
async fn unlock_and_increment_retry_already_exceeded(pool: PgPool) {
let id = helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
true,
5, // retry_count already past max
3, // max_retries
)
.await;
let has_retries = scrape_jobs::unlock_and_increment_retry(id, 3, &pool)
.await
.unwrap();
assert!(
!has_retries,
"should NOT have retries remaining (5→6, max=3)"
);
let (retry_count,): (i32,) =
sqlx::query_as("SELECT retry_count FROM scrape_jobs WHERE id = $1")
.bind(id)
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(retry_count, 6);
}
// ── find_existing_job_payloads ──────────────────────────────────────
#[sqlx::test]
async fn find_existing_payloads_returns_matching(pool: PgPool) {
let payload_a = json!({"subject": "CS"});
let payload_b = json!({"subject": "MAT"});
let payload_c = json!({"subject": "ENG"});
// Insert A and B as Subject jobs
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
payload_a.clone(),
ScrapePriority::Medium,
false,
0,
3,
)
.await;
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
payload_b.clone(),
ScrapePriority::Medium,
false,
0,
3,
)
.await;
// Insert C as a different target type
helpers::insert_scrape_job(
&pool,
TargetType::SingleCrn,
payload_c.clone(),
ScrapePriority::Medium,
false,
0,
3,
)
.await;
let candidates = vec![payload_a.clone(), payload_b.clone(), payload_c.clone()];
let existing = scrape_jobs::find_existing_job_payloads(TargetType::Subject, &candidates, &pool)
.await
.unwrap();
assert!(existing.contains(&payload_a.to_string()));
assert!(existing.contains(&payload_b.to_string()));
// payload_c is SingleCrn, not Subject — should not match
assert!(!existing.contains(&payload_c.to_string()));
}
#[sqlx::test]
async fn find_existing_payloads_ignores_locked(pool: PgPool) {
let payload = json!({"subject": "CS"});
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
payload.clone(),
ScrapePriority::Medium,
true, // locked
0,
3,
)
.await;
let candidates = vec![payload.clone()];
let existing = scrape_jobs::find_existing_job_payloads(TargetType::Subject, &candidates, &pool)
.await
.unwrap();
assert!(existing.is_empty(), "locked jobs should be ignored");
}
#[sqlx::test]
async fn find_existing_payloads_empty_candidates(pool: PgPool) {
// Insert a job so the table isn't empty
helpers::insert_scrape_job(
&pool,
TargetType::Subject,
json!({"subject": "CS"}),
ScrapePriority::Medium,
false,
0,
3,
)
.await;
let existing = scrape_jobs::find_existing_job_payloads(TargetType::Subject, &[], &pool)
.await
.unwrap();
assert!(
existing.is_empty(),
"empty candidates should return empty result"
);
}
// ── batch_insert_jobs ───────────────────────────────────────────────
#[sqlx::test]
async fn batch_insert_jobs_inserts_multiple(pool: PgPool) {
let jobs = vec![
(
json!({"subject": "CS"}),
TargetType::Subject,
ScrapePriority::High,
),
(
json!({"subject": "MAT"}),
TargetType::Subject,
ScrapePriority::Medium,
),
(
json!({"crn": "12345"}),
TargetType::SingleCrn,
ScrapePriority::Low,
),
];
scrape_jobs::batch_insert_jobs(&jobs, &pool).await.unwrap();
let (count,): (i64,) = sqlx::query_as("SELECT COUNT(*) FROM scrape_jobs")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count, 3);
}
#[sqlx::test]
async fn batch_insert_jobs_empty_slice(pool: PgPool) {
scrape_jobs::batch_insert_jobs(&[], &pool).await.unwrap();
let (count,): (i64,) = sqlx::query_as("SELECT COUNT(*) FROM scrape_jobs")
.fetch_one(&pool)
.await
.unwrap();
assert_eq!(count, 0);
}
+88
View File
@@ -0,0 +1,88 @@
use banner::banner::Course;
use banner::data::models::{ScrapePriority, TargetType};
use chrono::Utc;
use sqlx::PgPool;
/// Build a test `Course` (Banner API model) with sensible defaults.
///
/// Only the fields used by `batch_upsert_courses` need meaningful values;
/// the rest are filled with harmless placeholders.
pub fn make_course(
crn: &str,
term: &str,
subject: &str,
course_number: &str,
title: &str,
enrollment: i32,
max_enrollment: i32,
wait_count: i32,
wait_capacity: i32,
) -> Course {
Course {
id: 0,
term: term.to_owned(),
term_desc: String::new(),
course_reference_number: crn.to_owned(),
part_of_term: "1".to_owned(),
course_number: course_number.to_owned(),
subject: subject.to_owned(),
subject_description: subject.to_owned(),
sequence_number: "001".to_owned(),
campus_description: "Main Campus".to_owned(),
schedule_type_description: "Lecture".to_owned(),
course_title: title.to_owned(),
credit_hours: Some(3),
maximum_enrollment: max_enrollment,
enrollment,
seats_available: max_enrollment - enrollment,
wait_capacity,
wait_count,
cross_list: None,
cross_list_capacity: None,
cross_list_count: None,
cross_list_available: None,
credit_hour_high: None,
credit_hour_low: None,
credit_hour_indicator: None,
open_section: enrollment < max_enrollment,
link_identifier: None,
is_section_linked: false,
subject_course: format!("{subject}{course_number}"),
reserved_seat_summary: None,
instructional_method: "FF".to_owned(),
instructional_method_description: "Face to Face".to_owned(),
section_attributes: vec![],
faculty: vec![],
meetings_faculty: vec![],
}
}
/// Insert a scrape job row directly via SQL, returning the generated ID.
pub async fn insert_scrape_job(
pool: &PgPool,
target_type: TargetType,
payload: serde_json::Value,
priority: ScrapePriority,
locked: bool,
retry_count: i32,
max_retries: i32,
) -> i32 {
let locked_at = if locked { Some(Utc::now()) } else { None };
let (id,): (i32,) = sqlx::query_as(
"INSERT INTO scrape_jobs (target_type, target_payload, priority, execute_at, locked_at, retry_count, max_retries)
VALUES ($1, $2, $3, NOW(), $4, $5, $6)
RETURNING id",
)
.bind(target_type)
.bind(payload)
.bind(priority)
.bind(locked_at)
.bind(retry_count)
.bind(max_retries)
.fetch_one(pool)
.await
.expect("insert_scrape_job failed");
id
}