feat: add target deletion functionality with cascaded removal of associated jobs and tasks

internal/db/queries.go

@@ -2,6 +2,7 @@ package db
 
 import (
 	"database/sql"
+	"fmt"
 	"math"
 	"regexp"
 	"strings"
@@ -83,6 +84,86 @@ func (db *DB) UpdateTargetCrawledAt(id int64) error {
 	return err
 }
 
+func (db *DB) DeleteTarget(id int64) (*SearchTarget, int64, error) {
+	tx, err := db.Begin()
+	if err != nil {
+		return nil, 0, err
+	}
+	defer func() { _ = tx.Rollback() }()
+
+	// 1. Get target info before deleting
+	var target SearchTarget
+	var lastCrawled sql.NullTime
+	err = tx.QueryRow(`SELECT id, name, target_url, platform, is_active, last_crawled_at FROM search_targets WHERE id = ?`, id).Scan(
+		&target.ID, &target.Name, &target.TargetURL, &target.Platform, &target.IsActive, &lastCrawled,
+	)
+	if err != nil {
+		return nil, 0, err
+	}
+	if lastCrawled.Valid {
+		target.LastCrawledAt = &lastCrawled.Time
+	}
+
+	// 2. Find all raw_jobs IDs associated with this target
+	rows, err := tx.Query(`SELECT id, canonical_id FROM raw_jobs WHERE target_id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+	var rawJobIDs []int64
+	var canonicalIDs []string
+	for rows.Next() {
+		var rID int64
+		var cID string
+		if err := rows.Scan(&rID, &cID); err != nil {
+			_ = rows.Close()
+			return nil, 0, err
+		}
+		rawJobIDs = append(rawJobIDs, rID)
+		canonicalIDs = append(canonicalIDs, cID)
+	}
+	_ = rows.Close()
+
+	// 3. Delete tasks referencing this target or its discovered jobs
+	targetIDPattern := fmt.Sprintf(`%%"target_id":%d%%`, id)
+	targetIDPatternStr := fmt.Sprintf(`%%"target_id": %d%%`, id)
+	_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ? OR payload LIKE ?`, targetIDPattern, targetIDPatternStr)
+
+	for _, cid := range canonicalIDs {
+		cPattern := fmt.Sprintf(`%%"canonical_id":"%s"%%`, cid)
+		_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ?`, cPattern)
+	}
+	for _, rID := range rawJobIDs {
+		rPattern := fmt.Sprintf(`%%"raw_job_id":%d%%`, rID)
+		rPatternStr := fmt.Sprintf(`%%"raw_job_id": %d%%`, rID)
+		_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ? OR payload LIKE ?`, rPattern, rPatternStr)
+	}
+
+	// 4. Delete job_applications and job_analyses for these raw_jobs
+	for _, rID := range rawJobIDs {
+		_, _ = tx.Exec(`DELETE FROM job_applications WHERE raw_job_id = ?`, rID)
+		_, _ = tx.Exec(`DELETE FROM job_analyses WHERE raw_job_id = ?`, rID)
+	}
+
+	// 5. Delete raw_jobs for this target
+	res, err := tx.Exec(`DELETE FROM raw_jobs WHERE target_id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+	deletedJobsCount, _ := res.RowsAffected()
+
+	// 6. Delete search_target
+	_, err = tx.Exec(`DELETE FROM search_targets WHERE id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	if err := tx.Commit(); err != nil {
+		return nil, 0, err
+	}
+
+	return &target, deletedJobsCount, nil
+}
+
 // Raw Jobs Queries & Vector Similarity Deduplication
 func (db *DB) InsertRawJob(canonicalID string, targetID *int64, title, company, location, applyURL, rawContent string) (*RawJob, bool, error) {
 	// Deduplication Check 1 & 2: Check canonical_id exist AND vector similarity >= 99%
@@ -207,8 +288,6 @@ func tokenizeToVector(text string) map[string]int {
 	return freqMap
 }
 
-
-
 // Job Analyses Queries
 func (db *DB) CreateJobAnalysis(rawJobID int64, score float64, isQualified bool, status string) error {
 	query := `