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

2026-08-22 01:15:05 -07:00
parent af1ee1c9b3
commit 367929e7d0
9 changed files with 501 additions and 21 deletions
@@ -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 := `
@@ -1,7 +1,9 @@
 package db

 import (
+	"os"
 	"path/filepath"
+	"strings"
 	"testing"
 )

@@ -53,3 +55,130 @@ func TestVectorSimilarityDeduplication(t *testing.T) {
 		t.Errorf("first job mismatch")
 	}
 }
+
+func TestDeleteTargetCascade(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_delete_target.db")
+
+	database, err := InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	// 1. Create two search targets
+	target1, err := database.CreateTarget("Target 1", "https://example.com/target1", "google_jobs")
+	if err != nil {
+		t.Fatalf("failed to create target 1: %v", err)
+	}
+	target2, err := database.CreateTarget("Target 2", "https://example.com/target2", "apple")
+	if err != nil {
+		t.Fatalf("failed to create target 2: %v", err)
+	}
+
+	// 2. Insert raw jobs for target 1 and target 2
+	job1, _, err := database.InsertRawJob("canon-t1-j1", &target1.ID, "Job 1", "Google", "Seattle", "https://google.com/1", "Software Engineer Go")
+	if err != nil {
+		t.Fatalf("failed to insert job 1: %v", err)
+	}
+	job2, _, err := database.InsertRawJob("canon-t1-j2", &target1.ID, "Job 2", "Google", "Seattle", "https://google.com/2", "Software Engineer Python")
+	if err != nil {
+		t.Fatalf("failed to insert job 2: %v", err)
+	}
+	job3, _, err := database.InsertRawJob("canon-t2-j1", &target2.ID, "Job 3", "Apple", "Seattle", "https://apple.com/1", "Software Engineer iOS")
+	if err != nil {
+		t.Fatalf("failed to insert job 3: %v", err)
+	}
+
+	// 3. Create analyses and applications for target 1 jobs and target 2 jobs
+	_ = database.CreateJobAnalysis(job1.ID, 85.0, true, "DONE")
+	_ = database.CreateJobAnalysis(job2.ID, 40.0, false, "DONE")
+	_ = database.CreateJobAnalysis(job3.ID, 90.0, true, "DONE")
+
+	_, _ = database.CreateOrUpdateApplication(job1.ID, "APPLIED", "Sent resume")
+	_, _ = database.CreateOrUpdateApplication(job3.ID, "SAVED", "Review later")
+
+	// 4. Enqueue tasks
+	_ = database.EnqueueTask("DISCOVER", map[string]interface{}{"target_id": target1.ID, "target_url": target1.TargetURL})
+	_ = database.EnqueueTask("DISCOVER", map[string]interface{}{"target_id": target2.ID, "target_url": target2.TargetURL})
+	_ = database.EnqueueTask("PROCESS", map[string]interface{}{"raw_job_id": job1.ID})
+	_ = database.EnqueueTask("PROCESS", map[string]interface{}{"raw_job_id": job3.ID})
+
+	// 5. Delete target 1
+	deletedTarget, deletedJobsCount, err := database.DeleteTarget(target1.ID)
+	if err != nil {
+		t.Fatalf("DeleteTarget failed: %v", err)
+	}
+	if deletedTarget.ID != target1.ID {
+		t.Errorf("expected deleted target ID %d, got %d", target1.ID, deletedTarget.ID)
+	}
+	if deletedJobsCount != 2 {
+		t.Errorf("expected 2 deleted jobs, got %d", deletedJobsCount)
+	}
+
+	// 6. Verify Target 1 is gone, Target 2 remains
+	allTargets, err := database.GetAllTargets()
+	if err != nil {
+		t.Fatalf("GetAllTargets failed: %v", err)
+	}
+	if len(allTargets) != 1 || allTargets[0].ID != target2.ID {
+		t.Errorf("expected 1 remaining target (target 2), got %v", allTargets)
+	}
+
+	// 7. Verify Target 1's jobs are gone, Target 2's job remains
+	jobsWithAnalysis, err := database.GetJobsWithAnalysis()
+	if err != nil {
+		t.Fatalf("GetJobsWithAnalysis failed: %v", err)
+	}
+	if len(jobsWithAnalysis) != 1 || jobsWithAnalysis[0].ID != job3.ID {
+		t.Errorf("expected only job 3 to remain, got %v", jobsWithAnalysis)
+	}
+
+	// 8. Verify applications for target 1 are gone, target 2 application remains
+	apps, err := database.GetApplications()
+	if err != nil {
+		t.Fatalf("GetApplications failed: %v", err)
+	}
+	if len(apps) != 1 || apps[0].ID != job3.ID {
+		t.Errorf("expected only job 3 application to remain, got %v", apps)
+	}
+}
+
+func TestRemoveTargetFromYAML(t *testing.T) {
+	tempDir := t.TempDir()
+	yamlPath := filepath.Join(tempDir, "targets.yaml")
+
+	initialYAML := `targets:
+  - name: "Google Seattle"
+    target_url: "https://google.com/jobs"
+    platform: "google_jobs"
+    is_active: true
+  - name: "Apple Seattle"
+    target_url: "https://apple.com/jobs"
+    platform: "apple"
+    is_active: true
+`
+	if err := os.WriteFile(yamlPath, []byte(initialYAML), 0644); err != nil {
+		t.Fatalf("failed to write initial yaml: %v", err)
+	}
+
+	// Remove Google Seattle
+	if err := RemoveTargetFromYAML(yamlPath, "https://google.com/jobs"); err != nil {
+		t.Fatalf("RemoveTargetFromYAML failed: %v", err)
+	}
+
+	// Read back and verify
+	data, err := os.ReadFile(yamlPath)
+	if err != nil {
+		t.Fatalf("failed to read yaml back: %v", err)
+	}
+
+	content := string(data)
+	if strings.Contains(content, "https://google.com/jobs") {
+		t.Errorf("expected google target to be removed, but found in file:\n%s", content)
+	}
+	if !strings.Contains(content, "https://apple.com/jobs") {
+		t.Errorf("expected apple target to remain, but not found in file:\n%s", content)
+	}
+}
+
@@ -49,3 +49,44 @@ func (db *DB) LoadTargetsFromYAML(filepath string) error {
 	}
 	return nil
 }
+
+// RemoveTargetFromYAML removes a target from the target configuration YAML file if it exists
+func RemoveTargetFromYAML(filepath string, targetURL string) error {
+	if filepath == "" {
+		return nil
+	}
+	data, err := os.ReadFile(filepath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil
+		}
+		return fmt.Errorf("could not read target config: %w", err)
+	}
+
+	var config targetConfig
+	if err := yaml.Unmarshal(data, &config); err != nil {
+		return fmt.Errorf("could not parse target config: %w", err)
+	}
+
+	var newTargets []struct {
+		Name      string `yaml:"name"`
+		TargetURL string `yaml:"target_url"`
+		Platform  string `yaml:"platform"`
+		IsActive  bool   `yaml:"is_active"`
+	}
+
+	for _, t := range config.Targets {
+		if t.TargetURL != targetURL {
+			newTargets = append(newTargets, t)
+		}
+	}
+
+	config.Targets = newTargets
+	outData, err := yaml.Marshal(&config)
+	if err != nil {
+		return fmt.Errorf("could not marshal target config: %w", err)
+	}
+
+	return os.WriteFile(filepath, outData, 0644)
+}
+