Hongshi/mac sw inventory mem

pkg/inventory/software/collector_darwin.go

@@ -15,7 +15,6 @@ import (
 	"os/exec"
 	"path/filepath"
 	"regexp"
-	"strconv"
 	"strings"
 	"sync"
 	"time"
@@ -368,66 +367,6 @@ func getPublisherFromInfoPlist(bundlePath string) string {
 	return getPublisherFromPlistData(plistData)
 }
 
-// pkgInfo contains information about a package installation from pkgutil
-type pkgInfo struct {
-	// PkgID is the package identifier (e.g., "com.microsoft.Word")
-	PkgID string
-	// Volume is the install volume (e.g., "/")
-	Volume string
-	// InstallTime is the installation timestamp
-	InstallTime string
-}
-
-// getPkgInfo queries the macOS package receipt database to find which PKG installed
-// a specific file or directory. This uses `pkgutil --file-info` which is the official
-// way to link applications to their installer receipts.
-//
-// Parameters:
-//   - path: The path to query (e.g., "/Applications/Numbers.app")
-//
-// Returns:
-//   - *pkgInfo: Package information if the path was installed by a PKG, nil otherwise
-//
-// Note: Returns nil for apps installed via drag-and-drop (no PKG receipt) or
-// Mac App Store apps (receipt stored inside the app bundle, not in pkgutil database).
-func getPkgInfo(path string) *pkgInfo {
-	// Run pkgutil --file-info to query which package installed this path
-	cmd := exec.Command("pkgutil", "--file-info", path)
-	output, err := cmd.Output()
-	if err != nil {
-		// No package owns this path (drag-and-drop install or error)
-		return nil
-	}
-
-	// Parse the output which looks like:
-	// volume: /
-	// path: Applications/Numbers.app
-	// pkgid: com.apple.pkg.Numbers
-	// pkg-version: 14.0
-	// install-time: 1654432493
-	info := &pkgInfo{}
-	for _, line := range strings.Split(string(output), "\n") {
-		line = strings.TrimSpace(line)
-		if strings.HasPrefix(line, "pkgid: ") {
-			info.PkgID = strings.TrimPrefix(line, "pkgid: ")
-		} else if strings.HasPrefix(line, "volume: ") {
-			info.Volume = strings.TrimPrefix(line, "volume: ")
-		} else if strings.HasPrefix(line, "install-time: ") {
-			// Convert Unix timestamp to ISO 8601 format for cross-platform consistency
-			timestampStr := strings.TrimPrefix(line, "install-time: ")
-			if unixTime, err := strconv.ParseInt(timestampStr, 10, 64); err == nil {
-				info.InstallTime = time.Unix(unixTime, 0).Format(time.RFC3339)
-			}
-		}
-	}
-
-	// Only return if we found a package ID
-	if info.PkgID != "" {
-		return info
-	}
-	return nil
-}
-
 // entryWithPath pairs an Entry with its bundle path for parallel processing.
 // When plistData is non-nil, the worker uses it to compute publisher without reading the file.
 type entryWithPath struct {

pkg/inventory/software/collector_darwin_apps.go

@@ -13,7 +13,6 @@ import (
 	"path/filepath"
 	"runtime"
 	"strings"
-	"sync"
 	"time"
 )
 
@@ -69,47 +68,26 @@ func getLocalUsers() ([]string, []*Warning) {
 	return users, warnings
 }
 
-// appPkgLookup holds info needed for parallel pkgutil lookup
+// appPkgLookup holds info needed for PKG install source lookup
 type appPkgLookup struct {
 	entry   *Entry
 	appPath string
 }
 
-// populatePkgInfoParallel queries pkgutil for multiple apps in parallel
-// Uses a worker pool to limit concurrent pkgutil processes
-func populatePkgInfoParallel(items []appPkgLookup) {
-	const maxWorkers = 10 // Limit concurrent pkgutil processes
-
+// populatePkgInfoFromIndex looks up PKG install source for each app using
+// the BOM-derived reverse index, avoiding per-app subprocess spawning.
+func populatePkgInfoFromIndex(items []appPkgLookup) {
 	if len(items) == 0 {
 		return
 	}
 
-	jobs := make(chan *appPkgLookup, len(items))
+	idx := getGlobalAppToPkgIndex()
 	for i := range items {
-		jobs <- &items[i]
-	}
-	close(jobs)
-
-	var wg sync.WaitGroup
-	workerCount := maxWorkers
-	if len(items) < maxWorkers {
-		workerCount = len(items)
-	}
-
-	for i := 0; i < workerCount; i++ {
-		wg.Add(1)
-		go func() {
-			defer wg.Done()
-			for item := range jobs {
-				if pkgInfo := getPkgInfo(item.appPath); pkgInfo != nil {
-					item.entry.InstallSource = installSourcePkg
-					item.entry.PkgID = pkgInfo.PkgID
-				}
-			}
-		}()
+		if pkgID := idx.lookupPkgForApp(items[i].appPath); pkgID != "" {
+			items[i].entry.InstallSource = installSourcePkg
+			items[i].entry.PkgID = pkgID
+		}
 	}
-
-	wg.Wait()
 }
 
 // Collect scans the /Applications directory recursively for installed applications.
@@ -220,10 +198,8 @@ func (c *applicationsCollector) Collect() ([]*Entry, []*Warning, error) {
 				if _, err := os.Stat(masReceiptPath); err == nil {
 					source = softwareTypeMAS
 					installSource = installSourceMAS
-				} else {
-					// Not a MAS app or system app - will need to check pkgutil later (in parallel)
-					needsPkgLookup = true
 				}
+				needsPkgLookup = true
 			}
 
 			// Determine architecture
@@ -267,9 +243,8 @@ func (c *applicationsCollector) Collect() ([]*Entry, []*Warning, error) {
 		}
 	}
 
-	// Populate PKG info in parallel for non-MAS apps
-	// This queries pkgutil --file-info to determine if the app was installed via PKG
-	populatePkgInfoParallel(itemsForPkgLookup)
+	// Populate PKG install source from BOM-derived reverse index (zero subprocesses)
+	populatePkgInfoFromIndex(itemsForPkgLookup)
 
 	// Populate publisher info in parallel using Info.plist extraction
 	populatePublishersParallel(itemsForPublisher)