mediascanner

func AddMediaPath ¶

func AddMediaPath(
	db database.MediaDBI,
	ss *database.ScanState,
	systemID string,
	path string,
	noExt bool,
	stripLeadingNumbers bool,
	cfg *config.Instance,
	mediaType slugs.MediaType,
) (titleIndex, mediaIndex int, err error)

func FindPath ¶

func FindPath(ctx context.Context, path string) (string, error)

FindPath case-insensitively finds a file/folder at a path and returns the actual filesystem case. On case-insensitive filesystems (Windows, macOS), this ensures we get the real case from the filesystem rather than preserving the input case, which prevents case-mismatch issues during string comparisons.

This function recursively normalizes the entire path from root to leaf to ensure all components match the actual filesystem case.

Special handling:

• Linux: Prefers exact match before case-insensitive match (handles File.txt vs file.txt)
• Windows: Handles 8.3 short names (PROGRA~1) via fallback to os.Stat
• All platforms: Works with symlinks, UNC paths, network drives

func FlushScanStateMaps ¶

func FlushScanStateMaps(ss *database.ScanState)

FlushScanStateMaps clears the per-system in-memory maps (TitleIDs, MediaIDs) to free memory after a system finishes indexing.

IMPORTANT: This must only be called BETWEEN systems, never mid-system. Both TitleIDs and MediaIDs hold dedup state for the current system; clearing them while still inside the per-file loop would cause:

• InsertMediaTitle UNIQUE failures for multi-file titles (e.g. multi-disc games sharing one MediaTitle row), silently dropping the later files
• In persistent mode, InsertMedia UNIQUE failures for already-existing media, leaving stale MissingMedia entries that get falsely marked missing

Cross-system safety: MediaKey/TitleKey are systemID-prefixed, so clearing these maps between systems can never collide with the next system's keys. In resume/persistent mode the next system's data is reloaded by PopulateScanStateForSystem / PopulatePersistentScanStateForSystem.

Other maps are intentionally NOT cleared:

• SystemIDs preserved — DO NOT clear (UNIQUE violations with batch inserts)
• TagIDs / TagTypeIDs preserved — reused across systems
• MissingMedia preserved — reinitialized per system before loading that system's existing media and finalized immediately after that system ends

func GetFiles ¶

func GetFiles(
	ctx context.Context,
	cfg *config.Instance,
	platform platforms.Platform,
	systemID string,
	path string,
) ([]string, error)

GetFiles searches for all valid games in a given path and returns a list of files. Uses fastwalk for parallel directory traversal with built-in symlink cycle detection. Deep searches .zip files when ZipsAsDirs is enabled.

func NewNamesIndex ¶

func NewNamesIndex(
	ctx context.Context,
	platform platforms.Platform,
	cfg *config.Instance,
	systems []systemdefs.System,
	fdb *database.Database,
	update func(IndexStatus),
	pauser *syncutil.Pauser,
) (int, error)

NewNamesIndex takes a list of systems, indexes all valid game files on disk and writes a name index to the DB.

Overwrites any existing names index but does not clean up old missing files.

Takes a function which will be called with the current status of the index during key steps.

Returns the total number of files indexed.

func PopulatePersistentScanStateForSystem ¶ added in v2.11.0

func PopulatePersistentScanStateForSystem(
	ctx context.Context, db database.MediaDBI, ss *database.ScanState, systemID string,
) error

PopulatePersistentScanStateForSystem loads existing Media DBIDs for a system into ss.MissingMedia. As AddMediaPath processes files, found entries are removed. After the system is fully scanned, remaining entries represent missing media. This also populates MediaIDs and TitleIDs via PopulateScanStateForSystem.

func PopulateScanStateForSelectiveIndexing ¶ added in v2.7.0

func PopulateScanStateForSelectiveIndexing(
	ctx context.Context, db database.MediaDBI, ss *database.ScanState, systemsToReindex []string,
) error

PopulateScanStateForSelectiveIndexing populates scan state for selective indexing with optimized loading. Uses true lazy loading for Systems/TagTypes (via UNIQUE constraints) and minimal data loading for MediaTitles/Media (only systems NOT being reindexed) to dramatically improve performance.

func PopulateScanStateForSystem ¶ added in v2.9.1

func PopulateScanStateForSystem(
	ctx context.Context, db database.MediaDBI, ss *database.ScanState, systemID string,
) error

PopulateScanStateForSystem loads the existing titles and media for a single system into the scan state. This is called lazily during resume, just before processing each system.

This function is safe to call multiple times for different systems - it appends to the existing TitleIDs and MediaIDs maps.

func PopulateScanStateFromDB ¶ added in v2.7.0

func PopulateScanStateFromDB(ctx context.Context, db database.MediaDBI, ss *database.ScanState) error

PopulateScanStateFromDB initializes the scan state with existing database IDs when resuming an interrupted indexing operation

func SeedCanonicalTags ¶ added in v2.7.0

func SeedCanonicalTags(db database.MediaDBI, ss *database.ScanState) error

SeedCanonicalTags seeds the database with canonical GameDataBase-style hierarchical tags. Tags follow the format: category:subcategory:value (e.g., "genre:sports:wrestling", "players:2:vs") Tag definitions are in tags.go for centralized management.

NOTE: This function ALWAYS uses non-batch mode (prepared statements) because the canonical tag dataset contains many entries and using batch mode with fail-fast behavior would cause issues. Prepared statements handle this better.