Gather symlink targets on crawl/change processing

Summary:
The groundwork has been done to store symlink targets in `metro-file-map`'s in-memory (and persisted) map, but we're not yet reading link targets.

This diff adds a `readLink` worker option and sets it when appropriate within `_processFile`, and delegates to the worker for consistency with other operations that require a "visit" to gather metadata subsequent to a crawl result or change event.

Symlink targets will then be stored for all symlinks in the cache, and we'll avoid any IO if their modification time doesn't change.

Changelog: [Internal]

Reviewed By: motiz88

Differential Revision: D42546808

fbshipit-source-id: 5c9737971f50dc90ade43f709ef896124023203e

packages/metro-file-map/src/HasteFS.js

@@ -72,6 +72,10 @@ export default class HasteFS implements MutableFileSystem {
     if (visitResult.dependencies != null) {
       metadata[H.DEPENDENCIES] = visitResult.dependencies;
     }
+
+    if (visitResult.symlinkTarget != null) {
+      metadata[H.SYMLINK] = visitResult.symlinkTarget;
+    }
   }
 
   getSerializableSnapshot(): FileData {
@@ -93,6 +97,16 @@ export default class HasteFS implements MutableFileSystem {
     return (fileMetadata && fileMetadata[H.SIZE]) ?? null;
   }
 
+  getSymlinkTarget(file: Path): ?string {
+    const fileMetadata = this._getFileData(file);
+    if (fileMetadata == null) {
+      return null;
+    }
+    return typeof fileMetadata[H.SYMLINK] === 'string'
+      ? fileMetadata[H.SYMLINK]
+      : null;
+  }
+
   getType(file: Path): ?('f' | 'l') {
     const fileMetadata = this._getFileData(file);
     if (fileMetadata == null) {