From 3df24b0419a22e7b0ce4fcf60b3603575317db69 Mon Sep 17 00:00:00 2001
From: William Stein <wstein@sagemath.com>
Date: Sat, 18 Nov 2023 18:19:44 +0000
Subject: [PATCH] improve read tracking -- add excludes

---
 websocketfs/lib/mount.ts         | 18 +++++++++++++----
 websocketfs/lib/read-tracking.ts | 34 +++++++++++++++++++++++++++++---
 websocketfs/lib/sftp-fuse.ts     |  7 ++++++-
 websocketfs/package.json         |  2 +-
 4 files changed, 52 insertions(+), 9 deletions(-)

diff --git a/websocketfs/lib/mount.ts b/websocketfs/lib/mount.ts
index 1d67a69..6b8daed 100644
--- a/websocketfs/lib/mount.ts
+++ b/websocketfs/lib/mount.ts
@@ -20,11 +20,19 @@ interface Options {
   cacheLinkTimeout?: number;
 
   // Read Tracking
-  // write out filenames of files that were explicitly read to this file
-  // separated by nulls (and it is null terminated).  A file is added no more
-  // than once.  Delete the file to reset things.  The leading slash is
-  // removed from the filenames, so they are relative to the mount point.
+  // Write out filenames of files that were explicitly read to readTrackingFile
+  // terminated by NULL, like the --null option to tar expects:
+  //    readTrackingFile
+  // A file is added no more than once until readTrackingFile is deleted.  Delete
+  // readTrackingFile to reset the processs.
+  // The leading slash is removed from the filenames, so they are relative to
+  // the mount point.
   readTrackingFile?: string;
+  // exclude given top level directories from tracking, e.g.,
+  //    ['scratch', 'tmp']
+  // would exclude scratch and tmp.  We also allow ['.*'] to mean "exclude
+  // all top level hidden directories".
+  readTrackingExclude?: string[];
 
   // Metadata
   // If the metadataFile file path is given, we poll it for modification every few seconds.
@@ -68,6 +76,7 @@ export default async function mount(
     cacheDirTimeout,
     cacheLinkTimeout,
     readTrackingFile,
+    readTrackingExclude,
     metadataFile,
     hidePath,
   } = opts;
@@ -79,6 +88,7 @@ export default async function mount(
     cacheDirTimeout,
     cacheLinkTimeout,
     readTrackingFile,
+    readTrackingExclude,
     metadataFile,
     hidePath,
   });
diff --git a/websocketfs/lib/read-tracking.ts b/websocketfs/lib/read-tracking.ts
index 2492dd2..7d0eb50 100644
--- a/websocketfs/lib/read-tracking.ts
+++ b/websocketfs/lib/read-tracking.ts
@@ -15,9 +15,21 @@ const log = debug("websocketfs:read-tracking");
 export default class ReadTracking {
   private readTrackingFile: string;
   private history = new Set<string>();
+  private excludeHidden: boolean;
+  private excludePaths: string[];
 
-  constructor(readTrackingFile) {
+  constructor(readTrackingFile: string, readTrackingExclude: string[]) {
     this.readTrackingFile = readTrackingFile;
+    this.excludeHidden = readTrackingExclude.includes(".*");
+    this.excludePaths = readTrackingExclude
+      .filter((pattern) => pattern != ".*")
+      .map((pattern) => {
+        if (!pattern.endsWith("/")) {
+          return pattern + "/";
+        } else {
+          return pattern;
+        }
+      });
     this.init();
   }
 
@@ -30,7 +42,11 @@ export default class ReadTracking {
   };
 
   trackRead = async (filename: string) => {
-    log(`fileWasRead`, { filename });
+    filename = filename.slice(1);
+    if (this.isExcluded(filename)) {
+      return;
+    }
+    log(`trackRead`, { filename });
     try {
       await stat(this.readTrackingFile);
     } catch (_) {
@@ -40,7 +56,19 @@ export default class ReadTracking {
     if (this.history.has(filename)) {
       return;
     }
-    await appendFile(this.readTrackingFile, `${filename.slice(1)}\0`);
+    await appendFile(this.readTrackingFile, `${filename}\0`);
     this.history.add(filename);
   };
+
+  private isExcluded = (filename: string) => {
+    if (this.excludeHidden && filename.startsWith(".")) {
+      return true;
+    }
+    for (const pattern of this.excludePaths) {
+      if (filename.startsWith(pattern)) {
+        return true;
+      }
+    }
+    return false;
+  };
 }
diff --git a/websocketfs/lib/sftp-fuse.ts b/websocketfs/lib/sftp-fuse.ts
index 332ee7a..1554747 100644
--- a/websocketfs/lib/sftp-fuse.ts
+++ b/websocketfs/lib/sftp-fuse.ts
@@ -44,6 +44,7 @@ interface Options {
   cacheDirTimeout?: number;
   cacheLinkTimeout?: number;
   readTrackingFile?: string;
+  readTrackingExclude?: string[];
   metadataFile?: string;
   // reconnect -- defaults to true; if true, automatically reconnects
   // to server when connection breaks.
@@ -76,6 +77,7 @@ export default class SftpFuse {
       cacheLinkTimeout,
       reconnect = true,
       readTrackingFile,
+      readTrackingExclude = [],
       metadataFile,
       hidePath,
     } = options;
@@ -111,7 +113,10 @@ export default class SftpFuse {
       });
     }
     if (readTrackingFile) {
-      this.readTracking = new ReadTracking(readTrackingFile);
+      this.readTracking = new ReadTracking(
+        readTrackingFile,
+        readTrackingExclude,
+      );
     }
     if (metadataFile) {
       if (this.attrCache != null && this.dirCache != null && cacheTimeout) {
diff --git a/websocketfs/package.json b/websocketfs/package.json
index b145546..eab0d83 100644
--- a/websocketfs/package.json
+++ b/websocketfs/package.json
@@ -1,6 +1,6 @@
 {
   "name": "websocketfs",
-  "version": "0.16.0",
+  "version": "0.17.0",
   "description": "Like sshfs, but over a WebSocket",
   "main": "./dist/lib/index.js",
   "scripts": {