Feature/browser activity classifier by AryanSingh103 · Pull Request #1354 · ActivityWatch/activitywatch

AryanSingh103 · 2026-07-08T17:19:54Z

Classifies browser tabs into appropriate activity.

greptile-apps · 2026-07-08T17:22:59Z

Greptile Summary

This PR adds a new standalone aw-watcher-browser-classifier that periodically reads browser tab events from the existing aw-watcher-web bucket, classifies each URL against a hard-coded category ruleset, and writes classified events to a new output bucket via heartbeat.

The heartbeat call passes a plain dict instead of the required aw_core.models.Event object, causing an AttributeError crash on every classification cycle.
The 30-second lookback window with a 10-second poll interval means every source event is re-submitted approximately three times per cycle with no deduplication, inflating the output bucket.
progress.txt (a personal timestamp note) is included in the commit and should be removed.

Confidence Score: 2/5

Not ready to merge — the watcher crashes on its first heartbeat call and, once that is fixed, will silently write triplicate events to the output bucket on every poll cycle.

The heartbeat call passes a plain dict where aw_core.models.Event is required, causing an immediate crash on every classification cycle. Independently, the 30s lookback with a 10s sleep means every source event is reprocessed ~3 times with no deduplication. Both defects are on the hot path of every run.

aw-watcher-browser-classifier/classifier.py — the Event type mismatch and overlapping poll window both need to be addressed before the watcher produces correct output.

Important Files Changed

Filename	Overview
aw-watcher-browser-classifier/classifier.py	New browser-activity classifier watcher with two runtime bugs: `heartbeat` receives a plain dict instead of `aw_core.models.Event` (crashes at runtime), and the 30s lookback with a 10s poll interval causes every source event to be written ~3x to the output bucket.
progress.txt	Personal progress note with a single timestamp; should not be committed to the repository.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Main
    participant AW_Server as ActivityWatch Server
    participant SourceBucket as aw-watcher-web bucket
    participant OutputBucket as classifier output bucket

    Main->>AW_Server: get_buckets()
    AW_Server-->>Main: bucket list
    Main->>AW_Server: create_bucket(output_bucket)

    loop Every POLL_INTERVAL_SECONDS (10s)
        Main->>AW_Server: "get_events(source_bucket, start=now-30s, limit=50)"
        AW_Server-->>SourceBucket: query last 30s events
        SourceBucket-->>Main: recent_events

        loop For each event
            Main->>Main: "classify_url(url) -> category"
            Main->>AW_Server: "heartbeat(output_bucket, dict, pulsetime=15)"
            Note over Main,AW_Server: dict passed instead of Event object
            AW_Server-->>OutputBucket: write classified event
        end

        Main->>Main: sleep(10s)
    end

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Main
    participant AW_Server as ActivityWatch Server
    participant SourceBucket as aw-watcher-web bucket
    participant OutputBucket as classifier output bucket

    Main->>AW_Server: get_buckets()
    AW_Server-->>Main: bucket list
    Main->>AW_Server: create_bucket(output_bucket)

    loop Every POLL_INTERVAL_SECONDS (10s)
        Main->>AW_Server: "get_events(source_bucket, start=now-30s, limit=50)"
        AW_Server-->>SourceBucket: query last 30s events
        SourceBucket-->>Main: recent_events

        loop For each event
            Main->>Main: "classify_url(url) -> category"
            Main->>AW_Server: "heartbeat(output_bucket, dict, pulsetime=15)"
            Note over Main,AW_Server: dict passed instead of Event object
            AW_Server-->>OutputBucket: write classified event
        end

        Main->>Main: sleep(10s)
    end

_{Reviews (1): Last reviewed commit: "Add custom browser-activity classifier w..." | Re-trigger Greptile}

greptile-apps · 2026-07-08T17:23:06Z

+                classified_event = {
+                    "timestamp": event.timestamp,
+                    "duration": event.duration.total_seconds() if event.duration else 0,
+                    "data": {"url": url, "title": title, "category": category},
+                }
+                client.heartbeat(
+                    output_bucket, classified_event,
+                    pulsetime=POLL_INTERVAL_SECONDS + 5, queued=True,
+                )


heartbeat receives a plain dict instead of an Event object

The official aw-client API signature is heartbeat(bucket_id: str, event: Event, pulsetime: float, ...) where event must be an aw_core.models.Event instance. Passing a plain dict here will raise an AttributeError at runtime when the client internally accesses attributes like .timestamp and .data on the object.

greptile-apps · 2026-07-08T17:23:07Z

+        while True:
+            since = datetime.now(timezone.utc) - timedelta(seconds=LOOKBACK_SECONDS)
+            recent_events = client.get_events(source_bucket, start=since, limit=50)
+
+            for event in recent_events:
+                url = event.data.get("url", "")
+                title = event.data.get("title", "")
+                category = classify_url(url)
+
+                classified_event = {
+                    "timestamp": event.timestamp,
+                    "duration": event.duration.total_seconds() if event.duration else 0,
+                    "data": {"url": url, "title": title, "category": category},
+                }
+                client.heartbeat(
+                    output_bucket, classified_event,
+                    pulsetime=POLL_INTERVAL_SECONDS + 5, queued=True,
+                )
+                log.info("Classified [%s] -> %s", category, url or title)
+
+            time.sleep(POLL_INTERVAL_SECONDS)


Overlapping lookback window causes duplicate events in the output bucket

LOOKBACK_SECONDS = 30 but the loop sleeps only POLL_INTERVAL_SECONDS = 10 seconds between runs. Each poll fetches the last 30 s of source events, so roughly 20 s worth of events from the previous cycle are re-fetched and re-heartbeated every iteration — meaning each source event is written ~3 times to the output bucket. There is no deduplication guard (e.g., tracking the last-processed timestamp or event ID).

greptile-apps · 2026-07-08T17:23:08Z

+    for category, domains in CATEGORY_RULES.items():
+        if any(known_domain in domain for known_domain in domains):
+            return category


Substring domain check causes false positives

known_domain in domain checks whether the rule string is a substring of the parsed netloc. For example, "x.com" would match "notx.com" and "neetcode.io" would match "myneetcode.io". Anchoring the check (e.g., domain == known_domain or domain.endswith("." + known_domain)) would prevent these misclassifications.

greptile-apps · 2026-07-08T17:23:09Z

@@ -0,0 +1 @@
+Update: Mon Mar 23 03:20:19 EDT 2026


Personal progress note should not be committed to the repository

progress.txt is a personal bookkeeping file with no relevance to the codebase. It should be removed from the PR and added to .gitignore.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

AryanSingh103 added 3 commits March 23, 2026 03:20

add daily progress note

9e16e7b

Merge branch 'ActivityWatch:master' into master

9eccc7a

Add custom browser-activity classifier watcher

cf6b47c

greptile-apps Bot reviewed Jul 8, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

Feature/browser activity classifier#1354

Feature/browser activity classifier#1354
AryanSingh103 wants to merge 3 commits into
ActivityWatch:masterfrom
AryanSingh103:feature/browser-activity-classifier

AryanSingh103 commented Jul 8, 2026

Uh oh!

greptile-apps Bot commented Jul 8, 2026

Uh oh!

greptile-apps Bot Jul 8, 2026

Uh oh!

greptile-apps Bot Jul 8, 2026

Uh oh!

greptile-apps Bot Jul 8, 2026

Uh oh!

greptile-apps Bot Jul 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

AryanSingh103 commented Jul 8, 2026

Uh oh!

greptile-apps Bot commented Jul 8, 2026

Greptile Summary

Confidence Score: 2/5

Important Files Changed

Sequence Diagram

Uh oh!

greptile-apps Bot Jul 8, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Jul 8, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Jul 8, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Jul 8, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant