Recopila comentarios públicos, datos de usuarios, interacción y marcas de tiempo de un vídeo de Bilibili mediante su ID BV o av
YouTube Video Comments
Give one YouTube video link and an expected comment count to receive public comments, authors, engagement, and publish times
Original · en-USVolver a la traducción
Descripción general
Public comments on a YouTube video show how viewers react in their own words, which remarks attract replies, and whether the creator engages back. This Data App turns the public comment area of one video into a consistent dataset with comment text, author details, reply relationships, engagement snapshots, and timestamps.
The result is suited to repeatable analysis rather than manual scrolling. It keeps the original comment wording and connects each comment to its video, author, and parent comment, making it easier to compare reactions, identify recurring themes, and inspect conversations around a video.
Data notes
The data covers comments that are publicly visible on the supplied YouTube video at collection time. Records are ordered from newest to oldest by Published at. Each normal record is identified by its Comment ID, which is used to remove duplicates. Private, deleted, members-only, and closed-comment videos are outside the coverage.
Like count, Reply count, Dislike count, Comment count, Channel owner, and Creator heart are snapshots and may change after collection. Count values and classification values such as Comment type and Raw comment type are retained as source text rather than converted into numbers. Published at, Collected at, and Cleaned at preserve the source timestamps without inferring a timezone. Relative time keeps the on-page display text, such as 1年前. Keyword, Business ID, Video URL, and Source item may repeat the submitted video link when that is what the source record carries. No data retains the collection outcome when a video has no public comment result.
What the results look like
Each normal record represents one publicly visible comment from the selected YouTube video.
| Comment ID | Comment text | Author ID | Published at | Like count | Reply count |
|---|---|---|---|---|---|
| UgwnKwe6YJKooQl-JaZ4AaABAg | 年紀漸長覺得這好像不錯 可以有大按鈕也不用佔螢幕空間 | @attacksoil | 2025-08-25T14:52:01 | 1 | 0 |
| UgwYMOjMzFk7w-8L0Fl4AaABAg | Good job DR RICK 👍 | @JoeGentry-g7k | 2025-08-27T15:03:26 | 0 | 0 |
Additional values include avatar and channel details, video identifiers, reply relationships, creator interaction flags, collection timestamps, and the no-data outcome.
Use cases
- Compare Comment text with Like count and Reply count to identify remarks that generated stronger visible engagement.
- Group Comment text by recurring phrases, then use Published at to see how the discussion changed over time.
- Reconstruct reply threads with Parent comment ID and Comment type to distinguish top-level remarks from follow-up replies.
- Review Channel owner and Creator heart together with Author ID to see how the channel participated in the discussion.
Alcance y límites
Each run collects comments from one YouTube video. The expected comment count is 1 to 10000. Results are returned in complete batches of 20, so the actual count may be rounded up. Only publicly visible comments at collection time are covered.
- Adecuada para
- When you need public comment text and engagement details from a specific YouTube video
- When you need comment authors, reply relationships, timestamps, and creator interaction signals for analysis
- No usar para
- When you need comments from more than one video in a single run — submit a separate run for each video
- When you need private, deleted, members-only, or closed-comment data — this App covers publicly visible comments only
Gestión de fallos
Comportamiento de fallos y reintentos declarado por el desarrollador. Recomendamos incluirlo en tu prompt del sistema al integrar.
- 1If no comments are returned, verify the video link and confirm that the video and its comments are publicly visible
- 2Retry later after rate limiting, temporary service unavailability, or a timeout
- 3When the submission outcome is unknown, avoid immediate repeated submissions that could create duplicate tasks
- 4If authentication fails, verify the API Key configured for the execution environment
Entrada
Parámetros necesarios para llamar a esta App, generados a partir del input.schema del manifest.json.
- Video link
video_urlObligatoriostringThe full YouTube watch URL of the public video whose comments will be collected. Enter one video link only; an invalid or inaccessible video may return no comments.Ejemplohttps://www.youtube.com/watch?v=mpUwmWy_tlg - Expected comment count
comment_countObligatoriointegerThe expected number of comments to collect, from 1 to 10000. Results are returned and billed in complete batches of 20, so the actual count may be rounded up.Rango1–10000Ejemplo20
Salida
Estructura de campos de un único registro, generada a partir del output.schema del manifest.json.
- Comment ID
comment_idstringThe stable unique identifier of the YouTube commentEjemploUgwnKwe6YJKooQl-JaZ4AaABAg - Video ID
content_idstringThe identifier of the YouTube video associated with the commentEjemplompUwmWy_tlg - Record ID
cur_idstringThe source record identifier returned with this commentEjemplo6a8d3b9297884a4f18a27756 - Comment type
comment_typestringThe source value identifying the comment type, retained as source textEjemplo1
Datos de ejemplo
Generado a partir de output.schema. Los campos sin valor de ejemplo usan marcadores de estructura.
{
"comment_id": "UgwnKwe6YJKooQl-JaZ4AaABAg",
"content_id": "mpUwmWy_tlg",
"cur_id": "6a8d3b9297884a4f18a27756",
"comment_type": "1",
"parent_comment_id": "<parent_comment_id>",
"comment_user_id": "@attacksoil",
"comment_user_avatar_url": "https://yt3.ggpht.com/ytc/AIdro_nx14vEi8CIGWe-1TfJUlTVRCSMosevvM0iyj2yylc=s88-c-k-c0x00ffffff-no-rj",
"comment_text": "年紀漸長覺得這好像不錯 可以有大按鈕也不用佔螢幕空間",
"publish_time": "2025-08-25T14:52:01",
"crawl_time": "2026-08-25T14:52:02",
"like_count": "1",
"reply_count": "0",
"keyword": "https://www.youtube.com/watch?v=mpUwmWy_tlg",
"root_content_url": "https://www.youtube.com/watch?v=mpUwmWy_tlg",
"channel_name": "<channel_name>",
"comment_time": "1年前",
"raw_comment_type": "0",
"dislike_count": "<dislike_count>",
"clean_update_time": "2026-08-25T14:52:02",
"author_is_channel_owner": false,
"comment_count": "4",
"has_creator_heart": false,
"biz_id": "https://www.youtube.com/watch?v=mpUwmWy_tlg",
"source_item": "https://www.youtube.com/watch?v=mpUwmWy_tlg",
"no_data": false
}Integración
Esta App se puede integrar mediante MCP, API, SDK o exportación de archivos — todos los canales comparten las mismas capacidades y precios. Cada solicitud se autentica con la cabecera Authorization: Bearer usando una API Key (de larga duración, creada en la consola de Data Hub); los clientes MCP también pueden iniciar sesión con OAuth sin clave. Más opciones como CLI y Skill están en camino. Esta App se ejecuta de forma asíncrona y se puede cancelar a mitad de camino: Tras la cancelación, la ejecución se detiene. Los registros ya entregados se conservan y se facturan.
Con MCP (Model Context Protocol) puedes llamar a esta App directamente desde clientes de IA como Claude y Cursor. Elige tu cliente y el modo de autenticación, y copia la configuración de abajo.
Configuración del cliente
Sustituye el valor tras Bearer por tu API Key de larga duración. Funciona en cualquier cliente, CI o entorno sin interfaz.
{
"mcpServers": {
"octoparse_datahub": {
"type": "http",
"url": "https://mcp-v2.octoparse.com?pin=hello_moto/youtube-video-comments",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}Deja que la IA lo configure por ti
¿No quieres editar configuraciones a mano? Copia el prompt de instalación y pégalo en cualquier cliente de IA: completará la configuración a su manera. (De forma predeterminada te guía para iniciar sesión y nunca te pide que envíes tu API Key en el chat.)
detail.access.mcp.composeHint
Precios
Se cobra según el número de registros devueltos correctamente. Las ejecuciones fallidas no se cobran. Se factura en unidades de 20 registros; cualquier unidad parcial se redondea a 20 registros.
Se cobra una vez por cada ejecución enviada, independientemente del número de registros devueltos.
Los distintos eventos de facturación se acumulan de forma independiente. Consulta cada elemento para más detalles. Las ejecuciones fallidas no se cobran.
Ver créditosPruébalo ahora
Rellena los parámetros y ejecuta — los resultados provienen de una llamada real.