data-feeds（数据馈送）

Bright Data — Data Feeds (Pipelines)

Extract structured data from supported platforms via bdata pipelines. One call, clean JSON, no scraping logic. For unsupported URLs, hand off to scrape. To find target URLs first, hand off to search.

Setup gate (run first)

if ! command -v bdata >/dev/null 2>&1; then
    echo "bdata CLI not installed — see bright-data-best-practices/references/cli-setup.md"
elif ! bdata zones >/dev/null 2>&1; then
    echo "bdata not authenticated — run: bdata login  (or: bdata login --device for SSH)"
fi

Halt and route to skills/bright-data-best-practices/references/cli-setup.md if either check fails.

Supported pipeline types (verified 2026-04-19)

Always verify with bdata pipelines list before hardcoding names — they change. Current 43 types:

amazon_product, amazon_product_reviews, amazon_product_search, apple_app_store, bestbuy_products, booking_hotel_listings, crunchbase_company, ebay_product, etsy_products, facebook_company_reviews, facebook_events, facebook_marketplace_listings, facebook_posts, github_repository_file, google_maps_reviews, google_play_store, google_shopping, homedepot_products, instagram_comments, instagram_posts, instagram_profiles, instagram_reels, linkedin_company_profile, linkedin_job_listings, linkedin_people_search, linkedin_person_profile, linkedin_posts, reddit_posts, reuter_news, tiktok_comments, tiktok_posts, tiktok_profiles, tiktok_shop, walmart_product, walmart_seller, x_posts, yahoo_finance_business, youtube_comments, youtube_profiles, youtube_videos, zara_products, zillow_properties_listing, zoominfo_company_profile

Naming note: inconsistent across platforms. amazon_product (singular), tiktok_profiles (plural), linkedin_person_profile (not linkedin_profile). Always copy from bdata pipelines list.

Pick your path

Keyword- and multi-arg pipelines (do NOT take a single URL)

A few pipelines take non-URL or multi-positional inputs. Invoke with no args to see the exact usage line from the CLI:

All other 37 pipelines take a single URL.

Action

Core commands:

# List available pipeline types (source of truth)
bdata pipelines list

# Amazon product
bdata pipelines amazon_product \
    "https://www.amazon.com/dp/B08N5WRWNW" \
    --format json --pretty -o product.json

# Amazon product reviews (slower — reviews can be hundreds)
bdata pipelines amazon_product_reviews \
    "https://www.amazon.com/dp/B08N5WRWNW" \
    --timeout 1200 -o reviews.json

# Amazon product search (keyword + domain URL)
bdata pipelines amazon_product_search \
    "noise cancelling headphones" "https://www.amazon.com" \
    --format json --pretty -o search.json

# LinkedIn person profile
bdata pipelines linkedin_person_profile \
    "https://www.linkedin.com/in/example" -o person.json

# LinkedIn company
bdata pipelines linkedin_company_profile \
    "https://www.linkedin.com/company/example" -o company.json

# LinkedIn people search (url + first + last name)
bdata pipelines linkedin_people_search \
    "https://www.linkedin.com/company/example" "Jane" "Doe" \
    -o people.json

# Instagram posts
bdata pipelines instagram_posts \
    "https://www.instagram.com/example/" -o posts.json

# Google Maps reviews (url + days_limit, default 3)
bdata pipelines google_maps_reviews \
    "https://maps.google.com/?cid=1234567890" 90 -o reviews.json

# YouTube comments (url + num_comments, default 10)
bdata pipelines youtube_comments \
    "https://www.youtube.com/watch?v=abc123" 100 -o yt-comments.json

# NDJSON for big feeds (one record per line)
bdata pipelines linkedin_posts "https://www.linkedin.com/in/example" \
    --format ndjson -o posts.ndjson

# Raise polling timeout for long jobs
bdata pipelines amazon_product_reviews "<url>" --timeout 1800 -o out.json

Full flag reference + full type table: references/flags.md.

Verification gate

1. JSON parses cleanly: jq. <output> returns 0 (or for --format ndjson, each line parses).
2. Record count matches expected. One URL usually = one record, *but* reviews/posts/comments pipelines return arrays sized by what the platform shows. Always check: jq 'length' out.json # top-level array count # OR jq 'if type == "array" then length else 1 end' out.json
3. No top-level error: jq -e 'if type == "object" then has("error") | not else true end' out.json \ || {echo "pipeline reported error"; exit 1;}
4. No per-record error: for array results, ensure no record has an error field: jq -e 'if type == "array" then map(has("error")) | any | not else true end' out.json \ || echo "WARN: one or more records have error fields" Partial failures are silent — this check is non-optional.
5. Core fields present for the pipeline type (examples): Spot-check with jq keys on the first record to learn the exact schema.

- amazon_product → .title + .price (or .final_price) - linkedin_person_profile → .name + .headline (or .position) - instagram_posts → .caption or .description + .url or .post_id - youtube_videos → .title + .video_id or .url

1. On failure: double --timeout and retry once. If still failing, bdata pipelines list to confirm the type name hasn't changed.

Red flags

• Using bdata scrape on Amazon/LinkedIn/TikTok/etc. when bdata pipelines <type> returns structured fields in one call. Loses structure and costs more time.
• Looping bdata pipelines for large jobs without rate-limiting — each call can trigger a long-running pipeline on the server. Cap parallelism at 2–3.
• Claiming success without the record-count + per-record error check. Partial failures are silent in pipeline output.
• Hardcoding pipeline type names (amazon_products with an s, linkedin_profile without _person_, etc.) — they're inconsistent across platforms. Always copy from bdata pipelines list.
• Using a tight --timeout on pipelines that legitimately take 5–15 minutes (reviews, company employees, big post feeds). Default 600s is a floor for small inputs; raise for long ones.
• Calling a keyword- or multi-arg pipeline (amazon_product_search, linkedin_people_search, google_maps_reviews, facebook_company_reviews, youtube_comments) with URL-only args — will fail with "Usage:...". Always check bdata pipelines <type> error output when in doubt.
• Passing a pages_to_search third arg to amazon_product_search — it's hardcoded to 1 by the CLI and extra args are ignored.

References

• references/flags.md — full pipelines flags + complete table of all 43 types with input shapes.
• references/patterns.md — sync timeout tuning, shell-loop batching with parallelism cap, partial-failure detection, keyword-shaped pipeline cheatsheet, legacy curl fallback, shared verification checklist.
• references/examples.md — (1) single Amazon product, (2) batch LinkedIn companies, (3) long reviews job with raised timeout, (4) mixed-platform workflow calling pipelines list first, (5) keyword-shaped amazon_product_search.