Configuration

User Agent Configuration

Crawlers should always act responsibly and ethically when accessing websites. A key aspect of this is properly identifying themselves through the User-Agent header. By providing a clear and accurate user agent string, webmasters can understand who is visiting their site and why, and can apply rules in robots.txt accordingly. Respecting these rules, avoiding excessive request rates, and honoring content restrictions not only ensures legal compliance but also maintains a healthy relationship with the web community. Transparent identification is a fundamental part of ethical web crawling.

The configuration of the user agent in StormCrawler has two purposes:

  1. Identification of the crawler for webmasters

  2. Selection of rules from robots.txt

Crawler Identification

The politeness of a web crawler is not limited to how frequently it fetches pages from a site, but also in how it identifies itself to sites it crawls. This is done by setting the HTTP header User-Agent, just like your web browser does.

The full user agent string is built from the concatenation of the configuration elements:

  • http.agent.name: name of your crawler

  • http.agent.version: version of your crawler

  • http.agent.description: description of what it does

  • http.agent.url: URL webmasters can go to to learn about it

  • http.agent.email: an email so that they can get in touch with you

Whereas StormCrawler used to provide a default value for these, this is not the case since version 2.11 and you will now be asked to provide a value.

You can specify the user agent verbatim with the config http.agent but you will still need to provide a http.agent.name for parsing robots.txt files.

Robots Exclusion Protocol

This is also known as the robots.txt protocol, it is formalised in RFC 9309. Part of what the robots directives do is to define rules to specify which parts of a website (if any) are allowed to be crawled. The rules are organised by User-Agent, with a * to match any agent not otherwise specified explicitly, e.g.:

User-Agent: *
Disallow: *.gif$
Disallow: /example/
Allow: /publications/

In the example above the rule allows access to the URLs with the /publications/ path prefix, and it restricts access to the URLs with the /example/ path prefix and to all URLs with a .gif suffix. The "*" character designates any character, including the otherwise-required forward slash.

The value of http.agent.name is what StormCrawler looks for in the robots.txt. It MUST contain only uppercase and lowercase letters ("a-z" and "A-Z"), underscores ("_"), and hyphens ("-").

Unless you are running a well known web crawler, it is unlikely that its agent name will be listed explicitly in the robots.txt (if it is, well, congratulations!). While you want the agent name value to reflect who your crawler is, you might want to follow rules set for better known crawlers. For instance, if you were a responsible AI company crawling the web to build a dataset to train LLMs, you would want to follow the rules set for Google-Extended (see list of Google crawlers) if any were found.

This is what the configuration http.robots.agents allows you to do. It is a comma-separated string but can also take a list of values. By setting it alongside http.agent.name (which should also be the first value it contains), you are able to broaden the match rules based on the identity as well as the purpose of your crawler.

Proxy

StormCrawler’s proxy system is built on top of the SCProxy class and the ProxyManager interface. Every proxy used in the system is formatted as a SCProxy. The ProxyManager implementations handle the management and delegation of their internal proxies. At the call of Protocol#getProtocolOutput(), the ProxyManager.getProxy() is called to retrieve a proxy for the individual request.

The ProxyManager interface can be implemented in a custom class to create custom logic for proxy management and load balancing. The default ProxyManager implementation is SingleProxyManager. This ensures backwards compatibility for prior StormCrawler releases. To use MultiProxyManager or custom implementations, pass the class path and name via the config parameter http.proxy.manager:

http.proxy.manager: "org.apache.stormcrawler.proxy.MultiProxyManager"

StormCrawler implements two ProxyManager classes by default:

  • SingleProxyManager Manages a single proxy passed by the backwards compatible proxy fields in the configuration:

    ----
http.proxy.host
http.proxy.port
http.proxy.type
http.proxy.user (optional)
http.proxy.pass (optional)
----

  • MultiProxyManager Manages multiple proxies passed through a TXT file. The file should contain connection strings for all proxies including the protocol and authentication (if needed). The file supports comment lines (// or #) and empty lines. The file path should be passed via the config at the below field. The TXT file must be available to all nodes participating in the topology:

    ----
http.proxy.file
----

The MultiProxyManager load balances across proxies using one of the following schemes. The load balancing scheme can be passed via the config using http.proxy.rotation; the default value is ROUND_ROBIN:

  • ROUND_ROBIN Evenly distributes load across all proxies

  • RANDOM Randomly selects proxies using the native Java random number generator. RNG is seeded with the nanos at instantiation

  • LEAST_USED Selects the proxy with the least amount of usage. This is performed lazily for speed and therefore will not account for changes to usages during the selection process. If no custom implementations are made this should theoretically operate the same as ROUND_ROBIN

The SCProxy class contains all of the information associated with proxy connection. In addition, it tracks the total usage of the proxy and optionally tracks the location of the proxy IP. Usage information is used for the LEAST_USED load balancing scheme. The location information is currently unused but left to enable custom implementations the ability to select proxies by location.

Metadata

Registering Metadata for Kryo Serialization

If your Apache StormCrawler topology doesn’t extend org.apache.stormcrawler.ConfigurableTopology, you will need to manually register StormCrawler’s Metadata class for serialization in Storm. For more information on Kryo serialization in Apache Storm, you can refer to the documentation.

To register Metadata for serialization, you’ll need to import org.apache.storm.Config and org.apache.stormcrawler.Metadata. Then, in your topology class, you’ll register the class with:

Config.registerSerialization(conf, Metadata.class);

where conf is your Storm configuration for the topology.

Alternatively, you can specify in the configuration file:

topology.kryo.register:
  - org.apache.stormcrawler.Metadata

MetadataTransfer

The class MetadataTransfer is an important part of the framework and is used in key parts of a pipeline:

  • Fetching

  • Parsing

  • Updating bolts

An instance (or extension) of MetadataTransfer gets created and configured with the method:

public static MetadataTransfer getInstance(Map<String, Object> conf)

which takes as parameter the standard Storm .

A MetadataTransfer instance has mainly two methods, both returning Metadata objects:

  • getMetaForOutlink(String targetURL, String sourceURL, Metadata parentMD)

  • filter(Metadata metadata)

The former is used when creating Outlinks, i.e., in the parsing bolts but also for handling redirections in the [[FetcherBolt(s)]].

The latter is used by extensions of the AbstractStatusUpdaterBolt class to determine which Metadata should be persisted.

The behavior of the default MetadataTransfer class is driven by configuration only. It has the following options:

  • metadata.transfer:: list of metadata key values to filter or transfer to the outlinks. Please see the corresponding comments in crawler-default.yaml

  • metadata.persist:: list of metadata key values to persist in the status storage. Please see the corresponding comments in crawler-default.yaml

  • metadata.track.path:: whether to track the URL path or not. Boolean value, true by default.

  • metadata.track.depth:: whether to track the depth from seed. Boolean value, true by default.

Note that the method getMetaForOutlink calls filter to determine which key values to keep.

Configuration Options

The following tables describe all available configuration options and their default values. If one of the keys is not present in your YAML file, the default value will be taken.

Note: Some configuration options may not be applicable depending on the specific components and features you are using in your Apache StormCrawler topology. Some external modules might define additional options not listed here.

Fetching and Partitioning

key default value description

fetcher.max.crawl.delay

30

Maximum accepted Crawl-delay from robots.txt (seconds). If the crawl-delay exceeds this value the behavior depends on the value of fetcher.max.crawl.delay.force.

fetcher.max.crawl.delay.force

false

Configures the behavior of fetcher if the robots.txt crawl-delay exceeds fetcher.max.crawl.delay. If false: the tuple is emitted to the StatusStream as an ERROR. If true: the queue delay is set to fetcher.max.crawl.delay.

fetcher.max.queue.size

-1

The maximum length of the queue used to store items to be fetched by the FetcherBolt. A setting of -1 sets the length to Integer.MAX_VALUE.

fetcher.max.throttle.sleep

-1

The maximum amount of time to wait between fetches; if exceeded, the item is sent to the back of the queue. Used in SimpleFetcherBolt. -1 disables it.

fetcher.max.urls.in.queues

-1

Limits the number of URLs that can be stored in a fetch queue. -1 disables the limit.

fetcher.maxThreads.host/domain/ip

fetcher.threads.per.queue

Overwrites fetcher.threads.per.queue. Useful for crawling some domains/hosts/IPs more intensively.

fetcher.metrics.time.bucket.secs

10

Metrics events emitted every value seconds to the system stream.

fetcher.queue.mode

byHost

Possible values: byHost, byDomain, byIP. Determines queue grouping.

fetcher.server.delay

1

Delay between crawls in the same queue if no Crawl-delay is defined.

fetcher.server.delay.force

false

Defines fetcher behavior when the robots.txt crawl-delay is smaller than fetcher.server.delay.

fetcher.server.min.delay

0

Delay between crawls for queues with >1 thread. Ignores robots.txt.

fetcher.threads.number

10

Total concurrent threads fetching pages. Adjust carefully based on system capacity.

fetcher.threads.per.queue

1

Default number of threads per queue. Can be overridden.

fetcher.threads.start.delay

10

Delay (milliseconds) between starting next fetcher thread. Avoids overloading DNS or network resources during fetcher startup when all threads simultaneously start requesting pages.

fetcher.timeout.queue

-1

Maximum wait time (seconds) for items in the queue. -1 disables timeout.

fetcherbolt.queue.debug.filepath

""

Path to a debug log (e.g. /tmp/fetcher-dump-{port}).

http.agent.description

-

Description for the User-Agent header.

http.agent.email

-

Email address in User-Agent header.

http.agent.name

-

Name in User-Agent header.

http.agent.url

-

URL in User-Agent header.

http.agent.version

-

Version in User-Agent header.

http.basicauth.password

-

Password for http.basicauth.user.

http.basicauth.user

-

Username for Basic Authentication.

http.content.limit

-1

Maximum HTTP response body size (bytes). Default: no limit.

http.protocol.implementation

org.apache.stormcrawler.protocol.httpclient.HttpProtocol

HTTP Protocol implementation.

http.proxy.host

-

HTTP proxy host.

http.proxy.pass

-

Proxy password.

http.proxy.port

8080

Proxy port.

http.proxy.user

-

Proxy username.

http.retry.on.connection.failure

true

(OkHttp only) Retry fetching on connection failure. See OkHttp docs.

http.robots.403.allow

true

Allow crawling when robots.txt returns HTTP 403.

http.robots.5xx.allow

false

Allow crawling when robots.txt returns a server error (5xx).

http.robots.agents

''

Additional user-agent strings for interpreting robots.txt.

http.robots.content.limit

-1

Maximum bytes to fetch for robots.txt. -1 uses http.content.limit.

http.robots.file.skip

false

Ignore robots.txt rules entirely.

http.robots.headers.skip

false

Ignore robots directives from HTTP headers.

http.robots.meta.skip

false

Ignore robots directives from HTML meta tags.

http.skip.robots

false

Deprecated (replaced by http.robots.file.skip).

robots.noFollow.strict

true

If true, remove all outlinks from pages marked as noFollow.

http.store.headers

false

Whether to store response headers.

http.timeout

10000

Connection timeout (ms).

http.use.cookies

false

Use cookies in subsequent requests.

https.protocol.implementation

org.apache.stormcrawler.protocol.httpclient.HttpProtocol

HTTPS Protocol implementation.

partition.url.mode

byHost

Defines how URLs are partitioned: byHost, byDomain, or byIP.

protocols

http,https,file

Supported protocols.

redirections.allowed

true

If true, emit redirect target URLs as "outlinks" to the status stream. If false, do not follow redirects. See also http.allow.redirects.

http.allow.redirects

false

(OkHttp only) Follow HTTP redirects immediately in the HTTP protocol client. Note: if followed immediately, redirect target URLs are not emitted to the status stream, are not filtered, not deduplicated, and not checked against robots.txt.

sitemap.discovery

false

Enable automatic sitemap discovery.

urlbuffer.class

org.apache.stormcrawler.persistence.urlbuffer.SimpleURLBuffer

URL buffer implementation used by spouts.

Protocol

key default value description

file.protocol.implementation

org.apache.stormcrawler.protocol.file.FileProtocol

Protocol implementation for file:// URLs.

file.encoding

UTF-8

Encoding for FileProtocol.

protocol.instances.num

1

Number of instances per protocol implementation.

http.protocol.versions

-

HTTP protocol versions in order of preference (h2, http/1.1, http/1.0, h2c). If empty, uses implementation defaults.

robots.cache.spec

maximumSize=10000,expireAfterWrite=6h

CacheBuilder configuration for robots cache.

robots.error.cache.spec

maximumSize=10000,expireAfterWrite=1h

CacheBuilder configuration for error cache.

okhttp.protocol.connection.pool.max.idle.connections

5

OkHttp maximum number of idle connections.

okhttp.protocol.connection.pool.connection.keep.alive

300

OkHttp connection keep-alive time (seconds).

http.custom.headers

-

Custom HTTP headers.

http.accept

text/html,application/xhtml+xml,application/xml;q=0.9,/;q=0.8

HTTP Accept header.

http.accept.language

en-us,en-gb,en;q=0.7,*;q=0.3

HTTP Accept-Language header.

http.content.partial.as.trimmed

false

Accepts partially fetched content in OKHTTP.

http.trust.everything

true

If true, trust all SSL/TLS connections.

navigationfilters.config.file

-

JSON config for NavigationFilter (used by the Selenium protocol module).

selenium.addresses

-

WebDriver server addresses.

selenium.capabilities

-

Desired WebDriver capabilities .

selenium.delegated.protocol

-

Delegated protocol for selective Selenium usage.

selenium.implicitlyWait

0

WebDriver element search timeout.

selenium.instances.num

1

Number of instances per WebDriver connection.

selenium.pageLoadTimeout

0

WebDriver page load timeout.

selenium.setScriptTimeout

0

WebDriver script execution timeout.

topology.message.timeout.secs

-1

OKHTTP message timeout.

Indexing

The values below are used by sub-classes of AbstractIndexerBolt.

key default value description

indexer.canonical.name

canonical

Metadata key for the canonical URL. Used to replace the URL with its canonical form before indexing.

indexer.ignore.empty.fields

false

If true, skip fields with empty values when indexing.

indexer.md.filter

-

YAML list of key=value filters for metadata-based indexing.

indexer.md.mapping

-

YAML mapping from metadata fields to persistence layer fields.

indexer.text.fieldname

content

Field name for indexed HTML body text.

indexer.text.maxlength

-1

Maximum length of text to index. -1 means no limit.

indexer.url.fieldname

url

Field name for indexed URL.

Status Persistence

This refers to persisting the status of a URL (e.g. ERROR, DISCOVERED etc.) along with something like a nextFetchDate that is being calculated by a Scheduler.

key default value description

fetchInterval.default

1440

Default revisit interval (minutes). Used by DefaultScheduler.

fetchInterval.error

-1

Revisit interval for error pages (minutes). -1 means never refetch.

fetchInterval.fetch.error

120

Revisit interval for fetch errors (minutes).

max.fetch.errors

3

Maximum number of successive fetch errors before changing status to ERROR.

scheduler.class

org.apache.stormcrawler.persistence.DefaultScheduler

Scheduler implementation for computing next fetch dates. Use AdaptiveScheduler for change-rate-based intervals.

status.updater.cache.spec

maximumSize=10000, expireAfterAccess=1h

Cache specification for the status updater.

status.updater.unit.round.date

SECOND

Unit for rounding the next fetch date. Can also be MINUTE or HOUR.

status.updater.use.cache

true

Whether to use cache to avoid re-persisting URLs.

Parsing

Configures parsing of fetched text and the handling of discovered URIs

key default value description

collections.file

collections.json

Config for CollectionTagger.

collections.key

collections

Key under which tags are stored in metadata.

detect.charset.maxlength

10000

Maximum number of bytes used for charset detection.

detect.mimetype

true

Enable MIME type detection during parsing.

feed.filter.hours.since.published

-1

Discard feeds older than value hours.

feed.sniffContent

false

Try to detect feeds automatically.

jsoup.treat.non.html.as.error

true

If true, non-HTML content is treated as an error by JSoupParserBolt.

parsefilters.config.file

parsefilters.json

Path to JSON config defining ParseFilters. See default parsefilters.json.

parser.emitOutlinks

true

Emit discovered links as DISCOVERED tuples.

parser.emitOutlinks.max.per.page

-1

Limit number of emitted links per page.

sitemap.filter.hours.since.modified

-1

Filter URLs in sitemaps based on their modification date. -1 disables filtering.

sitemap.schedule.delay

-1

Staggered scheduling delay for sitemaps (minutes). -1 disables staggering.

sitemap.extensions

-

Sitemap extensions to parse (IMAGE, LINKS, MOBILE, NEWS, VIDEO).

textextractor.exclude.tags

""

HTML tags ignored by TextExtractor.

textextractor.include.pattern

""

Regex patterns to include for TextExtractor.

textextractor.no.text

false

Disable text extraction entirely.

textextractor.skip.after

-1

Stop text extraction after this many characters. -1 means no limit.

track.anchors

true

Add anchor text to outlink metadata.

urlfilters.config.file

urlfilters.json

JSON file defining URL filters. See default urlfilters.json.

Metadata

Options on how Storm Crawler should handle metadata tracking as well as minimising metadata clashes

key default value description

metadata.persist

-

Metadata to persist but not transfer to outlinks.

metadata.track.depth

true

Track crawl depth of URLs.

metadata.track.path

true

Track URL path history in metadata.

metadata.transfer

-

Metadata to transfer to outlinks.

metadata.transfer.class

org.apache.stormcrawler.util.MetadataTransfer

Class handling metadata transfer.

protocol.md.prefix

protocol.

Prefix for remote metadata keys to avoid collisions.

External Module Configuration

The following sections document configuration options for the external integration modules. Each module has its own Maven dependency and may require additional infrastructure (e.g., OpenSearch, Solr, a SQL database). For full setup instructions, see the README in each module’s directory.

OpenSearch

Integration with OpenSearch for indexing, URL status persistence, and metrics. See the opensearch module for full details.

key default value description

opensearch.addresses

-

OpenSearch server address(es).

opensearch.user

-

Username for authentication (optional).

opensearch.password

-

Password for authentication (optional).

opensearch.concurrentRequests

2

Number of concurrent bulk requests.

opensearch.indexer.index.name

content

Index name for crawled documents.

opensearch.indexer.create

false

Auto-create index if it does not exist.

opensearch.indexer.bulkActions

100

Number of documents to buffer before flushing.

opensearch.indexer.flushInterval

2s

Maximum time before flushing the bulk buffer.

opensearch.indexer.pipeline

-

Ingest pipeline name (optional).

opensearch.status.index.name

status

Index name for URL status.

opensearch.status.bulkActions

500

Batch size for status updates.

opensearch.status.flushInterval

5s

Flush interval for status updates.

opensearch.status.routing

true

Enable index routing for status documents.

opensearch.status.routing.fieldname

key

Field used for routing.

opensearch.status.max.buckets

50

Number of buckets for AggregationSpout.

opensearch.status.max.urls.per.bucket

2

URLs returned per bucket.

opensearch.status.bucket.field

key

Field used for bucketing.

opensearch.status.sample

false

Use random sampling in AggregationSpout.

opensearch.metrics.index.name

metrics

Index name for Storm metrics.

Solr

Integration with Apache Solr for indexing, URL status persistence, and metrics. See the solr module for full details. Supports both standalone Solr and SolrCloud (via ZooKeeper).

key default value description

solr.indexer.url

-

Solr collection URL for indexing.

solr.indexer.zkhost

-

ZooKeeper host for SolrCloud indexing (alternative to URL).

solr.indexer.collection

-

SolrCloud collection name for indexing.

solr.status.url

-

Solr collection URL for status storage.

solr.status.zkhost

-

ZooKeeper host for SolrCloud status (alternative to URL).

solr.status.collection

-

SolrCloud collection name for status.

solr.status.bucket.field

host

Field used for bucketing in SolrSpout.

solr.status.bucket.maxsize

5

Maximum URLs per bucket.

solr.status.max.results

10

Maximum results per spout query.

solr.status.metadata.prefix

metadata

Prefix for metadata fields in Solr.

solr.metrics.url

-

Solr collection URL for metrics.

SQL

Integration with relational databases via JDBC for URL status persistence, indexing, and metrics. See the sql module for full details and table creation scripts.

key default value description

sql.connection.url

-

JDBC connection URL.

sql.connection.user

-

Database username.

sql.connection.password

-

Database password.

sql.status.table

urls

Table name for URL status storage.

sql.max.urls.per.bucket

5

Maximum URLs per bucket in SQLSpout.

sql.spout.max.results

100

Maximum results per spout query.

sql.metrics.table

metrics

Table name for metrics storage.

sql.index.table

content

Table name for indexed content.

URLFrontier

Integration with URLFrontier, a language-agnostic API for URL management and scheduling. See the urlfrontier module for full details.

key default value description

urlfrontier.host

localhost

URLFrontier service hostname.

urlfrontier.port

7071

URLFrontier service port.

urlfrontier.max.buckets

10

Number of buckets to request from the frontier.

urlfrontier.max.urls.per.bucket

10

Maximum URLs per bucket.

Tika

Apache Tika integration for parsing non-HTML documents (PDF, Word, etc.). See the tika module for full details.

key default value description

parser.mimetype.whitelist

-

Regex list of allowed MIME types. If set, only matching types are parsed by the Tika ParserBolt.

parser.tika.config.file

tika-config.xml

Path to the Tika configuration XML file.
When using the Tika ParserBolt alongside JSoupParserBolt, set jsoup.treat.non.html.as.error to false so that non-HTML content is passed through to the Tika parser rather than being treated as an error.

AWS

Integration with AWS services: CloudSearch for indexing and S3 for content caching. See the aws module for full details.

key default value description

cloudsearch.endpoint

-

AWS CloudSearch document endpoint URL.

cloudsearch.region

-

AWS region (e.g., "eu-west-1").

cloudsearch.batch.maxSize

-1

Documents to buffer before sending a batch. -1 disables batching.

cloudsearch.batch.max.time.buffered

10

Maximum time (seconds) before flushing the buffer.

cloudsearch.batch.dump

false

Dump batch JSON to a temp directory for debugging.

s3.region

-

AWS region for S3 operations.

s3.bucket

-

S3 bucket name for content caching.

s3.endpoint

-

Custom S3 endpoint (optional, for S3-compatible services).

AI (LLM Text Extraction)

LLM-based text extraction using OpenAI-compatible APIs (including local models via Ollama). See the ai module for full details.

key default value description

textextractor.class

-

Set to org.apache.stormcrawler.ai.OpenAITextExtractor to enable LLM extraction.

textextractor.llm.api_key

-

API key for the LLM service.

textextractor.llm.url

-

LLM API endpoint URL (e.g., OpenAI or Ollama endpoint).

textextractor.llm.model

-

Model name to use (e.g., "gpt-4", "llama2").

textextractor.system.prompt

-

System prompt for the LLM (optional).

textextractor.llm.prompt

-

User prompt template. Use {HTML} and {REQUEST} as placeholders (optional).

textextractor.llm.user_request

-

Extra user request passed to the prompt template (optional).

textextractor.llm.listener.clazz

-

Listener class for tracking LLM response metrics (optional).

Playwright

Browser-based fetching using Playwright for JavaScript-rendered pages. See the playwright module for full details.

key default value description

playwright.cdp.url

-

Chrome DevTools Protocol URL for connecting to an existing browser instance (e.g. http://localhost:9222). Mutually exclusive with playwright.remote.ws.

playwright.remote.ws

-

Remote WebSocket URL for Playwright (alternative to CDP, e.g. ws://localhost:3000/).

playwright.skip.download

false

Skip automatic browser download. Implicitly forced to true when playwright.cdp.url or playwright.remote.ws is set.

playwright.load.event

load

Page load event to wait for. One of load, domcontentloaded, networkidle.

playwright.skip.resource.types

-

List of resource types aborted during navigation (document, stylesheet, image, media, font, script, texttrack, xhr, fetch, eventsource, websocket, manifest, other).

playwright.evaluations

-

List of JavaScript expressions evaluated after load; each JSON-serialised result is stored in response metadata under the expression itself.

playwright.capture.content.on.error

false

If true, also capture page.content() for non-2xx responses — useful for SPAs that return a stub then hydrate via JS.

playwright.override.status.on.content

false

When content was captured for a non-2xx response, override the reported HTTP status with 200. The original status is preserved under the playwright.origin.status response metadata key. No-op unless playwright.capture.content.on.error is also true.

playwright.page.actions.config.file

-

JSON file declaring an ordered chain of PageAction implementations applied after navigate succeeds and before content capture. See Page actions below.
Page actions

The Playwright protocol exposes a PageAction extension point — an ordered chain of post-navigate DOM transformations loaded from a JSON file referenced by playwright.page.actions.config.file. Use this to plug site-specific behaviour (tab/accordion expansion, cookie-banner dismissal, infinite-scroll, custom evaluate() calls, screenshotting, …​) into the protocol without subclassing it. The chain runs only when content would otherwise be captured (on 2xx, or on non-2xx when playwright.capture.content.on.error is true). Per-action failures are logged and swallowed so one bad action cannot abort the rest of the chain.

{
  "org.apache.stormcrawler.protocol.playwright.PageActions": [
    {
      "class": "org.apache.stormcrawler.protocol.playwright.actions.DismissOverlayAction",
      "name": "cookies",
      "params": { "selectors": ["#cookie-accept"] }
    },
    {
      "class": "org.apache.stormcrawler.protocol.playwright.actions.ExpandClickablesAction",
      "name": "tabs",
      "params": {
        "selectors": [".tab-widget .tab-header"],
        "root": ".tab-widget",
        "body": ".tab-widget-body",
        "waitMs": 300
      }
    }
  ]
}

Built-in actions:

Class Purpose

ExpandClickablesAction

Clicks every element matching the configured selectors and clones the resulting body container into a hidden cache under the same widget root, so page.content() ends up containing the HTML of every tab/accordion panel rather than only the active one.

EvaluateAction

Evaluates a list of JavaScript expressions and stores each JSON-serialised result in response metadata.

ScrollToBottomAction

Repeatedly scrolls to the bottom of the page until the document height stops growing, the step cap is reached, or the time budget elapses — useful for infinite-scroll feeds.

WaitForSelectorAction

Waits for a selector to reach an attached / detached / visible / hidden state. Soft-fails on timeout by default; set required: true to fail.

DismissOverlayAction

Dismisses cookie banners, GDPR walls, newsletter modals, etc. by clicking the first match of each selector, and optionally removes sticky overlays from the DOM via removeSelectors.

ScreenshotAction

Captures a screenshot of the page and stores it base64-encoded in response metadata. For diagnostics / small-volume use; larger crawls should write to a blob store.

See the playwright module README for the full parameter list of each built-in action and a guide on writing your own.

JS rendering detection

Browser fetching is much more expensive than a plain HTTP fetch, so most operators only want Playwright on URLs that actually need it. The JsRenderingDetector parse filter inspects the parsed page from a cheap fetch and sets a metadata flag (default fetch.with=playwright) on URLs that look JavaScript-rendered. Pair it with DelegatorProtocol to route subsequent fetches of those URLs to the Playwright protocol while leaving everything else on a fast HTTP client.

Detection signals (cheapest first, short-circuiting):

  • SPA framework fingerprints in raw HTML — data-reactroot, ng-version=, NEXT_DATA, window.NUXT, data-svelte-h=, data-vue-app, data-astro-cid, <router-outlet.

  • <noscript> blocks containing language like "enable JavaScript".

  • Empty SPA hydration roots: <div id="root"></div> / app / next / #nuxt.

  • Outcome-based fallback: at least one <script> is present and both text.length and the outlink count are below configurable thresholds.

Detection is skipped when playwright.protocol.end is already on the URL (i.e. it was just fetched by Playwright) or when the routing key is already set, so the filter is idempotent.

Register the filter in parsefilters.json:

{
  "class": "org.apache.stormcrawler.protocol.playwright.parsefilter.JsRenderingDetector",
  "name": "js-rendering-detector",
  "params": { "minTextLength": 200, "minOutlinks": 2 }
}

And route on the metadata key it sets:

http.protocol.implementation:  "org.apache.stormcrawler.protocol.DelegatorProtocol"
https.protocol.implementation: "org.apache.stormcrawler.protocol.DelegatorProtocol"
protocol.delegator.config:
  - className: "org.apache.stormcrawler.protocol.playwright.HttpProtocol"
    filters:
      "fetch.with": "playwright"
  - className: "org.apache.stormcrawler.protocol.okhttp.HttpProtocol"

The dotted metadata key is quoted in the YAML above for readability; SnakeYAML accepts the unquoted form too. Note that DelegatorProtocol requires the last entry in protocol.delegator.config to have no filters: — it acts as the fallback, so keep the cheap protocol at the bottom of the list.

The parse filter alone does not trigger an immediate refetch — it only sets the metadata flag on the current fetch, and DefaultScheduler reschedules the URL according to the FETCHED interval (fetchInterval.default, 24h by default). For faster turnaround, either add a per-metadata-key fetch interval (fetchInterval.fetch.with=playwright: 5) or drop JsRenderingRedirectionBolt between the parser and indexer. The bolt reads the routing flag and, on hit, emits only to the status stream with Status.FETCHED so the stub document never reaches the index. The full parameter list and tuning notes are in the playwright module README.

Language ID

Language identification for crawled documents using the lang-detect library. See the langid module source code for details.

The LanguageID parse filter is configured in parsefilters.json:

{
  "class": "org.apache.stormcrawler.parse.filter.LanguageID",
  "name": "LanguageID",
  "params": {
    "key": "lang",
    "minProb": 0.99,
    "extracted": "parse.lang"
  }
}

  • key — metadata key to store the detected language code (default: "lang")

  • minProb — minimum probability threshold for detection (default: 0.999)

  • extracted — metadata key to check for a pre-extracted language value (default: "parse.lang")

WARC

Reading and writing WARC (Web ARChive) files for archival and replay. See the warc module for full details.

key default value description

warc.metadata.keys

-

Metadata keys to include as WARC metadata records (optional list).
For complete WARC records, set http.store.headers to true. The OkHttp protocol (org.apache.stormcrawler.protocol.okhttp.HttpProtocol) is recommended for WARC generation as it provides verbatim HTTP headers.

Common Spout Configuration

The following options are shared across storage-backed spouts (OpenSearch, Solr, SQL) that extend AbstractQueryingSpout:

key default value description

spout.ttl.purgatory

30

Time (seconds) a URL remains in purgatory after ack/fail before it can be re-emitted.

spout.min.delay.queries

2000

Minimum delay (ms) between backend queries.

spout.max.delay.queries

20000

Maximum delay (ms) between backend queries (OpenSearch only).

spout.reset.fetchdate.after

120

Reset a URL’s fetch date if it has not been acked within this many seconds.