Thursday 14 June 2018

What's new in StormCrawler 1.10

StormCrawler 1.9 is only a couple of weeks old but the new functionalities added since justify a new release.

Dependency upgrades

  • Apache Storm 1.2.2 (#583)
  • Crawler-Commons 0.10 (#580)
  • Elasticsearch 6.3.0 (#587)


  • parsefilters: added CommaSeparatedToMultivaluedMetadata to split parse.keywords
  • bugfix: java topology in archetype does not use FeedParserBolt, fixes #551
  • bugfix: archetype - move SC dependency to first place to avoid STORM-2428, fixes #559


  • IndexerBolt set pipeline via config (#584)
  • Wrapper for loading JSON-based ParseFilters from ES (#569) - see below
  • SimpleFetcherBolt to send URLs back to its own queue if time to wait above threshold (#582)
  • ParseFilter to tag a document based on pattern matching on its URL (#577)
  • New URL filter implementation based on JSON file and organised per hostname or domain #578

Let's have a closer look at some of the points above.

The CollectionTagger is a ParseFilter provides a similar functionality to what Collections are in Google Search Appliance, namely the ability to add a key value in the metadata based on the URL of a document matching one or more regular expressions. The rules are expressed in a JSON file and look like 

   "collections": [{
            "name": "stormcrawler",
            "includePatterns": [""]
            "name": "crawler",
            "includePatterns": [".+crawler.+", ".+nutch.+"],
            "excludePatterns": [".+baby.+", ".+spider.+"]

Please note that the format is different from what GSA does but it can achieve the same thing. 

So far, nothing revolutionary, the resource file gets loaded from the uber-jar, just like any other resource. However, what we introduced at the same time is the interface JSONResource, which CollectionTagger implements. This interface defines how implementations load a JSON file to build their resources.

Here comes the interesting bit. We added a new resource for Elasticsearch in #569 called JSONResourceWrapper. As the name suggests, this wraps any ParseFilter implementing JSONResource and delegates the filtering to it. What it also does, is that it allows loading the JSON resource from an Elasticsearch document instead of the uber-jar and reloads it periodically. This allows you to update a resource without having to recompile the uber-jar and restart the topology

The wrapper is configured in the usual way i.e via the parsefilter.json file, like so

 "class": "com.digitalpebble.stormcrawler.elasticsearch.parse.filter.JSONResourceWrapper",
     "name": "ESCollectionTagger",
     "params": {
         "refresh": "60",
         "delegate": {
             "class": "com.digitalpebble.stormcrawler.parse.filter.CollectionTagger",
             "params": {
                 "file": "collections.json"

The JSONResourceWrapper also needs to know where Elasticsearch lives. This is set via the usual configuration file:

  es.config.addresses: "localhost" "config"
  es.config.doc.type: "config"
  es.config.settings: "elasticsearch"

You can then push a modified version of the resources to Elasticsearch e.g. with CURL

curl -XPUT 'localhost:9200/config/config/collections.json?pretty' -H 'Content-Type: application/json' -d @collections.json

Another resource we introduced in this release is the FastURLFilter, which also implements JSONResource (but as there isn't a Wrapper for URLFilters yet, can't be loaded from ES). This is similar to the existing URL filter we have in that it allows to remove URLs based on regular expressions, however, it organises the rules per domain or hostname which makes it more efficient as a URL doesn't have to be checked against all the patterns, just the ones for its domain. There is even a scope based on metadata key/values, for instance, if some of your seeds were organised by collection, as well as a global scope which is tried for all URLs if nothing else matched.

The resource file looks like 

"scope": "GLOBAL",
"patterns": [
"DenyPathQuery \\.jpg"
"scope": "",
"patterns": [
"AllowPath /digitalpebble/",
"DenyPath .+"
"scope": "metadata:key=value",
"patterns": [
"DenyPath .+"

where the Query suffix indicates whether the pattern should be matched against the path + query element or just the path.

I hope you like this new release of StormCrawler and the new features it brings. I would like to thank all the users and contributors and particularly the Government of Northwest Territories in Canada who kindly donated some of the code of the CollectionTagger.

Happy Crawling!

Friday 25 May 2018

What's new in StormCrawler 1.9

Dependency upgrades
  • Crawl-delay in robots.txt should optionally not shrink the configured delay #549
  • Optimisation: faster extraction of META tags #553
  • CollectionMetric synchronized access to List #555
  • Configurable Robots Caches #557
  • JSOUPParserBolt: lazy DOM conversion #563
  • Purge internal queues of tuples which have already reached timeout #564
  • Added ParseFilter to convert single valued Metadata to multi-valued ones #571
  • Caching of redirected robots.txt may overwrite correct robots.txt rules, fixes #573
  • WARCBolt to handle incorrect URIs gracefully #560
  • WARCRecordFormat use ByteBuffer instead of ByteArrayOutputStream #561
  • Uses flux-core 1.2.1 #559
  • Added FeedParser to archetype topology #551
  • Added .kml and .wmv to url filters
  • MetricsConsumer handles recursive values #554
  • MetricsConsumer handles recursive values #554
  • ES Indexer and Deletion Bolts to get index name from constructor #572
  • Added option to LanguageID to skip if metadata already set #570
As usual, we advise all users to move to this version as it fixes several bugs. Thanks to all contributors and users. Happy crawling!

Friday 23 March 2018

Grafana StormCrawler metrics v4

The Grafana dashboard for StormCrawler is a good starting point for monitoring the behaviour of your StormCrawler topology. This is typically used with Elasticsearch as a storage backend for the metrics generated by Storm but should work with any other Storm-compatible backend like Grafite or CloudWatch. 

Some of the metrics are specific to the components from the Elasticsearch module (spout, status, indexer) but you can simply remove or modify them if you use e.g. SOLR (NOTE: there was a feature request in Grafana to add SOLR as a datasource but to my knowledge, this is not yet available).

The latest version (4) brings the following changes.

  • URLs waiting in queues 

The recent 1.8 release of StormCrawler added a new metrics for the FetcherBolt which allows tracking the amount of time URLs spend in the internal queues. This has been added to the "URLs waiting in queues" panel alongside the average population of the queues.

Average time spent in queues + average queues population

  • ES StatusUpdater
Instead of tracking the number of bulk requests sent in the last minute, we now have a panel showing the evolution over time. This information is for the ES StatusUpdaterBolt only.

ES status updater bulk requests
  • Acked in StatusBolt
This is a brand new panel which is not specific to Elasticsearch but operates on any component with 'status' for id and shows the number of tuples acked over time, broken down by source.  

Tuples acked by StatusUpdater
In the graph above, we can see a peak early in the crawl where most of the tuples acked came from the sitemap bolt. Please note that the values are stacked in this graph. Sitemap files are typically discovered early in a crawl and generate a large number of discovered URLs; this is not the case later on when most tuples come from the HTML parser.
  • Robots panel
We removed the robots panel as the number of HTTP requests to robots files is shown in the "Fetcher: pages fetched" panel anyway and after the initial few minutes of a crawl, the panel simply indicated that the robots files were mostly cached.
  • ES Indexed 
This is a new panel showing the number of documents indexed into Elasticsearch as well as the documents filtered out during the indexing.

Tuesday 20 March 2018

What's new in StormCrawler 1.8

I have just released StormCrawler 1.8. As usual, here is a summary of the main changes:

Dependency updates
  • Add option to send only N bytes of text to indexers #476
  • BasicURLNormalizer to optionally convert IDN host names to ASCII/Punycode #522
  • MemorySpout to generate tuples with DISCOVERED status #529
  • OKHttp configure type of proxy #530
  • http.content.limit inconsistent default to -1 #534
  • Track time spent in the FetcherBolt queues #535
  • Increase detect.charset.maxlength default value #537
  • FeedParserBolt: metadata added by parse filters not passed forward in topology #541
  • Use UTF-8 for input encoding of seeds (FileSpout) #542
  • Default URL filter: exclude localhost and private address spaces #543
  • URLStreamGrouping returns the taskIDs and not their index #547
  • Upgrade WARC module to 1.1.0 version of storm-hdfs, fixes #520
  • Schema for status index needs date type for nextFetchDate #544
  • SOLR indexer: use field type text for content field #545
  • AggregationSpout fails with default value of es.status.bucket.field == _routing #521
  • Move to Elasticsearch RESTAPi #539
We recommend all users to move to this version as it fixes several bugs (#541#547) and adds some great new features. In particular, the use of the REST API for Elasticsearch, which makes the module future-proof but also easier to configure, but also #535 and #543.

As usual, thanks to all contributors and users. Happy crawling!

Tuesday 28 November 2017

What's new in StormCrawler 1.7

Amazingly this is the 20th release of StormCrawler! Here are the main changes:

Dependencies updates
  • crawler-commons 0.9 #513
  • (bugfix) ParserBolts should use outlinks from parsefilters #498
  • LD_JSON parsefilter #501
  • okhttp : store request and response headers verbatim in metadata #506
  • (bugfix) okhttp protocol does not store headers in metadata #507
  • HTTP clients should handle http.accept.language and http.accept #499
  • Selenium protocol follows redirections #514
  • RemoteDriverProtocol needs multiple instances #505
  • SitemapParserBolt should force mime-type based on the clue #515
  • ES Spout : define filter query via config #502
  • Upgrade to ES 6.0 #517
We recommend all users to move to this version. If you wish to remain on an older version of Elasticsearch, you can simply keep your existing version of the stormcrawler elasticsearch module while upgrading stormcrawler core.

This version improves the processing of sitemaps, via #515 and the use of the crawler-commons 0.9 where we fixed the SAX parsing and extended its coverage. We also added improvements to our okhttp-based protocol implementation. If your crawl is a wide one with potentially any sort of content then you should go for okhttp over the default httpclient one. See our comparison of protocol implementations on the WIKI.

Finally, if you want to extract semantic data represented in ld-json then you'll love #501.

As usual, thanks to all contributors and users. Happy crawling!

Friday 8 September 2017

What's new in StormCrawler 1.6

Dependencies updates

  • jsoup 1.10.3
  • crawler-commons 0.8


  • Use ISO representation of time for modifiedtime in adaptivescheduler  #496
  • Use ISO representation of time for discoveryDate and lastProcessedDate, #477
  • Improved Charset Detection #495
  • SitemapParserBolt configure use SAX or not
  • SitemapParserBolt generates metrics for average processing time
  • HTTP protocol based on OKHTTP #484 
  • Apache Http client can use HEAD method on a per URL basis #485
  • ContentFilter to leave trace of the pattern that matched #480
  • Metadata has a new public method for getting first non-empty value from a set of keys
  • Added ARTICLE to patterns for content filter


  • Can add more than one lang code based on configurable prob threshold. #481


  •  Added rotation policy based on time and filesize


  • ES: added es.status.reset.fetchdate.after #478
  • Removed Grafana resources - can be downloaded from Grafana portal

Monday 29 May 2017

What's new in StormCrawler 1.5

StormCrawler 1.5 has just been released! It is an important road mark with the move to Elasticsearch 5.x and the implementation of long-awaited features such as the Selenium-based protocol. The code has been improved in many ways and despite the seemingly low number of lines below, this new release is a mammoth one!

The project, in general, is in very good health, with more and more organisations using it in production, and an increased visibility, reflected by the growing number of questions on StackOverflow.

Here are the main changes in 1.5.

  • Apache Storm 1.1.0 (#450)

  • HTTP Protocol: implement cookie handling (#32)
  • Not in GZIP format thrown on redirs with httpclient (#455)
  • Selenium-based protocol implementation (#144) which I described in a separate blog post
  • Indicate whether RobotsRules come from cache or have been fetched (#460)
  • Memory issues when ByteArrayBuffer gets instantiated with a large value despite maxLength being set (#462)
  • FetcherBolt to dump URLs being fetched to log (#464)
  • Override sitemapsAutoDiscovery settings per URL (#469)

Knowing whether RobotsRules come from the cache gives us more insights into the behaviour of the crawlers as we can display the ratio of cache vs live (see illustration below)

as well as pages fetched vs robots fetched.


  • Utility class to export URL and metadata from ES index to file (#444)
  • Fixed sampling with aggregation spout in ES5
  • Upgrade to Elasticsearch 5.3 (#221 and  #451)
  • Optimise nextFetchDate to speed up queries to Elasticsearch (#429 and  #452)
  • Delete gone pages from index (#253)
  • metrics - remove filtering (#281)

One of the main changes related to Elasticsearch is the removal of ElasticsearchSpout and the introduction of CollapsingSpout, which uses the brand new FieldCollapsing in Elasticsearch. We also fixed a concurrency issue in the StatusUpdaterBolt (9fefac8), improved the efficiency of the spouts by getting them to process results in a separate thread (1b0fb42), which combined with the optimisation of nextFetchDate (see above) and the fix of the sampling in AggregationSpout, means that the Elasticsearch module is more efficient than ever.

The move to Elasticsearch 5.x was not without difficulties but the result justifies the effort. I described in a separate post the common pitfalls of upgrading an existing topology to Elasticsearch 5.

Coming next?
As usual, it is hard to guess what the next release will be made of as the project is driven by its community.

Having said that, I'd expect the Selenium-based protocol to get improved as users start to use it. It is also likely that we'll move away from Apache HttpClient library (#443). As mentioned in the previous release, we'll probably upgrade to the next release of crawler-commons, which will have a brand new SAX-based Sitemap parser.

In the meantime and as usual, thanks to all contributors and users and happy crawling!