- The core compositions of Hadoop are HDFS, Yarn, and other engines and App, like Mapreduce, Tez, Nutch, Pig, Hive, Spark, etc.
- HDFS is composed of NameNode and DataNode for data storage.
- Yarn is composed of Resource Manager and node Manager for resource assignment.
- APPs like Pig, Hive are higher level language processor. They can conduct mapreduce job much easier.
- Use a customized Nutch to crawl apps metadata in GooglePlay
- Inject seed to nutchDB
- Generate urls to crawl from nutchDB
- Fetch app meatadata from html pages
- parse extracted metadata and outlinks
- update nutchDB with new outlinks
- Pig Loadfunc transforms nutchDB to readable text file form
- Create table and manage data by Hive
- git clone
git clone https://github.com/apache/nutch
git checkout release-1.12
- customize nutch
patch -p1 < /googleplaycrawler/googleplaycrawler.patch
- run googleplaycrawler on single nutch cluster
echo "https://play.google.com/store/apps/details?id=com.facebook.orca" > seed
hadoop fs -put seed
hadoop jar build/apache-nutch-1.12.job org.apache.nutch.googleplay.GooglePlayCrawler seed -numFetchers 10
- check output
hadoop fs -text file:///xxxxxx/nutchdb/segments/xxxxx/parse_data/part-00000/data
- fix the skew data job
patch -p1 < fixskew.patch
-
uplode seeds file and run web scrawler in AWS EMR
-
Aws S3
register target/nutchdbloader-0.0.1-SNAPSHOT.jar
register /home/hadoop/hadoop-2.7.3/share/hadoop/tools/lib/hadoop-aws-2.7.3.jar register nutch-1.12.jar
loaded = load 's3n://test/nutchdb/segments/*/parse_data/part-*/data' using com.example.NutchParsedDataLoader();
filtered = filter loaded by $0 is not null;
store filtered into 'output';
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent