我从 node.js 中获得了巨大的价值,并且喜欢流处理模型。我主要将它用于具有数据丰富和 ETL 之类的作业的流处理。
为了充实,我可能有这样的记录......
{ "ip":"123.45.789.01", "productId": 12345 }
我想通过添加产品详细信息来丰富这一点
{ "ip":"123.45.789.01", "productId": 12345, "description" : "Coca-Cola 12Pk", "price":4.00 }
描述数据和价格数据都来自不同的流。在高地处理这种依赖关系的最佳方法是什么?
H = require('highland')
descriptionStream = H(['[{"productId":1,"description":"Coca-Cola 12Pk"},{"productId":2,"description":"Coca-Cola 20oz Bottle"}]'])
.flatMap(JSON.parse)
priceStream = H(['[{"productId":1,"price":4.00},{"productId":2,"price":1.25}]'])
.flatMap(JSON.parse)
# the file is a 10G file with a json record on each line
activityStream = H(fs.createReadStream('8-11-all.json',{flags:'r',encoding:'utf8'}))
.splitBy("\n")
.take(100000) # just take 100k for testing
.filter((line)-> line.trim().length > 0) # to prevent barfing on empty lines
.doto((v)->
# here i want to add the decription from the descriptionStream
# and i want to add the price from the price stream.
# in order to do that, i need to make the execution of this
# stream dependent on the completion of the first two and
# availability of that data. this is easy with declarative
# programming but less intuitive with functional programming
)
.toArray((results)->
# dump my results here
)
有什么想法吗?