如何在pyspark中找到两个rdd的交叉点？

Question

我有两个rdds:

rdd1 = sc.parallelize([("www.page1.html", "word1"), ("www.page2.html", "word1"), 
    ("www.page1.html", "word3")])

rdd2 = sc.parallelize([("www.page1.html", 7.3), ("www.page2.html", 1.25), 
    ("www.page3.html", 5.41)])

intersection_rdd = rdd1.keys().intersection(rdd2.keys())       

//当我这样做的时候,我只得到了交叉的关键点(www.page1.html,www.page2.html).

但我需要键和两个rdds的值.输出应如下所示:

[www.page1.html, (word1, word3, 7.3)]

[www.page2.html, (word1, 1.25)]

Answer 1

你可以举例说明cogroup并过滤:

## This depends on empty resultiterable.ResultIterable
## evaluating to False

intersection_rdd = rdd1.cogroup(rdd2).filter(lambda x: x[1][0] and x[1][1])
intersection_rdd.map(lambda x: (x[0], (list(x[1][0]), list(x[1][1])))).collect()

## [('www.page1.html', (['word1', 'word3'], [7.3])),
##  ('www.page2.html', (['word1'], [1.25]))]