加入 Pyspark 数据帧，其中两个列表共享一个值

Question

I have two dataframes of the form

df1 = 
+------+---------+
|group1|  members|
+------+---------+
|     1|[a, b, c]|
|     2|[d, e, f]|
|     3|[g, h, i]|
+------+---------+

df2 = 
+------+---------+
|group2|  members|
+------+---------+
|     4|[s, t, d]|
|     5|[u, v, w]|
|     6|[x, y, b]|
+------+---------+

I would like to perform a join on these dataframes based on a condition when the members lists share a common value. For example, group2 would map onto df1 as:

+------+---------+------+
|group1|  members|group2|
+------+---------+------+
|     1|[a, b, c]|     6|
|     2|[d, e, f]|     4|
|     3|[g, h, i]|      |
+------+---------+------+

Is there an efficient method for this? At the moment I am just looping through the rows of df2 and using f.array_intersect() to compare.

Answer 1

You can use a left join, the join condition is to use the size function to determine that the intersection of df1 and df2 is greater than 0.

df2 = df2.toDF('group2', 'members2')
df = df1.join(df2, F.size(F.array_intersect(df1.members, df2.members2)) > 0, 'left').drop('members2')
df.show(truncate=False)