pyspark join with multiple conditions for different rows in in table dfs

I want to join two tables based on some conditions in pyspark

I have 2 dataframe (GR_df and HK_df) for table A and table B
there are three columns which are common in both the dataframes.

GR_df.columnx = HK_df.columnh,

GR_df.columny = HK_df.columni,

GR_df.columnz = HK_df.columnj

now, join will be perform based on the join conditions mentioned below:

joincond1 = (GR_df.columnx = HK_df.columnh & GR_df.columny = HK_df.columni & GR_df.columnz = HK_df.columnj),

joincond2 = (GR_df.columnx = HK_df.columnh & GR_df.columny = HK_df.columni),

joincond3 = (GR_df.columnx = HK_df.columnh & GR_df.columnz = HK_df.columnj),

joincond4 = (GR_df.columnx = HK_df.columnh),

If there is only one matching row for joincond1 for each value of columns in GR_df and HK_df
then write that to final df

If there is multiple rows return by joincond1 for each value of columns in GR_df and HK_df
then write distinct records to final df

If there is no matching row for a value in GR_df and Hk_df need to change the condition to joincond2 and so on till condition 4

If a record doesn't satisfy any condition that record will be skipped.

I know I need to iterate over dataframe rows but I am not sure How I can write a logic in pyspark or spark-sql for this problem can anyone help me.

GR_df-

columnx	columny	columnz	Other columns

 1	      a	      aa	  -
 2	      b	      bb	  -
 3	      c	      cc	  -
 4	      d	      dd	  -
 5	      l	      uu	  -

HK_df -

columnh	 columni columnj Other columns

 1	       a	   aa	      -
 2	       b	   zz	      -
 3	       m	   cc	      -
 4	       i	   jj	      -

final df -

columnh  columni  columnj  columns_GR_df  columns_HK_df

   1	   a	    aa	       -	          -
   2	   b	    zz	       -	          -
   3	   m	    cc	       -	          -
   4	   i	    jj	       -	          -

I think you're looking for an inner join, with the conditions combined by the or operator. To drop the duplicates after, we use distinct() after the join:

output_df=GR_df.join(HK_df, joincond1 | joincond2 | joincond 3, "inner").distinct()

Depending on what you want to happen when both cond1 and cond2 are satisfied (exclusive or?), your join condition may become:

(joincond1 & ~joincond2 & ~joincond3) | (~joincond1 & joincond2 & ~joincond3) | (~joincond1 & ~joincond2 & joincond3)