#!/usr/bin/env python3
""" 
Builds a user level dataset. Requires a functional spark installation.

"""

import sys
# add pyspark to your python path e.g.
#sys.path.append("/home/nathante/sparkstuff/spark/python/pyspark")
#sys.path.append("/home/nathante/sparkstuff/spark/python/")
from pyspark import SparkConf
from pyspark.sql import SparkSession, SQLContext
from pyspark.sql import Window
import pyspark.sql.functions as f
from pyspark.sql import types
import argparse
import glob
from os import mkdir
from os import path

#read a table

def parse_args():

    parser = argparse.ArgumentParser(description='Create a dataset of edits by user.')
    parser.add_argument('-i', '--input-file', help='Tsv file of wiki edits. Supports wildcards ', required=True, type=str)
    parser.add_argument('-o', '--output-dir', help='Output directory', default='./output', type=str)
#    parser.add_argument('--wiki', help="Wiki name. If not provided, we will guess based on the filename.", type=str)
#    parser.add_argument('--urlencode', help="whether we need to decode urls",action="store_true")
    parser.add_argument('--output-format', help = "[csv, parquet] format to output",type=str)
    parser.add_argument('--num-partitions', help = "number of partitions to output",type=int, default=1)
#    parser.add_argument('--ignore-input-errors', help = "ignore bad lines in input",action="store_true")
#    parser.add_argument('--nodes', help = "how many hyak nodes to use", default=0, type=int)
    args = parser.parse_args()
    return(args)

if __name__ == "__main__":
    args = parse_args()
    conf = SparkConf().setAppName("Wiki Users Spark")
    spark = SparkSession.builder.getOrCreate()
    files = glob.glob(args.input_file)
    files = [path.abspath(p) for p in files]
    reader = spark.read


    # going to have to do some coercing of the schema

    # build a schema
    struct = types.StructType().add("anon",types.StringType(),True)
    struct = struct.add("articleid",types.LongType(),True)
    struct = struct.add("date_time",types.TimestampType(), True)
    struct = struct.add("deleted",types.BooleanType(), True)
    struct = struct.add("editor",types.StringType(),True)
    struct = struct.add("editor_id",types.LongType(), True)
    struct = struct.add("minor", types.BooleanType(), True)
    struct = struct.add("namespace", types.LongType(), True)
    struct = struct.add("revert", types.BooleanType(), True)
    struct = struct.add("reverteds", types.StringType(), True)
    struct = struct.add("revid", types.LongType(), True)
    struct = struct.add("sha1", types.StringType(), True)
    struct = struct.add("text_chars", types.LongType(), True)
    struct = struct.add("title",types.StringType(), True)
    df = reader.csv(files,
                    sep='\t',
                    inferSchema=False,
                    header=True,
                    mode="PERMISSIVE",
                    schema = struct)
    df = df.repartition(args.num_partitions)
    # replace na editor ids
    df = df.select('*',f.coalesce(df['editor_id'],df['editor']).alias('editor_id_or_ip'))

    # assign which edit reverted what edit
    reverteds_df = df.filter(~ df.reverteds.isNull()).select(['revid','reverteds'])
    reverteds_df = reverteds_df.select("*", f.split(reverteds_df.reverteds,',').alias("reverteds_new"))
    reverteds_df = reverteds_df.drop("reverteds")
    reverteds_df = reverteds_df.withColumnRenamed("reverteds_new", "reverteds")
    reverteds_df = reverteds_df.select(reverteds_df.revid.alias('reverted_by'),
                                       f.explode(reverteds_df.reverteds).alias('reverted_id'))
    df = df.join(reverteds_df, df.revid == reverteds_df.reverted_id, how='left_outer')
    df.drop("reverted_id")    

    # sort by datetime 
    df = df.orderBy(df.date_time.asc())
    win = Window.orderBy('date_time').partitionBy('editor_id_or_ip')

    # count reverts
    reverts_df = df.filter(df.revert==True).select(['revid','editor_id_or_ip','date_time','revert'])
    reverts_df = reverts_df.withColumn('editor_nth_revert',f.rank().over(win))
    df = df.join(reverts_df, ["revid",'editor_id_or_ip','date_time','revert'], how='left_outer')
    del(reverts_df)

    # count edits
    df = df.withColumn('year', f.year(df.date_time))
    df = df.withColumn('month',f.month(df.date_time))
    df = df.withColumn('editor_nth_edit',f.rank().over(win))
    
    # output
    if not path.exists(args.output_dir):
        mkdir(args.output_dir)
    if args.output_format == "csv" or args.output_format == "tsv":
        df.write.csv(args.output_dir, sep='\t', mode='overwrite',header=True,timestampFormat="yyyy-MM-dd HH:mm:ss")
            # format == "parquet"
    else:
        df.write.parquet(args.output_dir, mode='overwrite')

        # for writing to csv we need to urlencode