Parse cloudfront log for non-streaming distribution and count bandwidth per object that matches /uploads/assets/file/...

cloudfront_not_streaming_log.pig

register file:/home/hadoop/lib/pig/piggybank.jar

raw_logs = 
  LOAD '$INPUT' 
  USING PigStorage('\t') 
  AS (
    date: chararray, time: chararray, x_edge_location: chararray, sc_bytes: int, 
    c_ip: chararray, cs_method: chararray, cs_host: chararray, cs_uri_stem: chararray,
    sc_status: chararray, cs_referer: chararray, cs_user_agent:chararray, cs_uri_query: chararray
  );
  
raw_logs_without_comments = FILTER raw_logs BY (date matches '^[^#].*');

all_logs = FOREACH raw_logs_without_comments GENERATE cs_uri_stem, sc_bytes;

logs = FILTER all_logs BY (cs_url_stem matches '^/uploads/asset/file/.*');

by_uri_stem_bytes = 
  FOREACH 
    (GROUP logs BY cs_uri_stem) 
  GENERATE 
    $0, 
    SUM($1.sc_bytes) AS num_bytes
  ;

STORE by_uri_stem_bytes INTO '$OUTPUT';