(기여도 60%)

다중메타 check

YK_set_multi_process_20211005

rm(list=ls())
install.packages("dplyr")
library(dplyr)
input<-read.csv('C:\\Users\\이해리\\Desktop\\multi\\bosomi_01.28.csv', sep=",")
input
input_ref <-read.table('C:\\Users\\이해리\\Desktop\\multi\\bosomi_ref_01.28.csv',sep = ",", header=TRUE)
input_ref

seq 새로 생성 및 input에서 제외

input_seq <- input %>% 
select("seq","pid","hashcode")
input <- input[,c(2:10)] : seq 빼고 나머지칼럼으로 재추출

pid, pid_title 제외 후 중복제거

library(dplyr)
input_unique_hash <- input[,c(-1,-2)] %>%
distinct(hashcode, title, option_name, .keep_all = TRUE)​
input_trans <- data.frame(matrix(nrow=0, ncol=0))

column으로 되어있는 option을 row로 변환

for (loop_col in 4:6) {
input_col <- input_unique_hash[,c(1:3,loop_col)]
names(input_col)[4] <- "option_nm"
input_trans <- rbind(input_trans, input_col)
}

reference와 merge하여 pid_title을 포함한 meta값 부여

input_attr_merged <- merge(input_trans, input_ref, by=c('option_nm'), all.x = TRUE)
#str(input_attr_merged)
input_attr_merged[] <- lapply(input_attr_merged, function(x) if (is.factor(x)) as.character(x) else {x})
input_attr_merged <- input_attr_merged[input_attr_merged$pid_title_ref != "증정",]

동종 개수(option_nm 이 같은 것이 몇 개 있는지)

cntofopt_nm 은 hashcode, pid_title, title, option_name, option_nm 를 unique 하게 만들어 option_nm count

library(dplyr)
cntofopt_nm <- input_attr_merged %>%
#cntofopt_nm <- input_attr_merged[input_attr_merged$mark == 99,] %>%
group_by(hashcode, pid_title_ref, title, option_name, option_nm) %>%
summarise(cnt_opt_nm=n())

multi_meta

option_nm 을 제외 후 pid_title_ref count

library(dplyr)
cntofmulti_meta <- cntofopt_nm %>%
group_by(hashcode, pid_title_ref, title, option_name) %>%
summarise(cnt_multi_meta=n())

set

option_nm 을 제외하여 unique한 pid_title_ref count

cntofset <- cntofmulti_meta %>%
group_by(hashcode, title, option_name) %>%
summarise(cnt_set=n())

input_attr <- input_attr_merged %>%
distinct(hashcode, pid_title_ref, title, option_name, option_nm, .keep_all = TRUE)

해당 hashcode 에 meta, 동종개수, multimeta, set 값 merge

input_attr <- merge(input_attr, cntofopt_nm, by = c('hashcode', 'pid_title_ref', 'title', 'option_name', 'option_nm'), all.x = TRUE)
input_attr <- merge(input_attr, cntofset, by = c('hashcode', 'title', 'option_name'), all.x = TRUE)
input_attr <- merge(input_attr, cntofmulti_meta, by = c('hashcode', 'pid_title_ref', 'title', 'option_name'), all.x = TRUE)
str(input_attr)

동종인 것은 cs_group_count를 개수만큼 곱해 줌

input_attr$cs_group_count_ref <- as.integer(input_attr$cs_group_count_ref)
input_attr$cs_group_count_same <- input_attr$cs_group_count_ref * input_attr$cnt_opt_nm
> input_attr <- input_attr[order(input_attr$hashcode, input_attr$pid_title_ref, input_attr$title, input_attr$option_name, input_attr$option_nm),]
str(input_attr)
nrow(input_attr)

동일한 hashcode, pid_title_ref 안에서 numbering

input_attr$num_pid <- 1

for (loop_nrow in 2:nrow(input_attr)) {

  input_attr$num_pid[loop_nrow] <- ifelse((input_attr$hashcode[loop_nrow] == input_attr$hashcode[loop_nrow-1] 

                                           & input_attr$pid_title_ref[loop_nrow] == input_attr$pid_title_ref[loop_nrow-1]), input_attr$num_pid[loop_nrow-1] + 1, 1)

}

# cnt_pid <- 추출할 row건수 판별
# input_attr 은 정확한 것으로 모두 확정값 (mark = 99)
# set_gift, set_multi_meta <- column 이름 변경 및 정의
#input_attr$cs_unit_count_final <- NULL
input_attr$cnt_pid <- input_attr$cnt_multi_meta - input_attr$num_pid

input_attr$mark <- 99
input_attr$set_gift <- input_attr$cnt_set
input_attr$set_multi_meta <- ifelse(input_attr$cnt_multi_meta == 1, 0, 1)

# multi_meta 인 경우 meta값 생성

for (loop_nrow in 1:nrow(input_attr)) {
input_attr$cs_unit_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_unit_count_ref[loop_nrow],

                                         paste(input_attr$cs_unit_count[loop_nrow - 1],

                                                      input_attr$cs_unit_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_group_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_group_count_same[loop_nrow],

                                                 paste(input_attr$cs_group_count[loop_nrow - 1],

                                                       input_attr$cs_group_count_same[loop_nrow], sep ="|"))

  input_attr$cs_extra_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_extra_count_ref[loop_nrow],

                                                 paste(input_attr$cs_extra_count[loop_nrow - 1],

                                                       input_attr$cs_extra_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_goods_unit_base_size[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_goods_unit_base_size_ref[loop_nrow],

                                                          paste(input_attr$cs_goods_unit_base_size[loop_nrow - 1],

                                                                input_attr$cs_goods_unit_base_size_ref[loop_nrow], sep ="|"))

  input_attr$cs_type_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_type_count_ref[loop_nrow],

                                                paste(input_attr$cs_type_count[loop_nrow - 1],

                                                      input_attr$cs_type_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_grade_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_grade_count_ref[loop_nrow],

                                                 paste(input_attr$cs_grade_count[loop_nrow - 1],

                                                       input_attr$cs_grade_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_gender_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_gender_count_ref[loop_nrow],

                                                  paste(input_attr$cs_gender_count[loop_nrow - 1],

                                                        input_attr$cs_gender_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_weight_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_weight_count_ref[loop_nrow],

                                                  paste(input_attr$cs_weight_count[loop_nrow - 1],

                                                        input_attr$cs_weight_count_ref[loop_nrow], sep ="|"))

  input_attr$cs_box_count[loop_nrow] <- ifelse(input_attr$num_pid[loop_nrow] == 1, input_attr$cs_box_count_ref[loop_nrow],

                                               paste(input_attr$cs_box_count[loop_nrow - 1],

                                                     input_attr$cs_box_count_ref[loop_nrow], sep ="|"))

  

}

최종값만 추출

matched <- input_attr[input_attr$cnt_pid == 0,]
matched <- matched[c(1:4,21:32)]
str(matched)
names(matched)[2] <- 'pid'

입력값과 matched merge 하여 입력값에 meta 값 생성

sku tool 에서는 filter(필수, 제외 키워드 등) 및 노출상품명에 따라서 해당 노드에 들어오지 않는 경우도 발생

노출상품명에 브랜드등(필수 키워드 등)의 키워드가 없는 경우 들어오지 않기 때문에 강제적으로 복제해 줌)

생성된 matched table에는 위의 경우가 포함되어 있기 때문에 input값에 없는 row가 있을 수 있음

nput과 matched 교집합이 아닌 경우 중

1. input에만 있는 경우 -> 격리 (mark = 11), default 값으로 meta 입력

2. matched에만 있는 경우 -> 노출상품명에 브랜드 등 키워드가 없는 경우

input_matched <- merge(input, matched, by = c('hashcode', 'pid', 'title', 'option_name'), all = TRUE)

input_matched$mark <- ifelse(is.na(input_matched$mark), 11, input_matched$mark)

 

#export <- merge(input, matched, by = c('hashcode', 'pid_title', 'title', 'option_name'), all.x = TRUE)

#export$mark <- ifelse(is.na(export$mark),11,export$mark)

 

input_matched$set_gift <- ifelse(input_matched$mark == 11, 1, input_matched$set_gift)

input_matched$set_multi_meta <- ifelse(input_matched$mark == 11, 0, input_matched$set_multi_meta)

input_matched$cs_unit_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_unit_count)

input_matched$cs_group_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_group_count)

input_matched$cs_extra_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_extra_count)

input_matched$cs_goods_unit_base_size <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_goods_unit_base_size)

input_matched$cs_type_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_type_count)

input_matched$cs_grade_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_grade_count)

input_matched$cs_gender_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_gender_count)

input_matched$cs_weight_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_weight_count)

input_matched$cs_box_count <- ifelse(input_matched$mark == 11, 9999, input_matched$cs_box_count)

market_pids, pid_title, seq추가 및 colume 정렬

input_add <- input %>% 
   distinct(pid,pid_title,market_pids)
input_matched_1 <- merge(input_matched, input_add, by="pid", all.x=TRUE)
input_matched_1 <- input_matched_1[,-5]
names(input_matched_1)[(ncol(input_matched_1)-1)] <- 'pid_title'
names(input_matched_1)[ncol(input_matched_1)] <- 'market_pids'
input_matched_final <- merge(input_matched_1, input_seq, by=c("hashcode","pid"), all.x=TRUE)
input_matched_final <- input_matched_final %>%
select("seq","pid","pid_title","hashcode","title","option_name","market_pids","mark","set_gift","set_multi_meta","cs_unit_count","cs_group_count","cs_extra_count","cs_goods_unit_base_size","cs_type_count","cs_grade_count","cs_gender_count","cs_weight_count","cs_box_count")

export

write.csv(input_matched_final, "C:\\Users\\이해리\\Desktop\\multi\\bosomi_result_01.28.csv", row.names=FALSE ) ```