根据另一个定制顺序的列对数据框中的列对进行排序

我同意zx8754，您想要做的是在（V2，V3）和（V4，V5）的V1上进行合并（在sql中合并），如果您的数据最初是在单独的数据库中，其格式为：

    df$RowNum <- 1:nrow(df) # row numbers to sort at the end

1）使用data.table，如果您有巨大的表，则可能有用

    library("data.table")
    dt <- as.data.table(df)

    # your data are essentially three different tables
    # so let's split it up, removing useless empty cells
    ref <- dt[which(dt$V1!=""),c("V1","RowNum"),with=FALSE]
          # with=FALSE necessary for the second argument to 
          #    be understood as column names in a data table
    setkey(ref,"V1") # the column used for the merges

    tab1 <- dt[which(dt$V2!=""),c("V2","V3"),with=FALSE]
    setkey(tab1,"V2")

    tab2 <- dt[which(dt$V4!=""),c("V4","V5"),with=FALSE]
    setkey(tab2,"V4")

    # merge tab1 to ref and tab2 to the product
    # using data.table formalism
    df3 <- tab2[tab1[ref,allow.cartesian=T],allow.cartesian=T]
             # allow.cartesio=T important to keep all in ref

    # and to get exactly the same output
    setkey(df3,"RowNum") # order df3 by RowNum
    df3 <- within(df3,{
                  V2 <- V1 <- V4 # make the columns V1 and V2
                  V4[is.na(V5)]<-NA # put back NA
                  V2[is.na(V3)]<-NA
    })
    setcolorder(df3,sort(names(df3))) # sort V1 to V5

2）使用与上述相同的基数R＃，将df拆分为三个表ref <-df [which（df $ V1！=“”），c（“ V1”，“ RowNum”）] tab1 <-df [which（ df $ V2！=“ =”“，c（” V2“，” V3“）] tab2 <-df [which（df $ V4！=”“），c（” V4“，” V5“）]

然后，您合并：

    df2 <- merge(ref,tab1,by.x="V1",by.y="V2",all=TRUE)
    df2 <- merge(df2,tab2,by.x="V1",by.y="V4",all=TRUE)

之后，如果您想要完全相同的输出：

    df2 <- df2[order(df2$RowNum),] # order by RowNum
    # make the V2 column
    df2$V2 <- df2$V1
    df2$V2[which(is.na(df2$V3))] <- NA

    # make the V4 column
    df2$V4 <- df2$V1
    df2$V4[which(is.na(df2$V5))] <- NA

    # order the columns as wanted
    df2 <- df2[,c("V1","V2","V3","V4","V5")]