skip some rows in read.csv in R

Question

I have a csv file which I read using the following function:

csvData <- read.csv(file="pf.csv", colClasses=c(NA, NA,"NULL",NA,"NULL",NA,"NULL","NULL","NULL"))
dimnames(csvData)[[2]]<- c("portfolio", "date", "ticker", "quantity")

It reads all lines from that file. But i want to skip some rows from reading. The row should not read if the value of the ticker-column is: ABT or ADCT. Is it possible?

sample of my csv file is as follows:

RUS1000,01/29/1999,21st Centy Ins Group,TW.Z,90130N10,72096,1527.534,0.01,21.188
RUS1000,01/29/1999,3com Corp,COMS,88553510,358764,16861.908,0.16,47.000
RUS1000,01/29/1999,3m Co,MMM,88579Y10,401346,31154.482,0.29,77.625
RUS1000,01/29/1999,A D C Telecommunicat,ADCT,00088630,135114,5379.226,0.05,39.813
RUS1000,01/29/1999,Abbott Labs,ABT,00282410,1517621,70474.523,0.66,46.438
RUS1000,02/26/1999,21st Centy Ins Group,TW.Z,90130N10,72096,1378.836,0.01,19.125
RUS1000,02/26/1999,3com Corp,COMS,88553510,358764,11278.644,0.11,31.438
RUS1000,02/26/1999,3m Co,MMM,88579Y10,402146,29783.938,0.29,74.063 

Answer 1

It is possible using sqldf package, using read.csv.sql

Lets say the contents of sample.csv looks like this:

id,name,age
1,"a",23
2,"b",24
3,"c",23

Now to read only rows where age=23:

require(sqldf)

df <- read.csv.sql("sample.csv", "select * from file where age=23")

df
  id name age
1  1  "a"  23
2  3  "c"  23

It is possible to select necessary columns:

df <- read.csv.sql("sample.csv", "select id, name from file where age=23")
df
  id name
1  1  "a"
2  3  "c"

Answer 2

It is better to read all and subset later like suggested in the comment :

csvData [!csvData$ticker %in% c('ADCT','ABT'),]

EDIT

You can use fread from data.table package for more efficient method to read your file.

library(read.table)
fread(file="pf.csv")

Answer 3

For me the sqldf package's read.csv.sql looked great at first blush. But when I tried to use it, it failed to deal with "NULL" strings. (Others have found this out as well.) Unfortunately, it doesn't support all of read.csv features. So I had to write my own. I am surprised that there isn't a good package for this.

fetchLines=function(inputFile,match,fixed=T,n=100,maxlines=100000){ #inputFile='simple.csv'; match='APPLE';
  message('reading:',inputFile)
  n=min(n,maxlines)
  con  <- base::file(inputFile, open = "r",encoding = "UTF-8-BOM")
  data=c(readLines(con, n = 1, warn = FALSE))
  while (length(oneLine <- readLines(con, n = n, warn = FALSE)) > 0) {
    grab=grep(match,oneLine,value=T,fixed=fixed)
    if(length(grab)>0){
      data=c(data,grab)
      if(length(data)>maxlines){
        warning("bailing out too many");
        return(data);
      }
      cat('.')
    }
  } 
  close(con)
  gc()
  cat("\n")
  data;
}

#To avoid: argument 'object' must deparse to a single character string
fdata=textConnection( fetchLines("datafile.csv",'\\bP58\\b',fixed=F,maxlines = 100000))
df<-read.csv(fdata,header=T,sep=",",na.strings = c('NULL',''),fileEncoding = "UTF-8-BOM",stringsAsFactors = F)

R textConnection: "argument 'object' must deparse to a single character string"