plot2.R

## *** Coursera Exploratory Data Analysis - Project 1 ***
## 
## Script name: plot2.R  (created using R 3.1.1 on Windows 7)
##
## See README.md for details on the assignment and source data used.
## Download source data and extract household_power_consumption.txt file first
## 
## Important: The script expects data in working directory of R. The code prompts
## and checks existence of file first. It terminates if file is not found.
## 
## The main code is implemented in a function named run().  Source the script file 
## into R enviroment and execute run() to begin.  
## 
## The code first reads the data for the two days from file and creates a dataset (data.use)
## in memory, applying column classes and names to suit the analysis. 
## strptime function is used to convert the original character data in Date and Time columns,
## which is then added to the dataset as a new variable of class Date/Time and used later
## for the plot generation. The plot code targets png file device in working directory
## and it is repeated for screen display as well.

fileCheck <- function () {
    resp <- readline("Is household_power_consumption.txt in working directory? [y/n]? ")
    if(tolower(substr(resp,1,1)) == "n") 
        stop("Please see README.md before continuing. Ending script..")
    
    dir.use <- getwd()  # expects data in working directory
    cat(paste0("\nChecking in ", dir.use,".." ))
    
    if (!file.exists(paste0(dir.use, "/household_power_consumption.txt"))) {
        stop("Data file not found! Please see README.md for download instructions.")
    }
    dir.use  #return data directory
}

## Main function
run <- function() {
    ## test existence of data file to begin with
    dir.use <- tryCatch({fileCheck()}, error = function(e) { stop(e)} )
    
    ## Data is rectangular and contains header as in README.md. We can read a few rows
    #  at the start and use it to determine column classes and column names 
    #  for use later in the script. Could have created the vectors manually as well.
    top5rows <- read.table("household_power_consumption.txt", 
                           header=TRUE, nrows=5, sep=";", stringsAsFactors=FALSE, na.strings=c("?", "NA"))
    
    classes <- sapply(top5rows, class)  ## using R's class function
    cat ("\nDataset has columns with classes:\n ")
    #cat(names(top5rows), "\n")
    cat(classes, "\n")
    
    
    ## 1. Read data for use in analysis
    #  Read only two days data from the large dataset
    
    # First get a handle to file in working directory. Need read access only
    fc <- file("household_power_consumption.txt", "r")
    
    #  Read text and extract lines that match dd/mm/yyyy formated 
    #  days 1 and 2 of Feb, 2007 
    cat("\nReading large data....do not interrupt...\n")
    
    data.use <- read.table(text=grep("^(0?[1,2])/(0?2)/2007", readLines(fc), value=TRUE), 
                           header=F, sep=";", stringsAsFactors=FALSE, na.strings=c("?", "NA"),
                           colClasses=classes, col.names=names(top5rows), check.names=FALSE)
    
    close(fc) # close connection to file
    
    ## create new datetime variable from Date and Time character data
    # Note, strptime converts and new variable gets Date/Time class 
    data.use <- transform(data.use, datetime = strptime(paste(Date,Time), format="%d/%m/%Y %H:%M:%S"))
    
    
    ##  Data ready..begin plot
    cat ("\nCreating plot2.png in ", dir.use, "\n ")


    png(file="plot2.png", width=480, height=480,  units = "px")  ## Open PNG device

    with(data.use, plot(datetime, Global_active_power, type="l",
                              xlab="", ylab="Global Active Power (kilowatts)") )

    d <- dev.off() # turn PNG graphics device off

    cat("\n", normalizePath("plot2.png"),"..is ready for viewing!")

}  ## end of run()