R/pisa.select.merge.R

In eldafani/intsvy: International Assessment Data Manager

pisa.select.merge <- 
  function(folder=getwd(), student.file, parent.file=c(), school.file=c(), countries, student=c(), parent, school) {
    
    # Remove leading and trailing whitespaces in var labes  
    if(!missing(student) & !is.null(student)) {
      student = gsub("^[[:space:]]+|[[:space:]]+$", "", student)
      student =toupper(student)
    }
    
    if(!missing(parent)){
      parent = gsub("^[[:space:]]+|[[:space:]]+$", "", parent)
      parent = toupper(parent)
    }
    
    if(!missing(school)){
      school = gsub("^[[:space:]]+|[[:space:]]+$", "", school)
      school = toupper(school)
    }
    
    # No variables selected (error)  
    
    if (missing(student) & missing(parent) & missing(school)) {
      stop("no variables are selected")
    }
    
    # Creating student, parent, and school filepaths 
    
    # Student file required for country labels
    if(missing(student.file)) {
      stop("the student file is required")
    }
    
    if(!missing(folder) & !missing(student.file)) {
      student.file =   file.path(folder, paste(student.file, sep=""))
    }
    
    if(!missing(folder) & !missing(school.file)) {
      school.file =  file.path(folder, paste(school.file, sep=""))
    }
    
    if(!missing(folder) & !missing(parent.file)) {
      parent.file =  file.path(folder, paste(parent.file, sep=""))
    }
    
    # Retrieve file names
    files.all <- list(student.file, parent.file, school.file)
    names(files.all) <- c('Student', 'Parent', 'School')
    
    # Remove null elements in list
    files.all <- files.all[sapply(files.all, length)>0]
    
    # Participating countries (from student file)
    pisa.student <- spss.system.file(files.all[["Student"]], to.lower=FALSE)
    country <- unique(as_haven(pisa.student$CNT)$CNT)
    
    # If countries missing, all countries selected
    if (missing(countries)) {
      countries <- country
    }
    
    # If countries are entered numerically, change to ISO labels for file selection (next)
    if (is.numeric(countries)) {
      countries=pisa.country[pisa.country$Code %in% countries, "ISO"]
    }
    
    # Have to use spss.system.file, otherwise read.spss crashes
    
    # Student data (need for school and parent data too)
    
    if (!missing(student) | !missing(parent)) {
  
      student.data <- read_sav(files.all[["Student"]], col_select = c("CNT", 
      unique(grep("^PV|^W_F|ID$|STD$", names(pisa.student), value=TRUE)), unique(student)))
      
      student.data <- student.data[as.character(student.data$CNT) %in% countries, ]
      
    }
    
    
    # Parental questionnaire
    
    if (!missing(parent)) {
      
      if (is.null(files.all[["Parent"]])) {
        stop("cannot locate parental questionnaire data file")
      }
      
      pisa.parent <- spss.system.file(files.all[["Parent"]], to.lower=FALSE)
      
      pisa.parent <- read_sav(files.all[["Parent"]], col_select = c("CNT", 
      unique(grep("ID$|STD$", names(pisa.parent), value=TRUE)), unique(parent)))
      
      parent.data <- pisa.parent[as.character(pisa.parent$CNT) %in% countries, ]

    }
    
    
    # School data
    
    if (!missing(school)) {
      
      if (is.null(files.all[["School"]])) {
        stop("cannot locate school data file")
      }
      
      pisa.school <- spss.system.file(files.all[["School"]], to.lower=FALSE)
      
      pisa.school <- read_sav(files.all[["School"]], col_select = 
      c("CNT", unique(grep("^W_F|ID$", names(pisa.school), value=TRUE)), unique(school)))
      
      school.data <- pisa.school[as.character(pisa.school$CNT) %in% countries, ]

    }
    
    
    # Merging data depending on existing datasets/arguments
    
    # Student data available
    
    if (!missing(student) & missing(parent) & missing(school)) {
      pisa.all <- student.data
    }
    
    if (!missing(student) & !missing(parent) & missing(school)) {
      pisa.all <- merge(student.data, parent.data, all.x=TRUE, by=intersect(names(student.data), names(parent.data)))
      pisa.all <- pisa.all[grep("*.y", names(pisa.all), invert=TRUE)]
      names(pisa.all) <- gsub("*.x", "", names(pisa.all))
    }
    
    if (!missing(student) & missing(parent) & !missing(school)) {
      pisa.all <- merge(student.data, school.data, all.x=TRUE, 
      by=grep("CNT|SCH.*ID", intersect(names(student.data), names(school.data)), value = TRUE))
      pisa.all <- pisa.all[grep("*.y", names(pisa.all), invert=TRUE)]
      names(pisa.all) <- gsub("*.x", "", names(pisa.all))
    }
    
    if (!missing(student) & !missing(parent) & !missing(school)) {
      pisa.all <- merge(student.data, parent.data, all.x=TRUE)
      pisa.all <- merge(pisa.all, school.data, all.x=TRUE)
      pisa.all <- pisa.all[grep("*.y", names(pisa.all), invert=TRUE)]
      names(pisa.all) <- gsub("*.x", "", names(pisa.all))
    }
    
    # Parent data available
    
    if (is.null(student) & !missing(parent) & missing(school)) {
      pisa.all <- merge(student.data, parent.data)
      pisa.all <- pisa.all[grep("*.y", names(pisa.all), invert=TRUE)]
      names(pisa.all) <- gsub("*.x", "", names(pisa.all))
    }
    
    if (is.null(student) & !missing(parent) & !missing(school)) {
      pisa.all <- merge(school.data, parent.data)
      pisa.all <- pisa.all[grep("*.y", names(pisa.all), invert=TRUE)]
      names(pisa.all) <- gsub("*.x", "", names(pisa.all))
    }
    
    # School data available
    
    if (is.null(student) & missing(parent) & !missing(school)) {
      pisa.all <- school.data
    }
    
    # Create country label variable (not possible to add labels to numeric factor, see to do list)
    
    return(droplevels(pisa.all))
  }