parcsv is a parser which treats CSV files as tabular dataframes enabling you to manipulate them fairly easily.

parcsv is based on Either[Throwable, CSV] to enable you to do functional style CSV processing.

Typical flow:

• Read CSV from different sources (File, Raw String, Map etc..)
• Process it
• Save it

Scala code:

import com.ghurtchu.csv._

// csv as a raw string
val source =
  """food,calories,protein,carbs,isHealthy
    |apple,52,0.3,14,true
    |egg,155,13,26,true
    |potato,77,4.3,26,true
    |sugar,387,0,100,false
    |""".stripMargin

// only keep those columns which contain "o" in their header value
val containsLetter: Column => Boolean = col => col.header.value.contains("o")

// only keeps those rows which have a value less than 10 under "protein" header
val lowProteinFoodFilter: Cell => Boolean = cell => cell.belongsTo("protein") && cell.numericValue <= 10

// csv processing
val newCSV = for {
  csv      <- CSV.fromString(source) // read
  cleaned  <- csv.filterColumns(containsLetter) // drop "carbs" and "isHealthy"
  filtered <- cleaned.filterRows(lowProteinFoodFilter) // take rows with "protein" value less than 10
  sorted   <- filtered.sortByColumn("calories", SortOrdering.Desc) // sort by "calories" in descending numeric order
  _        <- sorted.display // print it
  _        <- sorted.save("data/processed_food.csv") //save it
} yield sorted

Print result:

Let's see an advanced example using Pipes.

Pipe is useful when there is a need to apply more than 1 transformation to CSV. They hold filter/map/reduce functions which will be applied sequentially to CSV files.

In this example we use:

• FilterColumnPipe - filters columns sequentially
• FilterRowPipe - filters rows sequentially
• TransformColumnPipe - transforms rows sequentially

Scala code:

import com.ghurtchu.csv._

val filterColumnPipe = FilterColumnPipe(
  col => Seq("name", "popularity", "creator").contains(col.header.value), // choose columns by names
  col => col.cells.forall(_.value.length <= 20) // keep columns with all values shorter than 20 characters
)

val filterRowPipe = FilterRowPipe(
  row => row.index % 2 == 1, // then take only odd-indexed rows
  row => row.isFull // then keep those which have no N/A-s
)

val transformColumnPipe = TransformColumnPipe(
  col => Column(col.header, col.cells.map(cell => cell.copy(value = cell.value.toUpperCase))) // make all values uppercase
)

// create bigger pipe by joining from left to right
val fullPipe = filterColumnPipe ~> filterRowPipe ~> transformColumnPipe

// processing
val transformedCSV = for {
  csv         <- CSV.fromFile("data/programming_languages.csv") // read
  transformed <- csv.transformVia(fullPipe) // apply the whole pipe
  _           <- transformed.display // print it 
  _           <- transformed.save("data/updated.csv") // save
} yield transformed

Print result:

TODO:

• slicing
• add CSVError hierarchy to replace Throwable in Either[Throwable, CSV]
• and much more...