Method: Rust::DataFrame#merge

Defined in:

lib/rust/core/types/dataframe.rb

#merge(other, by, first_alias = "x", second_alias = "y") ⇒ Object

Raises:

• (TypeError)

# File 'lib/rust/core/types/dataframe.rb', line 313

def merge(other, by, first_alias = "x", second_alias = "y")
    raise TypeError, "Expected Rust::DataFrame" unless other.is_a?(DataFrame)
    raise TypeError, "Expected list of strings" if !by.is_a?(Array) || !by.all? { |e| e.is_a?(String) }
    raise "This dataset should have all the columns in #{by}" unless (by & self.column_names).size == by.size
    raise "The passed dataset should have all the columns in #{by}" unless (by & other.column_names).size == by.size
    
    if first_alias == second_alias
        if first_alias == ""
            my_columns = self.column_names - by
            other_columns = other.column_names - by
            intersection = my_columns & other_columns
            raise "Cannot merge because the following columns would overlap: #{intersection}" if intersection.size > 0
        else
            raise "The aliases can not have the same value"
        end
    end
    
    my_keys = {}
    self.each_with_index do |row, i|
        key = []
        by.each do |colname|
            key << row[colname]
        end
        
        my_keys[key] = i
    end
    
    merged_column_self  = (self.column_names - by)
    merged_column_other = (other.column_names - by)
    
    first_alias =  first_alias + "."     if first_alias.length > 0
    second_alias = second_alias + "."    if second_alias.length > 0
    
    merged_columns = merged_column_self.map { |colname| "#{first_alias}#{colname}" } + merged_column_other.map { |colname| "#{second_alias}#{colname}" }
    columns = by + merged_columns
    result = DataFrame.new(columns)
    other.each do |other_row|
        key = []
        by.each do |colname|
            key << other_row[colname]
        end
        
        my_row_index = my_keys[key]
        if my_row_index
            my_row = self.row(my_row_index)
            
            to_add = {}
            by.each do |colname|
                to_add[colname] = my_row[colname]
            end
            
            merged_column_self.each do |colname|
                to_add["#{first_alias}#{colname}"] = my_row[colname]
            end
            
            merged_column_other.each do |colname|
                to_add["#{second_alias}#{colname}"] = other_row[colname]
            end
            
            result << to_add
        end
    end
    
    return result
end