class BBS::Store::CsvImport

The rows of a Store’s CSV file of earlier versions, read once for the import into SQLite. A broken row does not stop the import: a record that is not CSV (an unclosed quote runs until a line that is a whole row, at most MAX_LINES lines), has more fields than the header or no session_id is skipped, and its lines are set aside in <path>.rejected, under the header, for a sysop to look at.

csv = BBS::Store::CsvImport.new('data/callers.csv', %w[session_id name])
csv.rows                          # => [{ 'session_id' => 'a', 'name' => 'Ada' }, ...]
csv.finish('data/callers.sqlite3') # renames the file, writes the rejects

Constants

MAX_LINES: The most lines a quoted field may run over before its record counts as broken.

Attributes

rejected [R]

The broken records, each the lines it was read from.

rows [R]

The good rows, each a Hash of column name => String or nil.

Public Class Methods

new (path, columns)

Source

# File lib/bbs/store/csv_import.rb, line 25
def initialize(path, columns)
  @path     = path
  @columns  = columns
  @rows     = []
  @rejected = []
  read
end

Reads the CSV file at path (as bytes: invalid UTF-8 is dropped, not fatal), keeping the columns (Array of String) of each row.

Public Instance Methods

finish (database)

Source

# File lib/bbs/store/csv_import.rb, line 42
def finish(database)
  File.rename(@path, "#{@path}.imported")
  File.write("#{@path}.rejected", @header + @rejected.join) if @rejected.any?
  "#{@rows.size} rows of #{@path} imported into #{database}" +
    (@rejected.any? ? ", #{@rejected.size} set aside in #{@path}.rejected" : '')
end

Renames the CSV file to <path>.imported, writes the rejected records to <path>.rejected when there are any, and returns a line for the log naming database, where the rows went.