Lesson 25 · Writing with a dialect

Writing with a dialect

The writer should speak the same dialects the reader does, so it can emit tab- or semicolon-separated files. Today you thread the dialect through the writer, so its delimiter and its quoting decision stay in sync.

The goal

Write with a configurable delimiter, quoting fields that contain that delimiter rather than the comma.

Start here - the target

TO DO

Scenario: Writing a semicolon dialect

Givena dialect whose delimiter is a semicolon, and the record ["a,b", "c;d"]

Whenit is written with that dialect

Thenthe output is a,b;"c;d" followed by a newline, so the comma field is written bare and the semicolon field is quoted

Andwriting then parsing with the same dialect round-trips, so parsing the output with the semicolon dialect returns [["a,b", "c;d"]]

Background

A writer that only ever emits commas cannot produce the tab- and semicolon-separated files the reader happily consumes, so the writer takes a dialect too. The delimiter it joins with becomes the dialect’s delimiter, and, just as importantly, so does the delimiter it checks for when deciding whether to quote. Under a semicolon dialect a comma is ordinary content that needs no quoting, while a semicolon is the character that would break a field and so must be quoted. The quote trigger and the join character must come from the same dialect, or the writer would quote the wrong things.

This keeps the two halves of the library symmetric: whatever dialect you write with, reading the result with the same dialect returns the original records. That symmetry is the practical meaning of a dialect, one description of a file format that both directions obey. Redefine the plain Write as WriteWith using the default comma dialect, exactly as you did for Parse and ParseWith, so all your earlier writer behavior is preserved while the configurable path opens up. With this, the library can convert between dialects, read semicolons and write tabs, purely by pairing a reader dialect with a writer dialect.

Make it work

// WriteWith(d Dialect, records) joins with d.Delimiter and terminates with '\n'
// the quote trigger now uses the dialect delimiter, not a hard-coded comma:
//   quote if the field contains d.Delimiter, a '"', a '\r', or a '\n'
// Write(records) is WriteWith(DefaultDialect, records)

CheckpointDONE

The writer honors the dialect delimiter and quotes accordingly, round-tripping with the matching reader. Commit and stop here.