Lesson 25 · Parsing a unified diff

Parsing a unified diff

A patch is a file header plus a sequence of hunks, each a header and prefixed body lines. Today you parse the whole thing back into hunks of keep/delete/insert operations - the same Op list your diff produces.

The goal

Parse a unified diff string into a list of hunks, each with its ranges and its operations.

Start here - the target

TO DO

Scenario: A unified diff parses back into hunks and operations

Giventhe unified diff text for ["line1".."line5"] against the same list with "line3" changed to "CHANGED"

WhenParse reads it

Thenit returns 1 hunk with old start 1 and new start 1, holding 6 operations

Andthose operations are Keep "line1", Keep "line2", Delete "line3", Insert "CHANGED", Keep "line4", Keep "line5" (from the " ", "-", and "+" prefixes)

Background

Parsing a patch is a line-by-line classification, using the very prefixes you emit when writing one. The --- and +++ file-header lines are skipped, an @@ line opens a new hunk (parsed by the previous lesson’s helper), and the \ No newline at end of file marker is skipped for now since it carries no operation. Every other line is a body line whose first character - space, minus, or plus - tells you whether it is a Keep, Delete, or Insert, with the rest of the line being the content. The result is a hunk holding the same []Op your Diff produces, which is exactly what makes applying it straightforward.

Notice the symmetry: writing a diff turned operations into prefixed lines, and reading it turns prefixed lines back into operations. The unified format is really just a serialization of the edit script with enough location metadata (the @@ ranges) to apply it to a file that may have shifted. With a parsed patch in hand, the applier can walk the original document and the hunk together, and the round-trip is one lesson away.

Make it work

for _, line := range strings.Split(patch, "\n") {
  switch {
  case strings.HasPrefix(line, "--- "), strings.HasPrefix(line, "+++ "):
    continue // file header
  case strings.HasPrefix(line, "@@"):
    // start a new hunk from parseHunkHeader(line)
  case strings.HasPrefix(line, "\\"):
    continue // "\ No newline at end of file" marker
  case strings.HasPrefix(line, " "):
    // Keep line[1:]
  case strings.HasPrefix(line, "-"):
    // Delete line[1:]
  case strings.HasPrefix(line, "+"):
    // Insert line[1:]
  }
}

CheckpointDONE

You can parse a full unified diff into hunks of operations. Commit and stop here.