A grammar for diff output in normal, unified, and git formats

diff.ebnf

(* EBNF grammar for diff output in normal, unified, and git formats *)
(* Author: Pedro R. Borges *)
(* Read the accompanying article at https://www.prborges.com/2023/grammar-for-diff-output *)

diff = emptyDiff | normalDiff | uniDiff | gitDiff ;

(* 1. Empty diff *)

emptyDiff = "" ;

(* 2. Normal format *)

normalDiff = normalHunk , {normalHunk } ;

normalHunk = addHunk | deleteHunk | changeHunk ;

addHunk =
    number , 'a' , range , eol ,
    addedLines ,
    [ noteLine ] ;

deleteHunk =
    range , 'd' , number , eol ,
    deletedLines ,
    [ noteLine ] ;

changeHunk =
    range , 'c' , range , eol ,
    deletedLines ,
    [ noteLine ] ,
    "---" , eol ,
    addedLines ,
    [ noteLine ] ;

addedLines = addedLine , { addedLine } ;

addedLine = "> " , line , eol ;

deletedLines = deletedLine , { deletedLine } ;

deletedLine = "< " , line , eol ;

(* 3. Unified and Git formats *)

uniDiff =
    "--- " , fileInfo , eol ,
    "+++ " , fileInfo , eol ,
    uniHunk , { uniHunk } ;

gitDiff =
    "diff --git " , filePath , ' ' , filePath , eol ,
    "index " , commitHash , ' ' , commitHash , eol ,
    "--- " , filePath , eol ,
    "+++ " , filePath , eol ,
    uniHunk , { uniHunk } ;

fileInfo = filePath , '\t' , timestamp ;

timestamp = date , ' ' , time24 , ' ' , timeZone ;

date = fourDigits , '-' , twoDigits , '-' , twoDigits ;

time24 = twoDigits , ':' , twoDigits , ':' , twoDigits , '.' , nineDigits ;

timeZone = ('+' | '-') , fourDigits ;

uniHunk =
    uniHunkDescriptor , eol ,
    { contextLine } ,
    lrGroup ,
    {clrGroup } ,
    [ contextLines , [ noteLine ] ] ;

uniHunkDescriptor= "@@ +" , range , " -" , range , " @@" , [section] ;

section = ' ' , line ;

clrGroup = contextLines , lrGroup ;

contextLines = contextLine , { contextLine } ;

contextLine = ' ' , line , eol ;

lrGroup
    = leftLines
    | rightLines
    | leftLines , rightLines ;

leftLines =
    leftLine ,
    { leftLine } ,
    [ noteLine ] ;

leftLine = '-' , line , eol ;

rightLines =
    rightLine ,
    { rightLine } ,
    [ noteLine ] ;

rightLine = '+' , line , eol ;

(* 4. Common and auxiliary rules *)

range
    = number
    | number , ',' , number ;

(* Here, "\\ " is one backslash followed by a space *)
noteLine = "\\ " , note , eol ;

number
    = '0'
    | '1' .. '9' , { digit } ;

(* Ranges and * repetition are not allowed in some EBNF variants *)

digit = '0' .. '9' ;

twoDigits = 2 * digit ;

fourDigits = 4 * digit ;

nineDigits = 9 * digit ;

eol = '\n' ;

(* 5. Special rules *)

line = ? line from a diffed file without '\n' ? ;

note = ? note indicating no newline in last line of a file ? ;

filePath = ? file path string ? ;

commitHash = ? alphanumeric commit hash ? ;