Support range formatting (#120)

This patch implements the `--lines=a:b` command line argument for
limiting the formatting to the line range `a:b`. Multiple ranges are
supported. Closes #114.

Author: fredrikekre

CHANGELOG.md

@@ -5,6 +5,11 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## Unreleased
+### Added
+ - New command line option `--lines=a:b` for limiting formatting to lines `a` to `b`.
+   `--lines` can be repeated to specify multiple ranges ([#114], [#120]).
+
 ## [v1.1.0] - 2024-12-04
 ### Changed
  - Fix a bug that caused "single space after keyword" to not apply after the `function`

src/Runic.jl

@@ -131,6 +131,7 @@ mutable struct Context
     check::Bool
     diff::Bool
     filemode::Bool
+    line_ranges::Vector{UnitRange{Int}}
     # Global state
     indent_level::Int # track (hard) indentation level
     call_depth::Int # track call-depth level for debug printing
@@ -144,11 +145,104 @@ mutable struct Context
     lineage_macros::Vector{String}
 end
 
+const RANGE_FORMATTING_BEGIN = "#= RUNIC RANGE FORMATTING " * "BEGIN =#"
+const RANGE_FORMATTING_END = "#= RUNIC RANGE FORMATTING " * "END =#"
+
+function add_line_range_markers(str, line_ranges)
+    lines = collect(eachline(IOBuffer(str); keep = true))
+    sort!(line_ranges, rev = true)
+    for r in line_ranges
+        a, b = extrema(r)
+        if a < 1 || b > length(lines)
+            throw(MainError("`--lines` range out of bounds"))
+        end
+        if b == length(lines) && !endswith(lines[end], "\n")
+            lines[end] *= "\n"
+        end
+        insert!(lines, b + 1, RANGE_FORMATTING_END * "\n")
+        insert!(lines, a, RANGE_FORMATTING_BEGIN * "\n")
+    end
+    io = IOBuffer(; maxsize = sum(sizeof, lines; init = 0))
+    join(io, lines)
+    src_str = String(take!(io))
+    return src_str
+end
+
+function remove_line_range_markers(src_io, fmt_io)
+    src_lines = eachline(seekstart(src_io); keep = true)
+    fmt_lines = eachline(seekstart(fmt_io); keep = true)
+    io = IOBuffer()
+    # These can't fail because we will at the minimum have the begin/end comments
+    src_itr = iterate(src_lines)
+    @assert src_itr !== nothing
+    src_ln, src_token = src_itr
+    itr_fmt = iterate(fmt_lines)
+    @assert itr_fmt !== nothing
+    fmt_ln, fmt_token = itr_fmt
+    eof = false
+    while true
+        # Take source lines until range start or eof
+        while !occursin(RANGE_FORMATTING_BEGIN, src_ln)
+            if !occursin(RANGE_FORMATTING_END, src_ln)
+                write(io, src_ln)
+            end
+            src_itr = iterate(src_lines, src_token)
+            if src_itr === nothing
+                eof = true
+                break
+            end
+            src_ln, src_token = src_itr
+        end
+        eof && break
+        @assert occursin(RANGE_FORMATTING_BEGIN, src_ln) &&
+            strip(src_ln) == RANGE_FORMATTING_BEGIN
+        # Skip ahead in the source lines until the range end
+        while !occursin(RANGE_FORMATTING_END, src_ln)
+            src_itr = iterate(src_lines, src_token)
+            @assert src_itr !== nothing
+            src_ln, src_token = src_itr
+        end
+        @assert occursin(RANGE_FORMATTING_END, src_ln) &&
+            strip(src_ln) == RANGE_FORMATTING_END
+        # Skip ahead in the formatted lines until range start
+        while !occursin(RANGE_FORMATTING_BEGIN, fmt_ln)
+            fmt_itr = iterate(fmt_lines, fmt_token)
+            @assert fmt_itr !== nothing
+            fmt_ln, fmt_token = fmt_itr
+        end
+        @assert occursin(RANGE_FORMATTING_BEGIN, fmt_ln) &&
+            strip(fmt_ln) == RANGE_FORMATTING_BEGIN
+        # Take formatted lines until range end
+        while !occursin(RANGE_FORMATTING_END, fmt_ln)
+            if !occursin(RANGE_FORMATTING_BEGIN, fmt_ln)
+                write(io, fmt_ln)
+            end
+            fmt_itr = iterate(fmt_lines, fmt_token)
+            @assert fmt_itr !== nothing
+            fmt_ln, fmt_token = fmt_itr
+        end
+        @assert occursin(RANGE_FORMATTING_END, fmt_ln) &&
+            strip(fmt_ln) == RANGE_FORMATTING_END
+        eof && break
+    end
+    write(seekstart(fmt_io), take!(io))
+    truncate(fmt_io, position(fmt_io))
+    return
+end
+
 function Context(
         src_str::String; assert::Bool = true, debug::Bool = false, verbose::Bool = debug,
-        diff::Bool = false, check::Bool = false, quiet::Bool = false, filemode::Bool = true
+        diff::Bool = false, check::Bool = false, quiet::Bool = false, filemode::Bool = true,
+        line_ranges::Vector{UnitRange{Int}} = UnitRange{Int}[]
     )
+    if !isempty(line_ranges)
+        # If formatting is limited to certain line ranges we modify the source string to
+        # include begin and end marker comments.
+        src_str = add_line_range_markers(src_str, line_ranges)
+    end
     src_io = IOBuffer(src_str)
+    # TODO: If parsing here fails, and we have line ranges, perhaps try to parse without the
+    # markers to check whether the markers are the cause of the failure.
     src_tree = Node(
         JuliaSyntax.parseall(JuliaSyntax.GreenNode, src_str; ignore_warnings = true, version = v"2-")
     )
@@ -176,7 +270,7 @@ function Context(
     format_on = true
     return Context(
         src_str, src_tree, src_io, fmt_io, fmt_tree, quiet, verbose, assert, debug, check,
-        diff, filemode, indent_level, call_depth, format_on, prev_sibling, next_sibling,
+        diff, filemode, line_ranges, indent_level, call_depth, format_on, prev_sibling, next_sibling,
         lineage_kinds, lineage_macros
     )
 end
@@ -501,14 +595,20 @@ function format_tree!(ctx::Context)
     end
     # Truncate the output at the root span
     truncate(ctx.fmt_io, span(root′))
+    # Remove line range markers if any
+    if !isempty(ctx.line_ranges)
+        remove_line_range_markers(ctx.src_io, ctx.fmt_io)
+    end
     # Check that the output is parseable
     try
         fmt_str = String(read(seekstart(ctx.fmt_io)))
+        # TODO: parsing may fail here because of the removal of the range comments
         JuliaSyntax.parseall(JuliaSyntax.GreenNode, fmt_str; ignore_warnings = true, version = v"2-")
     catch
         throw(AssertionError("re-parsing the formatted output failed"))
     end
     # Set the final tree
+    # TODO: When range formatting this doesn't match the content of ctx.fmt_io
     ctx.fmt_tree = root′
     return nothing
 end

src/debug.jl

@@ -10,6 +10,13 @@ struct AssertionError <: RunicException
     msg::String
 end
 
+# Thrown from internal code when invalid CLI arguments can not be validated directly in
+# `Runic.main`: `throw(MainError("message"))` from internal code is like calling
+# `panic("message")` in `Runic.main`.
+struct MainError <: RunicException
+    msg::String
+end
+
 function Base.showerror(io::IO, err::AssertionError)
     print(
         io,

src/main.jl

@@ -196,6 +196,24 @@ function writeo(output::Output, iob)
     return
 end
 
+function insert_line_range(line_ranges, lines)
+    m = match(r"^(\d+):(\d+)$", lines)
+    if m === nothing
+        return panic("can not parse `--lines` argument as an integer range")
+    end
+    range_start = parse(Int, m.captures[1]::SubString)
+    range_end = parse(Int, m.captures[2]::SubString)
+    if range_start > range_end
+        return panic("empty `--lines` range")
+    end
+    range = range_start:range_end
+    if !all(x -> isdisjoint(x, range), line_ranges)
+        return panic("`--lines` ranges cannot overlap")
+    end
+    push!(line_ranges, range)
+    return 0
+end
+
 function main(argv)
     # Reset errno
     global errno = 0
@@ -210,6 +228,7 @@ function main(argv)
     diff = false
     check = false
     fail_fast = false
+    line_ranges = typeof(1:2)[]
 
     # Parse the arguments
     while length(argv) > 0
@@ -234,6 +253,10 @@ function main(argv)
             check = true
         elseif x == "-vv" || x == "--debug"
             debug = verbose = true
+        elseif (m = match(r"^--lines=(.*)$", x); m !== nothing)
+            if insert_line_range(line_ranges, m.captures[1]::SubString) != 0
+                return errno
+            end
         elseif x == "-o"
             if length(argv) < 1
                 return panic("expected output file argument after `-o`")
@@ -272,6 +295,9 @@ function main(argv)
     if outputfile != "" && length(inputfiles) > 1
         return panic("option `--output` can not be used together with multiple input files")
     end
+    if !isempty(line_ranges) && length(inputfiles) > 1
+        return panic("option `--lines` can not be used together with multiple input files")
+    end
     if length(inputfiles) > 1 && !(inplace || check)
         return panic("option `--inplace` or `--check` required with multiple input files")
     end
@@ -363,14 +389,17 @@ function main(argv)
 
         # Call the library to format the text
         ctx = try
-            ctx′ = Context(sourcetext; quiet, verbose, debug, diff, check)
+            ctx′ = Context(sourcetext; quiet, verbose, debug, diff, check, line_ranges)
             format_tree!(ctx′)
             ctx′
         catch err
             print_progress && errln()
             if err isa JuliaSyntax.ParseError
                 panic("failed to parse input: ", err)
                 continue
+            elseif err isa MainError
+                panic(err.msg)
+                continue
             end
             msg = "failed to format input: "
             @static if juliac

test/maintests.jl

@@ -343,7 +343,7 @@ function maintests(f::R) where {R}
     end
 
     # runic -o readonly.jl in.jl
-    return cdtmp() do
+    cdtmp() do
         f_in = "in.jl"
         write(f_in, bad)
         f_out = "readonly.jl"
@@ -356,6 +356,49 @@ function maintests(f::R) where {R}
         @test isempty(fd1)
         @test occursin("could not write to output file", fd2)
     end
+
+    # runic --lines
+    cdtmp() do
+        src = """
+        function f(a,b)
+            return a+b
+         end
+        """
+        rc, fd1, fd2 = runic(["--lines=1:1"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a, b)\n    return a+b\n end\n"
+        rc, fd1, fd2 = runic(["--lines=2:2"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a,b)\n    return a + b\n end\n"
+        rc, fd1, fd2 = runic(["--lines=3:3"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a,b)\n    return a+b\nend\n"
+        rc, fd1, fd2 = runic(["--lines=1:1", "--lines=3:3"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a, b)\n    return a+b\nend\n"
+        rc, fd1, fd2 = runic(["--lines=1:1", "--lines=2:2", "--lines=3:3"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a, b)\n    return a + b\nend\n"
+        rc, fd1, fd2 = runic(["--lines=1:2"], src)
+        @test rc == 0 && isempty(fd2)
+        @test fd1 == "function f(a, b)\n    return a + b\n end\n"
+        # Errors
+        rc, fd1, fd2 = runic(["--lines=1:2", "--lines=2:3"], src)
+        @test rc == 1
+        @test isempty(fd1)
+        @test occursin("`--lines` ranges cannot overlap", fd2)
+        rc, fd1, fd2 = runic(["--lines=0:1"], src)
+        @test rc == 1 && isempty(fd1)
+        @test occursin("`--lines` range out of bounds", fd2)
+        rc, fd1, fd2 = runic(["--lines=3:4"], src)
+        @test rc == 1 && isempty(fd1)
+        @test occursin("`--lines` range out of bounds", fd2)
+        rc, fd1, fd2 = runic(["--lines=3:4", "foo.jl", "bar.jl"], src)
+        @test rc == 1 && isempty(fd1)
+        @test occursin("option `--lines` can not be used together with multiple input files", fd2)
+    end
+
+    return
 end
 
 # rc = let argv = pushfirst!(copy(argv), "runic"), argc = length(argv) % Cint

test/runtests.jl

@@ -1541,6 +1541,52 @@ end
     end
 end
 
+# TODO: Support lines in format_string and format_file
+function format_lines(str, lines)
+    line_ranges = lines isa UnitRange ? [lines] : lines
+    ctx = Runic.Context(str; filemode = false, line_ranges = line_ranges)
+    Runic.format_tree!(ctx)
+    return String(take!(ctx.fmt_io))
+end
+
+@testset "--lines" begin
+    str = """
+    function f(a,b)
+        return a+b
+     end
+    """
+    @test format_lines(str, 1:1) == """
+        function f(a, b)
+            return a+b
+         end
+        """
+    @test format_lines(str, 2:2) == """
+        function f(a,b)
+            return a + b
+         end
+        """
+    @test format_lines(str, 3:3) == """
+        function f(a,b)
+            return a+b
+        end
+        """
+    @test format_lines(str, [1:1, 3:3]) == """
+        function f(a, b)
+            return a+b
+        end
+        """
+    @test format_lines(str, [1:1, 2:2, 3:3]) == """
+        function f(a, b)
+            return a + b
+        end
+        """
+    @test format_lines(str, [1:2]) == """
+        function f(a, b)
+            return a + b
+         end
+        """
+end
+
 module RunicMain1
     using Test: @testset
     using Runic: main