From f808a6a1fca5d82bffc6c540cdd263c5661c20ab Mon Sep 17 00:00:00 2001
From: c42f <chris42f@gmail.com>
Date: Wed, 23 Nov 2022 16:11:08 +1000
Subject: [PATCH] Fix source line reporting at EOF

---
 src/source_files.jl  |  9 +++++----
 test/runtests.jl     |  6 ++----
 test/source_files.jl | 12 ++++++++++++
 test/test_utils.jl   |  1 +
 4 files changed, 20 insertions(+), 8 deletions(-)
 create mode 100644 test/source_files.jl

diff --git a/src/source_files.jl b/src/source_files.jl
index fa95794a..bff823f9 100644
--- a/src/source_files.jl
+++ b/src/source_files.jl
@@ -23,7 +23,7 @@ function SourceFile(code::AbstractString; filename=nothing)
         code[i] == '\n' && push!(line_starts, i+1)
     end
     if isempty(code) || last(code) != '\n'
-        push!(line_starts, lastindex(code)+1)
+        push!(line_starts, ncodeunits(code)+1)
     end
     SourceFile(code, filename, line_starts)
 end
@@ -34,14 +34,15 @@ end
 
 # Get line number of the given byte within the code
 function source_line(source::SourceFile, byte_index)
-    searchsortedlast(source.line_starts, byte_index)
+    line = searchsortedlast(source.line_starts, byte_index)
+    return (line < lastindex(source.line_starts)) ? line : line-1
 end
 
 """
 Get line number and character within the line at the given byte index.
 """
 function source_location(source::SourceFile, byte_index)
-    line = searchsortedlast(source.line_starts, byte_index)
+    line = source_line(source, byte_index)
     i = source.line_starts[line]
     column = 1
     while i < byte_index
@@ -57,7 +58,7 @@ Get byte range of the source line at byte_index, buffered by
 """
 function source_line_range(source::SourceFile, byte_index;
                            context_lines_before=0, context_lines_after=0)
-    line = searchsortedlast(source.line_starts, byte_index)
+    line = source_line(source, byte_index)
     fbyte = source.line_starts[max(line-context_lines_before, 1)]
     lbyte = source.line_starts[min(line+1+context_lines_after, end)] - 1
     fbyte,lbyte
diff --git a/test/runtests.jl b/test/runtests.jl
index b08da533..fad4278d 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -17,12 +17,10 @@ include("parser.jl")
 include("diagnostics.jl")
 include("parser_api.jl")
 include("expr.jl")
-@testset "Parsing values from strings" begin
+@testset "Parsing literals from strings" begin
     include("value_parsing.jl")
 end
 include("hooks.jl")
 include("parse_packages.jl")
+include("source_files.jl")
 
-# Prototypes
-#include("syntax_interpolation.jl")
-#include("simple_parser.jl")
diff --git a/test/source_files.jl b/test/source_files.jl
new file mode 100644
index 00000000..99071926
--- /dev/null
+++ b/test/source_files.jl
@@ -0,0 +1,12 @@
+@testset "SourceFile lines and column indexing" begin
+    @test source_location(SourceFile("a"), 1) == (1,1)
+    @test source_location(SourceFile("a"), 2) == (1,2)
+
+    @test source_location(SourceFile("a\n"), 2) == (1,2)
+    @test source_location(SourceFile("a\n"), 3) == (1,3)
+
+    @test source_location(SourceFile("a\nb\n"), 2) == (1,2)
+    @test source_location(SourceFile("a\nb\n"), 3) == (2,1)
+    @test source_location(SourceFile("a\nb\n"), 4) == (2,2)
+    @test source_location(SourceFile("a\nb\n"), 5) == (2,3)
+end
diff --git a/test/test_utils.jl b/test/test_utils.jl
index 1cd77284..03bda337 100644
--- a/test/test_utils.jl
+++ b/test/test_utils.jl
@@ -7,6 +7,7 @@ using .JuliaSyntax:
     ParseState,
     Diagnostic,
     SourceFile,
+    source_location,
     parse!,
     parse,
     parseall,