build: add --fix option for check-whitespace

AGENTS.md

@@ -19,7 +19,7 @@ will not be reflected, unless you use `Revise`.
 
 ## For all changes
 
-1. Run `make check-whitespace` before creating the PR to make sure you're not committing any whitespace errors.
+1. Run `make fix-whitespace` before creating the PR to make sure you're not committing any whitespace errors.
 
 ## Building Julia
 

Makefile

@@ -136,6 +136,15 @@ else
 	$(warn "Skipping whitespace check because git is unavailable")
 endif
 
+fix-whitespace:
+ifneq ($(NO_GIT), 1)
+	@# Append the directory containing the julia we just built to the end of `PATH`,
+	@# to give us the best chance of being able to run this check.
+	@PATH="$(PATH):$(dir $(JULIA_EXECUTABLE))" julia $(call cygpath_w,$(JULIAHOME)/contrib/check-whitespace.jl) --fix
+else
+	$(warn "Skipping whitespace fix because git is unavailable")
+endif
+
 release-candidate: release testall
 	@$(JULIA_EXECUTABLE) $(JULIAHOME)/contrib/add_license_to_files.jl #add license headers
 	@#Check documentation
@@ -686,7 +695,7 @@ distcleanall: cleanall
 	@-$(MAKE) -C $(BUILDROOT)/doc cleanall
 
 .FORCE:
-.PHONY: .FORCE default debug release check-whitespace release-candidate \
+.PHONY: .FORCE default debug release check-whitespace fix-whitespace release-candidate \
 	julia-debug julia-release julia-stdlib julia-deps julia-deps-libs \
 	julia-cli-release julia-cli-debug julia-src-release julia-src-debug \
 	julia-symlink julia-base julia-sysimg julia-sysimg-ji julia-sysimg-release julia-sysimg-debug \

contrib/check-whitespace.jl

@@ -32,10 +32,39 @@ allow_tabs(path) =
     endswith(path, "test/syntax.jl") ||
     endswith(path, "test/triplequote.jl")
 
-const errors = Set{Tuple{String,Int,String}}()
-
 function check_whitespace()
-    for path in eachline(`git ls-files -- $patterns`)
+    # Get file list from ARGS if provided, otherwise use git ls-files
+    errors = Set{Tuple{String,Int,String}}()
+    files_to_check = filter(arg -> arg != "--fix", ARGS)
+    if isempty(files_to_check)
+        files_to_check = eachline(`git ls-files -- $patterns`)
+    end
+
+    files_fixed = 0
+    if "--fix" in ARGS
+        for path in files_to_check
+            content = newcontent = read(path, String)
+            isempty(content) && continue
+            if !allow_tabs(path)
+                tabpattern = r"^([ \t]+)"m => (x -> replace(x, r"((?: {4})*)( *\t)" => s"\1    ")) # Replace tab sequences at start of line after any number of 4-space groups
+                newcontent = replace(newcontent, tabpattern)
+            end
+            newcontent = replace(newcontent,
+                r"\s*$" => '\n',                # Remove trailing whitespace and normalize line ending at eof
+                r"\s*?[\r\n]" => '\n',          # Remove trailing whitespace and normalize line endings on each line
+                r"\xa0" => ' '                  # Replace non-breaking spaces
+            )
+            if content != newcontent
+                write(path, newcontent)
+                files_fixed += 1
+            end
+        end
+        if files_fixed > 0
+            println(stderr, "Fixed whitespace issues in $files_fixed files.")
+        end
+    end
+
+    for path in files_to_check
         lineno = 0
         non_blank = 0