Fix handling of version files that contain CRLFs

From looking at metrics, I noticed that the `cat --show-nonprinting`
approach added in #1783 to aid the visualisation of invisible characters
(such as ASCII control characters) also escapes carriage return
characters (to `^M`), which we don't want.

Instead, `sed` is now used which replaces anything not matching the
`:print:` and `:space:` groups with the Unicode substitution character
(`�`) - which means the classic buildpack now also matches the CNB's
behaviour:
https://github.com/heroku/buildpacks-python/blob/f2fdd00edf0f63b298cf88c82377885c88666440/src/python_version_file.rs#L16-L19

For a mapping of what `:print:` and `:space:` cover, see the chart at
the bottom of this page:
https://en.cppreference.com/w/cpp/string/byte/isprint

GUS-W-18225347.

Author: edmorley

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## [Unreleased]
 
+- Fix parsing of `runtime.txt` and `.python-version` files that contain CRLF characters. ([#1789](https://github.com/heroku/heroku-buildpack-python/pull/1789))
 
 ## [v283] - 2025-05-06
 

lib/python_version.sh

@@ -52,15 +52,15 @@ function python_version::read_requested_python_version() {
 
 	local runtime_txt_path="${build_dir}/runtime.txt"
 	if [[ -f "${runtime_txt_path}" ]]; then
-		contents="$(cat --show-nonprinting "${runtime_txt_path}")"
+		contents="$(utils::read_file_with_special_chars_substituted "${runtime_txt_path}")"
 		version="$(python_version::parse_runtime_txt "${contents}")"
 		origin="runtime.txt"
 		return 0
 	fi
 
 	local python_version_file_path="${build_dir}/.python-version"
 	if [[ -f "${python_version_file_path}" ]]; then
-		contents="$(cat --show-nonprinting "${python_version_file_path}")"
+		contents="$(utils::read_file_with_special_chars_substituted "${python_version_file_path}")"
 		version="$(python_version::parse_python_version_file "${contents}")"
 		origin=".python-version"
 		return 0

lib/utils.sh

@@ -16,6 +16,15 @@ function utils::get_requirement_version() {
 	echo "${requirement_version}"
 }
 
+# Replaces any invisible unwanted characters (such as ASCII control codes) with the Unicode
+# replacement character, so they are visible in error messages. Also removes any carriage
+# return characters, to prevent them interfering with the rendering of error messages that
+# include the raw file contents.
+function utils::read_file_with_special_chars_substituted() {
+	local file="${1}"
+	sed --regexp-extended --expression 's/[^[:print:][:space:]]/�/g' --expression 's/\r$//' "${file}"
+}
+
 # Python bundles pip within its standard library, which we can use to install our chosen
 # pip version from PyPI, saving us from having to download the usual pip bootstrap script.
 function utils::bundled_pip_module_path() {

spec/fixtures/python_version_file_invalid_version/.python-version

@@ -322,7 +322,7 @@
           remote:  !     isn't in the correct format.
           remote:  !     
           remote:  !     The following version was found:
-          remote:  !       3.12.0^[  
+          remote:  !       �3.�12.0�  
           remote:  !     
           remote:  !     However, the Python version must be specified as either:
           remote:  !     1. The major version only, for example: #{DEFAULT_PYTHON_MAJOR_VERSION} (recommended)
@@ -522,7 +522,7 @@
           remote:  !     in the correct format.
           remote:  !     
           remote:  !     The following file contents were found, which aren't valid:
-          remote:  !     python-3.12.0^[
+          remote:  !     python-3.12.0�
           remote:  !     
           remote:  !     However, the runtime.txt file is deprecated since it has been
           remote:  !     replaced by the more widely supported .python-version file: