insert-example-code.py: Specify file encoding

kjellahl · kjellahl · commit d9319e4724a5 · 2021-07-16T15:21:42.000+02:00
The default file encoding is platform dependent in Python.
Better tell which encoding is expected.
diff --git a/docs/manual/insert_example_code.py b/docs/manual/insert_example_code.py
@@ -19,7 +19,7 @@
 
 def process_source_file(source_directory, source_basename, outfile):
   source_filename = os.path.join(source_directory, source_basename)
-  with open(source_filename, mode='r') as srcfile:
+  with open(source_filename, mode='r', encoding='utf-8', errors='surrogateescape') as srcfile:
     outfile.write('<para>File: <filename>' + source_basename + '</filename></para>\n')
     outfile.write('<programlisting>\n<![CDATA[')
 
@@ -38,9 +38,13 @@ def insert_example_code(examples_base_dir, input_xml_files, output_xml_file):
   if not isinstance(input_xml_files, list):
     input_xml_files = [input_xml_files]
 
-  with open(output_xml_file, mode='w') as outfile:
+  # Assume that all files are UTF-8 encoded.
+  # If illegal UTF-8 bytes in the range 0x80..0xff are encountered, they are
+  # replaced by Unicode Private Use characters in the range 0xdc80..0xdcff
+  # and restored to their original values when the file is rewritten.
+  with open(output_xml_file, mode='w', encoding='utf-8', errors='surrogateescape') as outfile:
     for input_xml_file in input_xml_files:
-      with open(input_xml_file, mode='r') as infile:
+      with open(input_xml_file, mode='r', encoding='utf-8', errors='surrogateescape') as infile:
         for line in infile:
           # Print the line.
           outfile.write(line)
diff --git a/tools/build_scripts/tutorial-custom-cmd.py b/tools/build_scripts/tutorial-custom-cmd.py
@@ -13,8 +13,8 @@
 subcommand = sys.argv[1]
 
 def insert_ex_code():
-  #      argv[2]        argv[3]         argv[4]         argv[5]
-  # <py_script_file> <examples_dir> <input_xml_file> <output_xml_file>
+  #      argv[2]       argv[3]         argv[4]         argv[5]
+  # <py_script_dir> <examples_dir> <input_xml_file> <output_xml_file>
 
   # Search for insert_example_code.py first in <py_script_dir>.
   sys.path.insert(0, sys.argv[2])
