Enforce utf8 when opening files in python scripts (elastic#16376)

Since the migration to Python 3 (elastic#14798) these scripts fail in environments
not configured with unicode locales with this error. We saw this in the new
jenkins pipelines, and can be reproduced at least on Linux running make check
with LANG=C.

(cherry picked from commit c2f6358)

dev-tools/generate_notice.py

@@ -15,13 +15,8 @@ def read_file(filename):
         print("File not found {}".format(filename))
         return ""
 
-    try:
-        with open(filename, 'r') as f:
-            return f.read()
-    except UnicodeDecodeError:
-        # try latin-1
-        with open(filename, 'r', encoding="ISO-8859-1") as f:
-            return f.read()
+    with open(filename, 'r', encoding='utf_8') as f:
+        return f.read()
 
 
 def get_library_path(license):
@@ -37,7 +32,7 @@ def get_library_path(license):
 
 def read_versions(vendor):
     libs = []
-    with open(os.path.join(vendor, "vendor.json")) as f:
+    with open(os.path.join(vendor, "vendor.json"), encoding='utf_8') as f:
         govendor = json.load(f)
         for package in govendor["package"]:
             libs.append(package)
@@ -222,11 +217,11 @@ def get_url(repo):
 def create_notice(filename, beat, copyright, vendor_dirs, csvfile, overrides=None):
     dependencies = gather_dependencies(vendor_dirs, overrides=overrides)
     if not csvfile:
-        with open(filename, "w+") as f:
+        with open(filename, "w+", encoding='utf_8') as f:
             write_notice_file(f, beat, copyright, dependencies)
             print("Available at {}".format(filename))
     else:
-        with open(csvfile, "wb") as f:
+        with open(csvfile, "wb", encoding='utf_8') as f:
             csvwriter = csv.writer(f)
             write_csv_file(csvwriter, dependencies)
             print("Available at {}".format(csvfile))

filebeat/scripts/docs_collector.py

@@ -38,13 +38,13 @@ def collect(beat_name):
         module_file = generated_note
         module_file += "[[filebeat-module-" + module + "]]\n"
 
-        with open(module_doc) as f:
+        with open(module_doc, encoding='utf_8') as f:
             module_file += f.read()
 
         beat_path = os.path.join(module_dir, "_meta")
 
         # Load title from fields.yml
-        with open(beat_path + "/fields.yml") as f:
+        with open(beat_path + "/fields.yml", encoding='utf_8') as f:
             fields = yaml.load(f.read())
             title = fields[0]["title"]
 
@@ -61,8 +61,8 @@ def collect(beat_name):
 """
 
         # Write module docs
-        with open(os.path.abspath("docs") + "/modules/" +
-                  module + ".asciidoc", 'w') as f:
+        docs_path = os.path.join(os.path.abspath("docs"), "modules",  module + ".asciidoc")
+        with open(docs_path, 'w', encoding='utf_8') as f:
             f.write(module_file)
 
     module_list_output = generated_note
@@ -76,7 +76,7 @@ def collect(beat_name):
         module_list_output += "include::modules/" + m + ".asciidoc[]\n"
 
     # Write module link list
-    with open(os.path.abspath("docs") + "/modules_list.asciidoc", 'w') as f:
+    with open(os.path.abspath("docs") + "/modules_list.asciidoc", 'w', encoding='utf_8') as f:
         f.write(module_list_output)
 
 

libbeat/scripts/unpack_dashboards.py

@@ -27,7 +27,7 @@ def transform_data(data, method):
 
 
 def transform_file(path, method):
-    with open(path) as f:
+    with open(path, encoding='utf_8') as f:
         data = json.load(f)
 
     transform_data(data, method)
@@ -52,5 +52,5 @@ def transform_file(path, method):
         data = transform_file(path, method)
         new_data = json.dumps(data, sort_keys=True, indent=4)
 
-        with open(path, 'w') as f:
+        with open(path, 'w', encoding='utf_8') as f:
             f.write(new_data)