diff --git a/scripts/i18n_utils.py b/scripts/i18n_utils.py
new file mode 100644
--- /dev/null
+++ b/scripts/i18n_utils.py
@@ -0,0 +1,197 @@
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+from __future__ import print_function
+
+import os
+import re
+import shutil
+import subprocess
+import tempfile
+
+
+do_debug = False  # set from scripts/i18n --debug
+
+def debug(*args, **kwargs):
+    if do_debug:
+        print(*args, **kwargs)
+
+def runcmd(cmd, *args, **kwargs):
+    debug('... Executing command: %s' % ' '.join(cmd))
+    subprocess.check_call(cmd, *args, **kwargs)
+
+header_comment_strip_re = re.compile(r'''
+    ^
+    [#][ ]Translations[ ]template[ ]for[ ]Kallithea[.] \n
+    |
+    ^
+    [#][ ]FIRST[ ]AUTHOR[ ]<EMAIL@ADDRESS>,[ ]\d+[.] \n
+    (?:[#] \n)?
+    |
+    ^
+    (?:[#] \n)?
+    [#],[ ]fuzzy \n
+    |
+    ^
+    [#][ ][#],[ ]fuzzy \n
+    ''',
+    re.MULTILINE|re.VERBOSE)
+
+header_normalize_re = re.compile(r'''
+    ^ "
+    (POT-Creation-Date|PO-Revision-Date|Last-Translator|Language-Team|X-Generator|Generated-By|Project-Id-Version):
+    [ ][^\\]*\\n
+    " \n
+    ''',
+    re.MULTILINE|re.IGNORECASE|re.VERBOSE)
+
+def _normalize_po(raw_content):
+    r"""
+    >>> print(_normalize_po(r'''
+    ... # header comment
+    ...
+    ...
+    ... # comment before header
+    ... msgid ""
+    ... msgstr "yada"
+    ... "POT-Creation-Date: 2019-05-04 21:13+0200\n"
+    ... "MIME-Version: "
+    ... "1.0\n"
+    ... "Last-Translator: Jabba"
+    ... "the Hutt\n"
+    ... "X-Generator: Weblate 1.2.3\n"
+    ...
+    ... # comment, but not in header
+    ... msgid "None"
+    ... msgstr "Ingen"
+    ...
+    ...
+    ... line 2
+    ... # third comment
+    ...
+    ... msgid "Special"
+    ... msgstr ""
+    ...
+    ... msgid "Specialist"
+    ... # odd comment
+    ... msgstr ""
+    ... "Expert"
+    ...
+    ... # crazy fuzzy auto translation by msgmerge, using foo for bar
+    ... #, fuzzy
+    ... #| msgid "some foo string"
+    ... msgid "some bar string."
+    ... msgstr "translation of foo string"
+    ...
+    ... msgid "%d minute"
+    ... msgid_plural "%d minutes"
+    ... msgstr[0] "minut"
+    ... msgstr[1] "minutter"
+    ... msgstr[2] ""
+    ...
+    ... msgid "%d year"
+    ... msgid_plural "%d years"
+    ... msgstr[0] ""
+    ... msgstr[1] ""
+    ...
+    ... # last comment
+    ... ''') + '^^^')
+    # header comment
+    <BLANKLINE>
+    <BLANKLINE>
+    # comment before header
+    <BLANKLINE>
+    msgid ""
+    msgstr "yada"
+    "MIME-Version: "
+    "1.0\n"
+    <BLANKLINE>
+    msgid "None"
+    msgstr "Ingen"
+    <BLANKLINE>
+    line 2
+    <BLANKLINE>
+    msgid "Specialist"
+    msgstr ""
+    "Expert"
+    <BLANKLINE>
+    msgid "%d minute"
+    msgid_plural "%d minutes"
+    msgstr[0] "minut"
+    msgstr[1] "minutter"
+    msgstr[2] ""
+    ^^^
+    """
+    header_start = raw_content.find('\nmsgid ""\n') + 1
+    header_end = raw_content.find('\n\n', header_start) + 1 or len(raw_content)
+    chunks = [
+        header_comment_strip_re.sub('', raw_content[0:header_start])
+            .strip(),
+        '',
+        header_normalize_re.sub('', raw_content[header_start:header_end])
+            .replace(
+                r'"Content-Type: text/plain; charset=utf-8\n"',
+                r'"Content-Type: text/plain; charset=UTF-8\n"')  # maintain msgmerge casing
+            .strip(),
+        '']  # preserve normalized header
+    # all chunks are separated by empty line
+    for raw_chunk in raw_content[header_end:].split('\n\n'):
+        if '\n#, fuzzy' in raw_chunk:  # might be like "#, fuzzy, python-format"
+            continue  # drop crazy auto translation that is worse than useless
+        # strip all comment lines from chunk
+        chunk_lines = [
+            line
+            for line in raw_chunk.splitlines()
+            if line
+            and not line.startswith('#')
+        ]
+        if not chunk_lines:
+            continue
+        # check lines starting from first msgstr, skip chunk if no translation lines
+        msgstr_i = [i for i, line in enumerate(chunk_lines) if line.startswith('msgstr')]
+        if (
+            chunk_lines[0].startswith('msgid') and
+            msgstr_i and
+            all(line.endswith(' ""') for line in chunk_lines[msgstr_i[0]:])
+        ):  # skip translation chunks that doesn't have any actual translations
+            continue
+        chunks.append('\n'.join(chunk_lines) + '\n')
+    return '\n'.join(chunks)
+
+def _normalize_po_file(po_file, merge_pot_file=None, strip=False):
+    if merge_pot_file:
+        runcmd(['msgmerge', '--width=76', '--backup=none', '--previous',
+                '--update', po_file, '-q', merge_pot_file])
+    if strip:
+        po_tmp = po_file + '.tmp'
+        with open(po_file, 'r') as src, open(po_tmp, 'w') as dest:
+            raw_content = src.read()
+            normalized_content = _normalize_po(raw_content)
+            dest.write(normalized_content)
+        os.rename(po_tmp, po_file)
+
+def _normalized_diff(file1, file2, merge_pot_file=None, strip=False):
+    # Create temporary copies of both files
+    temp1 = tempfile.NamedTemporaryFile(prefix=os.path.basename(file1))
+    temp2 = tempfile.NamedTemporaryFile(prefix=os.path.basename(file2))
+    debug('normalized_diff: %s -> %s / %s -> %s' % (file1, temp1.name, file2, temp2.name))
+    shutil.copyfile(file1, temp1.name)
+    shutil.copyfile(file2, temp2.name)
+    # Normalize them in place
+    _normalize_po_file(temp1.name, merge_pot_file=merge_pot_file, strip=strip)
+    _normalize_po_file(temp2.name, merge_pot_file=merge_pot_file, strip=strip)
+    # Now compare
+    try:
+        runcmd(['diff', '-u', temp1.name, temp2.name])
+    except subprocess.CalledProcessError as e:
+        return e.returncode