diff --git a/_tools/pygments2chroma_xml.py b/_tools/pygments2chroma_xml.py
new file mode 100644
index 0000000..a824fee
--- /dev/null
+++ b/_tools/pygments2chroma_xml.py
@@ -0,0 +1,191 @@
+import functools
+import importlib
+import json
+import os
+import re
+import sys
+import types
+import html
+
+import pystache
+from pygments import lexer as pygments_lexer
+from pygments.token import _TokenType
+
+TEMPLATE = r'''
+<lexer>
+  <config>
+    <name>{{name}}</name>
+    {{#aliases}}
+    <alias>{{alias}}</alias>
+    {{/aliases}}
+    {{#filenames}}
+    <filename>{{filename}}</filename>
+    {{/filenames}}
+    {{#mimetypes}}
+    <mime_type>{{mimetype}}</mime_type>
+    {{/mimetypes}}
+    {{#re_ignorecase}}
+    <case_insensitive>true</case_insensitive>
+    {{/re_ignorecase}}
+    {{#re_dotall}}
+    <dot_all>true</dot_all>
+    {{/re_dotall}}
+    {{#re_not_multiline}}
+    <not_multiline>true</not_multiline>
+    {{/re_not_multiline}}
+  </config>
+  <rules>
+    {{#tokens}}
+    <state name="{{state}}">
+      {{#rules}}
+      {{{.}}}
+      {{/rules}}
+    </state>
+    {{/tokens}}
+  </rules>
+</lexer>
+'''
+
+
+def xml_regex(s):
+    return xml_string(s)
+
+def xml_string(s):
+    s = html.escape(s)
+    return '"' + s + '"'
+
+
+def to_camel_case(snake_str):
+    components = snake_str.split('_')
+    return ''.join(x.title() for x in components)
+
+
+def warning(message):
+    print('warning: ' + message, file=sys.stderr)
+
+
+def resolve_emitter(emitter):
+    if isinstance(emitter, types.FunctionType):
+        if repr(emitter).startswith('<function bygroups.'):
+            args = emitter.__closure__[0].cell_contents
+            emitter = '<bygroups>%s</bygroups>' % ''.join(resolve_emitter(e) for e in args)
+        elif repr(emitter).startswith('<function using.'):
+            args = emitter.__closure__[0].cell_contents
+            if isinstance(args, dict):
+                state = 'root'
+                if 'stack' in args:
+                    state = args['stack'][1]
+                    args.pop('stack')
+                assert args == {}, args
+                emitter = '<usingself state="%s"/>' % state
+            elif issubclass(args, pygments_lexer.Lexer):
+                name = args.__name__
+                if name.endswith('Lexer'):
+                    name = name[:-5]
+                emitter = '<using state="%s"/>' % state
+            else:
+                raise ValueError('only support "using" with lexer classes, not %r' % args)
+        else:
+            warning('unsupported emitter function %r' % emitter)
+            emitter = '?? %r ??' % emitter
+    elif isinstance(emitter, _TokenType):
+        emitter = '<token type="%s"/>' % str(emitter).replace('.', '')[5:]
+    elif emitter is None:
+        return 'None'
+    else:
+        raise ValueError('unsupported emitter type %r' % emitter)
+    assert isinstance(emitter, str)
+    return emitter
+
+
+def process_state_action(action):
+    if isinstance(action, tuple):
+        return functools.reduce(lambda a, b: a + b, (process_state_action(a) for a in action))
+    if action.startswith('#'):
+        action = action[1:]
+        if action== 'pop':
+            action = '<pop depth="1"/>'
+        elif action.startswith('pop:'):
+            action = '<pop depth="%s"/>' % action[4:]
+        elif action == 'push':
+            action = '<push/>'
+        elif action.startswith('push:'):
+            action = '<push state="%s"/>' % action[5:]
+        else:
+            raise ValueError('unsupported action %r' % (action,))
+    else:
+        action = '<push state="%s"/>' % action
+    return (action,)
+
+
+def translate_rules(rules):
+    out = []
+    for rule in rules:
+        if isinstance(rule, tuple):
+            regex = rule[0]
+            if isinstance(regex, str):
+                regex = xml_regex(regex)
+            elif isinstance(regex, pygments_lexer.words):
+                regex = xml_string('%s(%s)%s' % (regex.prefix,
+                                      '|'.join(w for w in regex.words),
+                                      regex.suffix))
+            else:
+                raise ValueError('expected regex string but got %r' % regex)
+            emitter = resolve_emitter(rule[1])
+            if len(rule) == 2:
+                modifier = ''
+            elif type(rule[2]) is str:
+                modifier = process_state_action(rule[2])[0]
+            elif isinstance(rule[2], pygments_lexer.combined):
+                modifier = '<combined state="%s"/>' % '" state="'.join(rule[2])
+            elif type(rule[2]) is tuple:
+                modifier = '<push state="%s"/>' % '" state="'.join(rule[2])
+            else:
+                raise ValueError('unsupported modifier %r' % (rule[2],))
+            out.append('<rule pattern={}>{}{}</rule>'.format(regex, emitter, modifier))
+        elif isinstance(rule, pygments_lexer.include):
+            out.append('<rule><include state="{}"/></rule>'.format(rule))
+        elif isinstance(rule, pygments_lexer.default):
+            process_state_action(rule.state)
+            out.append('<rule>{}</rule>'.format(''.join(process_state_action(rule.state))))
+        else:
+            raise ValueError('unsupported rule %r' % (rule,))
+    return out
+
+
+class TemplateView(object):
+    def __init__(self, **kwargs):
+        for key, value in kwargs.items():
+            setattr(self, key, value)
+
+    def re_not_multiline(self):
+        return not (self.regex_flags & re.MULTILINE)
+
+    def re_dotall(self):
+        return self.regex_flags & re.DOTALL
+
+    def re_ignorecase(self):
+        return self.regex_flags & re.IGNORECASE
+
+
+def main():
+    package_name, symbol_name = sys.argv[1].rsplit(sep=".", maxsplit=1)
+
+    package = importlib.import_module(package_name)
+
+    lexer_cls = getattr(package, symbol_name)
+
+    assert issubclass(lexer_cls, pygments_lexer.RegexLexer), 'can only translate from RegexLexer'
+
+    print(pystache.render(TEMPLATE, TemplateView(
+        name=lexer_cls.name,
+        regex_flags=lexer_cls.flags,
+        aliases=[{'alias': alias} for alias in lexer_cls.aliases],
+        filenames=[{'filename': filename} for filename in lexer_cls.filenames],
+        mimetypes=[{'mimetype': mimetype} for mimetype in lexer_cls.mimetypes],
+        tokens=[{'state': state, 'rules': translate_rules(rules)} for (state, rules) in lexer_cls.get_tokendefs().items()],
+    )))
+
+
+if __name__ == '__main__':
+    main()