Added optional compression to standalone parser

lark/tools/standalone.py

@@ -123,10 +123,19 @@ def main(fobj, start, print=print):
     lark_inst = Lark(fobj, parser="lalr", lexer="contextual", start=start)
     gen_standalone(lark_inst, print)
 
-def gen_standalone(lark_inst, output=None, out=sys.stdout):
+def gen_standalone(lark_inst, output=None, out=sys.stdout, compress=False):
     if output is None:
         output = partial(print, file=out)
 
+    import pickle, zlib, base64
+    def compressed_output(obj):
+        s = pickle.dumps(obj, pickle.HIGHEST_PROTOCOL)
+        c = zlib.compress(s)
+        output(repr(base64.b85encode(c)))
+
+    def output_decompress(name):
+        output('%(name)s = pickle.loads(zlib.decompress(base64.b85decode(%(name)s)))' % locals())
+
     output('# The file was automatically generated by Lark v%s' % lark.__version__)
     output('__version__ = "%s"' % lark.__version__)
     output()
@@ -139,13 +148,23 @@ def gen_standalone(lark_inst, output=None, out=sys.stdout):
             output(code)
 
     data, m = lark_inst.memo_serialize([TerminalDef, Rule])
-    output('DATA = (')
-    # pprint(data, width=160)
-    output(data)
-    output(')')
-    output('MEMO = (')
-    output(m)
-    output(')')
+    output('import pickle, zlib, base64')
+    if compress:
+        output('DATA = (')
+        compressed_output(data)
+        output(')')
+        output_decompress('DATA')
+        output('MEMO = (')
+        compressed_output(m)
+        output(')')
+        output_decompress('MEMO')
+    else:
+        output('DATA = (')
+        output(data)
+        output(')')
+        output('MEMO = (')
+        output(m)
+        output(')')
 
 
     output('Shift = 0')
@@ -156,16 +175,19 @@ def gen_standalone(lark_inst, output=None, out=sys.stdout):
 
 
 
-if __name__ == '__main__':
+def main():
     make_warnings_comments()
     parser = ArgumentParser(prog="prog='python -m lark.tools.standalone'", description="Lark Stand-alone Generator Tool",
                             parents=[lalr_argparser], epilog='Look at the Lark documentation for more info on the options')
     parser.add_argument("old_start", nargs='?', help=SUPPRESS)
+    parser.add_argument('-c', '--compress', action='store_true', default=0, help="Enable compression")
     ns = parser.parse_args()
     if ns.old_start is not None:
         warn('The syntax `python -m lark.tools.standalone <grammar-file> <start>` is deprecated. Use the -s option')
         ns.start.append(ns.old_start)
-    
+
     lark_inst, out = build_lalr(ns)
-    gen_standalone(lark_inst, out=out)
-    
+    gen_standalone(lark_inst, out=out, compress=ns.compress)
+
+if __name__ == '__main__':
+    main()