hackc: backslash escape extension

3 files changed, 19 insertions, 7 deletions

diff --git a/projects/hackc/__main__.py b/projects/hackc/__main__.py
index a7bc06f..1cb5c3f 100644
--- a/projects/hackc/__main__.py
+++ b/projects/hackc/__main__.py
@@ -4,7 +4,7 @@ import os
 from .parser import Parser
 
 
-def compile_jack(input_path: Path, verbose: bool):
+def compile_jack(input_path: Path, extensions: list, verbose: bool):
     try:
         filenames = os.listdir(input_path)
         files = [Path(input_path / f) for f in filenames]
@@ -16,7 +16,7 @@ def compile_jack(input_path: Path, verbose: bool):
         return
 
     for input_fn in jack_files:
-        parser = Parser(input_fn)
+        parser = Parser(input_fn, extensions=extensions)
         parser.tokenize()
         parser.print_tokens()
 
@@ -24,6 +24,13 @@ def compile_jack(input_path: Path, verbose: bool):
 if __name__ == "__main__":
     parser = ArgumentParser("hackc")
     parser.add_argument("-v", "--verbose", action="store_true", help="verbose mode")
+    parser.add_argument(
+        "-E",
+        "--extension",
+        metavar="extension",
+        action="append",
+        help="enable extension of standard Jack",
+    )
     parser.add_argument("input_path", help="Jack file or directory")
     args = parser.parse_args()
-    compile_jack(Path(args.input_path), args.verbose)
+    compile_jack(Path(args.input_path), args.extension or [], args.verbose)
diff --git a/projects/hackc/parser.py b/projects/hackc/parser.py
index 400b096..f73c3d1 100644
--- a/projects/hackc/parser.py
+++ b/projects/hackc/parser.py
@@ -29,8 +29,9 @@ SYMBOLS = "{}()[].,;+-*/&|<>=~"
 
 
 class Parser:
-    def __init__(self, fp):
+    def __init__(self, fp, extensions=[]):
         self._fp = fp
+        self._extensions = extensions
         self.tokens = []
 
     def print_tokens(self):
@@ -80,7 +81,7 @@ class Parser:
                     break
 
                 rem = line[pos:]  # remainder of line
-                token = Token.from_line(rem, line_no, pos)
+                token = Token.from_line(rem, line_no, pos, extensions=self._extensions)
                 if token is not None:
                     self.tokens.append(token)
                     pos += token.length()
diff --git a/projects/hackc/tokens.py b/projects/hackc/tokens.py
index 7ae37ce..50c4173 100644
--- a/projects/hackc/tokens.py
+++ b/projects/hackc/tokens.py
@@ -35,7 +35,7 @@ class Token:
         self.column = column
 
     @classmethod
-    def from_line(cls, line: str, line_no: int, column: int):
+    def from_line(cls, line: str, line_no: int, column: int, extensions=[]):
         """Extract first token from line and return it as an instance of Token."""
         if not line:
             return None
@@ -47,7 +47,11 @@ class Token:
         if int_match is not None:
             return Token("integer", int_match.group(1), line_no, column)
 
-        str_match = re.match('(".*")', line)
+        if "escape" in extensions:
+            str_match = re.match(r'("(.|\\")+?[^\\]")', line)
+        else:
+            str_match = re.match('(".*?")', line)
+
         if str_match is not None:
             return Token("string", str_match.group(1), line_no, column)