Fix handling of false-positive Unicode escapes in rxing-cli raw output

Author: dlenski

test/test_all.py

@@ -160,14 +160,14 @@ def test_rxing_parsing():
     stdout = (r"""
 [Barcode Format] qrcode
 [Points] [PointT { x: -0.123, y: 456 }, PointT { x: 1.5, y: 1.5 }, PointT { x: 2.5, y: -2.5 }, PointT { x: -0, y: 0.0 }]
-[Data] \u{a1}Atenci\u{f3}n
+[Data] \u{a1}Atenci\u{f3}n, \\u{f00} is not a Unicode escape
 """)
     dec = zxing.BarCode.parse_rxing(stdout, '/tmp/test.png')
     assert dec.uri == 'file:///tmp/test.png'
     assert dec.path == '/tmp/test.png'
     assert dec.format == 'QR_CODE'
     assert dec.type == 'TEXT'
-    assert dec.raw == dec.parsed == '\u00a1Atenci\u00f3n'
+    assert dec.raw == dec.parsed == '\u00a1Atenci\u00f3n, \\u{f00} is not a Unicode escape'
     assert dec.raw_bits is None
     assert dec.points == [(-0.123, 456.0), (1.5, 1.5), (2.5, -2.5), (0.0, 0.0)]
     r = repr(dec)
@@ -190,15 +190,6 @@ def test_rxing_parsing_not_found():
     assert r.startswith('BarCode(') and r.endswith(')')
 
 
-def test_rxing_parsing_WRONG():
-    # FIXME: This is wrong, should not be parsed as a Unicode escape
-    stdout = (r"""
-[Data] 00\\u{ff}11
-""")
-    dec = zxing.BarCode.parse_rxing(stdout, '/tmp/test.png')
-    assert dec.raw == dec.parsed == r'00\u{ff}11'
-
-
 def test_wrong_formats():
     all_test_formats = {fmt for fn, fmt, raw in test_barcodes}
     yield from ((_check_decoding, filename, expected_format, None, dict(possible_formats=all_test_formats - {expected_format}))

zxing/__init__.py

@@ -235,8 +235,17 @@ class CLROutputBlock(Enum):
     RAW_BITS = 4
 
 
+RUST_UNICODE_ESCAPE = re.compile(r'(\\+)u\{([a-f0-9]+)\}')
+def _unrustify_repr(rust_repr):
+    # Rust's repr of a Unicode string can include backslash escapes, which are mostly 
+    # compatible with Python's. Unlike Python, however, its Unicode character escapes
+    # are not always 4 hex digits, and have braces (e.g. '\u{123}')
+    # To handle the case of an escaped '\\' followed by 'u{...}', we ignore even
+    # numbers of backslashes.
+    unrust = RUST_UNICODE_ESCAPE.sub(lambda m: r'\u' + m.group(2).rjust(4, '0') if len(m.group(1)) % 2 else m.group(0), rust_repr)
+    return unrust.encode().decode('unicode_escape')
+
 class BarCode(object):
-    RUST_UNICODE_ESCAPE = re.compile(r'\\u\{([a-f0-9]+)\}')
     POINTS = re.compile(r'PointT\s*\{\s*x\:\s*([-\d.]+),\s*y\:\s*([-\d.]+)\s*\}')
 
     RXING_FORMAT_TO_ZXING = {'QRCODE': 'QR_CODE'}
@@ -260,11 +269,8 @@ def parse_rxing(cls, rxing_output, fn):
                 format = l.removeprefix('[Barcode Format] ').replace(' ', '_').upper()
                 format = cls.RXING_FORMAT_TO_ZXING.get(format, format)
             elif l.startswith('[Data] '):
-                # This is a Rust repr of a string, include backslash escapes. Unlike Python, its
-                # unicode character escapes are not always 4 digits, and have braces (e.g. '\u{123}')
-                # FIXME: This incorrectly handles the case of an escaped '\\' followed by 'u{...}'
-                raw = l.removeprefix('[Data] ')
-                raw = cls.RUST_UNICODE_ESCAPE.sub(lambda m: r'\u' + m.group(1).rjust(4, '0'), raw).encode().decode('unicode_escape')
+                # This is a Rust repr of a string, which is almost-but-not-quite handled by bytes.decode('unicode_escape')
+                raw = _unrustify_repr(l.removeprefix('[Data] '))
             elif l.startswith('[Points] '):
                 points = [((float(m[0]), float(m[1]))) for m in cls.POINTS.findall(l.removeprefix('[Points] '))]