Merge pull request #41 from goodmami/fix-38-pack-mach-diff

Don't reuse instructions in repetitions

CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased][unreleased]
 
+### Fixed
+
+* Repetitions with mincount > 0 no longer reuse the instruction object ([#38])
+
 
 ## [v0.5.0][]
 
@@ -164,3 +168,4 @@ descent parser and a work-in-progress state-machine parser.
 [#21]: https://github.com/goodmami/pe/issues/21
 [#23]: https://github.com/goodmami/pe/issues/23
 [#31]: https://github.com/goodmami/pe/issues/31
+[#38]: https://github.com/goodmami/pe/issues/38

pe/_cy_machine.pyx

@@ -123,6 +123,17 @@ cdef class Instruction:
         self.action = action
         self.name = name
 
+    def copy(self) -> Instruction:
+        return Instruction(
+            self.opcode,
+            self.oploc,
+            scanner=self.scanner,
+            marking=self.marking,
+            capturing=self.capturing,
+            action=self.action,
+            name=self.name,
+        )
+
 
 _Program = List[Instruction]
 
@@ -402,7 +413,7 @@ def _rpt(defn, mincount):
                             marking=pi.marking,
                             capturing=pi.capturing,
                             action=pi.action)]
-    return [*(pis * mincount),
+    return [*(pi.copy() for _ in range(mincount) for pi in pis),
             Instruction(BRANCH, len(pis) + 2),
             *pis,
             Instruction(UPDATE, -len(pis))]

test/test_regression.py

@@ -27,3 +27,16 @@ def test_multi_range_charclass(parser):
     assert p.match('"1"') is not None
     assert p.match('"\U0010ffff"') is not None
     assert p.match('"a1"') is not None
+
+
+@pytest.mark.parametrize('parser', ['packrat', 'machine', 'machine-python'])
+def test_capture_repeated(parser):
+    m1 = pe.match('"a"+', 'aaa', parser=parser)
+    assert m1.group() == 'aaa'
+    assert m1.groups() == ()
+    m2 = pe.match('~"a"+', 'aaa', parser=parser)
+    assert m2.group() == 'aaa'
+    assert m2.groups() == ('aaa',)
+    m3 = pe.match('(~"a")+', 'aaa', parser=parser)
+    assert m3.group() == 'aaa'
+    assert m3.groups() == ('a', 'a', 'a')