Add support for regex replacement backreferences

README.rst

@@ -129,6 +129,9 @@ Examples::
     # Case insensitive global replacement
     value =~ s/foo/bar/gi
 
+    # Backreferences
+    value =~ s/(.+?) (?<name>.+?)/$1 $name/
+
 
 Dollar variables
 ----------------

perl/translator.py

@@ -1,4 +1,5 @@
 import io
+import re
 import tokenize
 from enum import Enum
 
@@ -128,10 +129,10 @@ def translate(self, readline):
                 continue
 
             #
-            # Convert $1 (as long as we're not collecting)
+            # Convert $1 (as long as we're not in a regex)
             #
 
-            if self.collecting_match != CollectState.ACTIVE:
+            if self.collecting_match == CollectState.WAITING:
                 if tok.type == tokenize.ERRORTOKEN and tok.string == "$":
                     self.dollar = True
                     continue
@@ -327,7 +328,10 @@ def render(self):
             ]
 
         else:
+            # Build replace  and covert any backrefs
             replace = "".join(self.replace)
+            replace = re.sub(r"\$(\w+)", r"\\g\g<1>", replace)
+
             if self.is_global:
                 # By default the count is unlimited
                 count = ""
@@ -338,7 +342,7 @@ def render(self):
             # Regex needs to reset the vars first in case it's a None
             python = [
                 f"{whitespace}{variable} = __perl__reset_vars() or ",
-                f"re.sub(r'{match}', '{replace}', {variable}",
+                f"re.sub(r'{match}', r'{replace}', {variable}",
                 f"{count}{flags})",
             ]
 

tests/test_translator.py → tests/test_translator/match.py

@@ -14,20 +14,6 @@ def test_translate__match_all():
     )
 
 
-def test_translate__replace():
-    assert (
-        translate_string("var =~ s/foo/bar/")
-        == "var = __perl__reset_vars() or re.sub(r'foo', 'bar', var, count=1)"
-    )
-
-
-def test_translate__replace_all():
-    assert (
-        translate_string("var =~ s/foo/bar/g")
-        == "var = __perl__reset_vars() or re.sub(r'foo', 'bar', var)"
-    )
-
-
 def test_translate__escaped():
     assert (
         translate_string(r"var =~ /foo\/bar/")
@@ -74,3 +60,31 @@ def test_translate__match_assign():
         translate_string(r"match = var =~ /foo/")
         == "match = __perl__re_match(re.search(r'foo', var))"
     )
+
+
+def test_translate__replace():
+    assert (
+        translate_string("var =~ s/foo/bar/")
+        == "var = __perl__reset_vars() or re.sub(r'foo', r'bar', var, count=1)"
+    )
+
+
+def test_translate__replace_with_backref():
+    assert (
+        translate_string("var =~ s/^foo (.+?) bar/foo $1 bar/")
+        == "var = __perl__reset_vars() or re.sub(r'^foo (.+?)', r'foo \\g<1> bar')"
+    )
+
+
+def test_translate__replace_with_named_backref():
+    assert translate_string("var =~ s/^foo (?P<named>.+?) bar/foo $named bar/") == (
+        "var = __perl__reset_vars() or "
+        "re.sub(r'^foo (?P<named>.+?) bar', r'foo \\g<named> bar')"
+    )
+
+
+def test_translate__replace_all():
+    assert (
+        translate_string("var =~ s/foo/bar/g")
+        == "var = __perl__reset_vars() or re.sub(r'foo', r'bar', var)"
+    )

tests/test_translator/replace.py

@@ -0,0 +1,29 @@
+from perl.translator import translate_string
+
+
+def test_translate__replace():
+    assert (
+        translate_string("var =~ s/foo/bar/")
+        == "var = __perl__reset_vars() or re.sub(r'foo', r'bar', var, count=1)"
+    )
+
+
+def test_translate__replace_with_backref():
+    assert (
+        translate_string("var =~ s/^foo (.+?) bar/foo $1 bar/")
+        == "var = __perl__reset_vars() or re.sub(r'^foo (.+?)', r'foo \\g<1> bar')"
+    )
+
+
+def test_translate__replace_with_named_backref():
+    assert translate_string("var =~ s/^foo (?P<named>.+?) bar/foo $named bar/") == (
+        "var = __perl__reset_vars() or "
+        "re.sub(r'^foo (?P<named>.+?) bar', r'foo \\g<named> bar')"
+    )
+
+
+def test_translate__replace_all():
+    assert (
+        translate_string("var =~ s/foo/bar/g")
+        == "var = __perl__reset_vars() or re.sub(r'foo', r'bar', var)"
+    )