From b56f3c803573973fa8d754356ba7f074661aea46 Mon Sep 17 00:00:00 2001
From: Pascal Thomet <pthomet@gmail.com>
Date: Fri, 29 Nov 2024 16:18:23 +0100
Subject: [PATCH] Fix standard_value_replacements() for hexadecimal numbers 
 (cf #24)

---
 src/litgen/internal/cpp_to_python.py          | 25 +++----
 .../tests/internal/cpp_to_python_test.py      | 68 ++++++++++++++++---
 2 files changed, 69 insertions(+), 24 deletions(-)

diff --git a/src/litgen/internal/cpp_to_python.py b/src/litgen/internal/cpp_to_python.py
index 2bc57625..bfac1025 100644
--- a/src/litgen/internal/cpp_to_python.py
+++ b/src/litgen/internal/cpp_to_python.py
@@ -721,23 +721,16 @@ def standard_type_replacements() -> RegexReplacementList:
 
 
 def standard_value_replacements() -> RegexReplacementList:
-    """Replacements for C++ code when translating to python.
-
-    Consists mostly of
-    * number translation (e.g. `1.5f` -> `1.5`)
-    """
+    """Replacements for C++ code when translating to Python."""
     replacements_str = r"""
-
     \btrue\b -> True
     \bfalse\b -> False
     \bvoid\s*\* -> Any
     \bvoid\b -> None
     \bNULL\b -> None
     \bnullptr\b -> None
-    \std::nullopt\b -> None
-    ^std::nullopt\b$ -> None
-    \std.nullopt\b -> None
-    ^std.nullopt\b$ -> None
+    \bstd::nullopt\b -> None
+    \bstd.nullopt\b -> None
 
     \bstd::string\(\) -> ""
     \bstd.string\(\) -> ""
@@ -749,14 +742,14 @@ def standard_value_replacements() -> RegexReplacementList:
     \bLDBL_MIN\b -> sys.float_info.min
     \bLDBL_MAX\b -> sys.float_info.max
 
-
-    ([+-]?[0-9]+([.][0-9]*)?|[.][0-9]+)(d?) -> \1
-    ([+-]?[0-9]+([.][0-9]*)?|[.][0-9]+)(f?) -> \1
-
+    ([0-9])'([0-9]) -> \1_\2
+    (?<![a-zA-Z0-9])([+-]?[0-9]+(?:\.[0-9]*)?(?:[eE][+-]?[0-9]+)?)([fFdD])(?=\W|$) -> \1
     :: -> .
     """
-    # Note: the two last regexes replace C numbers like 1.5f or 1.5d by 1.5
-
+    # Explanation of the last regexes:
+    #   ([0-9])'([0-9]) -> \1_\2 # Replace 1'000 by 1_000
+    #   (?<![a-zA-Z0-9])([+-]?[0-9]+(?:\.[0-9]*)?(?:[eE][+-]?[0-9]+)?)([fFdD])(?=\W|$) -> \1
+    #      Replace C numbers like 1.5f or 1.5d by 1.5
     replaces = RegexReplacementList.from_string(replacements_str)
     return replaces
 
diff --git a/src/litgen/tests/internal/cpp_to_python_test.py b/src/litgen/tests/internal/cpp_to_python_test.py
index 5d430c2d..b7fa6243 100644
--- a/src/litgen/tests/internal/cpp_to_python_test.py
+++ b/src/litgen/tests/internal/cpp_to_python_test.py
@@ -3,19 +3,71 @@
 from litgen.internal import cpp_to_python
 
 
-def test_standard_replacements():
+def test_opencv_replacements():
     s = "cv::Sizeounette cv::Size s = cv::Size()"
     r = litgen.opencv_replacements().apply(s)
     assert r == "cv::Sizeounette Size s = (0, 0)"
 
-    s = "a = 1.5f;"
-    r = litgen.opencv_replacements().apply(s)
-    r = litgen.standard_value_replacements().apply(s)
-    assert r == "a = 1.5;"
 
-    s = "a = -1.5d;"
-    r = litgen.standard_value_replacements().apply(s)
-    assert r == "a = -1.5;"
+def test_standard_replacements():
+    replacements = litgen.standard_value_replacements()
+
+    # Basic float replacements
+    assert replacements.apply("a = 1.5f;") == "a = 1.5;"
+    assert replacements.apply("a = -1.5d;") == "a = -1.5;"
+    assert replacements.apply("a = 2.0;") == "a = 2.0;"
+
+    # Scientific notation
+    assert replacements.apply("a = 1.5e10f;") == "a = 1.5e10;"
+    assert replacements.apply("a = -1.5E-3d;") == "a = -1.5E-3;"
+    assert replacements.apply("a = .5e2;") == "a = .5e2;"
+    assert replacements.apply("a = -1E+2f;") == "a = -1E+2;"
+
+    # Hexadecimal literals
+    assert replacements.apply("a = 0x7000000e;" ) == "a = 0x7000000e;"
+    assert replacements.apply("a = 0x7000000f;" ) == "a = 0x7000000f;"
+    assert replacements.apply("a = 0XABCD;" ) == "a = 0XABCD;"
+
+    # Octal literals
+    assert replacements.apply("a = 0755;" ) == "a = 0755;"  # Classic octal
+    assert replacements.apply("a = 0o755;" ) == "a = 0o755;"  # Modern octal
+
+    # Binary literals
+    assert replacements.apply("a = 0b1010;" ) == "a = 0b1010;"
+    assert replacements.apply("a = 0B1010;" ) == "a = 0B1010;"
+
+    # True/false replacements
+    assert replacements.apply("flag = true;") == "flag = True;"
+    assert replacements.apply("flag = false;") == "flag = False;"
+
+    # nullptr/nullopt/void replacements
+    assert replacements.apply("p = nullptr;") == "p = None;"
+    assert replacements.apply("opt = std::nullopt;") == "opt = None;"
+    assert replacements.apply("auto value = void*();") == "auto value = Any();"
+
+    # String replacements
+    assert replacements.apply("s = std::string();") == 's = "";'
+
+    # Macros
+    assert replacements.apply("min = FLT_MIN; max = FLT_MAX;") == (
+        "min = sys.float_info.min; max = sys.float_info.max;"
+    )
+
+    # Large integers
+    assert replacements.apply("a = 0xFFFFFFFF;") == "a = 0xFFFFFFFF;"
+    assert replacements.apply("a = 12345678901234567890;") == "a = 12345678901234567890;"
+
+    # Numbers with underscores
+    assert replacements.apply("a = 1'000'000;") == "a = 1_000_000;"
+
+    # Numbers with single quotes -> Converted to underscores
+    assert replacements.apply("a = 1'000'000;") == "a = 1_000_000;"
+    assert replacements.apply("a = -1'234'567;") == "a = -1_234_567;"
+    assert replacements.apply("a = 1'2'3'4;") == "a = 1_2_3_4;"
+
+    # Verify valid Python syntax is preserved
+    assert replacements.apply("a = 1_000_000;") == "a = 1_000_000;"
+    assert replacements.apply("a = -1_234_567;") == "a = -1_234_567;"
 
 
 def test_std_array():