CI Errors

Author: joocer

opteryx/__version__.py

@@ -1,9 +1,9 @@
 # THIS FILE IS AUTOMATICALLY UPDATED DURING THE BUILD PROCESS
 # DO NOT EDIT THIS FILE DIRECTLY
 
-__build__ = 1935
+__build__ = 1936
 __author__ = "@joocer"
-__version__ = "0.26.2-beta.1935"
+__version__ = "0.26.2-beta.1936"
 
 # Store the version here so:
 # 1) we don't load dependencies by storing it in __init__.py

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "opteryx"
-version = "0.26.2-beta.1935"
+version = "0.26.2-beta.1936"
 description = "Query your data, where it lives"
 requires-python = '>=3.11'
 readme = {file = "README.md", content-type = "text/markdown"}

setup.py

@@ -180,7 +180,10 @@ def get_parquet_vendor_sources():
     C_COMPILE_FLAGS = ["-O3"]
     if is_mac():
         CPP_COMPILE_FLAGS += ["-std=c++17"]
-        C_COMPILE_FLAGS += ["-std=c17"]
+        # Use a C standard that is widely supported; avoid passing a C
+        # language flag that could be accidentally applied to C++ sources
+        # when the build system compiles mixed-language extensions.
+        C_COMPILE_FLAGS += ["-std=c11"]
     elif is_win():
         CPP_COMPILE_FLAGS += ["/std:c++17"]
         C_COMPILE_FLAGS += ["/std:c17"]
@@ -360,7 +363,7 @@ def make_draken_extension(module_path, source_file, depends=None, language=None,
             sources=[
                 "opteryx/third_party/tktech/csimdjson.pyx",
                 "third_party/tktech/simdjson/simdjson.cpp",
-                "third_party/tktech/simdjson/util.cpp",
+                "src/cpp/simdjson_error_shim.cpp",
             ],
             include_dirs=include_dirs + ["third_party/tktech/simdjson"],
             language="c++",
@@ -761,6 +764,45 @@ def make_draken_extension(module_path, source_file, depends=None, language=None,
             ext.language = "c++"
             ext.extra_compile_args = CPP_COMPILE_FLAGS
 
+    # Ensure extensions that include any C++ source file use the C++
+    # compile flags; some extensions earlier mistakenly received
+    # C flags which can include `-std=c17` and break when clang++ is
+    # used to compile/link the module.
+    for ext in extensions:
+        try:
+            srcs = ext.sources or []
+        except Exception:
+            srcs = []
+        # Determine if this extension has C and/or C++ sources.
+        has_cpp = any(str(s).endswith((".cpp", ".cxx", ".cc")) for s in srcs)
+        has_c = any(str(s).endswith(".c") for s in srcs)
+        # Extensions with Cython sources will generate C files at build
+        # time. Treat `.pyx` sources as C for the purposes of deciding
+        # whether to avoid passing C++-only std flags to the C compiler.
+        has_pyx = any(str(s).endswith(".pyx") for s in srcs)
+        if has_pyx:
+            has_c = True
+
+        # If the extension contains both C and C++ sources, avoid passing
+        # the C++ language standard flag (e.g. -std=c++17) to the C
+        # compiler. Distutils applies `extra_compile_args` to each
+        # compilation command, so a C compiler will reject C++-only
+        # flags. In that mixed-source case we filter out std flags but
+        # keep other safe optimizations.
+        if has_cpp:
+            if has_c:
+                def _filter_cpp_std(flags):
+                    out = []
+                    for flag in flags:
+                        if flag.startswith("-std=c++") or flag.startswith("/std:c++"):
+                            continue
+                        out.append(flag)
+                    return out
+
+                ext.extra_compile_args = _filter_cpp_std(CPP_COMPILE_FLAGS)
+            else:
+                ext.extra_compile_args = CPP_COMPILE_FLAGS
+
     extensions.append(
         Extension(
             name="opteryx.compiled.list_ops.function_definitions",

src/cpp/simdjson_error_shim.cpp

@@ -0,0 +1,90 @@
+// Small shim to provide simdjson_error_handler symbol when the vendored
+// simdjson/util.cpp is not compiled into this extension. This avoids
+// import-time undefined-symbol errors on macOS. The implementation is a
+// minimal translator mapping simdjson exceptions to Python exceptions.
+
+// Minimal shim that converts any active C++ exception into a Python exception.
+// Avoid including the full simdjson headers here to prevent symbol duplication
+// and large in-object code generation; catch std::exception to retrieve
+// the message when available.
+
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include <exception>
+
+void simdjson_error_handler() {
+    try {
+        if (PyErr_Occurred()) {
+            return; // preserve existing Python exception
+        } else {
+            throw; // rethrow the active C++ exception
+        }
+    } catch (const std::exception &e) {
+        const char *msg = e.what();
+        if (!msg) msg = "simdjson: unknown error";
+
+        // Map known simdjson error codes (present in exception message)
+        // to the appropriate Python exception types. This mirrors the
+        // behavior in the original vendored `util.cpp` without
+        // including simdjson headers here (avoids symbol duplication).
+        if (strstr(msg, "NO_SUCH_FIELD") != NULL) {
+            PyErr_SetString(PyExc_KeyError, msg);
+            return;
+        }
+        if (strstr(msg, "INDEX_OUT_OF_BOUNDS") != NULL) {
+            PyErr_SetString(PyExc_IndexError, msg);
+            return;
+        }
+        if (strstr(msg, "INCORRECT_TYPE") != NULL) {
+            PyErr_SetString(PyExc_TypeError, msg);
+            return;
+        }
+        if (strstr(msg, "MEMALLOC") != NULL) {
+            PyErr_SetNone(PyExc_MemoryError);
+            return;
+        }
+
+        // ValueError group
+        if (strstr(msg, "EMPTY") != NULL || strstr(msg, "STRING_ERROR") != NULL ||
+            strstr(msg, "T_ATOM_ERROR") != NULL || strstr(msg, "F_ATOM_ERROR") != NULL ||
+            strstr(msg, "N_ATOM_ERROR") != NULL || strstr(msg, "NUMBER_ERROR") != NULL ||
+            strstr(msg, "UNESCAPED_CHARS") != NULL || strstr(msg, "UNCLOSED_STRING") != NULL ||
+            strstr(msg, "NUMBER_OUT_OF_RANGE") != NULL || strstr(msg, "INVALID_JSON_POINTER") != NULL ||
+            strstr(msg, "INVALID_URI_FRAGMENT") != NULL || strstr(msg, "CAPACITY") != NULL ||
+            strstr(msg, "TAPE_ERROR") != NULL) {
+            PyErr_SetString(PyExc_ValueError, msg);
+            return;
+        }
+
+        if (strstr(msg, "IO_ERROR") != NULL) {
+            PyErr_SetString(PyExc_IOError, msg);
+            return;
+        }
+
+        if (strstr(msg, "UTF8_ERROR") != NULL) {
+            PyObject *unicode_error = PyObject_CallFunction(
+                PyExc_UnicodeDecodeError,
+                "sy#nns",
+                "utf-8",
+                "",
+                0,
+                0,
+                1,
+                msg
+            );
+            if (unicode_error) {
+                PyErr_SetObject(PyExc_UnicodeDecodeError, unicode_error);
+                Py_XDECREF(unicode_error);
+            } else {
+                PyErr_SetString(PyExc_UnicodeDecodeError, msg);
+            }
+            return;
+        }
+
+        PyErr_SetString(PyExc_RuntimeError, msg);
+        return;
+    } catch (...) {
+        PyErr_SetString(PyExc_RuntimeError, "simdjson: unknown error");
+        return;
+    }
+}