Add RedHat patches for big-endian CPUs

manually updated cfx_seekablestreamproxy.cpp to newer pdfium version

tested that patches apply and build on little endian

Author: mara004

REUSE.toml

@@ -143,6 +143,22 @@ SPDX-FileCopyrightText = [
 ]
 SPDX-License-Identifier = "BSD-3-Clause OR Apache-2.0"
 
+[[annotations]]
+path = [
+    "pdfium_patches/bigendian.patch",
+    "pdfium_patches/bigendian_test.patch",
+]
+precedence = "aggregate"
+SPDX-FileCopyrightText = [
+    "2025 Christian Heimes <[email protected]>",
+    "2025 geisserml <[email protected]>",
+]
+SPDX-License-Identifier = "LicenseRef-Ignore"
+SPDX-FileComment = '''
+Derived from RedHat packaging repository: https://github.com/tiran/libpdfium/blob/d99370b3ac7f0c9cd7222be2dfab2c7b648f2e9e/0001-bigendian.patch
+Repo does not have an own license file, but assumably the patches are being made available under pdfium's license, or leastways an open-source license.
+'''
+
 [[annotations]]
 path = [
     "docs/build/.gitkeep",

pdfium_patches/bigendian.patch

@@ -0,0 +1,40 @@
+diff --git a/core/fxcrt/cfx_seekablestreamproxy.cpp b/core/fxcrt/cfx_seekablestreamproxy.cpp
+index 33cc528..5fa42bf 100644
+--- a/core/fxcrt/cfx_seekablestreamproxy.cpp
++++ b/core/fxcrt/cfx_seekablestreamproxy.cpp
+@@ -89,11 +89,19 @@ void SwapByteOrder(pdfium::span<uint16_t> str) {
+ 
+ }  // namespace
+ 
++#if defined(ARCH_CPU_LITTLE_ENDIAN)
+ #define BOM_UTF8_MASK 0x00FFFFFF
+ #define BOM_UTF8 0x00BFBBEF
+ #define BOM_UTF16_MASK 0x0000FFFF
+ #define BOM_UTF16_BE 0x0000FFFE
+ #define BOM_UTF16_LE 0x0000FEFF
++#else
++#define BOM_UTF8_MASK 0xFFFFFF00
++#define BOM_UTF8 0xEFBBBF00
++#define BOM_UTF16_MASK 0xFFFF0000
++#define BOM_UTF16_BE 0xFEFF0000
++#define BOM_UTF16_LE 0xFFFE0000
++#endif
+ 
+ CFX_SeekableStreamProxy::CFX_SeekableStreamProxy(
+     const RetainPtr<IFX_SeekableReadStream>& stream)
+@@ -188,9 +196,15 @@ size_t CFX_SeekableStreamProxy::ReadBlock(pdfium::span<wchar_t> buffer) {
+     size_t bytes_read =
+         ReadData(pdfium::as_writable_bytes(buffer).first(bytes_to_read));
+     size_t elements = bytes_read / sizeof(uint16_t);
++#if defined(ARCH_CPU_LITTLE_ENDIAN)
+     if (code_page_ == FX_CodePage::kUTF16BE) {
+       SwapByteOrder(fxcrt::reinterpret_span<uint16_t>(buffer).first(elements));
+     }
++#else
++    if (m_wCodePage == FX_CodePage::kUTF16LE) {
++      SwapByteOrder(fxcrt::reinterpret_span<uint16_t>(buffer).first(elements));
++    }
++#endif
+     UTF16ToWChar(buffer.first(elements));
+     return elements;
+   }

pdfium_patches/bigendian_test.patch

@@ -0,0 +1,29 @@
+--- a/core/fxcrt/binary_buffer_unittest.cpp
++++ b/core/fxcrt/binary_buffer_unittest.cpp
+@@ -3,6 +3,7 @@
+ // found in the LICENSE file.
+ 
+ #include "core/fxcrt/binary_buffer.h"
++#include "core/fxcrt/byteorder.h"
+ 
+ #include <utility>
+ #include <vector>
+@@ -122,7 +123,7 @@
+ // Assumes little endian.
+ TEST(BinaryBuffer, AppendUint16) {
+   BinaryBuffer buffer;
+-  buffer.AppendUint16(0x4321);
++  buffer.AppendUint16(FromLE16(0x4321));
+   EXPECT_EQ(2u, buffer.GetSize());
+   EXPECT_EQ(2u, buffer.GetLength());
+   EXPECT_EQ(0x21u, buffer.GetSpan()[0]);
+@@ -132,7 +133,7 @@
+ // Assumes little endian.
+ TEST(BinaryBuffer, AppendUint32) {
+   BinaryBuffer buffer;
+-  buffer.AppendUint32(0x87654321);
++  buffer.AppendUint32(FromLE32(0x87654321));
+   EXPECT_EQ(4u, buffer.GetSize());
+   EXPECT_EQ(4u, buffer.GetLength());
+   EXPECT_EQ(0x21u, buffer.GetSpan()[0]);
+

setupsrc/pypdfium2_setup/build_native.py

@@ -188,6 +188,10 @@ def get_sources(deps_info, short_ver, with_tests, compiler, clang_path, single_l
                 "#if 1  // defined(COMPONENT_BUILD)",
                 is_regex=False, exp_count=1,
             )
+        if sys.byteorder == "big":
+            git_apply_patch(PatchDir/"bigendian.patch", cwd=PDFIUM_DIR)
+            if with_tests:
+                git_apply_patch(PatchDir/"bigendian_test.patch", cwd=PDFIUM_DIR)
 
     do_patches = _fetch_dep(deps_info, "build", PDFIUM_DIR/"build", reset=reset)
     if do_patches: