Split the main PDF converter code into its own class

This commit is contained in:
Leon Styhre 2023-12-11 18:31:38 +01:00
parent 8403b40d4c
commit ddaf2f01b7
4 changed files with 249 additions and 113 deletions

View file

@ -21,13 +21,27 @@ if(WIN32)
elseif(APPLE) elseif(APPLE)
set(POPPLER_CPP_INCLUDE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/cpp ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/build/cpp) set(POPPLER_CPP_INCLUDE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/cpp ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/build/cpp)
set(POPPLER_CPP_LIBRARY ${CMAKE_CURRENT_SOURCE_DIR}/../libpoppler-cpp.0.dylib) set(POPPLER_CPP_LIBRARY ${CMAKE_CURRENT_SOURCE_DIR}/../libpoppler-cpp.0.dylib)
elseif(ANDROID)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fvisibility=hidden")
set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -llog")
set(POPPLER_CPP_INCLUDE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/cpp ${CMAKE_CURRENT_SOURCE_DIR}/../external/poppler/build/cpp)
set(POPPLER_CPP_LIBRARY ${CMAKE_CURRENT_SOURCE_DIR}/../android/libs/${ANDROID_CPU_ARCH}/libpoppler-cpp.so)
else() else()
find_package(Poppler REQUIRED COMPONENTS cpp) find_package(Poppler REQUIRED COMPONENTS cpp)
endif() endif()
include_directories(${POPPLER_CPP_INCLUDE_DIR}) include_directories(${POPPLER_CPP_INCLUDE_DIR})
add_executable(es-pdf-convert WIN32 ${CMAKE_CURRENT_SOURCE_DIR}/src/main.cpp) if (ANDROID)
target_link_libraries(es-pdf-convert ${POPPLER_CPP_LIBRARY}) set(CONVERTER_SOURCE_FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/ConvertPDF.cpp
${CMAKE_CURRENT_SOURCE_DIR}/src/ConvertPDF.h)
add_library(es-pdf-convert SHARED ${CONVERTER_SOURCE_FILES})
else()
set(CONVERTER_SOURCE_FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/main.cpp
${CMAKE_CURRENT_SOURCE_DIR}/src/ConvertPDF.cpp
${CMAKE_CURRENT_SOURCE_DIR}/src/ConvertPDF.h)
add_executable(es-pdf-convert WIN32 ${CONVERTER_SOURCE_FILES})
endif()
target_link_libraries(es-pdf-convert PRIVATE ${POPPLER_CPP_LIBRARY})
if(WIN32) if(WIN32)
set_target_properties(es-pdf-convert PROPERTIES RUNTIME_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/es-pdf-converter" INSTALL_RPATH_USE_LINK_PATH TRUE) set_target_properties(es-pdf-convert PROPERTIES RUNTIME_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/es-pdf-converter" INSTALL_RPATH_USE_LINK_PATH TRUE)

View file

@ -0,0 +1,194 @@
// SPDX-License-Identifier: GPL-2.0-only
//
// ES-DE
// ConvertPDF.cpp
//
// Converts PDF document pages to raw ARGB32 pixel data for maximum performance.
// This needs to be separated into its own binary to get around the restrictive GPL
// license used by the Poppler PDF rendering library.
//
#include "ConvertPDF.h"
#include "poppler-document.h"
#include "poppler-image.h"
#include "poppler-page-renderer.h"
#include "poppler-page.h"
#include <cmath>
#include <fstream>
#include <iostream>
#if defined(__ANDROID__)
#include <android/log.h>
#endif
#if defined(_WIN64)
#include <fcntl.h>
#include <io.h>
#include <windows.h>
#endif
#if defined(_WIN64)
int ConvertPDF::processFile(
const std::wstring path, const std::wstring mode, int pageNum, int width, int height)
#elif defined(__ANDROID__)
int ConvertPDF::processFile(const std::string path,
const std::string mode,
int pageNum,
int width,
int height,
std::string& result)
#else
int ConvertPDF::processFile(
const std::string path, const std::string mode, int pageNum, int width, int height)
#endif
{
std::ifstream file;
file.open(path.c_str(), std::ifstream::binary);
if (file.fail()) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID,
"Error: Couldn't open PDF file, permission problems?");
#else
std::cerr << "Error: Couldn't open PDF file, permission problems?" << std::endl;
#endif
return (-1);
}
file.seekg(0, std::ios::end);
const long fileLength {static_cast<long>(file.tellg())};
file.seekg(0, std::ios::beg);
std::vector<char> fileData(fileLength);
file.read(&fileData[0], fileLength);
file.close();
const poppler::document* document {
poppler::document::load_from_raw_data(&fileData[0], fileLength)};
if (document == nullptr) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID,
"Error: Couldn't open document, invalid PDF file?");
#else
std::cerr << "Error: Couldn't open document, invalid PDF file?" << std::endl;
#endif
return (-1);
}
const int pageCount {document->pages()};
#if defined(_WIN64)
if (mode == L"-fileinfo") {
#else
if (mode == "-fileinfo") {
#endif
std::vector<std::string> pageInfo;
for (int i {0}; i < pageCount; ++i) {
std::string pageRow;
const poppler::page* page {document->create_page(i)};
if (page == nullptr) {
if (page == nullptr) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID,
"Error: Couldn't read page %i", i + 1);
#else
std::cerr << "Error: Couldn't read page " << i + 1 << std::endl;
#endif
return (-1);
}
}
std::string orientation;
if (page->orientation() == poppler::page::portrait)
orientation = "portrait";
else if (page->orientation() == poppler::page::upside_down)
orientation = "upside_down";
else if (page->orientation() == poppler::page::seascape)
orientation = "seascape";
else
orientation = "landscape";
const poppler::rectf pageRect {page->page_rect()};
pageRow.append(std::to_string(i + 1))
.append(";")
.append(orientation)
.append(";")
.append(std::to_string(pageRect.width()))
.append(";")
.append(std::to_string(pageRect.height()));
pageInfo.emplace_back(pageRow);
}
for (auto& row : pageInfo) {
#if defined(__ANDROID__)
result.append(row).append("\n");
#else
std::cout << row << std::endl;
#endif
}
return (0);
}
if (pageNum < 1 || pageNum > pageCount) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID,
"Error: Requested page %i does not exist in document", pageNum);
#else
std::cerr << "Error: Requested page " << pageNum << " does not exist in document"
<< std::endl;
#endif
return (-1);
}
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, "org.es_de.frontend", "PDF CONVERTING BREAK 10");
#endif
const poppler::page* page {document->create_page(pageNum - 1)};
if (page == nullptr) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID,
"Error: Couldn't read page %i", pageNum);
#else
std::cerr << "Error: Couldn't read page " << pageNum << std::endl;
#endif
return (-1);
}
poppler::page_renderer pageRenderer;
pageRenderer.set_render_hint(poppler::page_renderer::text_antialiasing);
pageRenderer.set_render_hint(poppler::page_renderer::antialiasing);
// pageRenderer.set_render_hint(poppler::page_renderer::text_hinting);
const poppler::rectf pageRect {page->page_rect()};
const bool rotate {page->orientation() == poppler::page::portrait ||
page->orientation() == poppler::page::upside_down};
const double pageHeight {pageRect.height()};
const double sizeFactor {static_cast<double>(rotate ? height : width) / pageHeight};
poppler::image image {
pageRenderer.render_page(page, 72.0 * sizeFactor, 72.0 * sizeFactor, 0, 0, width, height)};
if (!image.is_valid()) {
#if defined(__ANDROID__)
__android_log_print(ANDROID_LOG_ERROR, ANDROID_APPLICATION_ID, "Rendered image is invalid");
#else
std::cerr << "Rendered image is invalid" << std::endl;
#endif
return (-1);
}
#if defined(__ANDROID__)
result.insert(0, std::move(image.data()), width * height * 4);
#else
// Necessary as the image data stream may contain null characters.
std::string imageARGB32;
imageARGB32.insert(0, std::move(image.data()), width * height * 4);
std::cout << imageARGB32;
#endif
return 0;
}

View file

@ -0,0 +1,35 @@
// SPDX-License-Identifier: GPL-2.0-only
//
// ES-DE
// ConvertPDF.h
//
// Converts PDF document pages to raw ARGB32 pixel data for maximum performance.
// This needs to be separated into its own binary to get around the restrictive GPL
// license used by the Poppler PDF rendering library.
//
#include <string>
#ifndef ES_PDF_CONVERTER_CONVERT_PDF_H
#define ES_PDF_CONVERTER_CONVERT_PDF_H
class ConvertPDF
{
public:
#if defined(_WIN64)
static int processFile(
const std::wstring path, const std::wstring mode, int pageNum, int width, int height);
#elif defined(__ANDROID__)
__attribute__((visibility("default"))) static int processFile(const std::string path,
const std::string mode,
int pageNum,
int width,
int height,
std::string& result);
#else
static int processFile(
const std::string path, const std::string mode, int pageNum, int width, int height);
#endif
};
#endif // ES_PDF_CONVERTER_CONVERT_PDF_H

View file

@ -1,6 +1,6 @@
// SPDX-License-Identifier: GPL-2.0-only // SPDX-License-Identifier: GPL-2.0-only
// //
// EmulationStation Desktop Edition (ES-DE) PDF converter // ES-DE PDF converter
// main.cpp // main.cpp
// //
// Converts PDF document pages to raw ARGB32 pixel data for maximum performance. // Converts PDF document pages to raw ARGB32 pixel data for maximum performance.
@ -11,13 +11,8 @@
// All ES-DE C++ source code is formatted using clang-format. // All ES-DE C++ source code is formatted using clang-format.
// //
#include "poppler-document.h" #include "ConvertPDF.h"
#include "poppler-image.h"
#include "poppler-page-renderer.h"
#include "poppler-page.h"
#include <cmath>
#include <fstream>
#include <iostream> #include <iostream>
#if defined(_WIN64) #if defined(_WIN64)
@ -101,7 +96,7 @@ int main(int argc, char* argv[])
pageNum = atoi(argv[3]); pageNum = atoi(argv[3]);
width = atoi(argv[4]); width = atoi(argv[4]);
height = atoi(argv[5]); height = atoi(argv[5]);
#endif #endif // _WIN64
if (width < 1 || width > 7680) { if (width < 1 || width > 7680) {
std::cerr << "Invalid horizontal resolution defined: " << argv[3] << std::endl; std::cerr << "Invalid horizontal resolution defined: " << argv[3] << std::endl;
exit(-1); exit(-1);
@ -116,107 +111,5 @@ int main(int argc, char* argv[])
// << width << "x" << height << " pixels" << std::endl; // << width << "x" << height << " pixels" << std::endl;
} }
std::ifstream file; return ConvertPDF::processFile(path, mode, pageNum, width, height);
file.open(path.c_str(), std::ifstream::binary);
if (file.fail()) {
std::cerr << "Error: Couldn't open PDF file, permission problems?" << std::endl;
exit(-1);
}
file.seekg(0, std::ios::end);
const long fileLength {static_cast<long>(file.tellg())};
file.seekg(0, std::ios::beg);
std::vector<char> fileData(fileLength);
file.read(&fileData[0], fileLength);
file.close();
const poppler::document* document {
poppler::document::load_from_raw_data(&fileData[0], fileLength)};
if (document == nullptr) {
std::cerr << "Error: Couldn't open document, invalid PDF file?" << std::endl;
exit(-1);
}
const int pageCount {document->pages()};
#if defined(_WIN64)
if (mode == L"-fileinfo") {
#else
if (mode == "-fileinfo") {
#endif
std::vector<std::string> pageInfo;
for (int i {0}; i < pageCount; ++i) {
std::string pageRow;
const poppler::page* page {document->create_page(i)};
if (page == nullptr) {
if (page == nullptr) {
std::cerr << "Error: Couldn't read page " << i + 1 << std::endl;
exit(-1);
}
}
std::string orientation;
if (page->orientation() == poppler::page::portrait)
orientation = "portrait";
else if (page->orientation() == poppler::page::upside_down)
orientation = "upside_down";
else if (page->orientation() == poppler::page::seascape)
orientation = "seascape";
else
orientation = "landscape";
const poppler::rectf pageRect {page->page_rect()};
pageRow.append(std::to_string(i + 1))
.append(";")
.append(orientation)
.append(";")
.append(std::to_string(pageRect.width()))
.append(";")
.append(std::to_string(pageRect.height()));
pageInfo.emplace_back(pageRow);
}
for (auto& row : pageInfo)
std::cout << row << std::endl;
exit(0);
}
if (pageNum < 1 || pageNum > pageCount) {
std::cerr << "Error: Requested page " << pageNum << " does not exist in document"
<< std::endl;
exit(-1);
}
const poppler::page* page {document->create_page(pageNum - 1)};
if (page == nullptr) {
std::cerr << "Error: Couldn't read page " << pageNum << std::endl;
exit(-1);
}
poppler::page_renderer pageRenderer;
pageRenderer.set_render_hint(poppler::page_renderer::text_antialiasing);
pageRenderer.set_render_hint(poppler::page_renderer::antialiasing);
// pageRenderer.set_render_hint(poppler::page_renderer::text_hinting);
const poppler::rectf pageRect {page->page_rect()};
const bool rotate {page->orientation() == poppler::page::portrait ||
page->orientation() == poppler::page::upside_down};
const double pageHeight {pageRect.height()};
const double sizeFactor {static_cast<double>(rotate ? height : width) / pageHeight};
poppler::image image {
pageRenderer.render_page(page, 72.0 * sizeFactor, 72.0 * sizeFactor, 0, 0, width, height)};
if (!image.is_valid()) {
std::cerr << "Rendered image is invalid" << std::endl;
exit(-1);
}
// Necessary as the image data stream may contain null characters.
std::string imageARGB32;
imageARGB32.insert(0, std::move(image.data()), width * height * 4);
std::cout << imageARGB32;
return 0;
} }