C++实现批量提取PDF内容

sudo apt-get install libpoppler-cpp-dev

#include <iostream>
#include <fstream>
#include <vector>
#include <string>
#include <poppler/cpp/poppler-document.h>
#include <poppler/cpp/poppler-page.h>
#include "libxl.h"
 
using namespace libxl;
 
// 提取 PDF 文件中的文字内容
std::string extractTextFromPDF(const std::string& filePath) {
    poppler::document* doc = poppler::document::load_from_file(filePath);
    if (!doc || doc->is_locked()) {
        delete doc;
        return "";
    }
 
    std::string text;
    for (int i = 0; i < doc->pages(); ++i) {
        poppler::page* page = doc->create_page(i);
        if (page) {
            text += page->text().to_latin1();
            delete page;
        }
    }
 
    delete doc;
    return text;
}
 
// 批量提取 PDF 文件内容并导出到 Excel 表格
void batchExtractPDFsToExcel(const std::vector<std::string>& pdfFiles, const std::string& outputFilePath) {
    Book* book = xlCreateBook();
    if (book) {
        Sheet* sheet = book->addSheet("PDF Text");
        if (sheet) {
            for (size_t i = 0; i < pdfFiles.size(); ++i) {
                std::string text = extractTextFromPDF(pdfFiles[i]);
                sheet->writeStr(i, 0, pdfFiles[i].c_str());
                sheet->writeStr(i, 1, text.c_str());
            }
        }
        book->save(outputFilePath.c_str());
        book->release();
    }
}
 
int main() {
    std::vector<std::string> pdfFiles = {
        "file1.pdf",
        "file2.pdf",
        // 添加更多 PDF 文件路径
    };
    std::string outputFilePath = "output.xlsx";
    batchExtractPDFsToExcel(pdfFiles, outputFilePath);
    return 0;
}

g++ -o extract_pdf extract_pdf.cpp -lpoppler-cpp -lxl

./extract_pdf

#include <iostream>
#include <filesystem>
#include <string>
 
namespace fs = std::filesystem;
 
// 批量给 PDF 文件改名
void batchRenamePDFs(const std::string& directoryPath) {
    int counter = 1;
    for (const auto& entry : fs::directory_iterator(directoryPath)) {
        if (entry.is_regular_file() && entry.path().extension() == ".pdf") {
            fs::path newPath = entry.path().parent_path() / (std::to_string(counter) + ".pdf");
            fs::rename(entry.path(), newPath);
            std::cout << "Renamed " << entry.path() << " to " << newPath << std::endl;
            ++counter;
        }
    }
}
 
int main() {
    std::string directoryPath = "./pdfs"; // 替换为实际的 PDF 文件目录
    batchRenamePDFs(directoryPath);
    return 0;
}

g++ -std=c++17 -o rename_pdf rename_pdf.cpp

./rename_pdf