【PDF识别重命名】如何识别PDF区域内容对文件进行改名处理或将内容导出表格

#include <iostream>
#include <string>
#include <poppler/cpp/poppler-document.h>
#include <poppler/cpp/poppler-page.h>
#include "libxl.h"

// 提取PDF指定区域的文本内容
std::string extractTextFromPDF(const std::string& filePath, double x, double y, double width, double height) {
    poppler::document* doc = poppler::document::load_from_file(filePath);
    if (!doc || doc->is_locked()) {
        delete doc;
        return "";
    }

    poppler::page* page = doc->create_page(0);
    if (!page) {
        delete doc;
        return "";
    }

    poppler::rectf rect(x, y, x + width, y + height);
    poppler::text_box_layout layout;
    layout.set_rect(rect);
    std::string text = page->text(layout).to_latin1();

    delete page;
    delete doc;
    return text;
}

// 对文件进行重命名
bool renameFile(const std::string& oldPath, const std::string& newName) {
    std::string newPath = oldPath.substr(0, oldPath.find_last_of("/\\") + 1) + newName + ".pdf";
    return std::rename(oldPath.c_str(), newPath.c_str()) == 0;
}

// 将内容导出到Excel表格
bool exportToExcel(const std::string& filePath, const std::string& content) {
    libxl::Book* book = xlCreateBook();
    if (book) {
        libxl::Sheet* sheet = book->addSheet("Sheet1");
        if (sheet) {
            sheet->writeStr(0, 0, content.c_str());
            if (book->save(filePath.c_str())) {
                book->release();
                return true;
            }
        }
        book->release();
    }
    return false;
}

int main() {
    std::string pdfFilePath = "example.pdf";
    // 假设提取区域的坐标和尺寸
    double x = 100, y = 100, width = 200, height = 50;

    // 提取指定区域的文本内容
    std::string extractedText = extractTextFromPDF(pdfFilePath, x, y, width, height);
    if (!extractedText.empty()) {
        std::cout << "Extracted text: " << extractedText << std::endl;

        // 对文件进行重命名
        if (renameFile(pdfFilePath, extractedText)) {
            std::cout << "File renamed successfully." << std::endl;
        } else {
            std::cerr << "Failed to rename file." << std::endl;
        }

        // 将内容导出到Excel表格
        std::string excelFilePath = "output.xlsx";
        if (exportToExcel(excelFilePath, extractedText)) {
            std::cout << "Content exported to Excel successfully." << std::endl;
        } else {
            std::cerr << "Failed to export content to Excel." << std::endl;
        }
    } else {
        std::cerr << "Failed to extract text from PDF." << std::endl;
    }

    return 0;
}

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

./pdf_processing