fs_ocr.h

浏览该文件的文档.

#if (defined(_WIN32) || defined(_WIN64)) || defined(__linux__)

#ifndef FS_OCR_H_
#define FS_OCR_H_

#include "common/fs_common.h"
#include "pdf/fs_pdfdoc.h"
#include "pdf/fs_pdfpage.h"

namespace foxit {
namespace addon {
namespace ocr {
class OCRCallback {
 public:
  virtual bool NeedToCancelNow(const wchar_t* info) = 0;

  virtual bool IsImageIgnored(foxit::pdf::graphics::ImageObject* image_object) = 0;
};

class OCRProgressCallback {
 public:
  virtual void ProgressNotify(int current_rate) = 0;
};

class OCREngine FS_FINAL : public Object {
 public:
  static ErrorCode Initialize(const wchar_t* ocr_resource_path);

  static ErrorCode Initialize(const wchar_t* ocr_resource_path, bool is_shared_cpu_cores_mode);

  static void Release();

  static void SetLogFile(const char* log_file_path);

  static void SetLogFile(const wchar_t* log_file_path);

  static void SetLanguages(const wchar_t* languages);

  static void SetOCRCallback(OCRCallback* callback);
};

class OCRConfig FS_FINAL : public Object {
 public:
  OCRConfig()
  :is_detect_pictures(true)
  ,is_remove_noise(true)
  ,is_correct_skew(true)
  ,is_enable_text_extraction_mode(false)
  ,is_sequentially_process(false)
  ,is_auto_overwrite_resolution(true)
  ,resolution_to_overwrite(300)
  ,confidence(0){}

  OCRConfig(bool is_detect_pictures, bool is_remove_noise, bool is_correct_skew, bool is_enable_text_extraction_mode, bool is_sequentially_process,
            bool is_auto_overwrite_resolution, int resolution_to_overwrite, int confidence) {
    this->is_detect_pictures = is_detect_pictures;
    this->is_remove_noise = is_remove_noise;
    this->is_correct_skew = is_correct_skew;
    this->is_enable_text_extraction_mode = is_enable_text_extraction_mode;
    this->is_sequentially_process = is_sequentially_process;
    this->is_auto_overwrite_resolution = is_auto_overwrite_resolution;
    this->resolution_to_overwrite = resolution_to_overwrite;
    this->confidence = confidence;
  }

  void Set(bool is_detect_pictures, bool is_remove_noise, bool is_correct_skew, bool is_enable_text_extraction_mode, bool is_sequentially_process,
           bool is_auto_overwrite_resolution, int resolution_to_overwrite, int confidence) {
    this->is_detect_pictures = is_detect_pictures;
    this->is_remove_noise = is_remove_noise;
    this->is_correct_skew = is_correct_skew;
    this->is_enable_text_extraction_mode = is_enable_text_extraction_mode;
    this->is_sequentially_process = is_sequentially_process;
    this->is_auto_overwrite_resolution = is_auto_overwrite_resolution;
    this->resolution_to_overwrite = resolution_to_overwrite;
    this->confidence = confidence;
  }

  OCRConfig& operator=(const OCRConfig& other) {
    is_detect_pictures = other.is_detect_pictures;
    is_remove_noise = other.is_remove_noise;
    is_correct_skew = other.is_correct_skew;
    is_enable_text_extraction_mode = other.is_enable_text_extraction_mode;
    is_sequentially_process = other.is_sequentially_process;
    is_auto_overwrite_resolution = other.is_auto_overwrite_resolution;
    resolution_to_overwrite = other.resolution_to_overwrite;
    confidence = other.confidence;
    return (*this);
  }

  bool operator!=(const OCRConfig& other) {
    if (is_detect_pictures != other.is_detect_pictures || is_remove_noise != other.is_remove_noise ||
        is_correct_skew != other.is_correct_skew || is_enable_text_extraction_mode != other.is_enable_text_extraction_mode ||
      is_sequentially_process != other.is_sequentially_process || is_auto_overwrite_resolution != other.is_auto_overwrite_resolution ||
      resolution_to_overwrite != other.resolution_to_overwrite || confidence != other.confidence)
      return true;
    return false;
  }

  bool is_detect_pictures;

  bool is_remove_noise;

  bool is_correct_skew;

  bool is_enable_text_extraction_mode;

  bool is_sequentially_process;

  bool is_auto_overwrite_resolution;

  int resolution_to_overwrite;

  int confidence;
};

class OCRSettingData FS_FINAL : public Object {
 public:
  OCRSettingData(){}

  OCRSettingData(pdf::PDFDoc pdf_doc, const common::Range& page_range, bool is_editable, const OCRConfig& ocr_config) {
    this->pdf_doc = pdf_doc;
    this->page_range = page_range;
    this->is_editable = is_editable;
    this->ocr_config = ocr_config;
  }

  void Set(pdf::PDFDoc pdf_doc, const common::Range& page_range, bool is_editable, const OCRConfig& ocr_config) {
    this->pdf_doc = pdf_doc;
    this->page_range = page_range;
    this->is_editable = is_editable;
    this->ocr_config = ocr_config;
  }

  OCRSettingData& operator=(const OCRSettingData& data) {
    pdf_doc = data.pdf_doc;
    page_range = data.page_range;
    is_editable = data.is_editable;
    ocr_config = data.ocr_config;
    return (*this);
  }

  bool operator!=(const OCRSettingData& data) {
    if (pdf_doc != data.pdf_doc || page_range != data.page_range || is_editable != data.is_editable || ocr_config != data.ocr_config)
      return true;
    return false;
  }

  pdf::PDFDoc pdf_doc;

  common::Range page_range;

  bool is_editable;

  OCRConfig ocr_config;
};

FSDK_DEFINE_ARRAY(OCRSettingDataArray, OCRSettingData)


class OCRSuspectInfo FS_FINAL : public Object {
 public:
  int page_index;

  foxit::RectF words_rect;

  WString suspect_words;
};

FSDK_DEFINE_ARRAY(OCRSuspectInfoArray, OCRSuspectInfo)


class OCR FS_FINAL : public Base {
 public:
  typedef enum _OCRConvertFormat {
    e_OCRConvertFormatDOCX  = 0,
    e_OCRConvertFormatDOC = 1,
    e_OCRConvertFormatRTF = 2,
    e_OCRConvertFormatXLSX = 3,
    e_OCRConvertFormatXLS = 4,
    e_OCRConvertFormatPPTX = 5,
    e_OCRConvertFormatHTML = 6
  } OCRConvertFormat;
  
  OCR();

  OCR(const OCR& other);

  // 强烈建议用户不要使用此方法；否则可能出现未知情况。
  explicit OCR(FS_HANDLE handle);
  ~OCR();

  OCR& operator = (const OCR& other);
  bool operator == (const OCR& other) const;
  bool operator != (const OCR& other) const;

  bool IsEmpty() const;

  void OCRPDFPage(pdf::PDFPage pdf_page, bool is_editable, OCRProgressCallback* callback = NULL);

  void OCRPDFPage(pdf::PDFPage pdf_page, bool is_editable, const OCRConfig& config, OCRProgressCallback* callback = NULL);

  void OCRPDFDocument(pdf::PDFDoc pdf_doc, bool is_editable, OCRProgressCallback* callback = NULL);

  void OCRPDFDocument(pdf::PDFDoc pdf_doc, bool is_editable, const OCRConfig& config, OCRProgressCallback* callback = NULL);

  void OCRConvertTo(OCRConvertFormat format, const wchar_t* src_pdf_path, const wchar_t* password, const wchar_t* saved_file_path, common::Range page_range, bool is_retain_flowing_text, OCRProgressCallback* callback = NULL);

  void OCRConvertTo(OCRConvertFormat format, const wchar_t* src_pdf_path, const wchar_t* password, const wchar_t* saved_file_path, common::Range page_range, bool is_retain_flowing_text, const OCRConfig& config, OCRProgressCallback* callback = NULL);

#if (defined(_WIN32) || defined(_WIN64)) || defined(__linux__)

  void OCRPDFDocuments(const ocr::OCRSettingDataArray& settingdata_array, OCRProgressCallback* callback = NULL);
#endif

  OCRSuspectInfoArray GetOCRSuspectsInfo(pdf::PDFDoc ocred_pdf_doc);
};

}  // namespace ocr
}  // namespace addon
}  // namespace foxit

#endif  // FS_OCR_H_

#if (defined(_WIN32) || defined(_WIN64)) || defined(__linux__)