tesseract::DocumentCache Class Reference

#include <imagedata.h>

Public Member Functions
TESS_API	DocumentCache (int64_t max_memory)
TESS_API	~DocumentCache ()
void	Clear ()
TESS_API bool	LoadDocuments (const std::vector< std::string > &filenames, CachingStrategy cache_strategy, FileReader reader)
bool	AddToCache (DocumentData *data)
DocumentData *	FindDocument (const std::string &document_name) const
const ImageData *	GetPageBySerial (int serial)
const std::vector< DocumentData * > &	documents () const
TESS_API int	TotalPages ()

Detailed Description

Definition at line 288 of file imagedata.h.

Constructor & Destructor Documentation

DocumentCache()

tesseract::DocumentCache::DocumentCache ( int64_t  max_memory )

explicit

Definition at line 604 of file imagedata.cpp.

: max_memory_(max_memory) {}

~DocumentCache()

tesseract::DocumentCache::~DocumentCache

(

)

Definition at line 606 of file imagedata.cpp.

                              {
  for (auto *document : documents_) {
    delete document;
  }
}

Member Function Documentation

AddToCache()

bool tesseract::DocumentCache::AddToCache

(

DocumentData *

data

)

Definition at line 641 of file imagedata.cpp.

                                                 {
  documents_.push_back(data);
  return true;
}

Clear()

void tesseract::DocumentCache::Clear ( )

inline

Definition at line 296 of file imagedata.h.

               {
    for (auto *document : documents_) {
      delete document;
    }
    documents_.clear();
    num_pages_per_doc_ = 0;
  }

documents()

const std::vector< DocumentData * > & tesseract::DocumentCache::documents ( ) const

inline

Definition at line 325 of file imagedata.h.

                                                     {
    return documents_;
  }

FindDocument()

DocumentData * tesseract::DocumentCache::FindDocument

(

const std::string &

document_name

)

const

Definition at line 647 of file imagedata.cpp.

                                          {
  for (auto *document : documents_) {
    if (document->document_name() == document_name) {
      return document;
    }
  }
  return nullptr;
}

GetPageBySerial()

const ImageData * tesseract::DocumentCache::GetPageBySerial ( int  serial )

inline

Definition at line 317 of file imagedata.h.

                                               {
    if (cache_strategy_ == CS_SEQUENTIAL) {
      return GetPageSequential(serial);
    } else {
      return GetPageRoundRobin(serial);
    }
  }

LoadDocuments()

bool tesseract::DocumentCache::LoadDocuments	(	const std::vector< std::string > &	filenames,
		CachingStrategy	cache_strategy,
		FileReader	reader
	)

Definition at line 614 of file imagedata.cpp.

                                                     {
  cache_strategy_ = cache_strategy;
  int64_t fair_share_memory = 0;
  // In the round-robin case, each DocumentData handles restricting its content
  // to its fair share of memory. In the sequential case, DocumentCache
  // determines which DocumentDatas are held entirely in memory.
  if (cache_strategy_ == CS_ROUND_ROBIN) {
    fair_share_memory = max_memory_ / filenames.size();
  }
  for (const auto &filename : filenames) {
    auto *document = new DocumentData(filename);
    document->SetDocument(filename.c_str(), fair_share_memory, reader);
    AddToCache(document);
  }
  if (!documents_.empty()) {
    // Try to get the first page now to verify the list of filenames.
    if (GetPageBySerial(0) != nullptr) {
      return true;
    }
    tprintf("Load of page 0 failed!\n");
  }
  return false;
}

TotalPages()

int tesseract::DocumentCache::TotalPages

(

)

Definition at line 659 of file imagedata.cpp.

                              {
  if (cache_strategy_ == CS_SEQUENTIAL) {
    // In sequential mode, we assume each doc has the same number of pages
    // whether it is true or not.
    if (num_pages_per_doc_ == 0) {
      GetPageSequential(0);
    }
    return num_pages_per_doc_ * documents_.size();
  }
  int total_pages = 0;
  for (auto *document : documents_) {
    // We have to load a page to make NumPages() valid.
    document->GetPage(0);
    total_pages += document->NumPages();
  }
  return total_pages;
}

---

The documentation for this class was generated from the following files:

• /media/home/debian/src/github/tesseract-ocr/tesseract/src/ccstruct/imagedata.h
• /media/home/debian/src/github/tesseract-ocr/tesseract/src/ccstruct/imagedata.cpp