tesseract::TessdataManager Class Reference

#include <tessdatamanager.h>

Public Member Functions
	TessdataManager ()
	TessdataManager (FileReader reader)
	~TessdataManager ()
bool	swap () const
bool	is_loaded () const
void	LoadFileLater (const char *data_file_name)
bool	Init (const char *data_file_name)
bool	LoadMemBuffer (const char *name, const char *data, int size)
void	OverwriteEntry (TessdataType type, const char *data, int size)
bool	SaveFile (const STRING &filename, FileWriter writer) const
void	Serialize (GenericVector< char > *data) const
void	Clear ()
void	Directory () const
bool	IsComponentAvailable (TessdataType type) const
bool	GetComponent (TessdataType type, TFile *fp)
bool	GetComponent (TessdataType type, TFile *fp) const
std::string	VersionString () const
void	SetVersionString (const std::string &v_str)
bool	IsBaseAvailable () const
bool	IsLSTMAvailable () const
const STRING &	GetDataFileName () const
bool	CombineDataFiles (const char *language_data_path_prefix, const char *output_filename)
bool	OverwriteComponents (const char *new_traineddata_filename, char **component_filenames, int num_new_components)
bool	ExtractToFile (const char *filename)

Static Public Member Functions
static bool	TessdataTypeFromFileSuffix (const char *suffix, TessdataType *type)
static bool	TessdataTypeFromFileName (const char *filename, TessdataType *type)

Detailed Description

Definition at line 126 of file tessdatamanager.h.

Constructor & Destructor Documentation

TessdataManager() [1/2]

tesseract::TessdataManager::TessdataManager

(

)

Definition at line 40 of file tessdatamanager.cpp.

                                 : reader_(nullptr), is_loaded_(false), swap_(false) {
  SetVersionString(PACKAGE_VERSION);
}

TessdataManager() [2/2]

tesseract::TessdataManager::TessdataManager ( FileReader  reader )

explicit

Definition at line 44 of file tessdatamanager.cpp.

  : reader_(reader),
    is_loaded_(false),
    swap_(false) {
  SetVersionString(PACKAGE_VERSION);
}

~TessdataManager()

tesseract::TessdataManager::~TessdataManager ( )

inline

Definition at line 131 of file tessdatamanager.h.

{}

Member Function Documentation

Clear()

void tesseract::TessdataManager::Clear

(

)

Definition at line 151 of file tessdatamanager.cpp.

                            {
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    entries_[i].clear();
  }
  is_loaded_ = false;
}

CombineDataFiles()

bool tesseract::TessdataManager::CombineDataFiles	(	const char *	language_data_path_prefix,
		const char *	output_filename
	)

Reads all the standard tesseract config and data files for a language at the given path and bundles them up into one binary data file. Returns true if the combined traineddata file was successfully written.

Definition at line 201 of file tessdatamanager.cpp.

                                 {
  // Load individual tessdata components from files.
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    TessdataType type;
    ASSERT_HOST(TessdataTypeFromFileSuffix(kTessdataFileSuffixes[i], &type));
    STRING filename = language_data_path_prefix;
    filename += kTessdataFileSuffixes[i];
    FILE *fp = fopen(filename.string(), "rb");
    if (fp != nullptr) {
      fclose(fp);
      if (!LoadDataFromFile(filename, &entries_[type])) {
        tprintf("Load of file %s failed!\n", filename.string());
        return false;
      }
    }
  }
  is_loaded_ = true;

  // Make sure that the required components are present.
  if (!IsBaseAvailable() && !IsLSTMAvailable()) {
    tprintf(
        "Error: traineddata file must contain at least (a unicharset file"
        "and inttemp) OR an lstm file.\n");
    return false;
  }
  // Write updated data to the output traineddata file.
  return SaveFile(output_filename, nullptr);
}

Directory()

void tesseract::TessdataManager::Directory

(

)

const

Definition at line 159 of file tessdatamanager.cpp.

                                      {
  tprintf("Version string:%s\n", VersionString().c_str());
  int offset = TESSDATA_NUM_ENTRIES * sizeof(int64_t);
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    if (!entries_[i].empty()) {
      tprintf("%d:%s:size=%d, offset=%d\n", i, kTessdataFileSuffixes[i],
              entries_[i].size(), offset);
      offset += entries_[i].size();
    }
  }
}

ExtractToFile()

bool tesseract::TessdataManager::ExtractToFile

(

const char *

filename

)

Extracts tessdata component implied by the name of the input file from the combined traineddata loaded into TessdataManager. Writes the extracted component to the file indicated by the file name. E.g. if the filename given is somepath/somelang.unicharset, unicharset will be extracted from the data loaded into the TessdataManager and will be written to somepath/somelang.unicharset.

Returns

true if the component was successfully extracted, false if the component was not present in the traineddata loaded into TessdataManager.

Definition at line 251 of file tessdatamanager.cpp.

                                                        {
  TessdataType type = TESSDATA_NUM_ENTRIES;
  ASSERT_HOST(
      tesseract::TessdataManager::TessdataTypeFromFileName(filename, &type));
  if (entries_[type].empty()) return false;
  return SaveDataToFile(entries_[type], filename);
}

GetComponent() [1/2]

bool tesseract::TessdataManager::GetComponent	(	TessdataType	type,
		TFile *	fp
	)

Definition at line 173 of file tessdatamanager.cpp.

                                                               {
  if (!is_loaded_ && !Init(data_file_name_.string())) return false;
  const TessdataManager *const_this = this;
  return const_this->GetComponent(type, fp);
}

GetComponent() [2/2]

bool tesseract::TessdataManager::GetComponent	(	TessdataType	type,
		TFile *	fp
	)		const

Definition at line 181 of file tessdatamanager.cpp.

                                                                     {
  ASSERT_HOST(is_loaded_);
  if (entries_[type].empty()) return false;
  fp->Open(&entries_[type][0], entries_[type].size());
  fp->set_swap(swap_);
  return true;
}

GetDataFileName()

const STRING& tesseract::TessdataManager::GetDataFileName ( ) const

inline

Definition at line 186 of file tessdatamanager.h.

{ return data_file_name_; }

Init()

bool tesseract::TessdataManager::Init

(

const char *

data_file_name

)

Opens and reads the given data file right now.

Returns

true on success.

Definition at line 58 of file tessdatamanager.cpp.

                                                     {
  GenericVector<char> data;
  if (reader_ == nullptr) {
    if (!LoadDataFromFile(data_file_name, &data)) return false;
  } else {
    if (!(*reader_)(data_file_name, &data)) return false;
  }
  return LoadMemBuffer(data_file_name, &data[0], data.size());
}

is_loaded()

bool tesseract::TessdataManager::is_loaded ( ) const

inline

Definition at line 134 of file tessdatamanager.h.

{ return is_loaded_; }

IsBaseAvailable()

bool tesseract::TessdataManager::IsBaseAvailable ( ) const

inline

Definition at line 177 of file tessdatamanager.h.

                               {
    return !entries_[TESSDATA_UNICHARSET].empty() &&
           !entries_[TESSDATA_INTTEMP].empty();
  }

IsComponentAvailable()

bool tesseract::TessdataManager::IsComponentAvailable ( TessdataType  type ) const

inline

Definition at line 161 of file tessdatamanager.h.

                                                     {
    return !entries_[type].empty();
  }

IsLSTMAvailable()

bool tesseract::TessdataManager::IsLSTMAvailable ( ) const

inline

Definition at line 183 of file tessdatamanager.h.

{ return !entries_[TESSDATA_LSTM].empty(); }

LoadFileLater()

void tesseract::TessdataManager::LoadFileLater

(

const char *

data_file_name

)

Definition at line 53 of file tessdatamanager.cpp.

                                                              {
  Clear();
  data_file_name_ = data_file_name;
}

LoadMemBuffer()

bool tesseract::TessdataManager::LoadMemBuffer	(	const char *	name,
		const char *	data,
		int	size
	)

Definition at line 69 of file tessdatamanager.cpp.

                                              {
  Clear();
  data_file_name_ = name;
  TFile fp;
  fp.Open(data, size);
  int32_t num_entries = TESSDATA_NUM_ENTRIES;
  if (fp.FRead(&num_entries, sizeof(num_entries), 1) != 1) return false;
  swap_ = num_entries > kMaxNumTessdataEntries || num_entries < 0;
  fp.set_swap(swap_);
  if (swap_) ReverseN(&num_entries, sizeof(num_entries));
  if (num_entries > kMaxNumTessdataEntries || num_entries < 0) return false;
  GenericVector<int64_t> offset_table;
  offset_table.resize_no_init(num_entries);
  if (fp.FReadEndian(&offset_table[0], sizeof(offset_table[0]), num_entries) !=
      num_entries)
    return false;
  for (int i = 0; i < num_entries && i < TESSDATA_NUM_ENTRIES; ++i) {
    if (offset_table[i] >= 0) {
      int64_t entry_size = size - offset_table[i];
      int j = i + 1;
      while (j < num_entries && offset_table[j] == -1) ++j;
      if (j < num_entries) entry_size = offset_table[j] - offset_table[i];
      entries_[i].resize_no_init(entry_size);
      if (fp.FRead(&entries_[i][0], 1, entry_size) != entry_size) return false;
    }
  }
  if (entries_[TESSDATA_VERSION].empty()) {
    SetVersionString("Pre-4.0.0");
  }
  is_loaded_ = true;
  return true;
}

OverwriteComponents()

bool tesseract::TessdataManager::OverwriteComponents	(	const char *	new_traineddata_filename,
		char **	component_filenames,
		int	num_new_components
	)

Gets the individual components from the data_file_ with which the class was initialized. Overwrites the components specified by component_filenames. Writes the updated traineddata file to new_traineddata_filename.

Definition at line 232 of file tessdatamanager.cpp.

                            {
  // Open the files with the new components.
  for (int i = 0; i < num_new_components; ++i) {
    TessdataType type;
    if (TessdataTypeFromFileName(component_filenames[i], &type)) {
      if (!LoadDataFromFile(component_filenames[i], &entries_[type])) {
        tprintf("Failed to read component file:%s\n", component_filenames[i]);
        return false;
      }
    }
  }

  // Write updated data to the output traineddata file.
  return SaveFile(new_traineddata_filename, nullptr);
}

OverwriteEntry()

void tesseract::TessdataManager::OverwriteEntry	(	TessdataType	type,
		const char *	data,
		int	size
	)

Definition at line 104 of file tessdatamanager.cpp.

                                               {
  is_loaded_ = true;
  entries_[type].resize_no_init(size);
  memcpy(&entries_[type][0], data, size);
}

SaveFile()

bool tesseract::TessdataManager::SaveFile	(	const STRING &	filename,
		FileWriter	writer
	)		const

Definition at line 112 of file tessdatamanager.cpp.

                                                        {
  ASSERT_HOST(is_loaded_);
  GenericVector<char> data;
  Serialize(&data);
  if (writer == nullptr)
    return SaveDataToFile(data, filename);
  else
    return (*writer)(data, filename);
}

Serialize()

void tesseract::TessdataManager::Serialize

(

GenericVector< char > *

data

)

const

Definition at line 124 of file tessdatamanager.cpp.

                                                               {
  ASSERT_HOST(is_loaded_);
  // Compute the offset_table and total size.
  int64_t offset_table[TESSDATA_NUM_ENTRIES];
  int64_t offset = sizeof(int32_t) + sizeof(offset_table);
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    if (entries_[i].empty()) {
      offset_table[i] = -1;
    } else {
      offset_table[i] = offset;
      offset += entries_[i].size();
    }
  }
  data->init_to_size(offset, 0);
  int32_t num_entries = TESSDATA_NUM_ENTRIES;
  TFile fp;
  fp.OpenWrite(data);
  fp.FWrite(&num_entries, sizeof(num_entries), 1);
  fp.FWrite(offset_table, sizeof(offset_table), 1);
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    if (!entries_[i].empty()) {
      fp.FWrite(&entries_[i][0], entries_[i].size(), 1);
    }
  }
}

SetVersionString()

void tesseract::TessdataManager::SetVersionString

(

const std::string &

v_str

)

Definition at line 196 of file tessdatamanager.cpp.

                                                             {
  entries_[TESSDATA_VERSION].resize_no_init(v_str.size());
  memcpy(&entries_[TESSDATA_VERSION][0], v_str.data(), v_str.size());
}

swap()

bool tesseract::TessdataManager::swap ( ) const

inline

Definition at line 133 of file tessdatamanager.h.

{ return swap_; }

TessdataTypeFromFileName()

bool tesseract::TessdataManager::TessdataTypeFromFileName ( const char *  filename, TessdataType *  type  )

static

Tries to determine tessdata component file suffix from filename, returns true on success.

Definition at line 272 of file tessdatamanager.cpp.

                                                                   {
  // Get the file suffix (extension)
  const char *suffix = strrchr(filename, '.');
  if (suffix == nullptr || *(++suffix) == '\0') return false;
  return TessdataTypeFromFileSuffix(suffix, type);
}

TessdataTypeFromFileSuffix()

bool tesseract::TessdataManager::TessdataTypeFromFileSuffix ( const char *  suffix, TessdataType *  type  )

static

Fills type with TessdataType of the tessdata component represented by the given file name. E.g. tessdata/eng.unicharset -> TESSDATA_UNICHARSET.

Returns

true if the tessdata component type could be determined from the given file name.

Definition at line 259 of file tessdatamanager.cpp.

                                                                     {
  for (int i = 0; i < TESSDATA_NUM_ENTRIES; ++i) {
    if (strcmp(kTessdataFileSuffixes[i], suffix) == 0) {
      *type = static_cast<TessdataType>(i);
      return true;
    }
  }
  tprintf("TessdataManager can't determine which tessdata"
         " component is represented by %s\n", suffix);
  return false;
}

VersionString()

std::string tesseract::TessdataManager::VersionString

(

)

const

Definition at line 190 of file tessdatamanager.cpp.

                                               {
  return std::string(&entries_[TESSDATA_VERSION][0],
                     entries_[TESSDATA_VERSION].size());
}

---

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

• /home/stweil/src/github/tesseract-ocr/tesseract/ccutil/tessdatamanager.h
• /home/stweil/src/github/tesseract-ocr/tesseract/ccutil/tessdatamanager.cpp