Merge pull request #2668 from stweil/api

Remove STRING from the public Tesseract API

src/api/renderer.h

@@ -24,6 +24,7 @@
 #include <string>  // for std::string
 #include "genericvector.h"
 #include "platform.h"
+#include "strngs.h"             // for STRING
 
 struct Pix;
 

src/ccstruct/boxread.cpp

@@ -56,7 +56,7 @@ bool ReadAllBoxes(int target_page, bool skip_blanks, const STRING& filename,
                   GenericVector<STRING>* box_texts,
                   GenericVector<int>* pages) {
   GenericVector<char> box_data;
-  if (!tesseract::LoadDataFromFile(BoxFileName(filename), &box_data))
+  if (!tesseract::LoadDataFromFile(BoxFileName(filename).c_str(), &box_data))
     return false;
   // Convert the array of bytes to a string, so it can be used by the parser.
   box_data.push_back('\0');

src/ccutil/Makefile.am

@@ -16,7 +16,7 @@ endif
 
 pkginclude_HEADERS = \
     genericvector.h helpers.h \
-    ocrclass.h platform.h serialis.h strngs.h \
+    ocrclass.h platform.h serialis.h \
     unichar.h
 
 noinst_HEADERS = \
@@ -25,7 +25,8 @@ noinst_HEADERS = \
     genericheap.h globaloc.h host.h \
     kdpair.h lsterr.h \
     object_cache.h params.h qrsequence.h sorthelper.h \
-    scanutils.h tessdatamanager.h tprintf.h \
+    scanutils.h strngs.h \
+    tessdatamanager.h tprintf.h \
     unicharcompress.h unicharmap.h unicharset.h unicity_table.h unicodes.h
 
 if !DISABLED_LEGACY_ENGINE

src/ccutil/genericvector.h

@@ -21,13 +21,14 @@
 
 #include <algorithm>
 #include <cassert>
+#include <climits>      // for LONG_MAX
+#include <cstdint>      // for uint32_t
 #include <cstdio>
 #include <cstdlib>
 #include <functional>   // for std::function
 
 #include "helpers.h"
 #include "serialis.h"
-#include "strngs.h"
 
 // Use PointerVector<T> below in preference to GenericVector<T*>, as that
 // provides automatic deletion of pointers, [De]Serialize that works, and
@@ -361,16 +362,11 @@ inline bool LoadDataFromFile(const char* filename, GenericVector<char>* data) {
   return result;
 }
 
-inline bool LoadDataFromFile(const STRING& filename,
-                             GenericVector<char>* data) {
-  return LoadDataFromFile(filename.string(), data);
-}
-
 // The default FileWriter writes the vector of char to the filename file,
 // returning false on error.
 inline bool SaveDataToFile(const GenericVector<char>& data,
-                           const STRING& filename) {
+                           const char* filename) {
-  FILE* fp = fopen(filename.string(), "wb");
+  FILE* fp = fopen(filename, "wb");
   if (fp == nullptr) {
     return false;
   }
@@ -379,17 +375,6 @@ inline bool SaveDataToFile(const GenericVector<char>& data,
   fclose(fp);
   return result;
 }
-// Reads a file as a vector of STRING.
-inline bool LoadFileLinesToStrings(const STRING& filename,
-                                   GenericVector<STRING>* lines) {
-  GenericVector<char> data;
-  if (!LoadDataFromFile(filename.string(), &data)) {
-    return false;
-  }
-  STRING lines_str(&data[0], data.size());
-  lines_str.split('\n', lines);
-  return true;
-}
 
 template <typename T>
 bool cmp_eq(T const& t1, T const& t2) {

src/ccutil/serialis.cpp

@@ -2,7 +2,6 @@
  * File:        serialis.cpp  (Formerly serialmac.h)
  * Description: Inline routines and macros for serialisation functions
  * Author:      Phil Cheatle
- * Created:     Tue Oct 08 08:33:12 BST 1991
  *
  * (C) Copyright 1990, Hewlett-Packard Ltd.
  ** Licensed under the Apache License, Version 2.0 (the "License");
@@ -21,6 +20,7 @@
 #include <cstdio>
 #include "errcode.h"
 #include "genericvector.h"
+#include "strngs.h"             // for STRING
 
 namespace tesseract {
 
@@ -202,9 +202,9 @@ bool TFile::Open(const STRING& filename, FileReader reader) {
   is_writing_ = false;
   swap_ = false;
   if (reader == nullptr)
-    return LoadDataFromFile(filename, data_);
+    return LoadDataFromFile(filename.c_str(), data_);
   else
-    return (*reader)(filename, data_);
+    return (*reader)(filename.c_str(), data_);
 }
 
 bool TFile::Open(const char* data, int size) {
@@ -310,9 +310,9 @@ void TFile::OpenWrite(GenericVector<char>* data) {
 bool TFile::CloseWrite(const STRING& filename, FileWriter writer) {
   ASSERT_HOST(is_writing_);
   if (writer == nullptr)
-    return SaveDataToFile(*data_, filename);
+    return SaveDataToFile(*data_, filename.c_str());
   else
-    return (*writer)(*data_, filename);
+    return (*writer)(*data_, filename.c_str());
 }
 
 int TFile::FWrite(const void* buffer, size_t size, int count) {

src/ccutil/serialis.h

@@ -46,10 +46,10 @@ constexpr size_t countof(T const (&)[N]) noexcept {
 
 // Function to read a GenericVector<char> from a whole file.
 // Returns false on failure.
-using FileReader = bool (*)(const STRING&, GenericVector<char>*);
+using FileReader = bool (*)(const char* filename, GenericVector<char>* data);
 // Function to write a GenericVector<char> to a whole file.
 // Returns false on failure.
-using FileWriter = bool (*)(const GenericVector<char>&, const STRING&);
+using FileWriter = bool (*)(const GenericVector<char>& data, const char* filename);
 
 // Deserialize data from file.
 bool DeSerialize(FILE* fp, char* data, size_t n = 1);

src/ccutil/tessdatamanager.cpp

@@ -157,9 +157,9 @@ bool TessdataManager::SaveFile(const STRING &filename,
   GenericVector<char> data;
   Serialize(&data);
   if (writer == nullptr)
-    return SaveDataToFile(data, filename);
+    return SaveDataToFile(data, filename.c_str());
   else
-    return (*writer)(data, filename);
+    return (*writer)(data, filename.c_str());
 }
 
 // Serializes to the given vector.
@@ -253,7 +253,7 @@ bool TessdataManager::CombineDataFiles(
     FILE *fp = fopen(filename.string(), "rb");
     if (fp != nullptr) {
       fclose(fp);
-      if (!LoadDataFromFile(filename, &entries_[type])) {
+      if (!LoadDataFromFile(filename.c_str(), &entries_[type])) {
         tprintf("Load of file %s failed!\n", filename.string());
         return false;
       }

src/ccutil/tessdatamanager.h

@@ -2,7 +2,6 @@
 // File:        tessdatamanager.h
 // Description: Functions to handle loading/combining tesseract data files.
 // Author:      Daria Antonova
-// Created:     Wed Jun 03 11:26:43 PST 2009
 //
 // (C) Copyright 2009, Google Inc.
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -21,6 +20,7 @@
 #define TESSERACT_CCUTIL_TESSDATAMANAGER_H_
 
 #include "genericvector.h"
+#include "strngs.h"             // for STRING
 
 static const char kTrainedDataSuffix[] = "traineddata";
 

src/lstm/network.h

@@ -27,6 +27,7 @@
 #include "networkio.h"
 #include "serialis.h"
 #include "static_shape.h"
+#include "strngs.h"             // for STRING
 #include "tprintf.h"
 
 struct Pix;

src/training/fileio.h

@@ -2,7 +2,6 @@
  * File:        fileio.h
  * Description: File I/O utilities.
  * Author:      Samuel Charron
- * Created:     Tuesday, July 9, 2013
  *
  * (C) Copyright 2013, Google Inc.
  * Licensed under the Apache License, Version 2.0 (the "License"); you may not
@@ -21,10 +20,25 @@
 #include <cstdio>
 #include <string>
 
+#include "genericvector.h"      // for GenericVector
 #include "platform.h"
+#include "strngs.h"             // for STRING
 
 namespace tesseract {
 
+// Reads a file as a vector of STRING.
+// TODO: Use std::vector and std::string for LoadFileLinesToStrings.
+inline bool LoadFileLinesToStrings(const char* filename,
+                                   GenericVector<STRING>* lines) {
+  GenericVector<char> data;
+  if (!LoadDataFromFile(filename, &data)) {
+    return false;
+  }
+  STRING lines_str(&data[0], data.size());
+  lines_str.split('\n', lines);
+  return true;
+}
+
 // A class to manipulate FILE*s.
 class File {
  public:

src/training/lstmtester.cpp

@@ -16,6 +16,7 @@
 ///////////////////////////////////////////////////////////////////////
 
 #include <thread>               // for std::thread
+#include "fileio.h"             // for LoadFileLinesToStrings
 #include "lstmtester.h"
 #include "genericvector.h"
 
@@ -29,7 +30,7 @@ LSTMTester::LSTMTester(int64_t max_memory)
 // loaded. The arg is a filename of a file that lists the filenames.
 bool LSTMTester::LoadAllEvalData(const STRING& filenames_file) {
   GenericVector<STRING> filenames;
-  if (!LoadFileLinesToStrings(filenames_file, &filenames)) {
+  if (!LoadFileLinesToStrings(filenames_file.c_str(), &filenames)) {
     tprintf("Failed to load list of eval filenames from %s\n",
             filenames_file.string());
     return false;

src/training/lstmtrainer.cpp

@@ -320,7 +320,7 @@ bool LSTMTrainer::MaintainCheckpoints(TestCallback tester, STRING* log_msg) {
     SaveTrainingDump(NO_BEST_TRAINER, this, &best_trainer_);
     if (error_rate < error_rate_of_last_saved_best_ * kBestCheckpointFraction) {
       STRING best_model_name = DumpFilename();
-      if (!SaveDataToFile(best_trainer_, best_model_name)) {
+      if (!SaveDataToFile(best_trainer_, best_model_name.c_str())) {
         *log_msg += " failed to write best model:";
       } else {
         *log_msg += " wrote best model:";
@@ -358,7 +358,7 @@ bool LSTMTrainer::MaintainCheckpoints(TestCallback tester, STRING* log_msg) {
     // Write a current checkpoint.
     GenericVector<char> checkpoint;
     if (!SaveTrainingDump(FULL, this, &checkpoint) ||
-        !SaveDataToFile(checkpoint, checkpoint_name_)) {
+        !SaveDataToFile(checkpoint, checkpoint_name_.c_str())) {
       *log_msg += " failed to write checkpoint.";
     } else {
       *log_msg += " wrote checkpoint.";

src/training/lstmtraining.cpp

@@ -20,6 +20,7 @@
 #endif
 #include <cerrno>
 #include "commontraining.h"
+#include "fileio.h"             // for LoadFileLinesToStrings
 #include "lstmtester.h"
 #include "lstmtrainer.h"
 #include "params.h"