Added format function and nullptr check to worker::process

CMakeLists.txt

@@ -29,7 +29,7 @@ set(LIBS ${LIBS} ${CMAKE_THREAD_LIBS_INIT})
 find_package(PwrUtils REQUIRED)
 set(LIBS ${LIBS} pwrutils)
 
-add_executable(ediText ediText.cpp ConfigParser.cpp functions/Replace.cpp)
+add_executable(ediText ediText.cpp ConfigParser.cpp functions/Replace.cpp functions/Format.cpp)
 target_link_libraries(ediText ${LIBS})
 
 SET(EXECUTABLE_OUTPUT_PATH ${PROJECT_BINARY_DIR}/../bin)

ConfigParser.cpp

@@ -36,6 +36,8 @@ Function* ConfigParser::str2function(std::string name, ptree& config)
 {
     if(name == "replace")
         return new Replace(config);
+    if(name == "format")
+        return new Format(config);
     else 
         return nullptr;
 }

ConfigParser.hpp

 #pragma once
 #include "functions/Function.hpp"
 #include "functions/Replace.hpp"
+#include "functions/Format.hpp"
 #include "textIterators/TextIterator.hpp"
 #include "textIterators/LineIterator.hpp"
 #include "textIterators/TokenIterator.hpp"

README.md

-## ediText
+# ediText
 Highly configurable tool for processing text. Input is processed by chain of functions defined by the user as an argument in JSON format.
 
-# Usage 
+## Usage 
 Exeplary JSON:
 {
     "0":

ediText.cpp

@@ -46,8 +46,18 @@ void ediText_Worker::process(string task_path, ptree &config, string output_path
 
 	do
 	{
-		auto fun = conf.getFunction();
 		auto text = conf.getTextIterator(buffer1);
+		if(text == nullptr)
+		{
+			LOG(WARNING) << "Iterator does not exists";
+			continue;
+		}
+		auto fun = conf.getFunction();
+		if(fun == nullptr)
+		{
+			LOG(WARNING) << "Function does not exists";
+			continue;
+		}
 		while(text->has_more())
 		{
 			buffer2 << fun->process(text->get());

functions/Format.cpp

+#include "functions/Format.hpp"
+
+#include <iostream>
+
+using ptree = boost::property_tree::ptree;
+
+std::string remove_line_breaks(std::string str)
+{
+    static std::regex reg = std::regex("[-–]\n");
+    std::string to("");
+    return std::regex_replace(str, reg, to);
+}
+
+
+std::string remove_page_numbers(std::string str)// TODO
+{
+
+    // static std::regex reg = std::regex("[0-9]");
+    // std::string to("[|]");
+    // return std::regex_replace(str, reg, to);
+    return str;
+}
+
+Format::Format(ptree& config)// TODO
+{
+    tasks.emplace_back(remove_line_breaks);
+    // if(config.get<std::string>("function") == "all")
+    // {
+    //     tasks.emplace_back(remove_line_breaks);
+    //     tasks.emplace_back(remove_page_numbers);
+    // }
+    // else
+    // {
+    //     ;// config.get_child("functions");
+    // }
+        // std::string from = config.get<std::string>("from");
+        // auto flags = std::regex_constants::optimize;
+        // if(!config.get<bool>("case_sensitive", true))
+        //     flags = flags | std::regex_constants::icase;
+        // reg = std::regex(from, flags);
+}
+
+std::string Format::process(std::string str)
+{
+    for(auto& task : tasks)
+    {
+        str = task(str);
+    }
+    return str;
+}

functions/Format.hpp

+#pragma once
+#include <boost/property_tree/ptree.hpp>
+#include <queue>
+#include <functional>
+#include <regex>
+
+#include "functions/Function.hpp"
+
+
+using ptree = boost::property_tree::ptree;
+
+class Format : public Function
+{
+public:
+    Format(ptree& config);
+    std::string process(std::string str);
+private:
+    std::deque<std::function<std::string(std::string)>> tasks;
+};