From d27b8d75fab2059f8c72fd0d0a04d5613354ee2b Mon Sep 17 00:00:00 2001
From: "HuYingzhuo(hugo/hyzboy)" <hyzboy@gmail.com>
Date: Thu, 13 Jul 2023 14:30:08 +0800
Subject: [PATCH] added TextInputStream

---
 inc/hgl/io/TextInputStream.h |  78 +++++++++++++++++++++
 src/CMakeLists.txt           |   2 +
 src/IO/TextInputStream.cpp   | 132 +++++++++++++++++++++++++++++++++++
 3 files changed, 212 insertions(+)
 create mode 100644 inc/hgl/io/TextInputStream.h
 create mode 100644 src/IO/TextInputStream.cpp

diff --git a/inc/hgl/io/TextInputStream.h b/inc/hgl/io/TextInputStream.h
new file mode 100644
index 0000000..0411e12
--- /dev/null
+++ b/inc/hgl/io/TextInputStream.h
@@ -0,0 +1,78 @@
+#pragma once
+#include<hgl/type/DataType.h>
+#include<hgl/io/InputStream.h>
+
+namespace hgl
+{
+    namespace io
+    {
+        /**
+        * 文本输入流<br>
+        * 它与TextOutputStream并无对应关注，主要作用是方便超大文本的读取与解晰。
+        */
+        class TextInputStream
+        {
+        private:
+
+            InputStream *input_stream;                                                              ///<输入流
+
+            uint8 *buffer;                                                                          ///<缓冲区
+            int32 buffer_size;                                                                      ///<缓冲区大小
+            int32 cur_buf_size;                                                                     ///<当前缓冲区大小
+
+            int64 stream_pos,stream_size;                                                           ///<流当前位置/大小
+
+            ByteOrderMask bom;                                                                      ///<BOM头
+
+        private:
+
+            template<typename T> int Parse(const T *);
+
+            int TextBlockParse();                                                                   ///<文本块解析
+
+        public:
+
+            TextInputStream(InputStream *i,const int buf_size=HGL_SIZE_1MB);
+            virtual ~TextInputStream()
+            {
+                SAFE_CLEAR_ARRAY(buffer);
+            }
+
+            virtual bool OnBOM(const ByteOrderMask &){return true;}                                  ///<读取到BOM头的回调函数
+
+            /**
+            * 读取到一行文本的回调函数(ansi/utf8)
+            * @param text 读取到的文本内容
+            * @param len 读取到的文本字长度
+            * @param line_end 当前行是否结束
+            */
+            virtual bool OnLine(const char *text,const int len,const bool line_end){return true;}
+
+            /**
+            * 读取到一行文本的回调函数(utf16le/utf16be)
+            * @param text 读取到的文本内容
+            * @param len 读取到的文本字长度
+            * @param line_end 当前行是否结束
+            */
+            virtual bool OnLine(const u16char *text,const int len,const bool line_end){return true;}
+
+            /**
+            * 读取到一行文本的回调函数(utf32le/utf32be)
+            * @param text 读取到的文本内容
+            * @param len 读取到的文本字长度
+            * @param line_end 当前行是否结束
+            */
+            virtual bool OnLine(const u32char *text,const int len,const bool line_end){return true;}
+
+            virtual bool OnEnd(){return true;}                                                      ///<读取结束的回调函数
+            virtual bool OnReadError(){return true;}                                                ///<读取错误的回调函数
+            virtual bool OnParseError(){return true;}                                               ///<解析错误的回调函数
+
+            /**
+            * 运行并解晰文本
+            * @return 解析出的文本行数
+            */
+            virtual int Run();
+        };//class TextInputStream
+    }//namespace io
+}//namespace hgl
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 3255d4f..2e1e7d0 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -114,9 +114,11 @@ SET(IO_FILE_FILES   ${IO_INCLUDE_PATH}/FileAccess.h
                   ${IO_INCLUDE_PATH}/FileInputStream.h
                   ${IO_INCLUDE_PATH}/FileOutputStream.h
                   ${IO_INCLUDE_PATH}/RandomAccessFile.h
+                  ${IO_INCLUDE_PATH}/TextInputStream.h
                   ${IO_INCLUDE_PATH}/TextOutputStream.h
                   IO/FileAccess.cpp
                   IO/FileInputStream.cpp
+                  IO/TextInputStream.cpp
                   IO/FileOutputStream.cpp
                   IO/RandomAccessFile.cpp)
 
diff --git a/src/IO/TextInputStream.cpp b/src/IO/TextInputStream.cpp
new file mode 100644
index 0000000..efe15f9
--- /dev/null
+++ b/src/IO/TextInputStream.cpp
@@ -0,0 +1,132 @@
+#include<hgl/io/TextInputStream.h>
+
+namespace hgl
+{
+    namespace io
+    {
+        TextInputStream::TextInputStream(InputStream *i,const int buf_size)
+        {
+            input_stream=i;
+            buffer_size=buf_size;
+            buffer=new uint8[buffer_size];
+            cur_buf_size=0;
+
+            stream_pos=0;
+            stream_size=input_stream->Available();
+
+            bom=ByteOrderMask::NONE;
+        }
+
+        template<typename T> int TextInputStream::Parse(const T *p)
+        {
+            const T *sp=(const T *)p;
+            const T *end=(const T *)(buffer+cur_buf_size);
+
+            int line_count=0;
+
+            while(p<end)
+            {
+                if(*p=='\n')
+                {
+                    OnLine(sp,p-sp,true);
+                    ++line_count;
+                    ++p;
+                    sp=p;
+                }
+                else
+                if(*p=='\r')
+                {
+                    ++p;
+                    if(*p=='\n')
+                    {
+                        OnLine(sp,p-sp-1,true);
+                        ++line_count;
+                        ++p;
+                        sp=p;
+                    }
+                }
+                else
+                    ++p;
+            }
+
+            if(sp<end)
+            {
+                OnLine(sp,end-sp,false);
+                ++line_count;
+            }
+
+            return line_count;
+        }
+
+        int TextInputStream::TextBlockParse()
+        {
+            uint8 *p=buffer;
+
+            if(stream_pos==0)       //�ʼ���Ǽ��һ��BOMͷ
+            {
+                if(cur_buf_size>=2)
+                {
+                    bom=CheckBOM(p);
+
+                    const BOMFileHeader *bfh=GetBOM(bom);
+
+                    if(bfh)
+                    {
+                        if(bfh->size==cur_buf_size)
+                            return(0);
+
+                        p+=bfh->size;
+                    }
+                }
+            }
+
+            if(bom==ByteOrderMask::UTF16LE||bom==ByteOrderMask::UTF16BE)
+                return Parse<u16char>((u16char *)p);
+            else
+            if(bom==ByteOrderMask::UTF32LE||bom==ByteOrderMask::UTF32BE)
+                return Parse<u32char>((u32char *)p);
+            else
+                return Parse<char>((char *)p);
+        }
+
+        int TextInputStream::Run()
+        {
+            if(!input_stream)return(-1);
+
+            int64 read_size;
+
+            int result;
+            int line_count=0;
+
+            while(stream_pos<stream_size)
+            {
+                read_size=stream_size-stream_pos;
+
+                if(read_size>buffer_size)
+                    read_size=buffer_size;
+
+                cur_buf_size=input_stream->Read(buffer,read_size);
+
+                if(cur_buf_size!=read_size)
+                {
+                    OnReadError();
+                    return(-1);
+                }
+
+                result=TextBlockParse();
+
+                if(result<0)
+                {
+                    OnReadError();
+                    return(result);
+                }
+
+                line_count+=result;
+
+                stream_pos+=cur_buf_size;
+            }
+
+            return line_count;
+        }
+    }//namespace io
+}//namespace hgl