VS的VC提取PDF中stream流的程序

wxleasyland@139.com

2021.12

vs2017对话框MFC。

long myoutputSTREAM(long j, long lsize, char* buffer, int mode_return,  CString filename)

{

    long l_start;

    l_start = j;

while(1)

{

    if (j >= lsize) break;

    if (mode_return == 1 && buffer[j + 2] == 'e' && buffer[j + 3] == 'n' && buffer[j + 4] == 'd' && buffer[j + 5] == 's' && buffer[j + 6] == 't' &&buffer[j + 7] == 'r' &&buffer[j + 8] == 'e' &&buffer[j + 9] == 'a' &&buffer[j + 10] == 'm')  //找到结尾了

    {   //回车2字节 + 9个字符 + 回车2字节

       j += 13;

       goto myoutputSTREAM_out;

    }

    if (   ( mode_return ==0 || mode_return == 2 ) && buffer[j + 1] == 'e' && buffer[j + 2] == 'n' && buffer[j + 3] == 'd' && buffer[j + 4] == 's' && buffer[j + 5] == 't' &&buffer[j + 6] == 'r' &&buffer[j + 7] == 'e' &&buffer[j + 8] == 'a' &&buffer[j + 9] == 'm')  //找到结尾了

    {   //回车1字节 + 9个字符 + 回车1字节

       j += 11;

       goto myoutputSTREAM_out;

    }

    j++;

}

    return j;

myoutputSTREAM_out:

    FILE * STREAMfile;

   

    STREAMfile = fopen(filename, "wb");

    if (NULL == STREAMfile)

    {

       printf("ERROR output File open failed\n");

       return j;

    }

    if (mode_return == 1)    fwrite(buffer+ l_start, 1, j-13 - l_start, STREAMfile);    //输出时不含回车

    else fwrite(buffer + l_start, 1, j-11 - l_start, STREAMfile);     //输出时不含回车

    fclose(STREAMfile);

    return j;

}

然后，在按钮程序中，加入：

    // TODO: 在此添加控件通知处理程序代码

    int mode_return = 0;  //回车模式

    long lsize = 0;    //PDF文件大小

    char * buffer;     //存PDF文件

    int num_output=0;      //output文件的编号

    FILE * PDFfile;

    CString cstr, cstrin, cstrout;

    GetDlgItem(IDC_EDIT1)->GetWindowText(cstrin);

    PDFfile = fopen(cstrin, "rb");

    if (NULL == PDFfile)

    {

       printf("ERROR File open failed\n");

       return;

    }

    //获取文件大小

    fseek(PDFfile, 0, SEEK_END);

    lsize = ftell(PDFfile);

    rewind(PDFfile);

    if (lsize > 2000000000)

    {

       printf("ERROR File too large\n");

       fclose(PDFfile);

       return;

    }

    buffer = (char*)malloc(lsize+100);  //多申请一些保险一点

    if (NULL == buffer)

    {

       printf("ERROR Memory malloc failed\n");

       fclose(PDFfile);

       return;

    }

    fread(buffer, sizeof(char), lsize, PDFfile);

    fclose(PDFfile);

    //printf("%s\n",argv[i]);

    printf(" %ld  ", lsize);

    long j;

    j = 0;

    if (buffer[8] == (char)0x0A  )    //回车判断

       {

           mode_return = 0;

           j += 9;

       }

    else if (buffer[8] == (char)0x0D )

       {

              if (buffer[9] == (char)0x0A) { mode_return = 1; j += 10; }

              else { mode_return = 2;    j += 9; }

       }

    else

    {

       printf("ERROR return detect\n");

           return;

    }

    while (1)

    {

       if (j >= lsize)

           break;

       if (buffer[j - 1] != 'd' && buffer[j] == 's' && buffer[j + 1] == 't' &&buffer[j + 2] == 'r' &&buffer[j + 3] == 'e' &&buffer[j + 4] == 'a' &&buffer[j + 5] == 'm')  //找到了

       {

           if (mode_return == 1) j += 8;

           else j += 7;

           cstrout = cstrin;

           num_output++;      //output文件的编号

           cstr.Format("-%d", num_output);

           cstrout += cstr;

           j = myoutputSTREAM(j, lsize, buffer, mode_return, cstrout);

           continue;

       }

       j++;

    }