wchar_sort.c

PB 熟悉的哥们希望大家可以互相学习一下

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "prjdb.h"

#ifndef _WIN32_WCE
	#define WSCMP( s1, s2) wcscoll(s1, s2)
#else
	#define WSCMP( s1, s2) wcscmp(s1, s2)	
#endif

#ifndef MCO_PLATFORM_X64
    #define PAGE_SIZE                        (uint2)128
#else 
    #define PAGE_SIZE                        (uint2)256
#endif 

#define DATABASE_SIZE                    1024*1024*70
#define ADDRESS                          0x2000000

char* dbname = "dbname";
mco_db_h db = 0;

/* Fatal error trap */
static void errhandler(int n)
{

    fprintf(stdout, "\n*** eXtremeDB runtime fatal error: %d\n", n);
    exit( - 1);
}

#define OBJ1	1
#define OBJ2	2
#define OBJ3	4
#define OBJ4	8
#define OBJ5	0x10

typedef struct tagOne
{

    struct tagOne* next;
    uint4 flags;
    uint4 line_no;
    void* line;
    uint4 line_len;

} one_t, * one_h;

one_h root;
one_h* plast = &root;
uint4 line_cnt = 0;

void DumpShorts(char* rem, wchar_t* ptr, uint2 len)
{

    uint2 i;

    if (rem)
    {
        fprintf(stdout, "%s:", rem);
    }

    for (i = 0; i < len; i++)
    {
        fprintf(stdout, "0x%08hx ", ptr[i]);
    };

    fprintf(stdout, "\n");
};

int LoadFile(char* fn)
{

    FILE* f;
    void* mem_buf;
    uint4 mem_buf_sz;
    wchar_t* p, * n, * q;
    unsigned short* _p, * _n;
    MCO_RET rc;
    uint4 len_b;
    one_h one;

    if (0 == (f = fopen(fn, "rb")))
    {
        return 1;
    }

    fseek(f, 0, SEEK_END);
    mem_buf_sz = ftell(f);
    fseek(f, 0, SEEK_SET);
    if (mem_buf_sz < 2)
    {
        printf("The file is not a valid unicode text file.\n");
        return 1;
    };
    // if file contain wchar size(2)

    p = _p = mem_buf = malloc(mem_buf_sz + 2);

    fread(mem_buf, 1, mem_buf_sz, f);
    fclose(f);

    _p[(mem_buf_sz / 2)] = 0;


    if (_p[0] != 0xFEFF)
    {

        _n = _p;
        while (((unsigned long)_n - (unsigned long)_p) < mem_buf_sz)
        {
            *_n = (*_n &0xFF00) >> 8 | (*_n &0xFF) << 8;
            _n++;
        };
    };

    if (sizeof(wchar_t) == 4)
    {
        long ps = (mem_buf_sz / 2) + 1;
        p = malloc((mem_buf_sz / 2)* sizeof(wchar_t) + sizeof(wchar_t));
        _n = _p;
        n = p;
        while (ps)
        {
            *n = * _n;
            _n++;
            n++;
            ps--;
        }
        free(_p);
        mem_buf = _p = p;
    }

    if (p[0] != 0xFEFF)
    {
        printf("The file is not a valid unicode text file.\n");
        free(mem_buf);
        return 1;
    };

    p++;
    do
    {

        mco_trans_h t;
        Obj1 obj1;
        Obj2 obj2;
        Obj3 obj3;
        Obj4 obj4;
        Obj5 obj5;
        uint2 i, j, sz;
        wchar_t* pw;
        uint2 pw_len;

        n = p;
        while (*n != 0xD && * n != 0xA && * n != 0)n++;

        len_b = (unsigned long)n - (unsigned long)p;

        one = (one_h)malloc(sizeof(one_t));
        one->next = 0;
        one->flags = 0;
        one->line_len = len_b;
        one->line = malloc(len_b + sizeof(wchar_t));
        memcpy(one->line, p, len_b);

        q = (wchar_t*)(((mco_puint)one->line) + len_b);
        *q = 0;
        one->line_no = line_cnt;

        line_cnt++;
        *plast = one;
        plast = &one->next;

        pw = (wchar_t*)one->line;
        pw_len = (uint2)(one->line_len / sizeof(wchar_t));

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_WRITE, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            Obj1_new(t, &obj1);

            Obj1_line_no_put(&obj1, one->line_no);

            Obj1_line_put(&obj1, pw, pw_len);

            sz = pw_len;
            if (sz > 16)
            {
                sz = 16;
            }
            Obj1_part_put(&obj1, pw, sz);

            sz = pw_len / Obj1_parts_a_length;
            for (i = 0; i < Obj1_parts_a_length; i++)
            {
                Obj1_parts_a_put(&obj1, i, &pw[i* sz], sz);
            };

            sz = pw_len / Obj1_parts_s_length;
            for (i = 0; i < Obj1_parts_s_length; i++)
            {
                Obj1_parts_s_put(&obj1, i, &pw[i* sz], sz);
            };

            sz = pw_len / 3;
            Obj1_vect_a_alloc(&obj1, sz);
            for (i = 0; i < sz; i++)
            {
                Obj1_vect_a_put(&obj1, i, &pw[i* 3], 3);
            };

            sz = 1;
            for (i = 0; i < pw_len; i++)
            {
                if (pw[i] == 0x20)
                {
                    sz++;
                }
            };
            Obj1_vect_s_alloc(&obj1, sz);
            i = j = 0;
            while (i < pw_len)
            {

                sz = 0;
                while ((i + sz) < pw_len && pw[i + sz] != 0x20)
                {
                    sz++;
                }

                Obj1_vect_s_put(&obj1, j, &pw[i], sz);

                j++;
                i += sz + 1;
            };

            if (MCO_S_OK == (rc = mco_trans_commit(t)))
            {
                one->flags |= OBJ1;
            }
            else
            {
                printf("Error: %d at %d ", rc, __LINE__);
            };

        }
        else
        {
            printf("Can't open a transaction. Error code: %d\n", rc);
        };

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_WRITE, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            Obj2_new(t, &obj2);

            Obj2_line_put(&obj2, pw, pw_len);

            if (MCO_S_OK == (rc = mco_trans_commit(t)))
            {
                one->flags |= OBJ2;
            }
            else
            {
                if (rc != MCO_S_DUPLICATE)
                {
                    printf("Error: %d at %d ", rc, __LINE__);
                }
            };

        }
        else
        {
            printf("Can't open a transaction. Error code: %d\n", rc);
        };

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_WRITE, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            Obj3_new(t, &obj3);

            Obj3_line_put(&obj3, pw, pw_len);

            if (MCO_S_OK == (rc = mco_trans_commit(t)))
            {
                one->flags |= OBJ3;
            }
            else
            {
                if (rc != MCO_S_DUPLICATE)
                {
                    printf("Error: %d at %d ", rc, __LINE__);
                }
            };

        }
        else
        {
            printf("Can't open a transaction. Error code: %d\n", rc);
        };

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_WRITE, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            Obj4_new(t, &obj4);

            Obj4_line_put(&obj4, pw, pw_len);

            if (MCO_S_OK == (rc = mco_trans_commit(t)))
            {
                one->flags |= OBJ4;
            }
            else
            {
                if (rc != MCO_S_DUPLICATE)
                {
                    printf("Error: %d at %d ", rc, __LINE__);
                }
            };

        }
        else
        {
            printf("Can't open a transaction. Error code: %d\n", rc);
        };

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_WRITE, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            Obj5_new(t, &obj5);

            Obj5_line_put(&obj5, pw, pw_len);

            if (MCO_S_OK == (rc = mco_trans_commit(t)))
            {
                one->flags |= OBJ5;
            }
            else
            {
                if (rc != MCO_S_DUPLICATE)
                {
                    printf("Error: %d at %d ", rc, __LINE__);
                }
            };

        }
        else
        {
            printf("Can't open a transaction. Error code: %d\n", rc);
        };

        if (*n == 0)
        {
            break;
        }

        while (*n != 0xA)
        {
            n++;
        }

        p = n + 1;

    }
    while (1)
        ;

    free(mem_buf);
    /*
    one = root;
    while ( one ) {
    DumpShorts( "chain", one->line, one->line_len / 2 );
    one = one->next;
    };
     */
    return 0;
};

int CheckContent()
{

    MCO_RET rc = MCO_S_OK;
    mco_trans_h t;
    Obj1 obj;
    one_h one = root;


    while (rc == MCO_S_OK && one)
    {

        /* begin a read-write transaction */
        if ((rc = mco_trans_start(db, MCO_READ_ONLY, MCO_TRANS_FOREGROUND, &t)) == MCO_S_OK)
        {

            if (MCO_S_OK == (rc = Obj1_by_line_no_find(t, one->line_no, &obj)))
            {

                uint2 i, j, sz = 0;
                void* p;
                wchar_t* pw;
                wchar_t part[16];


                /* wstring */
                Obj1_line_size(&obj, &sz);

                if (sz* sizeof(wchar_t) != one->line_len)
                {
                    printf("Invalid string size detected at %d. LineNo:%d.\n", __LINE__, one->line_no);
                    printf("expected:%d found:%d\n", one->line_len / sizeof(wchar_t), sz);
                    break;
                };
                p = malloc((sz + 1)* sizeof(wchar_t));
                Obj1_line_get(&obj, p, (uint2)(sz + 1), &sz);
                if (sz* sizeof(wchar_t) != one->line_len || 0 != memcmp(one->line, p, one->line_len))
                {
                    printf("Invalid string detected at %d. LineNo:%d.\n", __LINE__, one->line_no);
                    printf("expected:%d found:%d\n", one->line_len / sizeof(wchar_t), sz);
                    DumpShorts("expected", one->line, (uint2)(one->line_len / 2));
                    DumpShorts("   found", p, sz);
                    break;
                };
                if (((wchar_t*)p)[sz] != 0)
                {
                    printf("No zero-terminator at the end of the string at %d. LineNo:%d. ", __LINE__, one->line_no);
                    break;
                };
                free(p);

                /* nchar */
                sz = sizeof(part) / sizeof(part[0]);
                Obj1_part_get(&obj, part, sz);
                if (sz > (uint2)(one->line_len / sizeof(wchar_t)))
                {
                    sz = (uint2)(one->line_len / sizeof(wchar_t));
                }
                if (0 != memcmp(part, one->line, sz* sizeof(wchar_t)))
                {
                    printf("Invalid array detected at %d. LineNo:%d.\n", __LINE__, one->line_no);
                    DumpShorts("expected", one->line, (uint2)(one->line_len / sizeof(wchar_t)));
                    DumpShorts("   found", part, sz);
                    break;
                };

                /* array of nchar */
                sz = (uint2)((one->line_len / sizeof(wchar_t)) / Obj1_parts_a_length);
                pw = one->line;
                for (i = 0; i < Obj1_parts_a_length; i++)
                {
                    Obj1_parts_a_at(&obj, i, part, sizeof(part) / sizeof(part[0]));

                    if (0 != memcmp(part, &pw[i* sz], sz > 10 ? 10 : sz))
                    {
                        printf("Invalid array item detected at %d. LineNo:%d.\n", __LINE__, one->line_no);
                        DumpShorts("expected", &pw[i* sz], sz);
                        DumpShorts("   found", part, sz);
                        break;
                    };
                };
                if (i < Obj1_parts_a_length)
                {
                    break;
                }

                /* array of nstring */
                sz = (uint2)((one->line_len / sizeof(wchar_t)) / Obj1_parts_s_length);
                pw = one->line;
                for (i = 0; i < Obj1_parts_s_length; i++)
                {

                    Obj1_parts_s_at_len(&obj, i, &j);
                    if (j != sz)
                    {
                        printf("Invalid string size detected at %d. LineNo:%d.\n", __LINE__, one->line_no);
                        printf("expected:%d found:%d\n", sz, j);
                        break;
                    };

                    p = malloc((j + 1)* sizeof(wchar_t));
                    Obj1_parts_s_at(&obj, i, p, (uint2)(j + 1), &j);

                    if (sz != j || 0 != memcmp(&pw[i* sz], p, sz))
                    {
                        printf("The strings are not equal at %d. LineNo:%d.\n", __LINE__, one->line_no);
                        printf("expected:%d found:%d\n", sz, j);
                        DumpShorts("expected", &pw[i* sz], sz);
                        DumpShorts("   found", p, sz);
                        break;
                    };
                    if (((wchar_t*)p)[sz] != 0)
                    {
                        printf("No zero-terminator at the end of the string at %d. LineNo:%d. ", __LINE__, one->line_no)
                               ;
                        break;
                    };
                    free(p);
                };
                if (i < Obj1_parts_s_length)
                {
                    break;
                }