// 将机内格式记录构造为ISO2709格式记录。
// parameters:
// baMARC [in]机内格式记录。已经通过适当Encoding对象转换为ByteArray了
// baResult [out]ISO2709格式记录。
// return:
// -1 error
// 0 succeed
public static int BuildISO2709Record(byte[] baMARC,
out byte[] baResult)
{
int nLen;
byte[] baMuci = null; // 目次区
byte[] baBody = null; // 数据区
byte[] baFldName = null;
string strFldLen;
string strFldStart;
byte[] baFldContent = null;
int nStartPos;
int nFldLen;
int nFldStart;
bool bEnd = false;
int nPos;
int nRecLen = 0;
baResult = null;
if (baMARC == null)
{
return(-1);
}
if (baMARC.Length < 24)
{
return(-1);
}
// 2018/3/8
if (baMARC[0] == 0 ||
baMARC[1] == 0)
{
throw new Exception("ISO2709 格式无法使用编码方式 UCS-2 (UTF-16)");
}
MarcHeaderStruct header = new MarcHeaderStruct(baMARC);
/*
* ISO2709ANSIHEADER header;
* memcpy(&header,
* (LPCSTR)advstrMARC,
* sizeof(header));
*/
nLen = baMARC.Length;
for (nStartPos = 24, nFldStart = 0; ;)
{
nPos = ByteArray.IndexOf(baMARC, (byte)FLDEND, nStartPos);
// nPos = FindCharInStringA((LPCSTR)advstrMARC, FLDEND, nStartPos);
if (nPos == -1)
{
nFldLen = nLen - nStartPos;
bEnd = true;
}
else
{
nFldLen = nPos - nStartPos + 1;
}
if (nFldLen < 3)
{
goto SKIP;
}
// strFldName = advstrMARC.MidA(nStartPos, 3);
baFldName = new byte[3];
Array.Copy(baMARC,
nStartPos,
baFldName, 0,
3);
// advstrFldContent = advstrMARC.MidA(nStartPos + 3, nFldLen - 3);
baFldContent = new byte[nFldLen - 3];
Array.Copy(baMARC,
nStartPos + 3,
baFldContent, 0,
nFldLen - 3);
//advstrFldLen.Format("%04d", nFldLen - 3);
strFldLen = Convert.ToString(nFldLen - 3);
strFldLen = strFldLen.PadLeft(4, '0');
// advstrFldStart.Format("%05d", nFldStart);
strFldStart = Convert.ToString(nFldStart);
strFldStart = strFldStart.PadLeft(5, '0');
nFldStart += nFldLen - 3;
// advstrMuci += (LPCSTR)advstrFldName;
baMuci = ByteArray.Add(baMuci, baFldName);
// advstrMuci += (LPCSTR)advstrFldLen;
baMuci = ByteArray.Add(baMuci, Encoding.UTF8.GetBytes(strFldLen));
// advstrMuci += (LPCSTR)advstrFldStart;
baMuci = ByteArray.Add(baMuci, Encoding.UTF8.GetBytes(strFldStart));
baBody = ByteArray.Add(baBody, baFldContent);
SKIP:
if (bEnd)
{
break;
}
nStartPos = nPos + 1;
}
nRecLen = baMuci.Length + 1
+ baBody.Length + 1 + 24;
/*
* advstrText.Format(
* "%05d",
* nRecLen);
*
* memcpy(header.reclen,
* (LPCSTR)advstrText,
* advstrText.GetLengthA());
*/
header.RecLength = nRecLen;
/*
* advstrText.Format(
* "%05d",
* sizeof(header) + advstrMuci.GetLengthA() + 1);
* memcpy(header.baseaddr,
* (LPCSTR)advstrText,
* advstrText.GetLengthA());
*/
header.BaseAddress = 24 + baMuci.Length + 1;
// ForceUNIMARCHeader(&header);
/*
* In USMARC format, leader postion 09, one character indicate the character coding scheme:
*
* 09 - Character coding scheme
* Identifies the character coding scheme used in the record.
# - MARC-8
# a - UCS/Unicode
# (http://lcweb.loc.gov/marc/bibliographic/ecbdldrd.html)
*/
//baTarget.SetSize(nRecLen);
/*
* memcpy(baTarget.GetData(),
* (char *)&header,
* sizeof(header));
*/
baResult = ByteArray.Add(baResult, header.GetBytes());
/*
* memcpy((char *)baTarget.GetData() + sizeof(header),
* (LPCSTR)advstrMuci,
* advstrMuci.GetLengthA());
*/
baResult = ByteArray.Add(baResult, baMuci);
/*
*((char *)baTarget.GetData() + sizeof(header) + advstrMuci.GetLengthA())
* = FLDEND;
*/
baResult = ByteArray.Add(baResult, (byte)FLDEND);
/*
* memcpy((char *)baTarget.GetData() + sizeof(header)+ advstrMuci.GetLengthA() + 1,
* (LPCSTR)advstrBody,
* advstrBody.GetLengthA());
*/
baResult = ByteArray.Add(baResult, baBody);
/*
*((char *)baTarget.GetData() + nRecLen - 1)
* = RECEND;
*/
baResult = ByteArray.Add(baResult, (byte)RECEND);
return(0);
}
// 将ISO2709格式记录转换为字段数组
// aResult的每个元素为byte[]型，内容是一个字段。第一个元素是头标区，一定是24bytes
// return:
// -1 一般性错误
// -2 MARC格式错误
public static int Cvt2709ToFieldArray(
Encoding encoding, // 2007/7/11
byte[] s,
out List <byte[]> aResult, // out
out string strErrorInfo)
{
strErrorInfo = "";
aResult = new List <byte[]>();
// const char *sopp;
int maxbytes = 2000000; // 约2000K，防止攻击
// const byte RECEND = 29;
// const byte FLDEND = 30;
// const byte SUBFLD = 31;
if (encoding.Equals(Encoding.Unicode) == true)
{
throw new Exception("UCS2编码方式应当使用 ForceCvt2709ToFieldArray()，而不是 Cvt2709ToFieldArray()");
}
MarcHeaderStruct header = new MarcHeaderStruct(encoding, s);
{
// 输出头标区
byte[] tarray = null;
tarray = new byte[24];
Array.Copy(s, 0, tarray, 0, 24);
// 2014/5/9
// 防范头标区出现 0 字符
for (int j = 0; j < tarray.Length; j++)
{
if (tarray[j] == 0)
{
tarray[j] = (byte)'*';
}
}
aResult.Add(tarray);
}
int somaxlen;
int reclen, baseaddr, lenoffld, startposoffld;
int len, startpos;
// char *dirp;
int offs = 0;
int t = 0;
int i;
// char temp[30];
somaxlen = s.Length;
try
{
reclen = header.RecLength;
}
catch (FormatException ex)
{
strErrorInfo = "头标区开始5字符 '" + header.RecLengthString + "' 不是纯数字 :" + ex.Message;
// throw(new MarcException(strErrorInfo));
goto ERROR2;
}
if (reclen > somaxlen)
{
strErrorInfo = "头标区头5字符表示的记录长度"
+ Convert.ToString(reclen)
+ "大于源缓冲区整个内容的长度"
+ Convert.ToString(somaxlen);
goto ERROR2;
}
if (reclen < 24)
{
strErrorInfo = "头标区头5字符表示的记录长度"
+ Convert.ToString(reclen)
+ "小于24";
goto ERROR2;
}
if (s[reclen - 1] != RECEND)
{
strErrorInfo = "头标区声称的结束位置不是MARC记录结束符";
goto ERROR2; // 结束符不正确
}
for (i = 0; i < reclen - 1; i++)
{
if (s[i] == RECEND)
{
strErrorInfo = "记录内容中不能有记录结束符";
goto ERROR2;
}
}
try
{
baseaddr = header.BaseAddress;
}
catch (FormatException ex)
{
strErrorInfo = "头标区数据基地址5字符 '" + header.BaseAddressString + " '不是纯数字 :" + ex.Message;
//throw(new MarcException(strErrorInfo));
goto ERROR2;
}
if (baseaddr > somaxlen)
{
strErrorInfo = "数据基地址值 "
+ Convert.ToString(baseaddr)
+ " 已经超出源缓冲区整个内容的长度 "
+ Convert.ToString(somaxlen);
goto ERROR2;
}
if (baseaddr <= 24)
{
strErrorInfo = "数据基地址值 "
+ Convert.ToString(baseaddr)
+ " 小于24";
goto ERROR2; // 数据基地址太小
}
if (s[baseaddr - 1] != FLDEND)
{
strErrorInfo = "没有在目次区尾部位置" + Convert.ToString(baseaddr) + "找到FLDEND符号";
goto ERROR2; //
}
try
{
lenoffld = header.WidthOfFieldLength;
}
catch (FormatException ex)
{
strErrorInfo = "头标区目次区字段长度1字符 '" + header.WidthOfFieldLengthString + " '不是纯数字 :" + ex.Message;
//throw(new MarcException(strErrorInfo));
goto ERROR2;
}
try
{
startposoffld = header.WidthOfStartPositionOfField;
}
catch (FormatException ex)
{
strErrorInfo = "头标区目次区字段起始位置1字符 '" + header.WidthOfStartPositionOfFieldString + " '不是纯数字 :" + ex.Message;
// throw(new MarcException(strErrorInfo));
goto ERROR2;
}
if (lenoffld <= 0 || lenoffld > 30)
{
strErrorInfo = "目次区中字段长度值占用字符数 "
+ Convert.ToString(lenoffld)
+ " 不正确，应在1和29之间...";
goto ERROR2;
}
if (lenoffld != 4)
{ // 2001/5/15
strErrorInfo = "目次区中字段长度值占用字符数 "
+ Convert.ToString(lenoffld)
+ " 不正确，应为4...";
goto ERROR2;
}
lenoffld = 4;
if (startposoffld <= 0 || startposoffld > 30)
{
strErrorInfo = "目次区中字段起始位置值占用字符数 "
+ Convert.ToString(startposoffld)
+ " 不正确，应在1到29之间...";
goto ERROR2;
}
startposoffld = 5;
// 开始处理目次区
// dirp = (char *)sopp;
t = 24;
offs = 24;
MyByteList baField = null;
for (i = 0; ; i++)
{
if (s[offs] == FLDEND)
{
break; // 目次区结束
}
// 将字段名装入目标
if (offs + 3 >= baseaddr)
{
break;
}
if (t + 3 >= maxbytes)
{
break;
}
/*
* baTarget.SetSize(t+3, CHUNK_SIZE);
* memcpy((char *)baTarget.GetData()+t,
* dirp+offs,
* 3);
* t+=3;
*/
baField = new MyByteList();
baField.AddRange(s, offs, 3);
t += 3;
// 得到字段长度
offs += 3;
if (offs + lenoffld >= baseaddr)
{
break;
}
len = MarcHeaderStruct.IntValue(s, offs, lenoffld);
// 得到字段内容开始地址
offs += lenoffld;
if (offs + startposoffld >= baseaddr)
{
break;
}
startpos = MarcHeaderStruct.IntValue(s, offs, startposoffld);
offs += startposoffld;
if (offs >= baseaddr)
{
break;
}
// 将字段内容装入目标
if (t + len >= maxbytes)
{
break;
}
if (s[baseaddr + startpos - 1] != FLDEND)
{
// errnoiso2709 = ERROR_BADFLDCONTENT;
strErrorInfo = "缺乏字段结束符";
goto ERROR2;
}
if (s[baseaddr + startpos + len - 1] != FLDEND)
{
//errnoiso2709 = ERROR_BADFLDCONTENT;
strErrorInfo = "缺乏字段结束符";
goto ERROR2;
}
/*
* baTarget.SetSize(t+len, CHUNK_SIZE);
* memcpy((char *)baTarget.GetData()+t,
* sopp+baseaddr+startpos,
* len);
* t += len;
*/
baField.AddRange(s, baseaddr + startpos, len == 0 ? len : len - 1);
t += len;
aResult.Add(baField.GetByteArray());
baField = null;
}
if (t + 1 >= maxbytes)
{
// errnoiso2709 = ERROR_TARGETBUFFEROVERFLOW;
strErrorInfo = "记录太大";
goto ERROR2; // 目标空间不够
}
/*
* baField.Add((char)RECEND);
* t ++;
*/
/*
* baTarget.SetSize(t+1, CHUNK_SIZE);
*((char *)baTarget.GetData() + t++) = RECEND;
* if (t+1>=maxbytes)
* {
* errnoiso2709 = ERROR_TARGETBUFFEROVERFLOW;
* goto ERROR1; // 目标空间不够
* }
*/
Debug.Assert(t != -2, "");
return(t);
//ERROR1:
// return -1; // 一般性错误
ERROR2:
// 调试用
Debug.Assert(false, "");
return(-2); // MARC格式错误
}
// 强制将ISO2709格式记录转换为字段数组
// 本函数采用的算法是将目次区的地址和长度忽略，只取3字符的字段名
// aResult的每个元素为byte[]型，内容是一个字段。第一个元素是头标区，一定是24bytes
// return:
// -1 一般性错误
// -2 MARC格式错误
public static int ForceCvt2709ToFieldArray(
ref Encoding encoding, // 2007/7/11 函数内可能发生变化
byte[] s,
out List <byte[]> aResult,
out string strErrorInfo)
{
strErrorInfo = "";
aResult = new List <byte[]>();
Debug.Assert(s != null, "");
List <MyByteList> results = new List <MyByteList>();
bool bUcs2 = false;
if (encoding.Equals(Encoding.Unicode) == true)
{
bUcs2 = true;
}
if (bUcs2 == true)
{
string strRecord = encoding.GetString(s);
// 变换成UTF-8编码方式处理
s = Encoding.UTF8.GetBytes(strRecord);
encoding = Encoding.UTF8;
}
MarcHeaderStruct header = null;
try
{
header = new MarcHeaderStruct(encoding, s);
}
catch (ArgumentException)
{
// 不足 24 字符的，给与宽容
header = new MarcHeaderStruct(Encoding.ASCII, Encoding.ASCII.GetBytes("012345678901234567890123"));
}
header.ForceUNIMARCHeader(); // 强制将某些位置设置为缺省值
results.Add(header.GetByteList());
int somaxlen;
int offs;
int i, j;
somaxlen = s.Length;
// 开始处理目次区
offs = 24;
MyByteList baField = null;
bool bFound = false;
for (i = 0; ; i++)
{
bFound = false;
for (j = offs; j < offs + 3 + 4 + 5; j++)
{
if (j >= somaxlen)
{
break;
}
if (s[j] == FLDEND)
{
bFound = true;
break;
}
}
if (j >= somaxlen)
{
offs = j;
break;
}
if (bFound == true)
{
if (j <= offs + 3)
{
offs = j + 1;
break;
}
}
// 将字段名装入目标
baField = new MyByteList();
baField.AddRange(s, offs, 3);
results.Add(baField);
baField = null;
// 得到字段内容开始地址
offs += 3;
offs += 4;
offs += 5;
if (bFound == true)
{
offs = j + 1;
break;
}
}
if (offs >= somaxlen)
{
return(0);
}
int nFieldNumber = 1;
baField = null;
// 加入对应的字段内容
for (; offs < somaxlen; offs++)
{
byte c = s[offs];
if (c == RECEND)
{
break;
}
if (c == FLDEND)
{
nFieldNumber++;
baField = null;
}
else
{
if (baField == null)
{
// 确保下标不越界
while (nFieldNumber >= results.Count)
{
MyByteList temp = new MyByteList();
temp.Add((byte)'?');
temp.Add((byte)'?');
temp.Add((byte)'?');
results.Add(temp);
}
baField = results[nFieldNumber];
}
baField.Add(c);
}
}
aResult = new List <byte[]>();
foreach (MyByteList list in results)
{
aResult.Add(list.GetByteArray());
}
return(0);
// ERROR1:
// return -1; // 一般性错误
// ERROR2:
// return -2; // MARC格式错误
}
