我正在编写比较两组二进制数据之间差异的软件。二进制数据包含坐标,它还从DBF文件中读取以获取有关每个对象的属性信息。我的软件指示对象是否已移动,如果其属性已更改,则删除/添加了对象。
我的自定义对象哈希码是根据每个实例中的坐标列表生成的。 每个对象中还有一个数据行。当我最初尝试找到“不完美的记录”(基本上任何基于坐标和数据路径没有匹配的对象)时,它会考虑数据行,并且我可以使用HashSet,因为数据行使其足够独特。
public override int GetHashCode()
{
if (considerAttrs)
{
return (value.GetHashCode() + dbString.GetHashCode());
}
else
{
return value.GetHashCode();
}
}
我在哪里:
此时软件返回没有完美匹配的对象(没有完全匹配的坐标和属性)。这个数据是正确的(至少我希望)
origUnfoundCount:114 modUnfoundCount:223
在原版中找不到114,因为113已修改属性,1已被移动。它在修改后找不到223,因为有113个修改后的属性,1个被移动,109个被添加。
我被困的地方:
该软件可以很好地处理较小的数据(每个列表中有几百个)
我将每个对象中的ConsideAttrs更改为false并使用List而不是HashSet。速度惩罚非常严重,我的应用程序只能用于比较数据之间的微小差异。
我需要使用List,因为你不能在hashset中有重复项,但是列表只是放慢速度的方法。由于密钥必须是唯一的
,因此无法使用字典我需要一种新方法,下面的逻辑应该为您提供我需要它做的一般要点。
我目前的比较代码
//ignored modified records
HashSet<int> ignoredRecNo = new HashSet<int>();
//ignoring moved records
HashSet<String> ignoredDBstrings = new HashSet<string>();
HashSet<String> columnNames = new HashSet<string>();
HashSet<int> modModdedRNs = new HashSet<int>();
columnNames = (HashSet<String>)HttpContext.Current.Session["columnNames"];
List<PolyLineZ> originalNFs = new List<PolyLineZ>();
List<PolyLineZ> modifiedNFs = new List<PolyLineZ>();
List<PolyLineZ> removedList = new List<PolyLineZ>();
List<PolyLineZ> movedList = new List<PolyLineZ>();
List<PolyLineZ> modifiedList = new List<PolyLineZ>();
List<PolyLineZ> modifiedMatchList = new List<PolyLineZ>();
List<PolyLineZ> movedOrDeleted = new List<PolyLineZ>();
if (HttpContext.Current.Session["origNFList"] != null)
{
origPolyLineZNFList = (HashSet<PolyLineZ>)HttpContext.Current.Session["origNFList"];
}
if (HttpContext.Current.Session["modNFList"] != null)
{
modPolyLineZNFList = (HashSet<PolyLineZ>)HttpContext.Current.Session["modNFList"];
}
//----------Generate Lists of string for each row---------------//
foreach (PolyLineZ polyLineZ in origPolyLineZNFList)
{
origDBStrings.Add(polyLineZ.dbString);
PolyLineZ temp = new PolyLineZ();
temp = polyLineZ;
temp.considerAttrs = false;
originalNFs.Add(temp);
}
foreach (PolyLineZ polyLineZ in modPolyLineZNFList)
{
modDBStrings.Add(polyLineZ.dbString);
PolyLineZ temp = new PolyLineZ();
temp = polyLineZ;
temp.considerAttrs = false;
modifiedNFs.Add(temp);
}
foreach (PolyLineZ modpolyLineZ in modifiedNFs)
{
bool foundAmatch = false;
foreach (PolyLineZ origPolyLineZ in originalNFs)
{
if (origPolyLineZ.Equals(modpolyLineZ))
{
if (!modDBStrings.Contains(origPolyLineZ.dbString))
{
//database modifications are in here
modModdedRNs.Add(origPolyLineZ.RecordNumber);
foundAmatch = true;
break;
}
}
}
}
foreach (PolyLineZ polyLineZ in originalNFs)
{
bool foundAmatch = false;
foreach (PolyLineZ modpolyLineZ in modifiedNFs)
{
if (foundAmatch)
{
break;
}
if (modpolyLineZ.Equals(polyLineZ))
{
if (!origDBStrings.Contains(modpolyLineZ.dbString))
{
foundAmatch = true;
//database modifications are in here
ignoredRecNo.Add(modpolyLineZ.RecordNumber);
ignoredDBstrings.Add(modpolyLineZ.dbString);
modifiedList.Add(polyLineZ);
modifiedMatchList.Add(modpolyLineZ);
break;
} // end db string comparison
} //end shape equals if
} //end modNF loop
if (!foundAmatch)
{
movedOrDeleted.Add(polyLineZ);
ignoredDBstrings.Add(polyLineZ.dbString);
ignoredRecNo.Add(polyLineZ.RecordNumber);
}
} //end origNF loop
result += "movedDeletedCount: " + movedOrDeleted.Count + "<br/>";
foreach (PolyLineZ polylineZ in movedOrDeleted)
{
if (!modDBStrings.Contains(polylineZ.dbString))
{
removedList.Add(polylineZ);
}
else
{
movedList.Add(polylineZ);
}
}
/*************************** ITERATE DATABASE CHANGES***********************************/
for(int i=0; i < modifiedList.Count;i++)
{
if (modModdedRNs.Contains(modifiedList[i].RecordNumber))
{
if (modifiedAttrs < 1001)
{
//database modifications are in here
ignoredRecNo.Add(modifiedMatchList[i].RecordNumber);
ignoredDBstrings.Add(modifiedMatchList[i].dbString);
modifiedAttrs++;
modifiedResults += "<div class='turnBlue'>";
//show where the change was made at
modifiedResults += "Change Detected at original FID# " + (modifiedList[i].RecordNumber - 1) + " and modified FID#";
HashSet<String> mismatchedColumns = new HashSet<String>();
modifiedResults += (modifiedMatchList[i].RecordNumber - 1);
modifiedResults += "</div>"; //end turnblue div
DataRow origRow = modifiedList[i].datarow;
DataRow modRow = modifiedMatchList[i].datarow;
foreach (String columnName in columnNames)
{
String origRowValue = "" + origRow.Field<Object>(columnName);
String modRowValue = "" + modRow.Field<Object>(columnName);
if (!modRowValue.Equals(origRowValue))
{
mismatchedColumns.Add(columnName);
}
}
foreach (String mismatchedColumn in mismatchedColumns)
{
//grab original attr value
String origMismatchedRowValue = "" + origRow.Field<Object>(mismatchedColumn);
//grab the modified value
String modMismatchedRowValue = "" + modRow.Field<Object>(mismatchedColumn);
//generate a heading, letting the user know about the situation
modifiedResults += "<div class='turnBlue'>Value at Column: <b>" + mismatchedColumn + "</b> has been modified<br/>";
modifiedResults += "<div class='pushLeft'>Original value: <b>" + origMismatchedRowValue + "</b><br/></div>";
modifiedResults += "<div class='pushLeft'>Modified value: <b>" + modMismatchedRowValue + "</b><br/></div>";
modifiedResults += "</div>"; //end modified div
}
}
else
{
modifiedAttrs++;
}
}
else
{
if (removed < 1001)
{
//iterate removed data here
removed++;
}
else
{
removed++;
}
}
}
//****************************this determines which ones have been added ***************************/
foreach (PolyLineZ modpolyLineZ in modifiedNFs)
{
if (!ignoredRecNo.Contains(modpolyLineZ.RecordNumber) && (!ignoredDBstrings.Contains(modpolyLineZ.dbString)))
{
//iterate added data here
}
}
foreach (PolyLineZ polylineZ in removedList)
{
//iterate removed data here
}
foreach (PolyLineZ polylineZ in movedList)
{
//iterate moved data here
}
result += "<div id='addedJump'></div>" + addedResult;
result += "<div id='moddedJump'></div>" + modifiedResults;
result += "<div id='removedJump'></div>" + removedResults;
result += "<div id='movedJump'></div>" + movedResults;
}
答案 0 :(得分:0)
我从将所有数据附加到单个字符串切换到使用stringbuilder ...
我的程序从30分钟开始,将3,200个已删除的记录检测到7秒钟。
有关详细信息,请参阅此问题: