public ChineseAddressWordParser(string[] citys)
{
if (ms_nationwordmatch == null)
{
InitBasicWordmatch();
}
ms_citywordmatch = NewElementsWordmatch("city", citys);
InitBasicWordmatchChain();
}
public string GetStringByWordmatch(AbstractWordmatch wm)
{
foreach (MatchingBlock mb in m_matchblocklist)
{
if (mb.wordmatch == wm)
{
return(MatchingBlockGetString(mb));
}
}
return(null);
}
public StringCollection GetStringsByWordmatch(AbstractWordmatch wm)
{
StringCollection sc = new StringCollection();
foreach (MatchingBlock mb in m_matchblocklist)
{
if (mb.wordmatch == wm)
{
sc.Add(this.MatchingBlockGetString(mb));
}
}
return(sc);
}
private static void InitBasicWordmatch()
{
//国名
ms_nationwordmatch = NewSingleElementWordmatch("nation", "中国");
//省和直辖市
ms_provincewordmatch = new AggregateWordmatch("province", "province.txt", true);
//地级市
ms_citywordmatch = new AggregateWordmatch("city", "city.txt", true);
//县级市
ms_countywordmatch = new RegexWordmatch("county", new Regex(@"[\u4e00-\u9fa5]+?(县|市)", RegexOptions.Compiled));
//工业区、高新区
ms_industrialparkwordmatch = new RegexWordmatch("plaza", new Regex(@"[\u4e00-\u9fa5]+(园区|工业区|工业城)", RegexOptions.Compiled));
//区
ms_districtwordmatch = new RegexWordmatch("district", new Regex(@"近郊|[\u4e00-\u9fa5]+?区", RegexOptions.Compiled));
//街道&道路
ms_streetwordmatch = new RegexWordmatch("street", new Regex(@"[\u4e00-\u9fa5]+街道", RegexOptions.Compiled));
ms_roadwordmatch = new RegexWordmatch("road", new Regex(@"[\u4e00-\u9fa5]+?(胡同|弄堂|街|巷|路|道)", RegexOptions.Compiled));
//门牌
ms_numberwordmatch = new RegexWordmatch("number", new Regex(@"(\d|-|甲|乙|丙)+?号(?!楼)", RegexOptions.Compiled));
//住宅区
ms_zonewordmatch = new RegexWordmatch("number", new Regex(@"[\u4e00-\u9fa5]+?(社区|小区)", RegexOptions.Compiled));
//广场/购物中心/酒店
ms_plazawordmatch = new RegexWordmatch("plaza", new Regex(@"[\u4e00-\u9fa5]+(层|楼|广场|商城|商场|酒店|购物中心|市场|大厦|校区|百货)([A-Z]座)?", RegexOptions.Compiled));
//镇
ms_townwordmatch = new RegexWordmatch("town", new Regex(@"[\u4e00-\u9fa5]+?(镇|乡)", RegexOptions.Compiled));
//村
ms_villagewordmatch = new RegexWordmatch("village", new Regex(@"[\u4e00-\u9fa5]+?村", RegexOptions.Compiled));
//链头
ms_headwordmatch = new HeadWordmatch();
//备注
ms_notewordmatch = new RegexWordmatch("note", new Regex(@"\(.+\)", RegexOptions.Compiled), false);
//噪音收集器
ms_noisecollector = new NoiseCollector();
}
private static void InitBasicWordmatch()
{
//国名
ms_nationwordmatch = NewSingleElementWordmatch("nation", "中国");
//省和直辖市
ms_provincewordmatch = new AggregateWordmatch("province", "province.txt", true);
//地级市
ms_citywordmatch = new AggregateWordmatch("city", "city.txt", true);
//县级市
ms_countywordmatch = new RegexWordmatch("county", new Regex(@"[\u4e00-\u9fa5]+?(县|市)", RegexOptions.Compiled));
//工业区、高新区
ms_industrialparkwordmatch = new RegexWordmatch("plaza", new Regex(@"[\u4e00-\u9fa5]+(园区|工业区|工业城)", RegexOptions.Compiled));
//区
ms_districtwordmatch = new RegexWordmatch("district", new Regex(@"近郊|[\u4e00-\u9fa5]+?区", RegexOptions.Compiled));
//街道&道路
ms_streetwordmatch = new RegexWordmatch("street", new Regex(@"[\u4e00-\u9fa5]+街道", RegexOptions.Compiled));
ms_roadwordmatch = new RegexWordmatch("road", new Regex(@"[\u4e00-\u9fa5]+?(胡同|弄堂|街|巷|路|道)", RegexOptions.Compiled));
//门牌
ms_numberwordmatch = new RegexWordmatch("number", new Regex(@"(\d|-|甲|乙|丙)+?号(?!楼)", RegexOptions.Compiled));
//住宅区
ms_zonewordmatch = new RegexWordmatch("number", new Regex(@"[\u4e00-\u9fa5]+?(社区|小区)", RegexOptions.Compiled));
//广场/购物中心/酒店
ms_plazawordmatch = new RegexWordmatch("plaza", new Regex(@"[\u4e00-\u9fa5]+(层|楼|广场|商城|商场|酒店|购物中心|市场|大厦|校区|百货)([A-Z]座)?", RegexOptions.Compiled));
//镇
ms_townwordmatch = new RegexWordmatch("town", new Regex(@"[\u4e00-\u9fa5]+?(镇|乡)", RegexOptions.Compiled));
//村
ms_villagewordmatch = new RegexWordmatch("village", new Regex(@"[\u4e00-\u9fa5]+?村", RegexOptions.Compiled));
//链头
ms_headwordmatch = new HeadWordmatch();
//备注
ms_notewordmatch = new RegexWordmatch("note", new Regex(@"\(.+\)", RegexOptions.Compiled),false);
//噪音收集器
ms_noisecollector = new NoiseCollector();
}
public bool AddMatchingBlock(AbstractWordmatch wordmatch, int absolutestartpoint, int length)
{
if (absolutestartpoint < 0 || length > this.m_string.Length)
{
return(false);
}
//查找它在哪个匹配块内
foreach (MatchingBlock mpexist in m_matchblocklist)
{
if (mpexist.startpoint <= absolutestartpoint && absolutestartpoint + length <= mpexist.startpoint + mpexist.length)
{
//拆分该块
SplitMatchingBlock(mpexist, absolutestartpoint, length, wordmatch);
return(true);
}
}
return(false);
}
public AbstractWordmatch SetNext(AbstractWordmatch next)
{
this.next = next;
return next;
}
public ChineseAddressWordParser(string[] citys)
{
if (ms_nationwordmatch == null)
InitBasicWordmatch();
ms_citywordmatch = NewElementsWordmatch("city", citys);
InitBasicWordmatchChain();
}
private void SplitMatchingBlock(MatchingBlock oldmb, int startpoint, int length, AbstractWordmatch wordmatch)
{
int oldstart = oldmb.startpoint;
int oldlenth = oldmb.length;
m_matchblocklist.Remove(oldmb);
m_matchblocklist.Add(new MatchingBlock(wordmatch, startpoint, length));
if (oldstart < startpoint)
{
m_matchblocklist.Add(new MatchingBlock(null, oldstart, startpoint - oldstart));
}
if ((oldstart + oldlenth) > (startpoint + length))
{
m_matchblocklist.Add(new MatchingBlock(null, startpoint + length, oldstart + oldlenth - startpoint - length));
}
}
public bool AddMatchingBlock(AbstractWordmatch wordmatch, MatchingBlock oldmb, int relativestartpoint, int length)
{
return(this.AddMatchingBlock(wordmatch, oldmb.startpoint + relativestartpoint, length));
}
public MatchingBlock(AbstractWordmatch wm, int startp, int len)
{
this.wordmatch = wm;
this.startpoint = startp;
this.length = len;
}
public AbstractWordmatch SetNext(AbstractWordmatch next)
{
this.next = next;
return(next);
}
