用爬虫建立全网最全疾病病症医学数据库

元始天尊 · 发表于 2014-4-13 11:57:39

欢迎访问技术宅的结界，请注册或者登录吧。

您需要登录才可以下载或查看，没有账号？立即注册→加入我们

×

做这个源于一个想法，比如你要查头痛，会列出所有可能导致头痛的疾病和该疾病其他症状，网上是不存在专业医学数据库这种东西的，因此需要自己从其他医学网站爬得，做到手机app中，并利用jsp网站进行大型数据库的查询是非常好的方法，输入一个词，可以搜索出医生/科室/位置/疾病/症状/相似信息。为了达到这个目的，必须有自己的医学数据库，而合作的医院并没提供，因此先自己实验一下了。现在的数据库已有180M，收录15000条内容

package symptom;
import java.io.IOException;
import java.util.LinkedList;
import java.util.Queue;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
import SQLite.Database;
import SQLite.Exception;
public class zzkfamilydoctorcomcn
{
public final static int SYMPTOM=0;
public final static int DISEASE=1;
public static int threadnum=32;
private Queue<ELE> dataQueue=new LinkedList<ELE>();//症状队列
public Database db=null;
public class ELE
{
public ELE(String url,int type)
{
this.url=url;
this.type=type;
}
/**数据库格式
* Name 症状/疾病名
Id 全局唯一标识
IsSymptom 症状/疾病？
BodyPart 身体部位 ----症状头:眼睛
Keshi 挂号科室 ----疾病内科:男科
Description 概述
DDescription详述
WithSymptom 伴随症状 ----疾病对应症状简述，逗号隔开
WithDSymptom详细症状 ----疾病对应症状详述
PossibleIll 相关疾病 ----症状可能疾病，大项逗号隔开，小项冒号隔开疾病:科室:伴随症状
Pathogeny 病因
Check 检查
Diagnosis 诊断鉴别
Popularity 疾病/症状搜索热度
*
*/
public String getName()
{
return name;
}
public void setName(String name)
{
this.name = name;
}
public String getUrl()
{
return url;
}
public void setUrl(String url)
{
this.url = url;
}
public int getType()
{
return type;
}
public void setType(int type)
{
this.type = type;
}
public String getBodypart()
{
return bodypart;
}
public void setBodypart(String bodypart)
{
this.bodypart = bodypart;
}
public void setwithsymptom(String withsymptom)
{
this.withsymptom=withsymptom;
}
public String getwithsymptom()
{
return this.withsymptom;
}
private String name="";//症状名
private String url="";//数据地址
private int type=SYMPTOM;//疾病还是症状
private String bodypart="";//身体部位头:眼睛
private String withsymptom="";
private String keshi="";
public void resolve()
{
String description="";
String ddescription="";
String withdsymptom="";
StringBuilder possibleill=new StringBuilder();
String pathogeny="";
String check="";
String diagnosis="";
threadnum--;
String sql="";
try
{
Document doc=null;
try
{
doc = Jsoup.connect(url).timeout(0).get();
}
catch (java.lang.Exception e)
{
e.printStackTrace();
return;
}
if(type == SYMPTOM)
{//如果是症状
String[] dataset=doc.select("div.brumbs").text().split(" > ");
if(dataset.length <= 2)
return;
name=dataset[dataset.length-1];
System.out.println(name);
bodypart=doc.select("div.jzxx a").get(2).text();
keshi=dataset[dataset.length-2];
description=doc.select("div.introductionTextBox").text();
ddescription=description;
Elements tagnames=doc.select("div.subNav a");
for(Element tag:tagnames)
{
if(tag.text().equals("病因"))
{
pathogeny=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.module.mPathogeny div.moduleContent").text();
}
else if(tag.text().equals("诊断"))
{
diagnosis=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.module.mFeature div.moduleContent").text();
}
}
//处理相关疾病
Elements ll=doc.select("div.area.aXgjb div.colMain tr");
if(!ll.isEmpty())
ll.remove(0);
if(!ll.isEmpty())
{
for(Element ele:ll)
{
Elements temp2=ele.children();
possibleill.append(temp2.get(0).text());
possibleill.append(":");
possibleill.append(temp2.get(1).text());
possibleill.append(":");
possibleill.append(temp2.get(2).text());
possibleill.append(",");
}
if(possibleill.length() >= 1)
{
possibleill.deleteCharAt(possibleill.length()-1);//去掉多余的','
}
}
description=description.replace('\'',' ').replace('"',' ');
ddescription=ddescription.replace('\'',' ').replace('"',' ');
withsymptom=withsymptom.replace('\'',' ').replace('"',' ');
withdsymptom=withdsymptom.replace('\'',' ').replace('"',' ');
String ill=possibleill.toString().replace('\'',' ').replace('"',' ');
pathogeny=pathogeny.replace('\'',' ').replace('"',' ');
check=check.replace('\'',' ').replace('"',' ');
diagnosis=diagnosis.replace('\'',' ').replace('"',' ');
sql="insert into symptomdata values('"+name+"','0','1','"+bodypart+"','"+keshi+"','"+description+
"','"+ddescription+"','"+withsymptom+"','"+withdsymptom+"','"+ill+"','"+
pathogeny+"','"+check+"','"+diagnosis+"','0')";
db.exec(sql,null);
System.out.println(name+" left:"+dataQueue.size());
}
else if(type == DISEASE)
{
String[] dataset=doc.select("div.brumbs").text().split(" > ");
if(dataset.length <= 2)
return;
name=dataset[dataset.length-1];
System.out.println(name);
bodypart=doc.select("span[itemprop=bodyLocation]").text();
keshi=dataset[dataset.length-2];
description=doc.select("div.introductionTextBox").text();
ddescription=description;
Elements tagnames=doc.select("div.subNav a");
for(Element tag:tagnames)
{
if(tag.text().equals("病因"))
{
pathogeny=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.module.mPathogeny div.moduleContent").text();
}
else if(tag.text().equals("诊断"))
{
diagnosis=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.module.mFeature div.moduleContent").text();
}
else if(tag.text().equals("症状"))
{
withdsymptom=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.module.mIntro.mSymptom div.moduleContent").text();
}
}
//处理相关疾病
Elements ll=doc.select("div.area.aXgjb div.colMain tr");
if(!ll.isEmpty())
ll.remove(0);
if(!ll.isEmpty())
{
for(Element ele:ll)
{
Elements temp2=ele.children();
possibleill.append(temp2.get(0).text());
possibleill.append(":");
possibleill.append(temp2.get(1).text());
possibleill.append(":");
possibleill.append(temp2.get(2).text());
possibleill.append(",");
}
if(possibleill.length() >= 1)
{
possibleill.deleteCharAt(possibleill.length()-1);//去掉多余的','
}
}
description=description.replace('\'',' ').replace('"',' ');
ddescription=ddescription.replace('\'',' ').replace('"',' ');
withsymptom=withsymptom.replace('\'',' ').replace('"',' ');
withdsymptom=withdsymptom.replace('\'',' ').replace('"',' ');
String ill=possibleill.toString().replace('\'',' ').replace('"',' ');
pathogeny=pathogeny.replace('\'',' ').replace('"',' ');
check=check.replace('\'',' ').replace('"',' ');
diagnosis=diagnosis.replace('\'',' ').replace('"',' ');
sql="insert into symptomdata values('"+name+"','0','0','"+bodypart+"','"+keshi+"','"+description+
"','"+ddescription+"','"+withsymptom+"','"+withdsymptom+"','"+ill+"','"+
pathogeny+"','"+check+"','"+diagnosis+"','0')";
db.exec(sql,null);
System.out.println(name+" left:"+dataQueue.size());
}
}
catch (IOException e)
{
System.out.println(sql);
System.out.println(url);
e.printStackTrace();
}
catch (Exception e)
{
System.out.println(sql);
System.out.println(url);
e.printStackTrace();
}
finally
{
threadnum++;
}
}
}
public zzkfamilydoctorcomcn()
{
try
{
db=new Database();
db.open("J:/symptom.db",0666);
// //按症状查询
// for(int i=1;i<8000;i++)
// {
// String cururl="http://zzk.familydoctor.com.cn/"+i;
// dataQueue.offer(new ELE(cururl,SYMPTOM));
// }
//按疾病查询
for(int i=6900;i<11000;i++)
{
String cururl="http://jbk.familydoctor.com.cn/info"+i;
dataQueue.offer(new ELE(cururl,DISEASE));
}
}
catch(Exception e)
{
e.printStackTrace();
}
new Thread()
{
@Override
public void run()
{
try
{
sleep(1000);
while(!dataQueue.isEmpty())
{
int num=32;
while(num-- > 0 && threadnum > 0 && !dataQueue.isEmpty())
{
new Thread()
{
@Override
public void run()
{
ELE curele=dataQueue.poll();
if(curele != null)
curele.resolve();
}
}.start();
}
sleep(200);
}
}
catch (InterruptedException e)
{
e.printStackTrace();
}
}
}.start();
}
public static void main(String[] args)
{
new zzkfamilydoctorcomcn();
}
}

复制代码

package symptom;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.util.LinkedList;
import java.util.Queue;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
import SQLite.Database;
import SQLite.Exception;
public class tag120askcom
{
public final static int SYMPTOM=0;
public final static int DISEASE=1;
public static int threadnum=10;
private String beginstr=null;
private Queue<ELE> dataQueue=new LinkedList<ELE>();//症状队列
public Database db=null;
private int total;
public class ELE
{
public ELE(String name,String url,int type,String bodypart,String keshi)
{
if(name == null || name.equals(""))
System.out.println("error");
this.name=name;
this.url=url;
this.type=type;
this.keshi=keshi;
this.bodypart=bodypart;
}
/**数据库格式
* Name 症状/疾病名
Id 全局唯一标识
IsSymptom 症状/疾病？
BodyPart 身体部位 ----症状头:眼睛
Keshi 挂号科室 ----疾病内科:男科
Description 概述
DDescription详述
WithSymptom 伴随症状 ----疾病对应症状简述，逗号隔开
WithDSymptom详细症状 ----疾病对应症状详述
PossibleIll 相关疾病 ----症状可能疾病，大项逗号隔开，小项冒号隔开疾病:科室:伴随症状
Pathogeny 病因
Check 检查
Diagnosis 诊断鉴别
Popularity 疾病/症状搜索热度
*
*/
public String getName()
{
return name;
}
public void setName(String name)
{
this.name = name;
}
public String getUrl()
{
return url;
}
public void setUrl(String url)
{
this.url = url;
}
public int getType()
{
return type;
}
public void setType(int type)
{
this.type = type;
}
public String getKeshi()
{
return keshi;
}
public void setKeshi(String keshi)
{
this.keshi = keshi;
}
public String getBodypart()
{
return bodypart;
}
public void setBodypart(String bodypart)
{
this.bodypart = bodypart;
}
private String name="";//症状名
private String url="";//数据地址
private int type=SYMPTOM;//疾病还是症状
private String keshi="";//疾病科室（大科室）keshi+":"+血液内科
private String bodypart="";//身体部位头:眼睛
public void resolve()
{
String description="";
String ddescription="";
String withsymptom="";
String withdsymptom="";
StringBuilder possibleill=new StringBuilder();
String pathogeny="";
String check="";
String diagnosis="";
threadnum--;
String sql="";
try
{
Document doc=Jsoup.connect(url).timeout(0).get();
if(type == SYMPTOM)
{//如果是症状
description=doc.select("dl.p_sibox1dl.clears p").text();
String temp=doc.select("dl.p_sibox1dl.clears p a").attr("abs:href");
ddescription=Jsoup.connect(temp).timeout(0).get().select("div.p_cleftartbox").text();
Elements tagnames=doc.select("div.w_headnav_div a");
for(Element tag:tagnames)
{
if(tag.text().equals("病因"))
{
pathogeny=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("检查"))
{
check=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("诊断鉴别"))
{
diagnosis=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("相关疾病"))
{
Elements ll=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("ul.p_cbox3ul");
if(!ll.isEmpty())
{
Elements eles=ll.get(0).children();
for(Element ele:eles)
{
possibleill.append(ele.select("span.p_cbox3li1 a").text());
possibleill.append(":");
possibleill.append(ele.select("span.p_cbox3li2 a").text());
possibleill.append(":");
possibleill.append(ele.select("span.p_cbox3li3 a").text());
possibleill.append(",");
}
if(possibleill.length() >= 1)
{
possibleill.deleteCharAt(possibleill.length()-1);//去掉多余的','
}
}
}
}
description=description.replace('\'',' ').replace('"',' ');
ddescription=ddescription.replace('\'',' ').replace('"',' ');
withsymptom=withsymptom.replace('\'',' ').replace('"',' ');
withdsymptom=withdsymptom.replace('\'',' ').replace('"',' ');
String ill=possibleill.toString().replace('\'',' ').replace('"',' ');
pathogeny=pathogeny.replace('\'',' ').replace('"',' ');
check=check.replace('\'',' ').replace('"',' ');
diagnosis=diagnosis.replace('\'',' ').replace('"',' ');
sql="insert into symptomdata values('"+name+"','0','1','"+bodypart+"','"+keshi+"','"+description+
"','"+ddescription+"','"+withsymptom+"','"+withdsymptom+"','"+ill+"','"+
pathogeny+"','"+check+"','"+diagnosis+"','0')";
db.exec(sql,null);
System.out.println(name+" left:"+dataQueue.size()+" total:"+total);
}
else if(type == DISEASE)
{
description=doc.select("div.p_lbox1_ab p").text();
String temp=doc.select("div.p_lbox1_ab p a").attr("abs:href");
Document doc2=Jsoup.connect(temp).timeout(0).get();
ddescription=doc2.select("div.p_cleftartbox").text();
Elements tagnames=doc.select("div.p_topbox a");
for(Element tag:tagnames)
{
if(tag.text().equals("病因"))
{
pathogeny=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("检查"))
{
check=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("鉴别"))
{
diagnosis=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("症状"))
{
withdsymptom=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
else if(tag.text().equals("并发症"))
{
withsymptom=Jsoup.connect(tag.attr("abs:href")).timeout(0).get().select("div.p_cleftartbox").text();
}
}
description=description.replace('\'',' ').replace('"',' ');
ddescription=ddescription.replace('\'',' ').replace('"',' ');
withsymptom=withsymptom.replace('\'',' ').replace('"',' ');
withdsymptom=withdsymptom.replace('\'',' ').replace('"',' ');
String ill=possibleill.toString().replace('\'',' ').replace('"',' ');
pathogeny=pathogeny.replace('\'',' ').replace('"',' ');
check=check.replace('\'',' ').replace('"',' ');
diagnosis=diagnosis.replace('\'',' ').replace('"',' ');
sql="insert into symptomdata values('"+name+"','0','0','"+bodypart+"','"+keshi+"','"+description+
"','"+ddescription+"','"+withsymptom+"','"+withdsymptom+"','"+ill+"','"+
pathogeny+"','"+check+"','"+diagnosis+"','0')";
db.exec(sql,null);
System.out.println(name+" left:"+dataQueue.size()+" total:"+total);
}
}
catch (IOException e)
{
System.out.println(sql);
System.out.println(url);
e.printStackTrace();
}
catch (Exception e)
{
System.out.println(sql);
System.out.println(url);
e.printStackTrace();
}
finally
{
threadnum++;
}
}
}
public tag120askcom()
{
try
{
db=new Database();
db.open("J:/symptom.db",0666);
Document doc=null;
Elements eles=null;
//按症状查询
doc=Jsoup.connect("http://tag.120ask.com/zhengzhuang/").timeout(0).get();
eles=doc.select("ul.p_leftdivnav").get(0).children();
for(Element ele:eles)
{
Elements test=ele.select("div.p_lnavdisdiv h3 a");
Document innerdoc=Jsoup.connect(test.attr("abs:href")).timeout(0).get();
Element toprocess=innerdoc.select("div.w_header").get(0);
String bigbodypart=toprocess.select("h3").text();
Elements subset=toprocess.select("a");
//先直接遍历症状
Elements innerele=innerdoc.select("div.w_neike.clears").get(0).children();
for(int i=0;i<innerele.size();i+=2)
{
String bigclass=innerele.get(i).text();//获取大科室名
for(Element tempele:innerele.get(i+1).select("a"))
{
dataQueue.offer(new ELE(tempele.attr("title"),tempele.attr("abs:href"),SYMPTOM,bigbodypart,bigclass));
}
}
if(!subset.isEmpty())
{//有子节点则需要连接到每个网址然后遍历症状
for(Element curele:subset)
{
Document temp=Jsoup.connect(curele.attr("abs:href")).timeout(0).get();
Elements innerelel=temp.select("div.w_neike.clears").get(0).children();
for(int i=0;i<innerelel.size();i+=2)
{
String bigclass=innerelel.get(i).text();//获取大科室名
for(Element tempele:innerelel.get(i+1).select("a"))
{
dataQueue.offer(new ELE(tempele.attr("title"),tempele.attr("abs:href"),SYMPTOM,
bigbodypart+":"+curele.text(),bigclass));
}
}
}
}
}
//按疾病查询
doc=Jsoup.connect("http://tag.120ask.com/jibing/").timeout(0).get();
eles=doc.select("ul.p_leftdivnav").get(0).children();
for(Element ele:eles)
{
Elements test=ele.select("div.p_lnavdisdiv h3 a");
Document innerdoc=Jsoup.connect(test.attr("abs:href")).timeout(0).get();
Element toprocess=innerdoc.select("div.w_header").get(0);
String bigbodypart=toprocess.select("h3").text();
Elements subset=toprocess.select("a");
//先直接遍历症状
Elements innerele=innerdoc.select("div.w_neike.clears").get(0).children();
for(int i=0;i<innerele.size();i+=2)
{
String bigclass=innerele.get(i).text();//获取大科室名
for(Element tempele:innerele.get(i+1).select("a"))
{
dataQueue.offer(new ELE(tempele.attr("title"),tempele.attr("abs:href"),DISEASE,bigbodypart,bigclass));
}
}
if(!subset.isEmpty())
{//有子节点则需要连接到每个网址然后遍历症状
for(Element curele:subset)
{
Document temp=Jsoup.connect(curele.attr("abs:href")).timeout(0).get();
Elements innerele1=innerdoc.select("div.w_neike.clears").get(0).children();
for(int i=0;i<innerele1.size();i+=2)
{
String bigclass=innerele1.get(i).text();//获取大科室名
for(Element tempele:innerele1.get(i+1).select("a"))
{
dataQueue.offer(new ELE(tempele.attr("title"),tempele.attr("abs:href"),DISEASE,bigbodypart,bigclass));
}
}
}
}
}
System.out.println(dataQueue.size());
for(ELE ele:dataQueue)
{
System.out.println(ele.name);
}
total=dataQueue.size();
if(beginstr != null)
{
while(!dataQueue.poll().getName().equals(beginstr))
{
;
}
}
}
catch (FileNotFoundException e)
{
e.printStackTrace();
}
catch (IOException e)
{
e.printStackTrace();
}
catch (Exception e)
{
e.printStackTrace();
}
new Thread()
{
@Override
public void run()
{
try
{
sleep(1000);
while(!dataQueue.isEmpty())
{
int num=1;
while(num-- > 0 && threadnum > 0 && !dataQueue.isEmpty())
{
new Thread()
{
@Override
public void run()
{
ELE curele=dataQueue.poll();
if(curele != null)
curele.resolve();
}
}.start();
}
sleep(200);
}
}
catch (InterruptedException e)
{
e.printStackTrace();
}
}
}.start();
}
public static void main(String[] args)
{
new tag120askcom();
}
}

复制代码

账号		自动登录	找回密码
密码			立即注册→加入我们