Wikipedysta:Olafbot/OlafbotBasicTopics.java
Z Wikipedii
Poniższy kod w Javie dodaje do wszystkich artykułów wymienionych w "podstawowych zagadnieniach" z jakiejś dziedziny linki do "podstawowych zagadnień".
/* */
/* Olafbot for Wikipedia */
/* */
/* Olaf Matyja 2004 */
/* */
import java.lang.*;
import java.net.*;
import java.io.*;
import java.io.Reader.*;
import java.util.*;
public class OlafbotBasicTopics {
final static String dirPath="d:/Olaf/Olafbot";
private static class OlafBotException extends Exception {
boolean fatal;
OlafBotException(String comment,boolean fatal) {
super(comment);
this.fatal=fatal;
}
};
//polish letters
final static String a="%C4%85";
final static String c="%C4%87";
final static String e="%C4%99";
final static String l="%C5%82";
final static String n="%C5%84";
final static String o="%C3%B3";
final static String s="%C5%9B";
final static String x="%C5%BA";
final static String z="%C5%BC";
final static String A="%C4%84";
final static String C="%C4%86";
final static String E="%C4%98";
final static String L="%C5%81";
final static String N="%C5%83";
final static String O="%C3%93";
final static String S="%C5%9A";
final static String X="%C5%B9";
final static String Z="%C5%BB";
final static String plus="%2B";
final static String ampersand="%26";
static InetAddress wikipedia=null;
static String session="800737744ba08333e8d3802105395f51";
static String editTime="";
static String replaceAll(String src,String from,String to) {
int start=0;
while (true) {
int n=src.indexOf(from,start);
if (n<0)
return src;
src=(n>0 ? src.substring(0,n) : "")+
to+
(n+from.length()<src.length() ? src.substring(n+from.length()) : "");
start=n+to.length();
}
}
//connect with the server
static String connect(String request,String content) throws Exception {
Socket socket=null;
try {
socket= new Socket(wikipedia,80);
DataInputStream input = new DataInputStream(socket.getInputStream());
PrintStream output = new PrintStream(socket.getOutputStream(),true);
output.println(
request+
"Content-type: application/x-www-form-urlencoded\n"+
"Accept: */*\n"+
"User-Agent: OlafBot\n"+
"Host: pl.wikipedia.org\n"+
"Connection: Keep-Alive\n"+
"Cache-Control: no-cache\n"+
"Cookie: plwikiUserName=Olafbot; "+
"plwikiSession="+session+"; "+
"plwikiUserID=2189; "+
"plwikiPassword=deleted\n"+
"\n"+
content
);
String text="", inputLine;
while ((inputLine = input.readLine()) != null) {
//if (inputLine.length()>0)
// text=inputLine;
text+=inputLine.trim()+"\n";
//System.out.println(inputLine);
}
text=replaceAll(text,"<","<");
text=replaceAll(text,">",">");
text=replaceAll(text,""","\"");
text=replaceAll(text,"&","&");
//text=URLDecoder.decode(text);
//System.out.println(text);
final String sessStr="Set-Cookie: plwikiSession=";
int sess=text.indexOf(sessStr);
if (sess>=0) {
int end=text.indexOf("; path=",sess);
if (end<0)
throw new RuntimeException("Invalid cookie\n"+text);
session=text.substring(sess+sessStr.length(),end);
}
final String editStr=" name=\"wpEdittime\"";
int edit=text.indexOf(editStr);
if (edit>=0)
editTime=text.substring(edit-15,edit-1);
//System.out.println(text+"\n*************************************************"); //TODO debug
if (text.indexOf("<strong>Masz <a href=\"/wiki/Dyskusja_wikipedysty:Olafbot\"")>=0) {
throw new OlafBotException("Program zatrzymany.",true);
}
if (text.toLowerCase().indexOf("konflikt edycji")>=0) {
System.out.println(text);
throw new Exception("Konflikt edycji\n"+text);
}
if (content.length()==0) {
int start=text.indexOf("wrap=\"virtual\">");
int end=text.indexOf("</textarea>");
if (start<0 || end<0)
throw new Exception("Nie znaleziony tekst\n"+text);
text=text.substring(start+15,end);
for (int n=0;n<text.length();n++) {
int z=(int)text.charAt(n);
if (z>127 || z<32 && z!=10 || z==(int)'%') {
text=(n>0 ? text.substring(0,n) : "")+
"%"+Integer.toHexString((int)text.charAt(n)).toUpperCase()+
(n<text.length()-1 ? text.substring(n+1) : "");
}
}
//System.out.println(text);//todo debug
return text;
}
else
return null;
} finally {
if (socket!=null)
socket.close();
}
}
static public String getPage(String page) {
String pageName=replaceAll(page," ","_");
String content=null;
while (true)
try {
content=connect(
"GET /w/wiki.phtml?title="+pageName+"&action=edit&redirect=no HTTP/1.1\n",
""
);
break;
} catch(Exception ex) {
ex.printStackTrace();
};
return content;
}
static void login() {
String toPost=
"wpName=Olafbot&"+
"wpLoginattempt=Zaloguj mnie&"+
"wpPassword=(...)&"+ //Wycięte hasło
"wpRemember=1&"+
"\n";
while (true)
try {
System.out.println("Logowanie...");
connect(
"POST /w/wiki.phtml?title=Specjalna:Userlogin&returnto=Specjalna:Userlogout HTTP/1.1\n"+
"Content-Length: "+String.valueOf(toPost.length())+"\n",
toPost
);
break;
} catch(Exception ex) {
ex.printStackTrace();
}
}
static void writeLog(String pageName,String topicsName,String text) {
try {
PrintWriter fw=new PrintWriter(new FileWriter(dirPath+"/podstawowe.txt",true));
fw.println(pageName+"\t"+topicsName+"\t"+text);
fw.close();
} catch (Throwable ex2) {
ex2.printStackTrace();
System.exit(0);
}
}
static void writeLog(String pageName,String topicsName,Throwable ex) {
try {
PrintWriter fw=new PrintWriter(new FileWriter(dirPath+"/Log "+pageName+".txt",true));
fw.println(pageName);
fw.println(topicsName);
ex.printStackTrace();
ex.printStackTrace(fw);
fw.close();
} catch (Throwable ex2) {
ex2.printStackTrace();
System.exit(0);
}
}
static String processPage(String pageName,String topicsName) throws Exception {
Thread.sleep(5000);
System.out.println("Czytanie "+pageName);
String content=getPage(pageName).trim();
//System.out.println(content);
String result="???";
if (content.length()==0) {
result="Strona nie istnieje";
System.out.println(result);
return result;
}
String content2=content.toLowerCase();
if (content2.indexOf("redirect")>=0) {
result="Jest redirect";
System.out.println(result);
return result;
}
String topics1=topicsName.toLowerCase();
String topics2=replaceAll(topics1," ","_");
if (content2.indexOf(topics1)>=0 || content2.indexOf(topics2)>=0) {
result="Jest link";
System.out.println(result);
return result;
}
if (content2.indexOf("ujednoznaczniaj")>=0 || content2.indexOf("msg:disamb")>=0) {
result="Strona ujednoznaczniajaca";
System.out.println(result);
return result;
}
int zobacz=content2.indexOf("zobacz");
if (zobacz>=2 && content.charAt(zobacz-2)=='(') {
result="Dziwne zobacz";
System.out.println(result);
return result;
}
if (zobacz<0) {
result="Dodawana sekcja";
System.out.println(result);
content+="\n\nZobacz te"+z+": [["+topicsName.toLowerCase()+"]]\n";
} else {
if (content2.indexOf("zobacz",zobacz+1)>=0) {
result="Dwukrotne zobacz";
System.out.println(result);
return result;
}
int k=content.indexOf("[[",zobacz);
if (k<0) {
k=content.indexOf("\n",zobacz);
if (k<0)
k=content.length();
else while (k<content.length() && content.charAt(k)<' ')
k++;
}
int gwiazdka=content.substring(zobacz,k).indexOf("*");
result=gwiazdka>=0 ? "Dodawany link po gwiazdce" : "Dodawany link w linii";
System.out.println(result);
content=content.substring(0,k)+"[["+topicsName.toLowerCase()+
(gwiazdka>=0 ?
("]]\n*"+ (content.charAt(gwiazdka+1)==' ' ? " " : ""))
: "]], ")+
(k<content.length() ? content.substring(k) : "");
}
content=replaceAll(content,"[[mat.]]","");
content=replaceAll(content,"&","%26");
content=replaceAll(content,"+","%2B");
Thread.sleep(15000);
System.out.println("Zapisywanie "+pageName);
pageName=replaceAll(pageName," ","_");
String content3=
"wpSummary=Dodanie linku do spisu [["+topicsName.toLowerCase()+"]]&"+
"wpMinoredit=1&"+
"wpWatchthis=0&"+
"wpEdittime="+editTime+"&"+
"wpSection=&"+
"wpTextbox1="+content+
"\n";
connect(
"POST /w/wiki.phtml?title="+pageName+"&action=submit&redirect=no HTTP/1.1\n"+
"Content-Type: application/x-www-form-urlencoded\n"+
"Content-Length: "+String.valueOf(content3.length())+"\n",
content3
);
System.out.println("OK");
return result;
}
static boolean started=false;
static void processTopics(String topicsName) {
System.out.println("Czytanie "+topicsName);
String topics=getPage(topicsName);
System.out.println("Odczytane");
int start=0;
Properties pages=new Properties();
while (true) {
int n=topics.indexOf("[[",start);
if (n>=0) {
int k1=topics.indexOf("|",n);
int k2=topics.indexOf("]]",n);
int k=k1<k2 && k1>=0 ? k1 : k2;
String pageName=topics.substring(n+2,k).trim();
if (pageName.indexOf("odstawowe")<0 && pageName.indexOf("losariusz")<0 && !pageName.equals("Wikipedia") &&
pageName.indexOf(":")<0 && pageName.indexOf("#")<0 && pages.getProperty(pageName)==null) {
while (true) {
try {
//if (pageName.startsWith("tablica ca"))
started=true;
if (started)
writeLog(pageName,topicsName,processPage(pageName,topicsName));
//System.out.println("<h1>"+yearName(year,false)+" / "+(year>0 ? romanNumber(year) : "")+"</h1>\n");
//System.out.println(sovereignTable(year));
break;
}
catch(OlafBotException ex) {
writeLog(pageName,topicsName,"Error: "+ex.getMessage());
writeLog(pageName,topicsName,ex);
if (ex.fatal)
System.exit(0);
break;
}
catch(Throwable ex) {
if (ex instanceof SocketException)
ex.printStackTrace();
else
writeLog(pageName,topicsName,ex);
try {
Thread.sleep(30000);
} catch (Exception ex2) {
writeLog(pageName,topicsName,ex2);
}
}
}
pages.put(pageName,"");
}
start=n+2;
}
else
break;
}
}
public static void main(String[] args) throws Exception {
wikipedia=InetAddress.getByName("pl.wikipedia.org");
login();
writeLog("","","Wystartowany");
processTopics("Podstawowe zagadnienia z zakresu matematyki");
//processTopics("Podstawowe zagadnienia z zakresu muzyki");
processTopics("Podstawowe zagadnienia z zakresu biologii");
processTopics("Podstawowe zagadnienia z zakresu architektury");
processTopics("Podstawowe zagadnienia z zakresu astronomii");
processTopics("Podstawowe zagadnienia z zakresu geografii");
processTopics("Podstawowe zagadnienia z zakresu ekologii");
processTopics("Podstawowe zagadnienia z zakresu psychologii");
processTopics("Podstawowe zagadnienia z zakresu filozofii");
}
}
