Code source - PirateB@y.org

Cette source est écrite en PL/SQL
Tourne sous Oracle 10g + (xe) [les autres j'ai pas testé]

Utilité : faire une sauvegarde de tout les liens présents sur les 100 pages des liens récents sur le site thepiratebay.org

Cette sopurce peut être reprise /distribuée/ etc etc etc ... :)

DECLARE
        v_stream CLOB;
        currentPosx NUMBER             := 0 ;

        posy        NUMBER             := 0 ;
        seqstring   VARCHAR2(400);
        amt         INTEGER            := 1;
        i           NUMBER             := 0;
        iRecord     NUMBER             := 0;
        iPage       NUMBER             := 0;
        countData   NUMBER             := 0;
        rawPerPage  NUMBER             := 30;
        maxPages    NUMBER             := 100;
        indexofRow  NUMBER             := 0;
        TYPE Trecord        IS TABLE OF      VARCHAR2(200)  ;
        TYPE v_arr          IS TABLE OF      VARCHAR2(200)  INDEX BY BINARY_INTEGER;
        Tvalues  v_arr ;
        Tbegin Trecord := Trecord('browse/',
                                  '" title="',
                                  '<a href="',
                                  'title="',
                                  '<td><b>',
                                  '<a href="',
                                  'right">',
                                  'right">',
                                  'right">' );

        Tend Trecord := Trecord ('"',
                                    '"',
                                    '"',
                                    '"',
                                    '&',
                                    '"',
                                    '&',
                                    '<',
                                    '<');



BEGIN
    -- up buffer size for output verification
      DBMS_OUTPUT.ENABLE(500000);
      DBMS_OUTPUT.PUT_LINE(' -- Start analyse script --');
    -- get the html stream source code

  FOR iPage IN 0..maxPages LOOP
    v_stream := HTTPURITYPE.GETCLOB(HTTPURITYPE.CREATEURI(
      'http://thepiratebay.org/recent/'||iPage ));

  -- chck position of the '</thead>' ------------------------------HEADER --
    currentPosx := DBMS_LOB.INSTR(v_stream, '</thead>', 1, 1);
    v_stream :=REGEXP_REPLACE(DBMS_LOB.SUBSTR(v_stream,length(v_stream),currentposx), '[[:cntrl:]]', '');

    FOR i IN 1..rawPerPage LOOP
     FOR iRecord IN 1..Tbegin.COUNT LOOP

      currentPosx := DBMS_LOB.INSTR(v_stream, Tbegin(iRecord), 1, 1);
      v_stream := REGEXP_REPLACE(DBMS_LOB.SUBSTR(v_stream,length(v_stream), currentposx+length(Tbegin(iRecord)) ), '[[:cntrl:]]', '');

      currentPosx := DBMS_LOB.INSTR(v_stream, Tend(iRecord), 1, 1);
      seqstring := DBMS_LOB.SUBSTR(v_stream, currentPosx-length(Tend(iRecord)), 1) ;
      Tvalues(iRecord) := seqstring ;
        -- delete the record
      v_stream :=  REGEXP_REPLACE(DBMS_LOB.SUBSTR(v_stream,length(v_stream),currentposx ), '[[:cntrl:]]', '');

     END LOOP ; -- end iRecord
     countData := countData +1;
     INSERT INTO torrent (id,
                           category_number ,
                           title_link ,
                           loc_link ,
                           detail_link ,
                           time_link ,
                           url_link ,
                           size_link ,
                           seed ,
                           leech  )
            VALUES ( ((iPage*100)+i),Tvalues(1),Tvalues(2),Tvalues(3),Tvalues(4),Tvalues(5)
                     ,Tvalues(6),Tvalues(7),Tvalues(8),Tvalues(9)  );
    END LOOP; -- end i
  END LOOP; -- end iPage
     DBMS_OUTPUT.PUT_LINE('Receved Data = '||countData);


--      DBMS_OUTPUT.PUT_LINE(v_stream);

      END;
/

Enjoy !