Macwelt 2/Macwelt DVD 2.cdr/System/Internet-Utilities/macosx/News Mac 1.1.dmg/Parser.class

Java Class File | 2002-04-08 | 6.9 KB | 235 lines

import java.io.BufferedReader;
import java.io.InputStreamReader;
import java.net.URL;
import java.util.StringTokenizer;
import java.util.Vector;

public class Parser {
   protected Channel channel = new Channel();
   protected Vector html = new Vector(500, 100);
   private int headlinesFound = 0;
   private String headlines = "";

   public Parser() {
   }

   public Parser(Channel theChannel) throws Exception {
      try {
         this.channel = theChannel;
         this.html = this.getPage(this.channel.getSource(), false);
      } catch (Exception var3) {
         throw new Exception("Could not read the HTML file.");
      }
   }

   public Parser(String theChannel, String htmlFile) throws Exception {
      try {
         this.channel = (Channel)IOUtil.openObject(theChannel);
      } catch (Exception var5) {
         throw new Exception("Could not open channel " + theChannel);
      }

      try {
         this.html = this.getPage(htmlFile, false);
      } catch (Exception var4) {
         throw new Exception("Could not open HTML file " + htmlFile);
      }
   }

   protected void saveToCache(String fileName) throws Exception {
      String temp = "";

      for(int i = 0; i < this.html.size(); ++i) {
         temp = temp + (String)this.html.elementAt(i);
      }

      try {
         IOUtil.saveTextFile(fileName, temp, false);
      } catch (Exception var5) {
         throw new Exception("Parser: Error saving cache file " + fileName + "\n" + var5);
      }
   }

   private String filterTags(String source) {
      String temp = "";
      boolean ignore = false;

      for(int i = 0; i < source.length(); ++i) {
         if (!ignore & source.charAt(i) != '<' & source.charAt(i) != '>') {
            temp = temp + source.charAt(i);
         }

         if (source.charAt(i) == '<') {
            ignore = true;
         } else if (source.charAt(i) == '>') {
            ignore = false;
         }
      }

      temp = this.charConvert(temp);
      return temp;
   }

   protected String charConvert(String source) {
      String[] charList = new String[]{"“", "#146;", "”", "&", " ", """, "’", "–", "“", "”", "£", "~", "¨", "©", "—", "<", ">", "£", "'"};
      String[] replaceList = new String[]{"\"", "'", "\"", "&", " ", "\"", "'", "-", "\"", "\"", "£", "~", "\"", "©", " ", "<", ">", "£", "'"};

      for(int i = 0; i < source.length(); ++i) {
         for(int n = 0; n < charList.length; ++n) {
            int wordLength = charList[n].length();
            if (i + wordLength < source.length() && source.substring(i, i + wordLength).equalsIgnoreCase(charList[n])) {
               source = source.substring(0, i) + replaceList[n] + source.substring(i + wordLength, source.length());
            }
         }
      }

      return source;
   }

   protected String[] findHeadlines() throws Exception {
      int number = this.channel.getHeadlineCount();
      String[] returnString = new String[number];
      boolean headline = false;
      String htmlData = "";
      String openingMarker = this.channel.getOpeningMarker().toLowerCase();
      String closingMarker = this.channel.getClosingMarker().toLowerCase();
      String startMarker = this.channel.getStartMarker().toLowerCase();
      String endMarker = this.channel.getEndMarker().toLowerCase();
      int n = 0;
      int startLength = startMarker.length();
      int endLength = endMarker.length();
      int openingMarkerLength = openingMarker.length();
      int closingMarkerLength = closingMarker.length();
      int startingPosition = 0;
      boolean headlineSection = false;
      String carryOver = "";

      for(int lineCount = 0; lineCount < this.html.size(); ++lineCount) {
         String thisLine = ((String)this.html.elementAt(lineCount)).trim();
         if (n < number) {
            for(int i = 0; i < thisLine.length(); ++i) {
               if (n < number) {
                  try {
                     if (i + startLength <= thisLine.length() & !startMarker.equals("") && thisLine.substring(i, i + startLength).equalsIgnoreCase(startMarker)) {
                        headlineSection = true;
                        i += startLength;
                     }
                  } catch (Exception var25) {
                     throw new Exception("Parser: Failed locating start marker.");
                  }

                  try {
                     if (i + endLength <= thisLine.length() & !endMarker.equals("") && thisLine.substring(i, i + endLength).equalsIgnoreCase(endMarker)) {
                        headlineSection = false;
                     }
                  } catch (Exception var24) {
                     throw new Exception("Parser: Failed locating end marker.\n" + var24);
                  }

                  if (startMarker.equals("")) {
                     headlineSection = true;
                  }

                  try {
                     if (i + openingMarkerLength <= thisLine.length() & headlineSection && thisLine.substring(i, i + openingMarkerLength).equalsIgnoreCase(openingMarker)) {
                        startingPosition = i + openingMarkerLength;
                        headline = true;
                     }
                  } catch (Exception var23) {
                     throw new Exception("Parser: Failed locating opening marker.");
                  }

                  try {
                     if (i + closingMarkerLength <= thisLine.length() & headlineSection && thisLine.substring(i, i + closingMarkerLength).equalsIgnoreCase(closingMarker)) {
                        if (!carryOver.equals("")) {
                           try {
                              returnString[n] = this.filterTags(carryOver + thisLine.substring(0, i)).trim();
                              ++n;
                           } catch (Exception var22) {
                              throw new Exception("Parser: Processing carry over headline.");
                           }
                        } else {
                           try {
                              if (startingPosition < i & i < thisLine.length()) {
                                 String theString = thisLine.substring(startingPosition, i);
                                 String temp = this.filterTags(theString);
                                 if (!this.filterTags(temp).equals("")) {
                                    returnString[n] = temp;
                                    ++n;
                                 }
                              }
                           } catch (Exception var27) {
                              throw new Exception("Parser: Processing normal headline failed.");
                           }
                        }

                        carryOver = "";
                        startingPosition = 0;
                        headline = false;
                     }
                  } catch (Exception var28) {
                     throw new Exception("Parser: Failed processing closing marker.");
                  }

                  try {
                     if (i == thisLine.length() - 1 & headline && startingPosition < i) {
                        if (carryOver.equals("")) {
                           carryOver = carryOver + thisLine.substring(startingPosition, i + 1) + " ";
                        } else {
                           carryOver = carryOver + thisLine;
                        }
                     }
                  } catch (Exception var26) {
                     throw new Exception("Parsing error: Failed managing an overflow line.");
                  }
               }
            }
         }
      }

      return returnString;
   }

   protected void getHTML() throws Exception {
      try {
         this.html = this.getPage(this.channel.getSource(), false);
      } catch (Exception var2) {
         throw new Exception("Error downloading " + this.channel.getName());
      }
   }

   protected Vector getPage(String theURL, boolean parse) throws Exception {
      try {
         URL theSite = new URL(theURL);
         BufferedReader in = new BufferedReader(new InputStreamReader(theSite.openStream()));
         Vector tempVector = new Vector(200, 10);
         int p = 0;
         if (parse) {
            throw new Exception("This function is not yet available!");
         } else {
            String inputLine;
            while((inputLine = in.readLine()) != null) {
               tempVector.add(inputLine);
            }

            in.close();
            return tempVector;
         }
      } catch (Exception var8) {
         throw new Exception("Could not get channel from " + theURL);
      }
   }

   protected String[] getHeadlines() {
      StringTokenizer st = new StringTokenizer(this.headlines, "~");
      int n = 0;

      String[] temp;
      for(temp = new String[this.channel.getHeadlineCount()]; n < this.channel.getHeadlineCount() & st.hasMoreTokens(); ++n) {
         temp[n] = st.nextToken();
      }

      return temp;
   }
}