jobconfigure.jsp

Heritrix是一个开源,可扩展的web爬虫项目。Heritrix设计成严格按照robots.txt文件的排除指示和META robots标签。

                                MBeanInfo mapInfo = map.getMBeanInfo(settings);
                                MBeanAttributeInfo mp[] = mapInfo.getAttributes();

                                // Printout modules in map.
                                boolean alt = true;
                                for(int n2=0; n2<mp.length; n2++) {
                                    ModuleAttributeInfo mapAtt = (ModuleAttributeInfo)mp[n2]; //The attributes of the current attribute.

                                    Object currentMapAttribute = null;
                                    Object localMapAttribute = null;
                        
                                    try {
                                        currentMapAttribute = map.getAttribute(settings,mapAtt.getName());
                                        localMapAttribute = map.getLocalAttribute(settings,mapAtt.getName());
                                    } catch (Exception e1) {
                                        p.append(e1.toString() + " " + e1.getMessage());
                                    }
                                    p.append("<tr " + (alt?"bgcolor='#EEEEFF'":"") + "><td>" + mapAtt.getName() + "</td><td>&nbsp;-&gt;&nbsp</td><td>" + currentMapAttribute + "</td><td>&nbsp;<a href=\"javascript:doDeleteMap('"+attAbsoluteName+"','"+mapAtt.getName()+"')\">Remove</a></td></tr>\n");
                                    alt = !alt;
                                }
                                p.append("<tr><td><input name='"+attAbsoluteName+".key' name='"+attAbsoluteName+".key'></td><td>&nbsp;->&nbsp;</td><td><input name='"+attAbsoluteName+".value' name='"+attAbsoluteName+".value'></td><td>&nbsp;<input type='button' value='Add' onClick=\"doAddMap('"+attAbsoluteName+"')\"></td></tr>");
                                p.append("</table>\n");
                            } else if(legalValues != null && legalValues.length > 0) {
                                //Have legal values. Build combobox.
                                p.append("<select name='" + attAbsoluteName + "' style='width: 440px' onChange=\"setEdited('" + attAbsoluteName + "')\">\n");
                                for(int i=0 ; i < legalValues.length ; i++){
                                    p.append("<option value='"+legalValues[i]+"'");
                                    if(currentAttribute.equals(legalValues[i])){
                                        p.append(" selected");
                                    }
                                    p.append(">"+legalValues[i]+"</option>\n");
                                }
                                p.append("</select>\n");
                            } else if (currentAttribute instanceof Boolean){
                                // Boolean value
                                p.append("<select name='" + attAbsoluteName + "' style='width: 440px' onChange=\"setEdited('" + attAbsoluteName + "')\">\n");
                                p.append("<option value='False'"+ (currentAttribute.equals(new Boolean(false))?" selected":"") +">False</option>\n");
                                p.append("<option value='True'"+ (currentAttribute.equals(new Boolean(true))?" selected":"") +">True</option>\n");
                                p.append("</select>\n");
                            } else if (currentAttribute instanceof TextField){
                                // Text area
                                p.append("<textarea name='" + attAbsoluteName + "' style='width: 440px' rows='4' onChange=\"setEdited('" + attAbsoluteName + "')\">");
                                p.append(currentAttribute + "\n");
                                p.append("</textarea>\n");
                            } else {
                                //Input box
                                String patchedAttribute = currentAttribute.toString().replaceAll("&","&amp;");
                                p.append("<input name='" + attAbsoluteName + "' value='" + patchedAttribute + "' style='width: 440px' onChange=\"setEdited('" + attAbsoluteName + "')\">\n");
                            }
                        } else {
                            // Display non editable
                            if(currentAttribute instanceof ListType){
                                // Print list
                                ListType list = (ListType)currentAttribute;
                                p.append("</td><td colspan='" + (settings==null?"1":"2") + "'>");
                                for(int i=0 ; i<list.size() ; i++){
                                    p.append(list.get(i)+"<br>\n");
                                }
                            } else if(simpleMap) {
                                // Simple map
                                MapType map = (MapType)currentAttribute;
                                p.append("</td><td '" + (settings==null?"1":"2") + "'><table border='0' cellspacing='0' cellpadding='0'>\n");
                                
                                MBeanInfo mapInfo = map.getMBeanInfo(settings);
                                MBeanAttributeInfo mp[] = mapInfo.getAttributes();

                                // Printout modules in map.
                                for(int n2=0; n2<mp.length; n2++) {
                                    ModuleAttributeInfo mapAtt = (ModuleAttributeInfo)mp[n2]; //The attributes of the current attribute.

                                    Object currentMapAttribute = null;
                                    Object localMapAttribute = null;
                        
                                    try {
                                        currentMapAttribute = map.getAttribute(settings,mapAtt.getName());
                                        localMapAttribute = map.getLocalAttribute(settings,mapAtt.getName());
                                    } catch (Exception e1) {
                                        p.append(e1.toString() + " " + e1.getMessage());
                                    }
                                    p.append("<tr><td>" + mapAtt.getName() + "</td><td>&nbsp;-&gt;&nbsp</td><td>" + currentMapAttribute + "</td></tr>\n");
                                }
                                p.append("</table>\n");
                            } else {
                                p.append("</td><td colspan='" + (settings==null?"1":"2") + "'>"+currentAttribute);                        
                            }
                        }
                        p.append("</td></tr>\n");
                    }
                }
            }
        }
        return p.toString();
    }
    
    /**
     * Checks if there is an error for a specific attribute for a given
     * CrawlerSettings
     *
     * @param key The absolutename of the attribute to check for.
     * @param errorHandler The errorHandler containing the errors
     * @param settings the CrawlerSettings that is the 'current' context
     *
     */
    public String checkError(String key, CrawlJobErrorHandler errorHandler,
            CrawlerSettings settings){
        Constraint.FailedCheck failedCheck = null;
        if (errorHandler != null) {
            failedCheck = (Constraint.FailedCheck)errorHandler.getError(key);
        }
        if (failedCheck != null) {
            boolean sameSetting = false;
            if(settings != null && failedCheck.getSettings() == settings) {
                sameSetting = true;
            } else if(settings == null) {
                // If failedCheck.getSettings is the global setting then true.
                if(failedCheck.getSettings().getScope() == null ||
                        failedCheck.getSettings().getScope().length() == 0) {
                    sameSetting = true;
                }
            }
            
            if(sameSetting){
                return "<a class='help' style='color: red' href=\"javascript:doPop('" + 
                    TextUtils.escapeForHTMLJavascript(failedCheck.getMessage()) + "')\">*</a>";
            }
        }
        return "";
    }
%>


<script type="text/javascript">
    function doAddList(listName){
        newItem = document.getElementById(listName+".add");
        theList = document.getElementById(listName);
        
        if(newItem.value.length > 0){
            insertLocation = theList.length;
            theList.options[insertLocation] = new Option(newItem.value, newItem.value, false, false);
            newItem.value = "";
        }
        setEdited(listName);
    }
    
    function doDeleteList(listName){
        theList = document.getElementById(listName);
        theList.options[theList.selectedIndex] = null;
        setEdited(listName);
    }

    function doAddMap(mapName){
        document.frmConfig.action.value = "addMap";
        document.frmConfig.update.value = mapName;
        doSubmit();
    }
    
    function doDeleteMap(mapName, key){
        document.frmConfig.action.value = "deleteMap";
        document.frmConfig.update.value = mapName;
        document.frmConfig.item.value = key;
        doSubmit();
    }            
</script>

<%
    // This code is shared by each of the configure.jsp pages. 
    // Sets up the CrawlJob, CrawlOrder, settingsHandler, 
    // the CrawlJobErrorHandler and sets the expert boolean.
    CrawlJob theJob = handler.getJob(request.getParameter("job"));
    if (theJob == null) {
        // Didn't find any job with the given UID or no UID given.
        response.sendRedirect(request.getContextPath() + "/jobs.jsp?message=" +
            "No job selected " + request.getParameter("job"));
        return;
    } else if(theJob.isReadOnly()) {
        // Can't edit this job.
        response.sendRedirect(request.getContextPath() + "/jobs.jsp?message=" +
            "Can't configure a read only job");
        return;
    }
    CrawlJobErrorHandler errorHandler = theJob.getErrorHandler();
    boolean expert = false;
    if(CookieUtils.getCookieValue(request.getCookies(), "expert", 
            "false").equals("true")) {
        expert = true;
    }
    // Get the settings objects.
    XMLSettingsHandler settingsHandler = theJob.getSettingsHandler();
    CrawlOrder crawlOrder = settingsHandler.getOrder();
%>