Refactored code for increased optimization. (#1139)
* refactoring by decompose conditional technique * refactoring by introduction explaining variable technique * refactoring by rename method/variable technique * refactoring by introducing explaining variable technique * Added Extract class refactoring to increase maintainablilty * Refactoring using replace conditional with polymorphismmaster
parent
9b9f173c1c
commit
f051d978e2
|
@ -169,18 +169,25 @@ public class Page {
|
|||
* @param priority Priority for the URL
|
||||
*/
|
||||
private void addRequestIfValid(String url, long priority) {
|
||||
if (StringUtils.isBlank(url) || url.equals("#") || url.startsWith("javascript:")) {
|
||||
return;
|
||||
boolean isBlankUrl = StringUtils.isBlank(url);
|
||||
boolean isHashSymbol = url.equals("#");
|
||||
boolean isJavaScript = url.startsWith("javascript:");
|
||||
|
||||
if (isBlankUrl || isHashSymbol || isJavaScript) {
|
||||
return; // Invalid URL, so no further processing is needed.
|
||||
}
|
||||
|
||||
String canonicalizedUrl = UrlUtils.canonicalizeUrl(url, this.url.toString());
|
||||
Request req = new Request(canonicalizedUrl);
|
||||
if(priority > 0) {
|
||||
req.setPriority(priority);
|
||||
Request request = new Request(canonicalizedUrl);
|
||||
|
||||
if (priority > 0) {
|
||||
request.setPriority(priority);
|
||||
}
|
||||
targetRequests.add(req);
|
||||
|
||||
targetRequests.add(request);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* add url to fetch
|
||||
*
|
||||
|
|
|
@ -40,13 +40,14 @@ public class HttpClientGenerator {
|
|||
|
||||
private PoolingHttpClientConnectionManager connectionManager;
|
||||
|
||||
private static final int DEFAULT_MAX_PER_ROUTE = 100;
|
||||
public HttpClientGenerator() {
|
||||
Registry<ConnectionSocketFactory> reg = RegistryBuilder.<ConnectionSocketFactory>create()
|
||||
.register("http", PlainConnectionSocketFactory.INSTANCE)
|
||||
.register("https", buildSSLConnectionSocketFactory())
|
||||
.build();
|
||||
connectionManager = new PoolingHttpClientConnectionManager(reg);
|
||||
connectionManager.setDefaultMaxPerRoute(100);
|
||||
connectionManager.setDefaultMaxPerRoute(DEFAULT_MAX_PER_ROUTE);
|
||||
}
|
||||
|
||||
private SSLConnectionSocketFactory buildSSLConnectionSocketFactory() {
|
||||
|
|
|
@ -64,7 +64,7 @@ public class HttpRequestBody implements Serializable {
|
|||
this.encoding = encoding;
|
||||
}
|
||||
|
||||
public static HttpRequestBody json(String json, String encoding) {
|
||||
public static HttpRequestBody createJsonRequestBody(String json, String encoding) {
|
||||
try {
|
||||
return new HttpRequestBody(json.getBytes(encoding), ContentType.JSON, encoding);
|
||||
} catch (UnsupportedEncodingException e) {
|
||||
|
|
|
@ -0,0 +1,53 @@
|
|||
package us.codecraft.webmagic.selector;
|
||||
|
||||
import org.jsoup.nodes.Document;
|
||||
import org.jsoup.nodes.Element;
|
||||
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
import java.util.ListIterator;
|
||||
|
||||
public class ElementsUtil {
|
||||
HtmlNode htmlNode = new HtmlNode();
|
||||
public Selectable selectElements(BaseElementSelector elementSelector) {
|
||||
ListIterator<Element> elementIterator = htmlNode.getElements().listIterator();
|
||||
if (!elementSelector.hasAttribute()) {
|
||||
List<Element> resultElements = new ArrayList<Element>();
|
||||
while (elementIterator.hasNext()) {
|
||||
Element element = checkElementAndConvert(elementIterator);
|
||||
List<Element> selectElements = elementSelector.selectElements(element);
|
||||
resultElements.addAll(selectElements);
|
||||
}
|
||||
return new HtmlNode(resultElements);
|
||||
} else {
|
||||
// has attribute, consider as plaintext
|
||||
List<String> resultStrings = new ArrayList<String>();
|
||||
while (elementIterator.hasNext()) {
|
||||
Element element = checkElementAndConvert(elementIterator);
|
||||
List<String> selectList = elementSelector.selectList(element);
|
||||
resultStrings.addAll(selectList);
|
||||
}
|
||||
return new PlainText(resultStrings);
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Only document can be select
|
||||
* See: https://github.com/code4craft/webmagic/issues/113
|
||||
*
|
||||
* @param elementIterator elementIterator
|
||||
* @return element element
|
||||
*/
|
||||
public Element checkElementAndConvert(ListIterator<Element> elementIterator) {
|
||||
Element element = elementIterator.next();
|
||||
if (!(element instanceof Document)) {
|
||||
Document root = new Document(element.ownerDocument().baseUri());
|
||||
Element clone = element.clone();
|
||||
root.appendChild(clone);
|
||||
elementIterator.set(root);
|
||||
return root;
|
||||
}
|
||||
return element;
|
||||
}
|
||||
}
|
|
@ -33,19 +33,22 @@ public class HtmlNode extends AbstractSelectable {
|
|||
|
||||
@Override
|
||||
public Selectable links() {
|
||||
return selectElements(new LinksSelector());
|
||||
ElementsUtil elementsUtil = new ElementsUtil();
|
||||
return elementsUtil.selectElements(new LinksSelector());
|
||||
}
|
||||
|
||||
@Override
|
||||
public Selectable xpath(String xpath) {
|
||||
ElementsUtil elementsUtil = new ElementsUtil();
|
||||
XpathSelector xpathSelector = Selectors.xpath(xpath);
|
||||
return selectElements(xpathSelector);
|
||||
return elementsUtil.selectElements(xpathSelector);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Selectable selectList(Selector selector) {
|
||||
if (selector instanceof BaseElementSelector) {
|
||||
return selectElements((BaseElementSelector) selector);
|
||||
ElementsUtil elementsUtil = new ElementsUtil();
|
||||
return elementsUtil.selectElements((BaseElementSelector) selector);
|
||||
}
|
||||
return selectList(selector, getSourceTexts());
|
||||
}
|
||||
|
@ -55,64 +58,18 @@ public class HtmlNode extends AbstractSelectable {
|
|||
return selectList(selector);
|
||||
}
|
||||
|
||||
/**
|
||||
* select elements
|
||||
*
|
||||
* @param elementSelector elementSelector
|
||||
* @return result
|
||||
*/
|
||||
protected Selectable selectElements(BaseElementSelector elementSelector) {
|
||||
ListIterator<Element> elementIterator = getElements().listIterator();
|
||||
if (!elementSelector.hasAttribute()) {
|
||||
List<Element> resultElements = new ArrayList<Element>();
|
||||
while (elementIterator.hasNext()) {
|
||||
Element element = checkElementAndConvert(elementIterator);
|
||||
List<Element> selectElements = elementSelector.selectElements(element);
|
||||
resultElements.addAll(selectElements);
|
||||
}
|
||||
return new HtmlNode(resultElements);
|
||||
} else {
|
||||
// has attribute, consider as plaintext
|
||||
List<String> resultStrings = new ArrayList<String>();
|
||||
while (elementIterator.hasNext()) {
|
||||
Element element = checkElementAndConvert(elementIterator);
|
||||
List<String> selectList = elementSelector.selectList(element);
|
||||
resultStrings.addAll(selectList);
|
||||
}
|
||||
return new PlainText(resultStrings);
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Only document can be select
|
||||
* See: https://github.com/code4craft/webmagic/issues/113
|
||||
*
|
||||
* @param elementIterator elementIterator
|
||||
* @return element element
|
||||
*/
|
||||
private Element checkElementAndConvert(ListIterator<Element> elementIterator) {
|
||||
Element element = elementIterator.next();
|
||||
if (!(element instanceof Document)) {
|
||||
Document root = new Document(element.ownerDocument().baseUri());
|
||||
Element clone = element.clone();
|
||||
root.appendChild(clone);
|
||||
elementIterator.set(root);
|
||||
return root;
|
||||
}
|
||||
return element;
|
||||
}
|
||||
|
||||
@Override
|
||||
public Selectable $(String selector) {
|
||||
ElementsUtil elementsUtil = new ElementsUtil();
|
||||
CssSelector cssSelector = Selectors.$(selector);
|
||||
return selectElements(cssSelector);
|
||||
return elementsUtil.selectElements(cssSelector);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Selectable $(String selector, String attrName) {
|
||||
ElementsUtil elementsUtil = new ElementsUtil();
|
||||
CssSelector cssSelector = Selectors.$(selector, attrName);
|
||||
return selectElements(cssSelector);
|
||||
return elementsUtil.selectElements(cssSelector);
|
||||
}
|
||||
|
||||
@Override
|
||||
|
|
|
@ -76,26 +76,27 @@ public class ExtractRule {
|
|||
}
|
||||
|
||||
private Selector compileSelector() {
|
||||
SelectorFactory factory;
|
||||
switch (expressionType) {
|
||||
case Css:
|
||||
if (expressionParams.length >= 1) {
|
||||
return $(expressionValue, expressionParams[0]);
|
||||
} else {
|
||||
return $(expressionValue);
|
||||
}
|
||||
factory = new CssSelectorFactory();
|
||||
break;
|
||||
case XPath:
|
||||
return xpath(expressionValue);
|
||||
factory = new XPathSelectorFactory();
|
||||
break;
|
||||
case Regex:
|
||||
if (expressionParams.length >= 1) {
|
||||
return regex(expressionValue, Integer.parseInt(expressionParams[0]));
|
||||
} else {
|
||||
return regex(expressionValue);
|
||||
}
|
||||
factory = new RegexSelectorFactory();
|
||||
break;
|
||||
case JsonPath:
|
||||
return new JsonPathSelector(expressionValue);
|
||||
factory = new JsonPathSelectorFactory();
|
||||
break;
|
||||
default:
|
||||
return xpath(expressionValue);
|
||||
factory = new XPathSelectorFactory(); // Default to XPath
|
||||
}
|
||||
|
||||
SelectorCompiler selectorCompiler = new SelectorCompiler(factory);
|
||||
Selector compiledSelector = selectorCompiler.compileSelector(expressionValue, expressionParams);
|
||||
return compiledSelector;
|
||||
}
|
||||
|
||||
public void setSelector(Selector selector) {
|
||||
|
|
|
@ -0,0 +1,57 @@
|
|||
package us.codecraft.webmagic.configurable;
|
||||
|
||||
import us.codecraft.webmagic.selector.JsonPathSelector;
|
||||
import us.codecraft.webmagic.selector.Selector;
|
||||
|
||||
import static us.codecraft.webmagic.selector.Selectors.*;
|
||||
public interface SelectorFactory {
|
||||
Selector compileSelector(String expressionValue, String[] expressionParams);
|
||||
}
|
||||
|
||||
class CssSelectorFactory implements SelectorFactory {
|
||||
@Override
|
||||
public Selector compileSelector(String expressionValue, String[] expressionParams) {
|
||||
if (expressionParams.length >= 1) {
|
||||
return $(expressionValue, expressionParams[0]);
|
||||
} else {
|
||||
return $(expressionValue);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
class XPathSelectorFactory implements SelectorFactory {
|
||||
@Override
|
||||
public Selector compileSelector(String expressionValue, String[] expressionParams) {
|
||||
return xpath(expressionValue);
|
||||
}
|
||||
}
|
||||
|
||||
class RegexSelectorFactory implements SelectorFactory {
|
||||
@Override
|
||||
public Selector compileSelector(String expressionValue, String[] expressionParams) {
|
||||
if (expressionParams.length >= 1) {
|
||||
return regex(expressionValue, Integer.parseInt(expressionParams[0]));
|
||||
} else {
|
||||
return regex(expressionValue);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
class JsonPathSelectorFactory implements SelectorFactory {
|
||||
@Override
|
||||
public Selector compileSelector(String expressionValue, String[] expressionParams) {
|
||||
return new JsonPathSelector(expressionValue);
|
||||
}
|
||||
}
|
||||
|
||||
class SelectorCompiler {
|
||||
private final SelectorFactory selectorFactory;
|
||||
|
||||
public SelectorCompiler(SelectorFactory selectorFactory) {
|
||||
this.selectorFactory = selectorFactory;
|
||||
}
|
||||
|
||||
public Selector compileSelector(String expressionValue, String[] expressionParams) {
|
||||
return selectorFactory.compileSelector(expressionValue, expressionParams);
|
||||
}
|
||||
}
|
Loading…
Reference in New Issue