上传文件至 'w10'

w10/CNKIStrategy.java

@@ -0,0 +1,82 @@
+package strategy;
+
+import model.Paper;
+import org.jsoup.nodes.Document;
+import org.jsoup.nodes.Element;
+import org.jsoup.select.Elements;
+import utils.Utils;
+import java.util.ArrayList;
+import java.util.List;
+
+public class CNKIStrategy extends AbstractCrawlerStrategy {
+    @Override
+    public String getPlatformName() {
+        return "中国知网 (CNKI)";
+    }
+    
+    @Override
+    public boolean supportsUrl(String url) {
+        return url != null && (url.contains("cnki.net") || url.contains("cnki.cn"));
+    }
+    
+    @Override
+    protected List<Paper> fetchPapers(String url, int count) throws Exception {
+        List<Paper> papers = new ArrayList<>();
+        System.out.println("=== 开始使用中国知网获取论文 ===");
+        
+        String html = Utils.sendGetRequest(url);
+        if (html.isEmpty()) return papers;
+        
+        Document doc = Utils.parseHtml(html);
+        
+        String[] selectors = {".list-item", ".article-item", ".result-item", "tr[class*='item']", "div[class*='result']", "li[class*='result']"};
+        
+        Elements paperElements = null;
+        for (String selector : selectors) {
+            paperElements = doc.select(selector);
+            if (paperElements.size() > 0) {
+                break;
+            }
+        }
+        
+        if (paperElements != null && paperElements.size() > 0) {
+            int collected = 0;
+            for (Element element : paperElements) {
+                if (collected >= count) break;
+                
+                try {
+                    Element titleElement = element.select("a").first();
+                    if (titleElement == null) continue;
+                    
+                    String title = titleElement.text();
+                    String paperUrl = titleElement.attr("href");
+                    
+                    if (title.length() < 10 || paperUrl.isEmpty()) continue;
+                    
+                    if (!paperUrl.startsWith("http")) {
+                        paperUrl = "https://kns.cnki.net" + paperUrl;
+                    }
+                    
+                    String authors = "";
+                    Elements authorElements = element.select(".author");
+                    if (!authorElements.isEmpty()) {
+                        authors = authorElements.first().text();
+                    }
+                    
+                    String abstractText = "";
+                    Elements abstractElements = element.select(".abstract");
+                    if (!abstractElements.isEmpty()) {
+                        abstractText = abstractElements.first().text();
+                    }
+                    
+                    papers.add(new Paper(title, authors, abstractText, paperUrl, getPlatformName()));
+                    collected++;
+                } catch (Exception e) {
+                    continue;
+                }
+            }
+        }
+        
+        return papers;
+    }
+}

w10/CrawlerStrategy.java

@@ -0,0 +1,10 @@
+package strategy;
+
+import model.Paper;
+import java.util.List;
+
+public interface CrawlerStrategy {
+    String getPlatformName();
+    List<Paper> crawl(String url, int count) throws Exception;
+    boolean supportsUrl(String url);
+}

w10/IEEEStrategy.java

@@ -0,0 +1,64 @@
+package strategy;
+
+import model.Paper;
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.jsoup.nodes.Element;
+import org.jsoup.select.Elements;
+import utils.Utils;
+import java.util.ArrayList;
+import java.util.List;
+
+public class IEEEStrategy extends AbstractCrawlerStrategy {
+    @Override
+    public String getPlatformName() {
+        return "IEEE Xplore";
+    }
+    
+    @Override
+    public boolean supportsUrl(String url) {
+        return url != null && url.contains("ieeexplore.ieee.org");
+    }
+    
+    @Override
+    protected List<Paper> fetchPapers(String url, int count) throws Exception {
+        List<Paper> papers = new ArrayList<>();
+        System.out.println("=== 开始使用IEEE Xplore获取论文 ===");
+        
+        addDelay(2000, 3000);
+        
+        String html = Utils.sendGetRequest(url);
+        if (html.isEmpty()) return papers;
+        
+        Document doc = Jsoup.parse(html);
+        
+        Elements paperElements = doc.select(".List-results-items li");
+        
+        int collected = 0;
+        for (Element element : paperElements) {
+            if (collected >= count) break;
+            
+            try {
+                Element titleElement = element.selectFirst("h2 a");
+                String title = titleElement != null ? titleElement.text() : "";
+                
+                String paperUrl = titleElement != null ? titleElement.attr("href") : "";
+                if (!paperUrl.startsWith("http")) {
+                    paperUrl = "https://ieeexplore.ieee.org" + paperUrl;
+                }
+                
+                Element authorsElement = element.selectFirst(".authors");
+                String authors = authorsElement != null ? authorsElement.text() : "";
+                
+                if (title.length() < 5 || paperUrl.isEmpty()) continue;
+                
+                papers.add(new Paper(title, authors, "", paperUrl, getPlatformName()));
+                collected++;
+            } catch (Exception e) {
+                continue;
+            }
+        }
+        
+        return papers;
+    }
+}

w10/ScienceDirectStrategy.java

@@ -0,0 +1,64 @@
+package strategy;
+
+import model.Paper;
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.jsoup.nodes.Element;
+import org.jsoup.select.Elements;
+import utils.Utils;
+import java.util.ArrayList;
+import java.util.List;
+
+public class ScienceDirectStrategy extends AbstractCrawlerStrategy {
+    @Override
+    public String getPlatformName() {
+        return "ScienceDirect";
+    }
+    
+    @Override
+    public boolean supportsUrl(String url) {
+        return url != null && url.contains("sciencedirect.com");
+    }
+    
+    @Override
+    protected List<Paper> fetchPapers(String url, int count) throws Exception {
+        List<Paper> papers = new ArrayList<>();
+        System.out.println("=== 开始使用ScienceDirect获取论文 ===");
+        
+        addDelay(2000, 3000);
+        
+        String html = Utils.sendGetRequest(url);
+        if (html.isEmpty()) return papers;
+        
+        Document doc = Jsoup.parse(html);
+        
+        Elements paperElements = doc.select(".result-item-content");
+        
+        int collected = 0;
+        for (Element element : paperElements) {
+            if (collected >= count) break;
+            
+            try {
+                Element titleElement = element.selectFirst("h2 a");
+                String title = titleElement != null ? titleElement.text() : "";
+                
+                String paperUrl = titleElement != null ? titleElement.attr("href") : "";
+                if (!paperUrl.startsWith("http")) {
+                    paperUrl = "https://www.sciencedirect.com" + paperUrl;
+                }
+                
+                Element authorsElement = element.selectFirst(".author-group");
+                String authors = authorsElement != null ? authorsElement.text() : "";
+                
+                if (title.length() < 5 || paperUrl.isEmpty()) continue;
+                
+                papers.add(new Paper(title, authors, "", paperUrl, getPlatformName()));
+                collected++;
+            } catch (Exception e) {
+                continue;
+            }
+        }
+        
+        return papers;
+    }
+}

w10/SemanticScholarStrategy.java

@@ -0,0 +1,102 @@
+package strategy;
+
+import model.Paper;
+import com.fasterxml.jackson.databind.ObjectMapper;
+import utils.Utils;
+import java.util.ArrayList;
+import java.util.List;
+
+public class SemanticScholarStrategy extends AbstractCrawlerStrategy {
+    @Override
+    public String getPlatformName() {
+        return "Semantic Scholar";
+    }
+    
+    @Override
+    public boolean supportsUrl(String url) {
+        return url != null && url.contains("semanticscholar.org");
+    }
+    
+    @Override
+    protected List<Paper> fetchPapers(String url, int count) throws Exception {
+        List<Paper> papers = new ArrayList<>();
+        System.out.println("=== 开始使用Semantic Scholar获取论文 ===");
+        
+        addDelay(1000, 1500);
+        
+        try {
+            String response = Utils.sendGetRequest(url);
+            if (response.isEmpty()) return papers;
+            
+            ObjectMapper objectMapper = new ObjectMapper();
+            SemanticScholarResponse apiResponse = objectMapper.readValue(response, SemanticScholarResponse.class);
+            
+            if (apiResponse != null && apiResponse.getItems() != null) {
+                int collected = 0;
+                for (SemanticScholarPaper apiPaper : apiResponse.getItems()) {
+                    if (collected >= count) break;
+                    
+                    String title = apiPaper.getTitle();
+                    String paperUrl = apiPaper.getUrl();
+                    String abstractText = apiPaper.getAbstractText();
+                    
+                    StringBuilder authorsBuilder = new StringBuilder();
+                    if (apiPaper.getAuthors() != null) {
+                        for (SemanticScholarAuthor author : apiPaper.getAuthors()) {
+                            if (author != null && author.getName() != null) {
+                                if (authorsBuilder.length() > 0) {
+                                    authorsBuilder.append(", ");
+                                }
+                                authorsBuilder.append(author.getName());
+                            }
+                        }
+                    }
+                    String authors = authorsBuilder.toString();
+                    
+                    if (title == null || title.length() < 5 || paperUrl == null || paperUrl.isEmpty()) continue;
+                    
+                    papers.add(new Paper(title, authors, abstractText != null ? abstractText : "", paperUrl, getPlatformName()));
+                    collected++;
+                }
+            }
+        } catch (Exception e) {
+            System.out.println("Semantic Scholar解析失败: " + e.getMessage());
+        }
+        
+        return papers;
+    }
+    
+    private static class SemanticScholarResponse {
+        private List<SemanticScholarPaper> items;
+        public List<SemanticScholarPaper> getItems() { return items; }
+        @SuppressWarnings("unused")
+        public void setItems(List<SemanticScholarPaper> items) { this.items = items; }
+    }
+    
+    private static class SemanticScholarPaper {
+        private String title;
+        private String url;
+        private List<SemanticScholarAuthor> authors;
+        private String abstractText;
+        
+        public String getTitle() { return title; }
+        @SuppressWarnings("unused")
+        public void setTitle(String title) { this.title = title; }
+        public String getUrl() { return url; }
+        @SuppressWarnings("unused")
+        public void setUrl(String url) { this.url = url; }
+        public List<SemanticScholarAuthor> getAuthors() { return authors; }
+        @SuppressWarnings("unused")
+        public void setAuthors(List<SemanticScholarAuthor> authors) { this.authors = authors; }
+        public String getAbstractText() { return abstractText; }
+        @SuppressWarnings("unused")
+        public void setAbstractText(String abstractText) { this.abstractText = abstractText; }
+    }
+    
+    private static class SemanticScholarAuthor {
+        private String name;
+        public String getName() { return name; }
+        @SuppressWarnings("unused")
+        public void setName(String name) { this.name = name; }
+    }
+}