[Update] 使用搜索推荐接口优化搜索功能对搜索内容的相关性;

[Add] PixivURL 增加搜索推荐接口Url; [Add] PixivSearchLinkBuilder 增加`getContent`方法以允许外部获取搜索原始内容; [Add] CacheStoreCentral 增加`getSearchBody(PixivSearchLinkBuilder)`方法以允许直接提供builder进行请求; [Change] CacheStoreCentral, PixivUtils 移除`getSearchBody(String, ...)`方法对PixivSearchLinkBuilder的构建, 将该部分转移至`PixivUtils.buildSearchLinkBuilder`方法; [Update] BotCommandProcess 增加`addRecommendKeywords(PixivSearchLinkBuilder, int)`方法以针对PixivSearchLinkBuilder的搜索原始内容增加指定数量的搜索推荐关键词;
2025-04-29 14:17:33 +00:00 · 2020-07-15 09:00:00 +08:00 · 2020-07-15 09:00:00 +08:00 · f976017a89
commit f976017a89
parent 575dc0c7fb
5 changed files with 181 additions and 65 deletions
--- a/src/main/java/net/lamgc/cgj/bot/BotCommandProcess.java
+++ b/src/main/java/net/lamgc/cgj/bot/BotCommandProcess.java
@ -11,12 +11,17 @@ import net.lamgc.cgj.bot.cache.JsonRedisCacheStore;
 import net.lamgc.cgj.bot.event.BufferedMessageSender;
 import net.lamgc.cgj.bot.sort.PreLoadDataAttribute;
 import net.lamgc.cgj.bot.sort.PreLoadDataAttributeComparator;
+import net.lamgc.cgj.bot.util.PixivUtils;
 import net.lamgc.cgj.pixiv.PixivDownload;
 import net.lamgc.cgj.pixiv.PixivDownload.PageQuality;
 import net.lamgc.cgj.pixiv.PixivSearchLinkBuilder;
 import net.lamgc.cgj.pixiv.PixivURL;
 import net.lamgc.utils.base.runner.Argument;
 import net.lamgc.utils.base.runner.Command;
+import org.apache.http.HttpHeaders;
+import org.apache.http.HttpResponse;
+import org.apache.http.client.methods.HttpGet;
+import org.apache.http.util.EntityUtils;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

@ -315,6 +320,16 @@ public class BotCommandProcess {
            @Argument(name = "p", force = false, defaultValue = "1") int pagesIndex
    ) throws IOException, InterruptedException {
        log.debug("正在执行搜索...");
+        PixivSearchLinkBuilder linkBuilder = PixivUtils.buildSearchLinkBuilder(content, type, area, includeKeywords,
+                excludeKeywords, contentOption, pagesIndex);
+        int recommendKeywordCount = 0;
+        try {
+            recommendKeywordCount = Integer.parseInt(SettingProperties.getProperties(fromGroup)
+                    .getProperty("search.recommendKeywordCount", "0"));
+        } catch(NumberFormatException e) {
+            log.warn("配置项 search.recommendKeywordCount 的值无效");
+        }
+        addRecommendKeywords(linkBuilder, recommendKeywordCount);
        JsonObject resultBody = CacheStoreCentral.getCentral()
                .getSearchBody(content, type, area, includeKeywords, excludeKeywords, contentOption, pagesIndex);

@ -525,6 +540,43 @@ public class BotCommandProcess {
        return "色图姬收到了你的报告，将屏蔽该作品并对作品违规情况进行核实，感谢你的反馈！";
    }

+    /**
+     * 根据Pixiv搜索推荐列表补充关键词.
+     * <p>该操作可能会优化搜索效果.</p>
+     * @param searchLinkBuilder PixivSearchLinkBuilder对象
+     * @param includeKeywordsCount 需要添加的关键词数量
+     * @throws IOException 当获取推荐列表发生异常时抛出.
+     */
+    private static void addRecommendKeywords(PixivSearchLinkBuilder searchLinkBuilder, int includeKeywordsCount)
+            throws IOException {
+        if(includeKeywordsCount <= 0) {
+            return;
+        }
+        HttpGet request = BotGlobal.getGlobal().getPixivDownload()
+                .createHttpGetRequest(PixivURL.PIXIV_SEARCH_RECOMMENDS
+                    .replace("{content}", searchLinkBuilder.getContent()));
+        request.addHeader(HttpHeaders.REFERER, "https://www.pixiv.net/");
+
+        HttpResponse response = BotGlobal.getGlobal().getPixivDownload().getHttpClient().execute(request);
+        String responseBody = EntityUtils.toString(response.getEntity());
+        if(response.getStatusLine().getStatusCode() != 200) {
+            throw new IOException("Interface request failure: " + response.getStatusLine() +
+                    ", response body: '" + responseBody + "'");
+        }
+
+        JsonObject resultObject = BotGlobal.getGlobal().getGson()
+                .fromJson(responseBody, JsonObject.class);
+        if(!resultObject.has("candidates")) {
+            return;
+        }
+
+        JsonArray recommendsArr = resultObject.getAsJsonArray("candidates");
+        for (int count = 0; count < includeKeywordsCount && count < recommendsArr.size(); count++) {
+            searchLinkBuilder.addIncludeKeyword(
+                    recommendsArr.get(count).getAsJsonObject().get("tag_name").getAsString());
+        }
+    }
+
    /**
     * 检查某一作品是否被报告
     * @param illustId 作品Id
--- a/src/main/java/net/lamgc/cgj/bot/cache/CacheStoreCentral.java
+++ b/src/main/java/net/lamgc/cgj/bot/cache/CacheStoreCentral.java
@ -8,6 +8,7 @@ import net.lamgc.cgj.bot.BotCode;
 import net.lamgc.cgj.bot.BotCommandProcess;
 import net.lamgc.cgj.bot.SettingProperties;
 import net.lamgc.cgj.bot.boot.BotGlobal;
+import net.lamgc.cgj.bot.util.PixivUtils;
 import net.lamgc.cgj.exception.HttpRequestException;
 import net.lamgc.cgj.pixiv.PixivDownload;
 import net.lamgc.cgj.pixiv.PixivSearchLinkBuilder;
@ -405,73 +406,11 @@ public final class CacheStoreCentral {
        return PixivDownload.getRanking(result, start - 1, range);
    }

-    /**
-     * 获取搜索结果
-     * @param content 搜索内容
-     * @param type 类型
-     * @param area 范围
-     * @param includeKeywords 包含关键词
-     * @param excludeKeywords 排除关键词
-     * @param contentOption 内容类型
-     * @return 返回完整搜索结果
-     * @throws IOException 当请求发生异常, 或接口返回异常信息时抛出.
-     */
-    public JsonObject getSearchBody(
-            String content,
-            String type,
-            String area,
-            String includeKeywords,
-            String excludeKeywords,
-            String contentOption,
-            int pageIndex
-    ) throws IOException {
-        PixivSearchLinkBuilder searchBuilder = new PixivSearchLinkBuilder(Strings.isNullOrEmpty(content) ? "" : content);
-        if (type != null) {
-            try {
-                searchBuilder.setSearchType(PixivSearchLinkBuilder.SearchType.valueOf(type.toUpperCase()));
-            } catch (IllegalArgumentException e) {
-                log.warn("不支持的SearchType: {}", type);
-            }
-        }
-        if (area != null) {
-            try {
-                searchBuilder.setSearchArea(PixivSearchLinkBuilder.SearchArea.valueOf(area));
-            } catch (IllegalArgumentException e) {
-                log.warn("不支持的SearchArea: {}", area);
-            }
-        }
-        if (contentOption != null) {
-            try {
-                searchBuilder.setSearchContentOption(
-                        PixivSearchLinkBuilder.SearchContentOption.valueOf(contentOption.trim().toUpperCase()));
-            } catch (IllegalArgumentException e) {
-                log.warn("不支持的SearchContentOption: {}", contentOption);
-            }
-        }
-
-        if (!Strings.isNullOrEmpty(includeKeywords)) {
-            for (String keyword : includeKeywords.split(";")) {
-                searchBuilder.removeExcludeKeyword(keyword.trim());
-                searchBuilder.addIncludeKeyword(keyword.trim());
-                log.trace("已添加关键字: {}", keyword);
-            }
-        }
-        if (!Strings.isNullOrEmpty(excludeKeywords)) {
-            for (String keyword : excludeKeywords.split(";")) {
-                searchBuilder.removeIncludeKeyword(keyword.trim());
-                searchBuilder.addExcludeKeyword(keyword.trim());
-                log.trace("已添加排除关键字: {}", keyword);
-            }
-        }
-
-        if(pageIndex > 0) {
-            searchBuilder.setPage(pageIndex);
-        }
-
-        log.debug("正在搜索作品, 条件: {}", searchBuilder.getSearchCondition());
+    public JsonObject getSearchBody(PixivSearchLinkBuilder searchLinkBuilder) throws IOException {
+        log.debug("正在搜索作品, 条件: {}", searchLinkBuilder.getSearchCondition());

        Locker<String> locker
-                = buildSyncKey(searchBuilder.buildURL());
+                = buildSyncKey(searchLinkBuilder.buildURL());
        String requestUrl = locker.getKey();
        log.debug("RequestUrl: {}", requestUrl);
        JsonObject resultBody = null;
@ -524,6 +463,31 @@ public final class CacheStoreCentral {
        return resultBody;
    }

+
+    /**
+     * 获取搜索结果
+     * @param content 搜索内容
+     * @param type 类型
+     * @param area 范围
+     * @param includeKeywords 包含关键词
+     * @param excludeKeywords 排除关键词
+     * @param contentOption 内容类型
+     * @return 返回完整搜索结果
+     * @throws IOException 当请求发生异常, 或接口返回异常信息时抛出.
+     */
+    public JsonObject getSearchBody(
+            String content,
+            String type,
+            String area,
+            String includeKeywords,
+            String excludeKeywords,
+            String contentOption,
+            int pageIndex
+    ) throws IOException {
+        return getSearchBody(PixivUtils.buildSearchLinkBuilder(content, type, area, includeKeywords,
+                excludeKeywords, contentOption, pageIndex));
+    }
+
    protected ImageChecksum getImageChecksum(int illustId, int pageIndex) {
        String cacheKey = illustId + ":" + pageIndex;
        if(!imageChecksumCache.exists(cacheKey)) {
--- a/src/main/java/net/lamgc/cgj/bot/util/PixivUtils.java
+++ b/src/main/java/net/lamgc/cgj/bot/util/PixivUtils.java
@ -0,0 +1,86 @@
+package net.lamgc.cgj.bot.util;
+
+import com.google.common.base.Strings;
+import net.lamgc.cgj.pixiv.PixivSearchLinkBuilder;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Pixiv工具类
+ */
+public final class PixivUtils {
+
+    private final static Logger log = LoggerFactory.getLogger(PixivUtils.class);
+
+    private PixivUtils() {}
+
+    /**
+     * 快速构造一个PixivSearchLinkBuilder
+     * @param content 搜索内容
+     * @param type 搜索类型
+     * @param area 搜索范围
+     * @param includeKeywords 包含关键词
+     * @param excludeKeywords 排除关键词
+     * @param contentOption 内容级别选项
+     * @param pageIndex 搜索页数
+     * @return 返回PixivSearchLinkBuilder对象
+     * @see PixivSearchLinkBuilder
+     */
+    public static PixivSearchLinkBuilder buildSearchLinkBuilder(
+            String content,
+            String type,
+            String area,
+            String includeKeywords,
+            String excludeKeywords,
+            String contentOption,
+            int pageIndex
+    ) {
+        PixivSearchLinkBuilder searchBuilder = new PixivSearchLinkBuilder(Strings.isNullOrEmpty(content) ? "" : content);
+        if (type != null) {
+            try {
+                searchBuilder.setSearchType(PixivSearchLinkBuilder.SearchType.valueOf(type.toUpperCase()));
+            } catch (IllegalArgumentException e) {
+                log.warn("不支持的SearchType: {}", type);
+            }
+        }
+        if (area != null) {
+            try {
+                searchBuilder.setSearchArea(PixivSearchLinkBuilder.SearchArea.valueOf(area));
+            } catch (IllegalArgumentException e) {
+                log.warn("不支持的SearchArea: {}", area);
+            }
+        }
+        if (contentOption != null) {
+            try {
+                searchBuilder.setSearchContentOption(
+                        PixivSearchLinkBuilder.SearchContentOption.valueOf(contentOption.trim().toUpperCase()));
+            } catch (IllegalArgumentException e) {
+                log.warn("不支持的SearchContentOption: {}", contentOption);
+            }
+        }
+
+        if (!Strings.isNullOrEmpty(includeKeywords)) {
+            for (String keyword : includeKeywords.split(";")) {
+                searchBuilder.removeExcludeKeyword(keyword.trim());
+                searchBuilder.addIncludeKeyword(keyword.trim());
+                log.trace("已添加关键字: {}", keyword);
+            }
+        }
+        if (!Strings.isNullOrEmpty(excludeKeywords)) {
+            for (String keyword : excludeKeywords.split(";")) {
+                searchBuilder.removeIncludeKeyword(keyword.trim());
+                searchBuilder.addExcludeKeyword(keyword.trim());
+                log.trace("已添加排除关键字: {}", keyword);
+            }
+        }
+
+        if(pageIndex > 0) {
+            searchBuilder.setPage(pageIndex);
+        }
+
+        return searchBuilder;
+    }
+
+
+
+}
--- a/src/main/java/net/lamgc/cgj/pixiv/PixivSearchLinkBuilder.java
+++ b/src/main/java/net/lamgc/cgj/pixiv/PixivSearchLinkBuilder.java
@ -161,6 +161,10 @@ public class PixivSearchLinkBuilder {
                '}';
    }

+    public String getContent() {
+        return content;
+    }
+
    public PixivSearchLinkBuilder setSearchArea(SearchArea searchArea) {
        this.searchArea = Objects.requireNonNull(searchArea);
        return this;
--- a/src/main/java/net/lamgc/cgj/pixiv/PixivURL.java
+++ b/src/main/java/net/lamgc/cgj/pixiv/PixivURL.java
@ -194,6 +194,16 @@ public final class PixivURL {
     */
    public final static String PIXIV_USER_COLLECTION_PAGE = "https://www.pixiv.net/bookmark.php?rest=show&p={pageIndex}";

+    /**
+     * 搜索推荐接口.
+     * <p>可返回与搜索内容相关的标签信息, 以优化搜索效果.</p>
+     * 需要替换的文本:
+     * <ul>
+     *     <li>{content} - 搜索内容</li>
+     * </ul>
+     */
+    public final static String PIXIV_SEARCH_RECOMMENDS = "https://www.pixiv.net/rpc/cps.php?keywork={content}";
+
    /**
     * 获取排名榜
     * @param mode 查询类型, 详细信息看{@link RankingMode}, 如本参数为null, 则为每天