feat: enhance semantic search with multiple paths and pattern filtering

src/dfm-search/dfm-search-lib/semantic/extractors/actionextractor.cpp

@@ -7,9 +7,10 @@
 #include "semantic/semanticruleengine.h"
 #include "locationextractor.h"
 
 #include <QDir>
 #include <QFileInfo>
 #include <QDebug>
+#include <QVariant>
 
 DFM_SEARCH_BEGIN_NS
 
@@ -137,16 +138,29 @@
         }
 
         const QString xdgType = meta.value("xdg_type").toString();
-        const QString customSubdir = meta.value("custom_path").toString();
+        const QVariant customPathVar = meta.value("custom_path");
 
         QStringList resolved;
-        if (!customSubdir.isEmpty()) {
-            resolved = LocationExtractor::expandCustomPath(
-                    QDir::homePath(), customSubdir);
-        } else if (!xdgType.isEmpty()) {
-            const QString path = LocationExtractor::resolveXdgPath(xdgType);
-            if (!path.isEmpty()) {
-                resolved = { path };
+        if (customPathVar.userType() == QMetaType::QVariantList) {
+            for (const QVariant &pathVar : customPathVar.toList()) {
+                const QString relPath = pathVar.toString();
+                if (relPath.isEmpty()) continue;
+                const QStringList expanded = LocationExtractor::expandCustomPath(
+                        QDir::homePath(), relPath);
+                for (const QString &p : expanded) {
+                    if (!resolved.contains(p)) resolved.append(p);
+                }
+            }
+        } else {
+            const QString customSubdir = customPathVar.toString();
+            if (!customSubdir.isEmpty()) {
+                resolved = LocationExtractor::expandCustomPath(
+                        QDir::homePath(), customSubdir);
+            } else if (!xdgType.isEmpty()) {
+                const QString path = LocationExtractor::resolveXdgPath(xdgType);
+                if (!path.isEmpty()) {
+                    resolved = { path };
+                }
             }
         }
 

src/dfm-search/dfm-search-lib/semantic/extractors/locationextractor.cpp

@@ -6,9 +6,10 @@
 
 #include "semantic/semanticruleengine.h"
 
 #include <QDir>
 #include <QFileInfo>
 #include <QStandardPaths>
+#include <QVariant>
 
 DFM_SEARCH_BEGIN_NS
 
@@ -34,7 +35,7 @@
         const QVariantMap metadata = m_engine->ruleMetadata("location", ruleIds[i]);
 
         const QString xdgType = metadata.value("xdg_type").toString();
-        const QString customSubdir = metadata.value("custom_path").toString();
+        const QVariant customPathVar = metadata.value("custom_path");
         const bool includeHidden = metadata.value("include_hidden", false).toBool();
 
         // Always record the consumed span when a rule matches,
@@ -47,12 +48,24 @@
         intent.consumedSpans().append(span);
 
         QStringList resolvedPaths;
-        if (!customSubdir.isEmpty()) {
-            resolvedPaths = expandCustomPath(QDir::homePath(), customSubdir);
+        if (customPathVar.userType() == QMetaType::QVariantList) {
+            for (const QVariant &pathVar : customPathVar.toList()) {
+                const QString relPath = pathVar.toString();
+                if (relPath.isEmpty()) continue;
+                const QStringList expanded = expandCustomPath(QDir::homePath(), relPath);
+                for (const QString &p : expanded) {
+                    if (!resolvedPaths.contains(p)) resolvedPaths.append(p);
+                }
+            }
         } else {
-            const QString path = resolveXdgPath(xdgType);
-            if (!path.isEmpty()) {
-                resolvedPaths = { path };
+            const QString customSubdir = customPathVar.toString();
+            if (!customSubdir.isEmpty()) {
+                resolvedPaths = expandCustomPath(QDir::homePath(), customSubdir);
+            } else {
+                const QString path = resolveXdgPath(xdgType);
+                if (!path.isEmpty()) {
+                    resolvedPaths = { path };
+                }
             }
         }
         if (resolvedPaths.isEmpty()) {

src/dfm-search/dfm-search-lib/semantic/rules/zh_CN/location_rules.json

@@ -79,7 +79,11 @@
                     "enabled": true,
                     "priority": 180,
                     "metadata": {
-                        "custom_path": "Documents/xwechat_files/*/msg/file",
+                        "custom_path": [
+                            "Documents/xwechat_files/*/msg/file",
+                            "Documents/xwechat_files/*/msg/video"
+                        ],
+                        "exclude_patterns": ["*_thumb.jpg"],
                         "include_hidden": false,
                         "im_received": true
                     }

src/dfm-search/dfm-search-lib/semantic/semanticruleengine.cpp

@@ -178,6 +178,18 @@
     return {};
 }
 
+QVariantMap SemanticRuleEngine::ruleMetadataById(const QString &ruleId) const
+{
+    for (const RuleGroup &group : m_groups) {
+        for (const Rule &rule : group.rules) {
+            if (rule.id == ruleId) {
+                return rule.metadata;
+            }
+        }
+    }
+    return {};
+}
+
 bool SemanticRuleEngine::hasGroup(const QString &group) const
 {
     return m_groups.contains(group);

src/dfm-search/dfm-search-lib/semantic/semanticruleengine.h

@@ -86,6 +86,17 @@ class SemanticRuleEngine : public QObject
      */
     QVariantMap ruleMetadata(const QString &group, const QString &ruleId) const;
 
+    /**
+     * @brief Get a rule's metadata by rule ID, searching across all groups.
+     * @param ruleId The rule ID to look up
+     * @return The metadata map, or empty if not found
+     *
+     * Unlike ruleMetadata(group, ruleId), this searches all groups.
+     * Useful when the caller has a ruleId from consumedSpans but doesn't
+     * know which group it belongs to.
+     */
+    QVariantMap ruleMetadataById(const QString &ruleId) const;
+
     /**
      * @brief Get all rule IDs in a group.
      */

src/dfm-search/dfm-search-lib/semantic/semanticsearcher.cpp

@@ -11,10 +11,14 @@
 #include <dfm-search/dsearch_global.h>
 #include <dfm-search/semantic_types.h>
 
 #include <QDebug>
 #include <QDir>
 #include <QEventLoop>
+#include <QFileInfo>
 #include <QTimer>
+#include <QVariant>
+
+#include <algorithm>
 
 DFM_SEARCH_BEGIN_NS
 
@@ -47,13 +51,13 @@
 {
     int count = 0;
     if (!intent.searchDirectories().isEmpty()) ++count;   // location
-    if (intent.hiddenOnly()) ++count;                     // hiddenOnly 属性
-    if (intent.recentOnly()) ++count;                     // recentOnly 属性
-    if (intent.timeConstraint().isValid()) ++count;       // time 约束
-    if (intent.sizeConstraint().isValid()) ++count;       // size 约束
-    if (!intent.fileExtensions().isEmpty()) ++count;      // fileType
-    if (!intent.keywords().isEmpty()) ++count;            // keyword (unconsumed text)
-    if (hasTargetRuleSpan(intent)) ++count;               // target 名词 (文件/文件名/文件夹)
+    if (intent.hiddenOnly()) ++count;   // hiddenOnly 属性
+    if (intent.recentOnly()) ++count;   // recentOnly 属性
+    if (intent.timeConstraint().isValid()) ++count;   // time 约束
+    if (intent.sizeConstraint().isValid()) ++count;   // size 约束
+    if (!intent.fileExtensions().isEmpty()) ++count;   // fileType
+    if (!intent.keywords().isEmpty()) ++count;   // keyword (unconsumed text)
+    if (hasTargetRuleSpan(intent)) ++count;   // target 名词 (文件/文件名/文件夹)
 
     // 当 specific filetype 与 general filetype 同时命中（如 "PPT文档"、"pdf文档"），
     // general 类型词（文档/表格/幻灯片）充当目标名词，贡献额外维度。
@@ -161,48 +165,22 @@
         dirs = QStringList { QDir::homePath() };
     }
 
-    // Step 5: Set up signal/slot handlers
-    auto onFinished = [this](const SearchResultList &results) {
-        // Collect and deduplicate results from each engine's final result list
-        for (const SearchResult &r : results) {
-            if (!seenPaths.contains(r.path())) {
-                seenPaths.insert(r.path());
-                allResults.append(r);
-            }
-        }
-
-        if (pendingFinishCount.fetch_sub(1) == 1) {
-            // All engines finished
-            timeoutTimer->stop();
-
-            // Truncate final deduplicated results to maxResults
-            if (maxResults > 0 && allResults.size() > maxResults) {
-                allResults = allResults.mid(0, maxResults);
-            }
-
-            if (cancelled.load()) {
-                status.store(SearchStatus::Cancelled);
-                Q_EMIT q->statusChanged(SearchStatus::Cancelled);
-                Q_EMIT q->searchCancelled();
-            } else {
-                status.store(SearchStatus::Finished);
-                Q_EMIT q->statusChanged(SearchStatus::Finished);
-                Q_EMIT q->searchFinished(allResults);
+    excludeNamePatterns.clear();
+    for (const MatchSpan &span : intent.consumedSpans()) {
+        const QVariantMap meta = ruleEngine->ruleMetadataById(span.ruleId());
+        const QVariant ep = meta.value("exclude_patterns");
+        if (ep.userType() == QMetaType::QVariantList) {
+            for (const auto &v : ep.toList()) {
+                const QString p = v.toString();
+                if (!p.isEmpty() && !excludeNamePatterns.contains(p))
+                    excludeNamePatterns.append(p);
             }
         }
-    };
+    }
 
-    // 引擎错误也必须触发计数归零检查，否则上层会一直阻塞到 60s 超时。
-    // 校验失败（如 KeywordIsEmpty）会让 GenericSearchEngine::search 在
-    // emit errorOccurred 后直接 return，不会 emit searchFinished ——
-    // 此时必须由 onError 兜底减计数，把错误视为引擎完成。
-    auto onError = [this, onFinished](const SearchError &error) {
-        qWarning() << "Search error:" << error.message();
-        // Don't propagate individual engine errors to caller
-        // The other engines may still produce valid results
-        // 但仍需减计数，否则 pendingFinishCount 永不到 0
-        onFinished({});
-    };
+    // Step 5: Set up signal/slot handlers
+    auto onFinished = [this](const SearchResultList &r) { onEngineFinished(r); };
+    auto onError = [this](const SearchError &e) { onEngineError(e); };
 
     // Step 6: Clean up any previous search engines
     pendingFinishCount.store(0);
@@ -307,6 +285,48 @@
     engine->search(query);
 }
 
+void SemanticSearcherData::onEngineFinished(const SearchResultList &results)
+{
+    for (const SearchResult &r : results) {
+        if (!excludeNamePatterns.isEmpty()) {
+            const QString fileName = QFileInfo(r.path()).fileName();
+            bool excluded = std::any_of(excludeNamePatterns.cbegin(), excludeNamePatterns.cend(),
+                                        [&fileName](const QString &pattern) {
+                                            return QDir::match(pattern, fileName);
+                                        });
+            if (excluded) continue;
+        }
+        if (!seenPaths.contains(r.path())) {
+            seenPaths.insert(r.path());
+            allResults.append(r);
+        }
+    }
+
+    if (pendingFinishCount.fetch_sub(1) == 1) {
+        timeoutTimer->stop();
+
+        if (maxResults > 0 && allResults.size() > maxResults) {
+            allResults = allResults.mid(0, maxResults);
+        }
+
+        if (cancelled.load()) {
+            status.store(SearchStatus::Cancelled);
+            Q_EMIT q->statusChanged(SearchStatus::Cancelled);
+            Q_EMIT q->searchCancelled();
+        } else {
+            status.store(SearchStatus::Finished);
+            Q_EMIT q->statusChanged(SearchStatus::Finished);
+            Q_EMIT q->searchFinished(allResults);
+        }
+    }
+}
+
+void SemanticSearcherData::onEngineError(const SearchError &error)
+{
+    qWarning() << "Search error:" << error.message();
+    onEngineFinished({});
+}
+
 void SemanticSearcherData::doCancel()
 {
     cancelled.store(true);

src/dfm-search/dfm-search-lib/semantic/semanticsearcher_p.h

@@ -62,6 +62,9 @@ class SemanticSearcherData
                                std::function<void(const SearchResultList &)> onFinished,
                                std::function<void(const SearchError &)> onError);
 
+    void onEngineFinished(const SearchResultList &results);
+    void onEngineError(const SearchError &error);
+
     SemanticSearcher *q = nullptr;
 
     // State
@@ -89,6 +92,7 @@ class SemanticSearcherData
     bool detailedResultsEnabled = false;
     int maxResults = 0;   // 0 = unlimited
     QStringList excludedPaths;   // directories to exclude from search
+    QStringList excludeNamePatterns;   // glob patterns to filter from results (e.g. *_thumb.jpg)
 };
 
 DFM_SEARCH_END_NS