fix(search): return plain verbose snippets for text search

autotests/dfm-search-tests/tst_content_retriever.cpp

@@ -99,6 +99,7 @@ private Q_SLOTS:
     void fetchPreview_offsetBeyondContent();
     void fetchPreview_unlimitedNoKeyword();
     void previewSnippet_basic();
+    void plainSnippet_basic();
 };
 
 void tst_ContentRetriever::fetchContent_single()
@@ -407,6 +408,32 @@ void tst_ContentRetriever::previewSnippet_basic()
     QCOMPARE(kwOff, 6);
 }
 
+void tst_ContentRetriever::plainSnippet_basic()
+{
+    using DFMSEARCH::ContentHighlighter::PlainSnippetResult;
+    using DFMSEARCH::ContentHighlighter::plainSnippet;
+
+    const QString content = QStringLiteral("0123456789abcdefghijKEYWORDklmnopqrstuvwxyz");
+
+    {
+        const PlainSnippetResult result = plainSnippet({ "keyword" }, content, 20, 20);
+        QCOMPARE(result.content, QString("abcdefghijKEYWORDklm"));
+        QCOMPARE(result.snippetOffset, 10);
+    }
+
+    {
+        const PlainSnippetResult result = plainSnippet({ "keyword", "0123" }, content, 20, 20);
+        QCOMPARE(result.content, QString("0123456789abcdefghij"));
+        QCOMPARE(result.snippetOffset, 0);
+    }
+
+    {
+        const PlainSnippetResult result = plainSnippet({ "missing" }, content, 20, 20);
+        QVERIFY(result.content.isEmpty());
+        QCOMPARE(result.snippetOffset, -1);
+    }
+}
+
 QObject *create_tst_ContentRetriever()
 {
     return new tst_ContentRetriever();

autotests/dfm-search-tests/tst_textsearch_api.cpp

@@ -31,6 +31,7 @@ private Q_SLOTS:
     void textSearchResult_isHidden();
     void textSearchResult_modifyTimestamp();
     void textSearchResult_birthTimestamp();
+    void textSearchResult_plainSnippetAttributes();
 
     // ContentOptionsAPI tests
     void contentOptions_inheritance();
@@ -184,6 +185,20 @@ void tst_TextSearchAPI::textSearchResult_birthTimestamp()
     QCOMPARE(api.birthTimestamp(), 0);
 }
 
+void tst_TextSearchAPI::textSearchResult_plainSnippetAttributes()
+{
+    SearchResult result("/test/path");
+
+    QVERIFY(!result.hasCustomAttribute("plainContentMatch"));
+    QVERIFY(!result.hasCustomAttribute("snippetOffset"));
+
+    result.setCustomAttribute("plainContentMatch", QString("plain snippet"));
+    result.setCustomAttribute("snippetOffset", 42);
+
+    QCOMPARE(result.customAttribute("plainContentMatch").toString(), QString("plain snippet"));
+    QCOMPARE(result.customAttribute("snippetOffset").toInt(), 42);
+}
+
 // ==================== ContentOptionsAPI Tests ====================
 
 void tst_TextSearchAPI::contentOptions_inheritance()

src/dfm-search/dfm-search-client/output/json_output.cpp

@@ -5,14 +5,37 @@
 #include "json_output.h"
 #include <dfm-search/filenamesearchapi.h>
 #include <dfm-search/contentsearchapi.h>
 #include <dfm-search/ocrtextsearchapi.h>
 
 #include <QJsonDocument>
+#include <QRegularExpression>
 #include <iostream>
 
 using namespace dfmsearch;
 using namespace std;
 
+namespace {
+
+QString plainContentMatch(const SearchResult &result)
+{
+    if (result.hasCustomAttribute("plainContentMatch")) {
+        return result.customAttribute("plainContentMatch").toString();
+    }
+
+    QString content = result.customAttribute("highlightedContent").toString();
+    static const QRegularExpression kHtmlTagPattern(QStringLiteral("<[^>]+>"));
+    return content.remove(kHtmlTagPattern);
+}
+
+int snippetOffset(const SearchResult &result)
+{
+    return result.hasCustomAttribute("snippetOffset")
+            ? result.customAttribute("snippetOffset").toInt()
+            : -1;
+}
+
+}   // namespace
+
 void JsonOutput::setSearchContext(const QString &keyword, const QString &searchPath,
                                   SearchType searchType, SearchMethod searchMethod)
 {
@@ -204,7 +227,8 @@
         obj["path"] = result.path();
 
         ContentResultAPI resultAPI(const_cast<SearchResult &>(result));
-        obj["contentMatch"] = resultAPI.highlightedContent();
+        obj["contentMatch"] = plainContentMatch(result);
+        obj["snippetOffset"] = snippetOffset(result);
 
         QString filename = resultAPI.filename();
         if (!filename.isEmpty()) {
@@ -245,7 +269,8 @@
 
         OcrTextResultAPI resultAPI(const_cast<SearchResult &>(result));
 
-        obj["contentMatch"] = resultAPI.highlightedContent();
+        obj["contentMatch"] = plainContentMatch(result);
+        obj["snippetOffset"] = snippetOffset(result);
 
         QString ocrContent = resultAPI.ocrContent();
         if (!ocrContent.isEmpty()) {

src/dfm-search/dfm-search-client/output/text_output.cpp

@@ -4,14 +4,38 @@
 
 #include "text_output.h"
 #include <dfm-search/filenamesearchapi.h>
 #include <dfm-search/contentsearchapi.h>
 #include <dfm-search/ocrtextsearchapi.h>
 
+#include <QRegularExpression>
+
 #include <iostream>
 
 using namespace dfmsearch;
 using namespace std;
 
+namespace {
+
+QString plainContentMatch(const SearchResult &result)
+{
+    if (result.hasCustomAttribute("plainContentMatch")) {
+        return result.customAttribute("plainContentMatch").toString();
+    }
+
+    QString content = result.customAttribute("highlightedContent").toString();
+    static const QRegularExpression kHtmlTagPattern(QStringLiteral("<[^>]+>"));
+    return content.remove(kHtmlTagPattern);
+}
+
+int snippetOffset(const SearchResult &result)
+{
+    return result.hasCustomAttribute("snippetOffset")
+            ? result.customAttribute("snippetOffset").toInt()
+            : -1;
+}
+
+}   // namespace
+
 void TextOutput::setSearchContext(const QString &keyword, const QString &searchPath,
                                   SearchType searchType, SearchMethod searchMethod)
 {
@@ -128,7 +152,8 @@
     } else if (m_searchType == SearchType::Content) {
         ContentResultAPI resultAPI(const_cast<SearchResult &>(result));
 
-        std::cout << "  Content match: " << resultAPI.highlightedContent().toStdString() << std::endl;
+        std::cout << "  Content match: " << plainContentMatch(result).toStdString() << std::endl;
+        std::cout << "  Snippet offset: " << snippetOffset(result) << std::endl;
 
         // 文件名
         QString filename = resultAPI.filename();
@@ -161,7 +186,8 @@
     } else if (m_searchType == SearchType::Ocr) {
         OcrTextResultAPI resultAPI(const_cast<SearchResult &>(result));
 
-        std::cout << "  Content match: " << resultAPI.highlightedContent().toStdString() << std::endl;
+        std::cout << "  Content match: " << plainContentMatch(result).toStdString() << std::endl;
+        std::cout << "  Snippet offset: " << snippetOffset(result) << std::endl;
 
         QString ocrContent = resultAPI.ocrContent();
         if (!ocrContent.isEmpty()) {

src/dfm-search/dfm-search-lib/contentsearch/contentstrategies/indexedstrategy.cpp

@@ -405,14 +405,22 @@ void ContentIndexedStrategy::processSearchResults(const Lucene::IndexSearcherPtr
 
             SearchResult result(QString::fromStdWString(pathField));
             ContentResultAPI resultApi(result);
+            result.setCustomAttribute("plainContentMatch", QString());
+            result.setCustomAttribute("snippetOffset", -1);
 
             if (enableRetrieval) {
                 try {
                     Lucene::String contentField = doc->get(LuceneFieldNames::Content::kContents);
                     if (!contentField.empty()) {
                         const QString content = QString::fromStdWString(contentField);
+                        // Fix: keep a dedicated plain-text snippet for CLI verbose output
+                        // instead of reusing HTML-oriented highlightedContent.
+                        const ContentHighlighter::PlainSnippetResult plainSnippet = ContentHighlighter::plainSnippet(
+                                m_keywords, content, previewLen, previewLen);
                         const QString highlightedContent = ContentHighlighter::customHighlight(m_keywords, content, previewLen, enableHTML);
                         resultApi.setHighlightedContent(highlightedContent);
+                        result.setCustomAttribute("plainContentMatch", plainSnippet.content);
+                        result.setCustomAttribute("snippetOffset", plainSnippet.snippetOffset);
                     }
                 } catch (const Lucene::LuceneException &e) {
                     qWarning() << "Exception retrieving content field:" << QString::fromStdWString(e.getError());

src/dfm-search/dfm-search-lib/ocrtextsearch/ocrtextstrategies/indexedstrategy.cpp

@@ -404,19 +404,27 @@ void OcrTextIndexedStrategy::processSearchResults(const Lucene::IndexSearcherPtr
 
             // 设置 OCR 内容结果
             OcrTextResultAPI resultApi(result);
+            result.setCustomAttribute("plainContentMatch", QString());
+            result.setCustomAttribute("snippetOffset", -1);
 
             // 使用ContentHighlighter命名空间进行高亮
             if (enableRetrieval) {
                 try {
                     Lucene::String ocrContentField = doc->get(LuceneFieldNames::OcrText::kOcrContents);
                     if (!ocrContentField.empty()) {
                         const QString content = QString::fromStdWString(ocrContentField);
+                        // Fix: keep a dedicated plain-text snippet for CLI verbose output
+                        // instead of reusing HTML-oriented highlightedContent.
+                        const ContentHighlighter::PlainSnippetResult plainSnippet = ContentHighlighter::plainSnippet(
+                                m_keywords, content, previewLen, previewLen);
                         // 设置原始 OCR 内容
                         resultApi.setOcrContent(content);
                         // 设置高亮内容
                         const QString highlightedContent = ContentHighlighter::customHighlight(
                                 m_keywords, content, previewLen, enableHTML);
                         resultApi.setHighlightedContent(highlightedContent);
+                        result.setCustomAttribute("plainContentMatch", plainSnippet.content);
+                        result.setCustomAttribute("snippetOffset", plainSnippet.snippetOffset);
                     }
                 } catch (const Lucene::LuceneException &e) {
                     qWarning() << "Exception retrieving OCR content field:" << QString::fromStdWString(e.getError());

src/dfm-search/dfm-search-lib/utils/contenthighlighter.cpp

@@ -29,6 +29,15 @@ struct KeywordMatch
     QString keyword;
 };
 
+int positioningBeforeLength(int maxLength, int positioningMaxLength)
+{
+    const int effectivePositioningLength = qMax(30, positioningMaxLength);
+    if (maxLength > 0) {
+        return qMin(effectivePositioningLength / 2, maxLength / 2);
+    }
+    return effectivePositioningLength / 2;
+}
+
 /**
  * @brief 检查给定位置是否是段落的开头
  * @param content 原始文本内容
@@ -153,6 +162,24 @@ KeywordMatch findFirstKeywordMatch(const QString &content, const QStringList &ke
     }
     return match;
 }
+
+KeywordMatch findEarliestKeywordMatch(const QString &content, const QStringList &keywords)
+{
+    KeywordMatch match = { -1, 0, QString() };
+    for (const QString &keyword : keywords) {
+        if (keyword.isEmpty()) {
+            continue;
+        }
+
+        const int pos = content.indexOf(keyword, 0, Qt::CaseInsensitive);
+        if (pos != -1 && (match.position == -1 || pos < match.position)) {
+            match.position = pos;
+            match.length = keyword.length();
+            match.keyword = keyword;
+        }
+    }
+    return match;
+}
 }   // namespace
 
 QString customHighlight(const QStringList &keywords, const QString &content, int maxLength, bool enableHtml, int positioningMaxLength)
@@ -242,6 +269,33 @@ QString customHighlight(const QStringList &keywords, const QString &content, int
     return resultSnippet;
 }
 
+PlainSnippetResult plainSnippet(const QStringList &keywords, const QString &content,
+                                int maxLength, int positioningMaxLength)
+{
+    PlainSnippetResult result;
+    if (content.isEmpty() || keywords.isEmpty()) {
+        return result;
+    }
+
+    const KeywordMatch match = findEarliestKeywordMatch(content, keywords);
+    if (match.position == -1) {
+        return result;
+    }
+
+    int snippetStart = qMax(0, match.position - positioningBeforeLength(maxLength, positioningMaxLength));
+    if (maxLength > 0 && content.length() - snippetStart < maxLength) {
+        snippetStart = qMax(0, content.length() - maxLength);
+    }
+
+    const int snippetLength = maxLength > 0
+            ? qMin(maxLength, content.length() - snippetStart)
+            : content.length() - snippetStart;
+
+    result.content = content.mid(snippetStart, snippetLength);
+    result.snippetOffset = snippetStart;
+    return result;
+}
+
 QString previewSnippet(const QString &content, int offset, int maxLength,
                        const QString &keyword, int *keywordOffset)
 {

src/dfm-search/dfm-search-lib/utils/contenthighlighter.h

@@ -15,6 +15,12 @@ DFM_SEARCH_BEGIN_NS
 
 namespace ContentHighlighter {
 
+struct PlainSnippetResult
+{
+    QString content;
+    int snippetOffset = -1;
+};
+
 /**
  * @brief Extracts and highlights the most relevant text snippet from content based on the provided keywords.
  *
@@ -40,6 +46,23 @@ namespace ContentHighlighter {
  */
 QString customHighlight(const QStringList &keywords, const QString &content, int maxLength, bool enableHtml, int positioningMaxLength = 30);
 
+/**
+ * @brief Extracts a plain-text snippet for CLI verbose output.
+ *
+ * Unlike customHighlight(), this helper preserves the original content bytes
+ * in the returned snippet: it does not simplify whitespace, insert HTML tags,
+ * or prepend ellipsis. The snippet start is returned so callers can expose the
+ * offset in the original full text.
+ *
+ * @param keywords Search keywords used to anchor the snippet.
+ * @param content Original document content.
+ * @param maxLength Maximum length of the returned snippet. <= 0 means until EOF.
+ * @param positioningMaxLength Window used to position the first content match.
+ * @return PlainSnippetResult with snippet text and snippetOffset, or empty/-1 if no content match exists.
+ */
+PlainSnippetResult plainSnippet(const QStringList &keywords, const QString &content,
+                                int maxLength, int positioningMaxLength = 30);
+
 /**
  * @brief 精确预览：从 offset 位置截取内容，或从 offset 搜索 keyword 后从匹配位置截取
  *