Add configurable maximum length for RAG search documents

2 files changed, 24 insertions, 2 deletions

diff --git a/source/net/yacy/ai/RAGAugmentor.java b/source/net/yacy/ai/RAGAugmentor.java
index 71b285504..a2dde813c 100644
--- a/source/net/yacy/ai/RAGAugmentor.java
+++ b/source/net/yacy/ai/RAGAugmentor.java
@@ -68,11 +68,28 @@ import net.yacy.search.snippet.TextSnippet;
  */
 public final class RAGAugmentor {
 
+    public static final String SEARCH_DOCUMENT_MAX_LENGTH_CONFIG = "ai.rag.search-document-maxlength";
+    public static final int SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT = 30000;
+
     /**
      * Utility class; not instantiable.
      */
     private RAGAugmentor() {}
 
+    public static int configuredSearchDocumentMaxLength() {
+        final Switchboard sb = Switchboard.getSwitchboard();
+        if (sb == null) return SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT;
+        final long configured = sb.getConfigLong(SEARCH_DOCUMENT_MAX_LENGTH_CONFIG, SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT);
+        if (configured <= 0) return SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT;
+        return configured > Integer.MAX_VALUE ? Integer.MAX_VALUE : (int) configured;
+    }
+
+    private static String truncateSearchDocument(final String markdown) {
+        if (markdown == null || markdown.isEmpty()) return "";
+        final int maxLength = configuredSearchDocumentMaxLength();
+        return markdown.length() <= maxLength ? markdown : markdown.substring(0, maxLength);
+    }
+
     /**
      * Executes local index search.
      *
@@ -117,8 +134,9 @@ public final class RAGAugmentor {
             } catch (JSONException e) {}
         }
 
-        ConcurrentLog.info("RAGProxy", "markdownChars=" + sb.length() + " resultCount=" + searchResults.length());
-        return sb.toString();
+        final String markdown = truncateSearchDocument(sb.toString());
+        ConcurrentLog.info("RAGProxy", "markdownChars=" + markdown.length() + " resultCount=" + searchResults.length());
+        return markdown;
     }
 
     /**
diff --git a/source/net/yacy/htroot/RAGConfig_p.java b/source/net/yacy/htroot/RAGConfig_p.java
index 819d9832d..23a5beabe 100644
--- a/source/net/yacy/htroot/RAGConfig_p.java
+++ b/source/net/yacy/htroot/RAGConfig_p.java
@@ -18,10 +18,13 @@ public class RAGConfig_p {
             final String systemPrompt = post.get("ai.system-prompt", "").trim();
             final String userPrefix = post.get("ai.llm-user-prefix", "").trim();
             final String queryPrefix = post.get("ai.llm-query-generator-prefix", "").trim();
+            int searchDocumentMaxLength = post.getInt(net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_CONFIG, net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT);
+            if (searchDocumentMaxLength <= 0) searchDocumentMaxLength = net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT;
 
             sb.setConfig("ai.system-prompt", systemPrompt);
             sb.setConfig("ai.llm-user-prefix", userPrefix);
             sb.setConfig("ai.llm-query-generator-prefix", queryPrefix);
+            sb.setConfig(net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_CONFIG, Integer.toString(searchDocumentMaxLength));
         }
 
         // mark page as visited for gamification/quest tracking
@@ -30,6 +33,7 @@ public class RAGConfig_p {
         prop.put("ai.system-prompt", sb.getConfig("ai.system-prompt", net.yacy.http.servlets.RAGProxyServlet.LLM_SYSTEM_PROMPT_DEFAULT));
         prop.put("ai.llm-user-prefix", sb.getConfig("ai.llm-user-prefix", "\n\nAdditional Information:\n\nbelow you find a collection of texts that might be useful to generate a response. Do not discuss these documents, just use them to answer the question above.\n\n"));
         prop.put("ai.llm-query-generator-prefix", sb.getConfig("ai.llm-query-generator-prefix", "Make a list of search words with low document frequency for the following prompt; use a JSON Array: "));
+        prop.put("ai.rag.search-document-maxlength", sb.getConfig(net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_CONFIG, Integer.toString(net.yacy.ai.RAGAugmentor.SEARCH_DOCUMENT_MAX_LENGTH_DEFAULT)));
 
         return prop;
     }