postprocessing data2summary prompt (#436)

* postprocessing data2summary prompt Signed-off-by: hutiechuan <[email protected]> * adding unit tests for function postprocessing Signed-off-by: hutiechuan <[email protected]> * modify for camel cases Signed-off-by: hutiechuan <[email protected]> * updating the function name Signed-off-by: hutiechuan <[email protected]> * add changelog Signed-off-by: hutiechuan <[email protected]> * fixing unit test issues Signed-off-by: hutiechuan <[email protected]> * fixing unit test Signed-off-by: hutiechuan <[email protected]> * adding more unit test Signed-off-by: hutiechuan <[email protected]> * fixing unit test issues Signed-off-by: hutiechuan <[email protected]> * clean up the code Signed-off-by: hutiechuan <[email protected]> * clean up code and replace regex Signed-off-by: hutiechuan <[email protected]> --------- Signed-off-by: hutiechuan <[email protected]> Signed-off-by: Hailong Cui <[email protected]> Co-authored-by: hutiechuan <[email protected]> Co-authored-by: Hailong Cui <[email protected]>
opensearch-project · Feb 20, 2025 · 3c73361 · 3c73361
1 parent 2819352
commit 3c73361
Show file tree

Hide file tree

Showing 3 changed files with 92 additions and 1 deletion.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,8 @@ Inspired from [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 - expose chatEnabled flag to capabilities ([#398](https://github.com/opensearch-project/dashboards-assistant/pull/398))
 - update chatbot UI to align with new look ([#435](https://github.com/opensearch-project/dashboards-assistant/pull/435))
 - add flag to control if display conversation list ([#438](https://github.com/opensearch-project/dashboards-assistant/pull/438))
+- add data to summary response post processing ([#436](https://github.com/opensearch-project/dashboards-assistant/pull/436))
+
 
 ### Enhancements
 
@@ -20,6 +22,7 @@ Inspired from [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 - Improve alert summary with backend log pattern experience ([#389](https://github.com/opensearch-project/dashboards-assistant/pull/389))
 - fixed in context feature returning 500 error if workspace is invalid to returning 4XX [#429](https://github.com/opensearch-project/dashboards-assistant/pull/429)
 
+
 ### Infrastructure
 
 ### Documentation

diff --git a/server/routes/summary_routes.test.ts b/server/routes/summary_routes.test.ts
@@ -13,6 +13,8 @@ import { SUMMARY_ASSISTANT_API } from '../../common/constants/llm';
 import { registerData2SummaryRoutes, registerSummaryAssistantRoutes } from './summary_routes';
 import { AssistantClient } from '../services/assistant_client';
 import { RequestHandlerContext } from '../../../../src/core/server';
+import { postProcessing } from './summary_routes';
+
 import * as AgentHelpers from './get_agent';
 const mockedLogger = loggerMock.create();
 
@@ -326,4 +328,75 @@ describe('test summary route', () => {
       }
     `);
   });
+
+  describe('postprocessing', () => {
+    it('returns combined summarization and final insights when all tags exist', () => {
+      const input = `
+        Some text <summarization>This is the summary   </summarization> 
+        random info <final insights>   These are the insights  </final insights>
+        extra text
+      `;
+      const output = postProcessing(input);
+
+      expect(output).toEqual(`This is the summary\nThese are the insights`);
+    });
+
+    it('returns original output if <summarization> tag is missing', () => {
+      const input = `Hello world <final insights>Insights here</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(input);
+    });
+
+    it('returns original output if </summarization> closing tag is missing', () => {
+      const input = `<summarization>Partial Summarization <final insights>Insights</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(input);
+    });
+
+    it('returns original output if <final insights> or </final insights> tags are missing', () => {
+      const input = `<summarization>Summary only</summarization>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(input);
+    });
+
+    it('handles empty summarization or empty insights gracefully', () => {
+      const inputWithEmptySummary = `<summarization></summarization><final insights>Some insights</final insights>`;
+      const outputWithEmptySummary = postProcessing(inputWithEmptySummary);
+      expect(outputWithEmptySummary).toEqual(`\nSome insights`);
+
+      const inputWithEmptyInsights = `<summarization>Some summary</summarization><final insights></final insights>`;
+      const outputWithEmptyInsights = postProcessing(inputWithEmptyInsights);
+      expect(outputWithEmptyInsights).toEqual(`Some summary\n`);
+    });
+
+    it('returns original output if none of the special tags are found', () => {
+      const input = `Just a normal string without any special tags.`;
+      const output = postProcessing(input);
+      expect(output).toEqual(input);
+    });
+
+    it('handles malformed or mixed case tags by returning the original string', () => {
+      const input = `<Summarization>Summary</summarization><FINAL INSIGHTS>Insights</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(input);
+    });
+
+    it('handles special characters in content', () => {
+      const input = `<summarization>Summary with <>&"'</summarization><final insights>Insights with <>& "'</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(`Summary with <>&"'\nInsights with <>& "'`);
+    });
+
+    it('handles unicode characters correctly', () => {
+      const input = `<summarization>Summary with emoji 😊</summarization><final insights>Insights with unicode 你好</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(`Summary with emoji 😊\nInsights with unicode 你好`);
+    });
+
+    it('handles invisible characters', () => {
+      const input = `<summarization>Sum\u200Bmary</summarization><final insights>In\u200Bsights</final insights>`;
+      const output = postProcessing(input);
+      expect(output).toEqual(`Sum\u200Bmary\nIn\u200Bsights`);
+    });
+  });
 });
diff --git a/server/routes/summary_routes.ts b/server/routes/summary_routes.ts
@@ -16,6 +16,17 @@ const SUMMARY_AGENT_CONFIG_ID = 'os_summary';
 const LOG_PATTERN_SUMMARY_AGENT_CONFIG_ID = 'os_summary_with_log_pattern';
 const DATA2SUMMARY_AGENT_CONFIG_ID = 'os_data2summary';
 
+export function postProcessing(output: string) {
+  const pattern = /<summarization>(.*?)<\/summarization>.*?<final insights>(.*?)<\/final insights>/s;
+  const match = output.match(pattern);
+  if (match) {
+    const [, summarization, finalInsights] = match;
+    const processedOutput = `${summarization.trim()}\n${finalInsights.trim()}`;
+    return processedOutput;
+  }
+  return output;
+}
+
 export function registerSummaryAssistantRoutes(
   router: IRouter,
   assistantService: AssistantServiceSetup
@@ -96,6 +107,7 @@ export function registerSummaryAssistantRoutes(
         }),
       },
     },
+
     router.handleLegacyErrors(async (context, req, res) => {
       try {
         const client = await getOpenSearchClientTransport({
@@ -190,7 +202,10 @@ export function registerData2SummaryRoutes(
           }
         );
 
-        const result = response.body.inference_results[0].output[0].result;
+        let result = response.body.inference_results[0].output[0].result;
+
+        result = postProcessing(result);
+
         if (result) {
           return res.ok({ body: result });
         } else {