feat(confluence): return page content in get page version tool (#3344)

* feat(confluence): return page content in get page version tool * lint
2026-09-24 15:45:35 +08:00 · 2026-02-25 13:45:19 -08:00
parent 16f337f6fd
commit f625482bcb
4 changed files with 110 additions and 26 deletions
@@ -326,6 +326,8 @@ Get details about a specific version of a Confluence page.
 | --------- | ---- | ----------- |
 | `ts` | string | ISO 8601 timestamp of the operation |
 | `pageId` | string | ID of the page |
+| `title` | string | Page title at this version |
+| `content` | string | Page content with HTML tags stripped at this version |
 | `version` | object | Detailed version information |
 |   ↳ `number` | number | Version number |
 |   ↳ `message` | string | Version message |
@@ -336,6 +338,9 @@ Get details about a specific version of a Confluence page.
 |   ↳ `collaborators` | array | List of collaborator account IDs for this version |
 |   ↳ `prevVersion` | number | Previous version number |
 |   ↳ `nextVersion` | number | Next version number |
+| `body` | object | Raw page body content in storage format at this version |
+|   ↳ `value` | string | The content value in the specified format |
+|   ↳ `representation` | string | Content representation type |

 ### `confluence_list_page_properties`

@@ -2,7 +2,7 @@ import { createLogger } from '@sim/logger'
 import { type NextRequest, NextResponse } from 'next/server'
 import { checkSessionOrInternalAuth } from '@/lib/auth/hybrid'
 import { validateAlphanumericId, validateJiraCloudId } from '@/lib/core/security/input-validation'
-import { getConfluenceCloudId } from '@/tools/confluence/utils'
+import { cleanHtmlContent, getConfluenceCloudId } from '@/tools/confluence/utils'

 const logger = createLogger('ConfluencePageVersionsAPI')

@@ -55,42 +55,79 @@ export async function POST(request: NextRequest) {
      return NextResponse.json({ error: cloudIdValidation.error }, { status: 400 })
    }

-    // If versionNumber is provided, get specific version
+    // If versionNumber is provided, get specific version with page content
    if (versionNumber !== undefined && versionNumber !== null) {
-      const url = `https://api.atlassian.com/ex/confluence/${cloudId}/wiki/api/v2/pages/${pageId}/versions/${versionNumber}`
+      const versionUrl = `https://api.atlassian.com/ex/confluence/${cloudId}/wiki/api/v2/pages/${pageId}/versions/${versionNumber}`
+      const pageUrl = `https://api.atlassian.com/ex/confluence/${cloudId}/wiki/api/v2/pages/${pageId}?version=${versionNumber}&body-format=storage`

      logger.info(`Fetching version ${versionNumber} for page ${pageId}`)

-      const response = await fetch(url, {
-        method: 'GET',
-        headers: {
-          Accept: 'application/json',
-          Authorization: `Bearer ${accessToken}`,
-        },
-      })
+      const [versionResponse, pageResponse] = await Promise.all([
+        fetch(versionUrl, {
+          method: 'GET',
+          headers: {
+            Accept: 'application/json',
+            Authorization: `Bearer ${accessToken}`,
+          },
+        }),
+        fetch(pageUrl, {
+          method: 'GET',
+          headers: {
+            Accept: 'application/json',
+            Authorization: `Bearer ${accessToken}`,
+          },
+        }),
+      ])

-      if (!response.ok) {
-        const errorData = await response.json().catch(() => null)
+      if (!versionResponse.ok) {
+        const errorData = await versionResponse.json().catch(() => null)
        logger.error('Confluence API error response:', {
-          status: response.status,
-          statusText: response.statusText,
+          status: versionResponse.status,
+          statusText: versionResponse.statusText,
          error: JSON.stringify(errorData, null, 2),
        })
-        const errorMessage = errorData?.message || `Failed to get page version (${response.status})`
-        return NextResponse.json({ error: errorMessage }, { status: response.status })
+        const errorMessage =
+          errorData?.message || `Failed to get page version (${versionResponse.status})`
+        return NextResponse.json({ error: errorMessage }, { status: versionResponse.status })
      }

-      const data = await response.json()
+      const versionData = await versionResponse.json()
+
+      let title: string | null = null
+      let content: string | null = null
+      let body: Record<string, unknown> | null = null
+
+      if (pageResponse.ok) {
+        const pageData = await pageResponse.json()
+        title = pageData.title ?? null
+        body = pageData.body ?? null
+
+        const rawContent =
+          pageData.body?.storage?.value ||
+          pageData.body?.view?.value ||
+          pageData.body?.atlas_doc_format?.value ||
+          ''
+        if (rawContent) {
+          content = cleanHtmlContent(rawContent)
+        }
+      } else {
+        logger.warn(
+          `Could not fetch page content for version ${versionNumber}: ${pageResponse.status}`
+        )
+      }

      return NextResponse.json({
        version: {
-          number: data.number,
-          message: data.message ?? null,
-          minorEdit: data.minorEdit ?? false,
-          authorId: data.authorId ?? null,
-          createdAt: data.createdAt ?? null,
+          number: versionData.number,
+          message: versionData.message ?? null,
+          minorEdit: versionData.minorEdit ?? false,
+          authorId: versionData.authorId ?? null,
+          createdAt: versionData.createdAt ?? null,
        },
        pageId,
+        title,
+        content,
+        body,
      })
    }
    // List all versions
@@ -1,4 +1,8 @@
-import { DETAILED_VERSION_OUTPUT_PROPERTIES, TIMESTAMP_OUTPUT } from '@/tools/confluence/types'
+import {
+  BODY_FORMAT_PROPERTIES,
+  DETAILED_VERSION_OUTPUT_PROPERTIES,
+  TIMESTAMP_OUTPUT,
+} from '@/tools/confluence/types'
 import type { ToolConfig } from '@/tools/types'

 export interface ConfluenceGetPageVersionParams {
@@ -14,6 +18,8 @@ export interface ConfluenceGetPageVersionResponse {
  output: {
    ts: string
    pageId: string
+    title: string | null
+    content: string | null
    version: {
      number: number
      message: string | null
@@ -25,6 +31,12 @@ export interface ConfluenceGetPageVersionResponse {
      prevVersion: number | null
      nextVersion: number | null
    }
+    body: {
+      storage?: {
+        value: string
+        representation: string
+      }
+    } | null
  }
 }

@@ -100,6 +112,8 @@ export const confluenceGetPageVersionTool: ToolConfig<
      output: {
        ts: new Date().toISOString(),
        pageId: data.pageId ?? '',
+        title: data.title ?? null,
+        content: data.content ?? null,
        version: data.version ?? {
          number: 0,
          message: null,
@@ -107,6 +121,7 @@ export const confluenceGetPageVersionTool: ToolConfig<
          authorId: null,
          createdAt: null,
        },
+        body: data.body ?? null,
      },
    }
  },
@@ -114,10 +129,29 @@ export const confluenceGetPageVersionTool: ToolConfig<
  outputs: {
    ts: TIMESTAMP_OUTPUT,
    pageId: { type: 'string', description: 'ID of the page' },
+    title: { type: 'string', description: 'Page title at this version', optional: true },
+    content: {
+      type: 'string',
+      description: 'Page content with HTML tags stripped at this version',
+      optional: true,
+    },
    version: {
      type: 'object',
      description: 'Detailed version information',
      properties: DETAILED_VERSION_OUTPUT_PROPERTIES,
    },
+    body: {
+      type: 'object',
+      description: 'Raw page body content in storage format at this version',
+      properties: {
+        storage: {
+          type: 'object',
+          description: 'Body in storage format (Confluence markup)',
+          properties: BODY_FORMAT_PROPERTIES,
+          optional: true,
+        },
+      },
+      optional: true,
+    },
  },
 }
@@ -56,13 +56,21 @@ function stripHtmlTags(html: string): string {
  return text.trim()
 }

+/**
+ * Strips HTML tags and decodes HTML entities from raw Confluence content.
+ */
+export function cleanHtmlContent(rawContent: string): string {
+  let content = stripHtmlTags(rawContent)
+  content = decodeHtmlEntities(content)
+  content = content.replace(/\s+/g, ' ').trim()
+  return content
+}
+
 export function transformPageData(data: any) {
  const rawContent =
    data.body?.storage?.value || data.body?.view?.value || data.body?.atlas_doc_format?.value || ''

-  let cleanContent = stripHtmlTags(rawContent)
-  cleanContent = decodeHtmlEntities(cleanContent)
-  cleanContent = cleanContent.replace(/\s+/g, ' ').trim()
+  const cleanContent = cleanHtmlContent(rawContent)

  return {
    success: true,