fix(s3): support get-object region override and robust S3 URL parsing

Author: waleedlatif1

apps/sim/blocks/blocks/s3.ts

@@ -58,6 +58,16 @@ export const S3Block: BlockConfig<S3Response> = {
       },
       required: true,
     },
+    {
+      id: 'getObjectRegion',
+      title: 'AWS Region',
+      type: 'short-input',
+      placeholder: 'Used when S3 URL does not include region',
+      condition: {
+        field: 'operation',
+        value: ['get_object'],
+      },
+    },
     {
       id: 'bucketName',
       title: 'Bucket Name',
@@ -291,34 +301,11 @@ export const S3Block: BlockConfig<S3Response> = {
             if (!params.s3Uri) {
               throw new Error('S3 Object URL is required')
             }
-
-            // Parse S3 URI for get_object
-            try {
-              const url = new URL(params.s3Uri)
-              const hostname = url.hostname
-              const bucketName = hostname.split('.')[0]
-              const regionMatch = hostname.match(/s3[.-]([^.]+)\.amazonaws\.com/)
-              const region = regionMatch ? regionMatch[1] : params.region
-              const objectKey = url.pathname.startsWith('/')
-                ? url.pathname.substring(1)
-                : url.pathname
-
-              if (!bucketName || !objectKey) {
-                throw new Error('Could not parse S3 URL')
-              }
-
-              return {
-                accessKeyId: params.accessKeyId,
-                secretAccessKey: params.secretAccessKey,
-                region,
-                bucketName,
-                objectKey,
-                s3Uri: params.s3Uri,
-              }
-            } catch (_error) {
-              throw new Error(
-                'Invalid S3 Object URL format. Expected: https://bucket-name.s3.region.amazonaws.com/path/to/file'
-              )
+            return {
+              accessKeyId: params.accessKeyId,
+              secretAccessKey: params.secretAccessKey,
+              region: params.getObjectRegion || params.region,
+              s3Uri: params.s3Uri,
             }
           }
 
@@ -401,6 +388,7 @@ export const S3Block: BlockConfig<S3Response> = {
     acl: { type: 'string', description: 'Access control list' },
     // Download inputs
     s3Uri: { type: 'string', description: 'S3 object URL' },
+    getObjectRegion: { type: 'string', description: 'Optional AWS region override for downloads' },
     // List inputs
     prefix: { type: 'string', description: 'Prefix filter' },
     maxKeys: { type: 'number', description: 'Maximum results' },

apps/sim/tools/s3/get_object.ts

@@ -26,6 +26,13 @@ export const s3GetObjectTool: ToolConfig = {
       visibility: 'user-only',
       description: 'Your AWS Secret Access Key',
     },
+    region: {
+      type: 'string',
+      required: false,
+      visibility: 'user-only',
+      description:
+        'Optional region override when URL does not include region (e.g., us-east-1, eu-west-1)',
+    },
     s3Uri: {
       type: 'string',
       required: true,
@@ -37,7 +44,7 @@ export const s3GetObjectTool: ToolConfig = {
   request: {
     url: (params) => {
       try {
-        const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri)
+        const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri, params.region)
 
         params.bucketName = bucketName
         params.region = region
@@ -46,7 +53,7 @@ export const s3GetObjectTool: ToolConfig = {
         return `https://${bucketName}.s3.${region}.amazonaws.com/${encodeS3PathComponent(objectKey)}`
       } catch (_error) {
         throw new Error(
-          'Invalid S3 Object URL format. Expected format: https://bucket-name.s3.region.amazonaws.com/path/to/file'
+          'Invalid S3 Object URL. Use a valid S3 URL and optionally provide region if the URL omits it.'
         )
       }
     },
@@ -55,7 +62,7 @@ export const s3GetObjectTool: ToolConfig = {
       try {
         // Parse S3 URI if not already parsed
         if (!params.bucketName || !params.region || !params.objectKey) {
-          const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri)
+          const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri, params.region)
           params.bucketName = bucketName
           params.region = region
           params.objectKey = objectKey
@@ -102,7 +109,7 @@ export const s3GetObjectTool: ToolConfig = {
   transformResponse: async (response: Response, params) => {
     // Parse S3 URI if not already parsed
     if (!params.bucketName || !params.region || !params.objectKey) {
-      const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri)
+      const { bucketName, region, objectKey } = parseS3Uri(params.s3Uri, params.region)
       params.bucketName = bucketName
       params.region = region
       params.objectKey = objectKey

apps/sim/tools/s3/utils.ts

@@ -20,18 +20,61 @@ export function getSignatureKey(
   return kSigning
 }
 
-export function parseS3Uri(s3Uri: string): {
+export function parseS3Uri(
+  s3Uri: string,
+  fallbackRegion?: string
+): {
   bucketName: string
   region: string
   objectKey: string
 } {
   try {
     const url = new URL(s3Uri)
     const hostname = url.hostname
-    const bucketName = hostname.split('.')[0]
-    const regionMatch = hostname.match(/s3[.-]([^.]+)\.amazonaws\.com/)
-    const region = regionMatch ? regionMatch[1] : 'us-east-1'
-    const objectKey = url.pathname.startsWith('/') ? url.pathname.substring(1) : url.pathname
+    const normalizedPath = url.pathname.startsWith('/') ? url.pathname.slice(1) : url.pathname
+    const pathSegments = normalizedPath.split('/').filter(Boolean)
+
+    const virtualHostedDualstackMatch = hostname.match(
+      /^(.+)\.s3\.dualstack\.([^.]+)\.amazonaws\.com(?:\.cn)?$/
+    )
+    const virtualHostedRegionalMatch = hostname.match(
+      /^(.+)\.s3[.-]([^.]+)\.amazonaws\.com(?:\.cn)?$/
+    )
+    const virtualHostedGlobalMatch = hostname.match(/^(.+)\.s3\.amazonaws\.com(?:\.cn)?$/)
+
+    const pathStyleDualstackMatch = hostname.match(
+      /^s3\.dualstack\.([^.]+)\.amazonaws\.com(?:\.cn)?$/
+    )
+    const pathStyleRegionalMatch = hostname.match(/^s3[.-]([^.]+)\.amazonaws\.com(?:\.cn)?$/)
+    const pathStyleGlobalMatch = hostname.match(/^s3\.amazonaws\.com(?:\.cn)?$/)
+
+    const isPathStyleHost = Boolean(
+      pathStyleDualstackMatch || pathStyleRegionalMatch || pathStyleGlobalMatch
+    )
+
+    const bucketName = isPathStyleHost
+      ? pathSegments[0]
+      : (virtualHostedDualstackMatch?.[1] ??
+        virtualHostedRegionalMatch?.[1] ??
+        virtualHostedGlobalMatch?.[1] ??
+        '')
+
+    const rawObjectKey = isPathStyleHost ? pathSegments.slice(1).join('/') : normalizedPath
+    const objectKey = (() => {
+      try {
+        return decodeURIComponent(rawObjectKey)
+      } catch {
+        return rawObjectKey
+      }
+    })()
+
+    const region =
+      virtualHostedDualstackMatch?.[2] ??
+      virtualHostedRegionalMatch?.[2] ??
+      pathStyleDualstackMatch?.[1] ??
+      pathStyleRegionalMatch?.[1] ??
+      fallbackRegion ??
+      'us-east-1'
 
     if (!bucketName || !objectKey) {
       throw new Error('Invalid S3 URI format')
@@ -40,7 +83,7 @@ export function parseS3Uri(s3Uri: string): {
     return { bucketName, region, objectKey }
   } catch (_error) {
     throw new Error(
-      'Invalid S3 Object URL format. Expected format: https://bucket-name.s3.region.amazonaws.com/path/to/file'
+      'Invalid S3 Object URL format. Expected S3 virtual-hosted or path-style URL with object key.'
     )
   }
 }