feat: ✨ add file size limit and improve buffer handling

microsoft · Feb 5, 2025 · 4b671f8 · 4b671f8
1 parent 61123b3
commit 4b671f8
Show file tree

Hide file tree

Showing 5 changed files with 17 additions and 12 deletions.
diff --git a/packages/core/src/bufferlike.ts b/packages/core/src/bufferlike.ts
@@ -18,12 +18,12 @@ export async function resolveBufferLike(
     else if (bufferLike instanceof Uint8Array) return Buffer.from(bufferLike)
     else if (
         typeof bufferLike === "object" &&
-        (bufferLike as WorkspaceFile).content
-    )
+        (bufferLike as WorkspaceFile).filename
+    ) {
         return Buffer.from(
-            (bufferLike as WorkspaceFile).content,
-            (bufferLike as WorkspaceFile).encoding || "utf-8"
+            await resolveFileBytes(bufferLike as WorkspaceFile, options)
         )
+    }
     console.log(bufferLike)
     throw new Error("Unsupported buffer-like object")
 }

diff --git a/packages/core/src/constants.ts b/packages/core/src/constants.ts
@@ -173,7 +173,7 @@ export const MODEL_PROVIDER_OPENAI_HOSTS = Object.freeze([
     MODEL_PROVIDER_OPENAI,
     MODEL_PROVIDER_GITHUB,
     MODEL_PROVIDER_AZURE_OPENAI,
-    MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI
+    MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
 ])
 
 export const TRACE_FILE_PREVIEW_MAX_LENGTH = 240
@@ -344,3 +344,5 @@ export const WS_MAX_FRAME_LENGTH = 1000000
 
 export const SCHEMA_DEFAULT_FORMAT = "json"
 export const THINK_REGEX = /<think>(.*?)($|<\/think>)/gis
+
+export const MAX_FILE_CONTENT_SIZE = 1024 * 1024 * 2 // 2MB
diff --git a/packages/core/src/file.ts b/packages/core/src/file.ts
@@ -21,6 +21,7 @@ import {
     DOCX_MIME_TYPE,
     DOCX_REGEX,
     HTTPS_REGEX,
+    MAX_FILE_CONTENT_SIZE,
     PDF_MIME_TYPE,
     PDF_REGEX,
     XLSX_MIME_TYPE,
@@ -37,9 +38,9 @@ import { tidyData } from "./tidy"
  */
 export async function resolveFileContent(
     file: WorkspaceFile,
-    options?: TraceOptions
+    options?: TraceOptions & { maxFileSize?: number }
 ) {
-    const { trace } = options || {}
+    const { trace, maxFileSize = MAX_FILE_CONTENT_SIZE } = options || {}
 
     // decode known files
     if (file.encoding === "base64") {
@@ -120,7 +121,7 @@ export async function resolveFileContent(
         if (!isBinary) file.content = await readText(filename)
         else {
             const info = await host.statFile(filename)
-            if (info.size < 1000000) {
+            if (!maxFileSize || info.size < maxFileSize) {
                 const bytes: Uint8Array = await host.readFile(filename)
                 file.encoding = "base64"
                 file.content = toBase64(bytes)
@@ -200,8 +201,10 @@ export async function resolveFileBytes(
     options?: TraceOptions
 ): Promise<Uint8Array> {
     if (typeof filename === "object") {
-        if (filename.encoding === "base64" && filename.content)
-            return fromBase64(filename.content)
+        if (filename.encoding && filename.content)
+            return new Uint8Array(
+                Buffer.from(filename.content, filename.encoding)
+            )
         filename = filename.filename
     }
 

diff --git a/packages/sample/genaisrc/defimages.genai.mjs b/packages/sample/genaisrc/defimages.genai.mjs
@@ -2,12 +2,12 @@ script({
     files: "src/robots.jpg",
     model: "vision",
 })
-defImages(env.files, { detail: "low", autoCrop: true })
+defImages(env.files, { detail: "low",})
 $`give keywords describing for each image`
 defImages(env.files, {
     autoCrop: true,
     greyscale: true,
-    maxHeight: 1024,
+    maxHeight: 400,
     rotate: 90,
     scale: 0.9,
     crop: { x: 0, y: 0, w: 400, h: 400 },

diff --git a/packages/sample/src/vision/galaxy.jpg b/packages/sample/src/vision/galaxy.jpg