refactor: extract url parsing logic to utilities

dcs-soni · dcs-soni · commit 7e9055d93a2e · 2025-10-02T14:48:28.000+05:30
diff --git a/core/llm/llms/Anthropic.ts b/core/llm/llms/Anthropic.ts
@@ -28,6 +28,7 @@ import {
 } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
 import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import { extractBase64FromDataUrl } from "../../util/url.js";
 import { DEFAULT_REASONING_TOKENS } from "../constants.js";
 import { BaseLLM } from "../index.js";
 
@@ -107,14 +108,7 @@ class Anthropic extends BaseLLM {
         source: {
           type: "base64",
           media_type: getAnthropicMediaTypeFromDataUrl(part.imageUrl.url),
-          data: (() => {
-            const urlParts = part.imageUrl.url.split(",");
-            if(urlParts.length < 2) {
-              throw new Error("Invalid data URL format: missing comma separator");
-            }
-            const [...base64Parts] = urlParts;
-            return base64Parts.join(",");
-          })(),
+          data: extractBase64FromDataUrl(part.imageUrl.url),
         },
       };
     });
diff --git a/core/llm/llms/Bedrock.ts b/core/llm/llms/Bedrock.ts
@@ -21,6 +21,7 @@ import type { CompletionOptions } from "../../index.js";
 import { ChatMessage, Chunk, LLMOptions, MessageContent } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
 import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import { parseDataUrl } from "../../util/url.js";
 import { BaseLLM } from "../index.js";
 import { PROVIDER_TOOL_SUPPORT } from "../toolSupport.js";
 import { getSecureID } from "../utils/getSecureID.js";
@@ -546,12 +547,7 @@ class Bedrock extends BaseLLM {
           blocks.push({ text: part.text });
         } else if (part.type === "imageUrl" && part.imageUrl) {
           try {
-            const urlParts = part.imageUrl.url.split(",");
-            if(urlParts.length < 2) {
-              throw new Error("Invalid data URL format: missing comma separator");
-            }
-            const [mimeType, ...base64Parts] = urlParts;
-            const base64Data = base64Parts.join(",");
+            const { mimeType, base64Data } = parseDataUrl(part.imageUrl.url);
             const format = mimeType.split("/")[1]?.split(";")[0] || "jpeg";
             if (
               format === ImageFormat.JPEG ||
diff --git a/core/llm/llms/Gemini.ts b/core/llm/llms/Gemini.ts
@@ -11,6 +11,7 @@ import {
 } from "../../index.js";
 import { safeParseToolCallArgs } from "../../tools/parseArgs.js";
 import { renderChatMessage, stripImages } from "../../util/messageContent.js";
+import { extractBase64FromDataUrl } from "../../util/url.js";
 import { BaseLLM } from "../index.js";
 import {
   GeminiChatContent,
@@ -191,14 +192,7 @@ class Gemini extends BaseLLM {
       : {
           inlineData: {
             mimeType: "image/jpeg",
-            data: (() => {
-              const urlParts = part.imageUrl.url.split(",");
-              if (urlParts.length < 2) {
-                throw new Error("Invalid data URL format: missing comma separator");
-              }
-              const [, ...base64Parts] = urlParts;
-              return base64Parts.join(",");
-            })(),
+            data: part.imageUrl?.url ? extractBase64FromDataUrl(part.imageUrl.url) : ""
           },
         };
   }
diff --git a/core/llm/llms/Ollama.ts b/core/llm/llms/Ollama.ts
@@ -13,6 +13,7 @@ import {
 } from "../../index.js";
 import { renderChatMessage } from "../../util/messageContent.js";
 import { getRemoteModelInfo } from "../../util/ollamaHelper.js";
+import { extractBase64FromDataUrl } from "../../util/url.js";
 import { BaseLLM } from "../index.js";
 
 type OllamaChatMessage = {
@@ -303,15 +304,7 @@ class Ollama extends BaseLLM implements ModelInstaller {
       const images: string[] = [];
       message.content.forEach((part) => {
         if (part.type === "imageUrl" && part.imageUrl) {
-          const image = (() => {
-            if (!part.imageUrl?.url) return undefined;
-            const urlParts = part.imageUrl.url.split(",");
-            if (urlParts.length < 2) {
-              throw new Error("Invalid data URL format: missing comma separator");
-            }
-            const [, ...base64Parts] = urlParts;
-            return base64Parts.join(",");
-          })();
+          const image = part.imageUrl?.url ? extractBase64FromDataUrl(part.imageUrl.url) : undefined
           if (image) {
             images.push(image);
           }
diff --git a/core/util/url.ts b/core/util/url.ts
@@ -9,3 +9,34 @@ export function canParseUrl(url: string): boolean {
     return false;
   }
 }
+
+export function parseDataUrl(dataUrl: string): {
+  mimeType: string;
+  base64Data: string
+} {
+  const urlParts = dataUrl.split(",");
+
+  if(urlParts.length < 2 ) {
+    throw new Error("Invalid data URL format: expected 'data:type;base64,data' format")
+  }
+
+  const [mimeType, ...base64Parts] = urlParts;
+  const base64Data = base64Parts.join(",");
+  
+  return { mimeType, base64Data }
+}
+
+export function extractBase64FromDataUrl(dataUrl: string): string {
+  return parseDataUrl(dataUrl).base64Data
+}
+
+export function safeSplit(input: string, delimiter: string, expectedParts: number, errorContext: string = "input") : string[] {
+
+  const parts = input.split(delimiter);
+
+  if(parts.length !== expectedParts) {
+    throw new Error(`Invalid ${errorContext} format: expected ${expectedParts} parts separated by "${delimiter}", got ${parts.length}`)
+  }
+
+  return parts;
+}
diff --git a/packages/openai-adapters/src/apis/Anthropic.ts b/packages/openai-adapters/src/apis/Anthropic.ts
@@ -33,6 +33,7 @@ import {
   usageChatChunk,
 } from "../util.js";
 import { EMPTY_CHAT_COMPLETION } from "../util/emptyChatCompletion.js";
+import { extractBase64FromDataUrl } from "../../../../core/util/url.js"
 import { safeParseArgs } from "../util/parseArgs.js";
 import {
   CACHING_STRATEGIES,
@@ -199,14 +200,7 @@ export class AnthropicApi implements BaseLlmApi {
               source: {
                 type: "base64",
                 media_type: getAnthropicMediaTypeFromDataUrl(dataUrl),
-                data: (() => {
-                  const urlParts = dataUrl.split(",");
-                  if (urlParts.length < 2) {
-                    throw new Error("Invalid data URL format: missing comma separator");
-                  }
-                  const [, ...base64Parts] = urlParts;
-                  return base64Parts.join(",");
-                })(),
+                data: extractBase64FromDataUrl(dataUrl);
               },
             });
           }
diff --git a/packages/openai-adapters/src/apis/Bedrock.ts b/packages/openai-adapters/src/apis/Bedrock.ts
@@ -30,6 +30,7 @@ import {
 } from "openai/resources/index";
 
 import { fromNodeProviderChain } from "@aws-sdk/credential-providers";
+import { parseDataUrl } from "../../../../core/util/url.js";
 import { BedrockConfig } from "../types.js";
 import { chatChunk, chatChunkFromDelta, embedding, rerank } from "../util.js";
 import { safeParseArgs } from "../util/parseArgs.js";
@@ -121,12 +122,7 @@ export class BedrockApi implements BaseLlmApi {
       case "image_url":
       default:
         try {
-          const urlParts = (part as ChatCompletionContentPartImage).image_url.url.split(",");
-          if(urlParts.length < 2) {
-            throw new Error("Invalid data URL format: missing comma separator");
-          }
-          const [mimeType, ...base64Parts] = urlParts;
-          const base64Data = base64Parts.join(",");
+          const { mimeType, base64Data } = parseDataUrl((part as ChatCompletionContentPartImage).image_url.url);
           const format = mimeType.split("/")[1]?.split(";")[0] || "jpeg";
           if (
             format === ImageFormat.JPEG ||
