refactor: rename HtmlSmartProcessorMiddleware to HtmlSelectProcessorMiddleware

Author: arabold

…nts/HtmlSmartProcessorMiddleware.test.ts …ts/HtmlSelectProcessorMiddleware.test.tssrc/scraper/middleware/components/HtmlSmartProcessorMiddleware.test.ts renamed to src/scraper/middleware/components/HtmlSelectProcessorMiddleware.test.ts src/scraper/middleware/components/HtmlSmartProcessorMiddleware.test.ts renamed to src/scraper/middleware/components/HtmlSelectProcessorMiddleware.test.ts

@@ -2,7 +2,7 @@ import { beforeEach, describe, expect, it, vi } from "vitest";
 import type { ContentProcessingContext } from "../types";
 import { HtmlDomParserMiddleware } from "./HtmlDomParserMiddleware";
 import { HtmlPlaywrightMiddleware } from "./HtmlPlaywrightMiddleware";
-import { HtmlSmartProcessorMiddleware } from "./HtmlSmartProcessorMiddleware";
+import { HtmlSelectProcessorMiddleware } from "./HtmlSelectProcessorMiddleware";
 
 // Mock the underlying processors
 vi.mock("./HtmlDomParserMiddleware");
@@ -12,7 +12,7 @@ vi.mock("./HtmlPlaywrightMiddleware");
 const MockedHtmlDomParserMiddleware = vi.mocked(HtmlDomParserMiddleware);
 const MockedHtmlPlaywrightMiddleware = vi.mocked(HtmlPlaywrightMiddleware);
 
-describe("HtmlSmartProcessorMiddleware", () => {
+describe("HtmlSelectProcessorMiddleware", () => {
   const mockNext = vi.fn().mockResolvedValue(undefined);
 
   beforeEach(() => {
@@ -45,7 +45,7 @@ describe("HtmlSmartProcessorMiddleware", () => {
 
   it("should use HtmlDomParserMiddleware when scrapeMode is 'fetch'", async () => {
     const context = createContext("fetch");
-    const middleware = new HtmlSmartProcessorMiddleware();
+    const middleware = new HtmlSelectProcessorMiddleware();
 
     await middleware.process(context, mockNext);
 
@@ -65,7 +65,7 @@ describe("HtmlSmartProcessorMiddleware", () => {
 
   it("should use HtmlPlaywrightMiddleware when scrapeMode is 'playwright'", async () => {
     const context = createContext("playwright");
-    const middleware = new HtmlSmartProcessorMiddleware();
+    const middleware = new HtmlSelectProcessorMiddleware();
 
     await middleware.process(context, mockNext);
 
@@ -85,7 +85,7 @@ describe("HtmlSmartProcessorMiddleware", () => {
 
   it("should use HtmlPlaywrightMiddleware when scrapeMode is 'auto'", async () => {
     const context = createContext("auto");
-    const middleware = new HtmlSmartProcessorMiddleware();
+    const middleware = new HtmlSelectProcessorMiddleware();
 
     await middleware.process(context, mockNext);
 
@@ -105,7 +105,7 @@ describe("HtmlSmartProcessorMiddleware", () => {
 
   it("should default to 'auto' (Playwright) when scrapeMode is undefined", async () => {
     const context = createContext(undefined); // Explicitly undefined
-    const middleware = new HtmlSmartProcessorMiddleware();
+    const middleware = new HtmlSelectProcessorMiddleware();
 
     await middleware.process(context, mockNext);
 

…mponents/HtmlSmartProcessorMiddleware.ts …ponents/HtmlSelectProcessorMiddleware.tssrc/scraper/middleware/components/HtmlSmartProcessorMiddleware.ts renamed to src/scraper/middleware/components/HtmlSelectProcessorMiddleware.ts src/scraper/middleware/components/HtmlSmartProcessorMiddleware.ts renamed to src/scraper/middleware/components/HtmlSelectProcessorMiddleware.ts

@@ -11,7 +11,7 @@ import { HtmlPlaywrightMiddleware } from "./HtmlPlaywrightMiddleware";
  * - 'auto': Currently defaults to using `HtmlPlaywrightMiddleware`. Future implementations
  *   might add more sophisticated logic for 'auto' mode.
  */
-export class HtmlSmartProcessorMiddleware implements ContentProcessorMiddleware {
+export class HtmlSelectProcessorMiddleware implements ContentProcessorMiddleware {
   private readonly domProcessor: HtmlDomParserMiddleware;
   private readonly playwrightProcessor: HtmlPlaywrightMiddleware;
 

src/scraper/middleware/components/index.ts

@@ -4,7 +4,7 @@ export * from "./HtmlLinkExtractorMiddleware";
 export * from "./HtmlMetadataExtractorMiddleware";
 export * from "./HtmlPlaywrightMiddleware";
 export * from "./HtmlSanitizerMiddleware";
-export * from "./HtmlSmartProcessorMiddleware";
+export * from "./HtmlSelectProcessorMiddleware";
 export * from "./HtmlToMarkdownMiddleware";
 export * from "./MarkdownLinkExtractorMiddleware";
 export * from "./MarkdownMetadataExtractorMiddleware";

src/scraper/strategies/WebScraperStrategy.ts

@@ -10,7 +10,7 @@ import {
   HtmlLinkExtractorMiddleware,
   HtmlMetadataExtractorMiddleware,
   HtmlSanitizerMiddleware,
-  HtmlSmartProcessorMiddleware, // Import the new middleware
+  HtmlSelectProcessorMiddleware, // Import the new middleware
   HtmlToMarkdownMiddleware,
   MarkdownLinkExtractorMiddleware,
   MarkdownMetadataExtractorMiddleware,
@@ -99,7 +99,7 @@ export class WebScraperStrategy extends BaseScraperStrategy {
       if (initialContext.contentType.startsWith("text/html")) {
         // Updated HTML pipeline order
         pipeline = new ContentProcessingPipeline([
-          new HtmlSmartProcessorMiddleware(), // Use the smart processor
+          new HtmlSelectProcessorMiddleware(), // Use the smart processor
           new HtmlMetadataExtractorMiddleware(),
           new HtmlLinkExtractorMiddleware(), // Extract links before cleaning
           new HtmlSanitizerMiddleware(),