Project Files

.claude

settings.local.json

src

bing

build-urls.ts

index.ts

parse-results.ts

search-images.ts

cache

image-search-results-payload.ts

index.ts

search-cache-key.ts

ttl-cache.ts

web-search-results-payload.ts

config

auto-sentinel.ts

config-defaults.ts

config-schematics.ts

resolve-config.ts

duckduckgo

build-urls.ts

index.ts

parse-results.ts

search-web.ts

web-search-result.ts

enrichment

enrich-search-results.ts

index.ts

metascraper-helpers.d.ts

metascraper.ts

errors

abort-error.ts

error-message.ts

index.ts

no-results-error.ts

tool-error.ts

unsupported-content-type-error.ts

index.ts

lmstudio-home.ts

markdown-path.ts

url-filename.ts

http

challenge.ts

decode.ts

fetch-error.ts

fetch-retry.ts

fetch.ts

impit-client.ts

impit-error.ts

index.ts

parse-content-type.ts

redirects.ts

response-body.ts

ssrf.ts

url-schema.ts

url.ts

images

download-image.ts

download-images.ts

index.ts

page

fetch-page.ts

fetched-page.ts

index.ts

page-kind.ts

parsers

image-extensions.ts

index.ts

page-images.ts

page-text.ts

parse-html.ts

pdf-text.ts

renderers

image-results.ts

index.ts

page-result.ts

retrieval

chunking.ts

excerpt.ts

index.ts

relevance.ts

selection.ts

index.ts

safe-search.ts

search-page-parameter.ts

text

escape-markdown.ts

html-to-markdown.ts

html-to-text.ts

index.ts

normalize-blank-lines.ts

normalize-text.ts

timing

index.ts

per-host-rate-limiter.ts

rate-limiter.ts

tools

fetch-images-tool.ts

image-search-tool.ts

visit-website-tool.ts

web-search-tool.ts

index.ts

tools-provider.ts

.gitignore

.prettierrc.json

CLAUDE.md

eslint.config.mjs

knip.json

LICENSE

manifest.json

package-lock.json

package.json

QWEN.md

README.md

thumbnail.png

tsconfig.json

tsdoc.json

src / text / html-to-markdown.ts

/**
 * HTML to Markdown conversion.
 */

import TurndownService from "turndown"

import { normalizeBlankLines } from "./normalize-blank-lines"

/**
 * Shared Turndown service, built once at module load.
 */
const service = createService()

/**
 * Convert an HTML fragment to Markdown, preserving headings, lists, code blocks,
 * emphasis, inline links, and inline images.
 *
 * @param html - HTML fragment to convert.
 * @returns The markdown representation, with runs of blank lines collapsed and trailing whitespace trimmed.
 */
export function htmlToMarkdown(html: string): string {
  return normalizeBlankLines(service.turndown(html))
}

/**
 * Build the shared Turndown service with project-wide formatting preferences.
 *
 * @returns The configured service instance.
 */
function createService(): TurndownService {
  const turndown = new TurndownService({
    headingStyle: "atx",
    bulletListMarker: "-",
    codeBlockStyle: "fenced",
    emDelimiter: "*",
    strongDelimiter: "**",
    linkStyle: "inlined",
  })

  turndown.remove(["script", "style", "noscript", "template"])

  return turndown
}

web-tools