1import type { Adapter } from "../types";
2import { detectInteractionGate } from "../../browser/interaction-gates";
3import { captureNormalizedPageSnapshot } from "../../browser/page-snapshot";
4import { convertHtmlToMarkdown } from "../../extract/html-to-markdown";
5 
6export const genericAdapter: Adapter = {
7  name: "generic",
8  match() {
9    return true;
10  },
11  async process(context) {
12    context.log.info(`Loading ${context.input.url.toString()} with generic adapter`);
13    await context.browser.goto(context.input.url.toString(), context.timeoutMs);
14 
15    try {
16      await context.network.waitForIdle({
17        idleMs: 1_200,
18        timeoutMs: Math.min(context.timeoutMs, 15_000),
19      });
20    } catch {
21      context.log.debug("Network idle timed out on initial load; continuing.");
22    }
23 
24    await context.browser.scrollToEnd({ maxSteps: 4, delayMs: 300 });
25 
26    try {
27      await context.network.waitForIdle({
28        idleMs: 900,
29        timeoutMs: Math.min(context.timeoutMs, 10_000),
30      });
31    } catch {
32      context.log.debug("Network idle timed out after scrolling; continuing.");
33    }
34 
35    const interaction = await detectInteractionGate(context.browser);
36    if (interaction) {
37      return {
38        status: "needs_interaction",
39        interaction,
40      };
41    }
42 
43    const snapshot = await captureNormalizedPageSnapshot(context.browser);
44    const converted = await convertHtmlToMarkdown(snapshot.html, snapshot.finalUrl, {
45      enableRemoteMarkdownFallback: context.outputFormat === "markdown",
46      preserveBase64Images: context.downloadMedia,
47    });
48    const document = {
49      url: snapshot.finalUrl,
50      canonicalUrl: converted.metadata.canonicalUrl,
51      title: converted.metadata.title,
52      author: converted.metadata.author,
53      siteName: converted.metadata.siteName,
54      publishedAt: converted.metadata.publishedAt,
55      summary: converted.metadata.summary,
56      adapter: "generic",
57      metadata: {
58        coverImage: converted.metadata.coverImage,
59        language: converted.metadata.language,
60        capturedAt: converted.metadata.capturedAt,
61        conversionMethod: converted.conversionMethod,
62        fallbackReason: converted.fallbackReason,
63        kind: "generic/article",
64      },
65      content: converted.markdown ? [{ type: "markdown" as const, markdown: converted.markdown }] : [],
66    };
67 
68    return {
69      status: "ok",
70      document,
71      media: converted.media,
72    };
73  },
74};
75