feedscout 2.0.0-beta.1 → 2.0.0-beta.2

package/dist/blogrolls/defaults.d.ts

@@ -4,7 +4,6 @@ import { HtmlMethodOptions } from "../common/uris/html/types.js";
 import { LinkSelector } from "../common/types.js";
 
 //#region src/blogrolls/defaults.d.ts
-declare const mimeTypes: string[];
 declare const urisMinimal: string[];
 declare const urisBalanced: string[];
 declare const urisComprehensive: string[];
@@ -14,4 +13,4 @@ declare const defaultHtmlOptions: Omit<HtmlMethodOptions, 'baseUrl'>;
 declare const defaultHeadersOptions: Omit<HeadersMethodOptions, 'baseUrl'>;
 declare const defaultGuessOptions: Omit<GuessMethodOptions, 'baseUrl'>;
 //#endregion
-export { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal };
+export { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal };

package/dist/blogrolls/defaults.js

@@ -1,9 +1,4 @@
 //#region src/blogrolls/defaults.ts
-const mimeTypes = [
-	"text/x-opml",
-	"application/xml",
-	"text/xml"
-];
 const urisMinimal = [
 	"/.well-known/recommendations.opml",
 	"/blogroll.opml",
@@ -27,10 +22,22 @@ const anchorLabels = [
 	"subscriptions",
 	"reading list"
 ];
-const linkSelectors = [{ rel: "blogroll" }, {
-	rel: "outline",
-	types: mimeTypes
-}];
+const linkSelectors = [
+	{ rel: "blogroll" },
+	{
+		rel: "outline",
+		types: [
+			"text/x-opml",
+			"application/opml+xml",
+			"application/xml",
+			"text/xml"
+		]
+	},
+	{
+		rel: "alternate",
+		types: ["text/x-opml", "application/opml+xml"]
+	}
+];
 const defaultHtmlOptions = {
 	linkSelectors,
 	anchorUris: urisComprehensive,
@@ -40,4 +47,4 @@ const defaultHtmlOptions = {
 const defaultHeadersOptions = { linkSelectors };
 const defaultGuessOptions = { uris: urisBalanced };
 //#endregion
-export { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal };
+export { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal };

package/dist/blogrolls.d.ts

@@ -1,4 +1,4 @@
-import { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal } from "./blogrolls/defaults.js";
+import { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal } from "./blogrolls/defaults.js";
 import { BlogrollResult } from "./blogrolls/types.js";
 import { defaultExtractFn } from "./blogrolls/extractors.js";
-export { BlogrollResult, anchorLabels, defaultExtractFn, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal };
+export { BlogrollResult, anchorLabels, defaultExtractFn, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal };

package/dist/blogrolls.js

@@ -1,3 +1,3 @@
-import { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal } from "./blogrolls/defaults.js";
+import { anchorLabels, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal } from "./blogrolls/defaults.js";
 import { defaultExtractFn } from "./blogrolls/extractors.js";
-export { anchorLabels, defaultExtractFn, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, mimeTypes, urisBalanced, urisComprehensive, urisMinimal };
+export { anchorLabels, defaultExtractFn, defaultGuessOptions, defaultHeadersOptions, defaultHtmlOptions, linkSelectors, urisBalanced, urisComprehensive, urisMinimal };

package/dist/common/discover/defaults.js

@@ -17,9 +17,7 @@ const defaultFetchFn = async (url, options) => {
 const defaultResolveUrlFn = (url, baseUrl) => {
 	try {
 		return new URL(url, baseUrl).href;
-	} catch {
-		return;
-	}
+	} catch {}
 };
 const defaultResolveSiteUrlFn = (input, resolveUrlFn) => {
 	if (!input.content) return;

package/dist/common/uris/html/handlers.js

@@ -1,6 +1,10 @@
 import { endsWithAnyOf, includesAnyOf, matchesAnyOfLinkSelectors } from "../../utils.js";
 //#region src/common/uris/html/handlers.ts
 const handleOpenTag = (context, name, attribs, _isImplied) => {
+	if (name === "base" && context.baseHref === void 0) {
+		const href = attribs.href?.trim();
+		if (href) context.baseHref = href;
+	}
 	if (name === "link" && attribs.href) {
 		const rel = attribs.rel?.toLowerCase();
 		if (!rel) return;
@@ -16,6 +20,16 @@ const handleOpenTag = (context, name, attribs, _isImplied) => {
 		context.currentAnchor.href = attribs.href;
 		context.currentAnchor.text = "";
 		if (endsWithAnyOf(lowerHref, context.options.anchorUris)) context.discoveredUris.add(attribs.href);
+		if (context.options.anchorPathSegments?.length) {
+			let pathname;
+			try {
+				pathname = new URL(attribs.href, "https://feedscout.invalid").pathname;
+			} catch {}
+			if (pathname && includesAnyOf(pathname, context.options.anchorPathSegments)) context.discoveredUris.add(attribs.href);
+		}
+		const ariaLabel = attribs["aria-label"];
+		const title = attribs.title;
+		if (ariaLabel && includesAnyOf(ariaLabel, context.options.anchorLabels) || title && includesAnyOf(title, context.options.anchorLabels)) context.discoveredUris.add(attribs.href);
 	}
 };
 const handleText = (context, text) => {

package/dist/common/uris/html/index.js

@@ -13,7 +13,23 @@ const discoverUrisFromHtml = (html, options) => {
 	const parser = new Parser(createHtmlUrisHandlers(context), { decodeEntities: true });
 	parser.write(html);
 	parser.end();
-	return [...context.discoveredUris];
+	const uris = [...context.discoveredUris];
+	if (context.baseHref) {
+		let base;
+		try {
+			base = options.baseUrl ? new URL(context.baseHref, options.baseUrl).href : context.baseHref;
+		} catch {
+			base = options.baseUrl;
+		}
+		return uris.map((uri) => {
+			try {
+				return new URL(uri, base).href;
+			} catch {
+				return uri;
+			}
+		});
+	}
+	return uris;
 };
 //#endregion
 export { discoverUrisFromHtml };

package/dist/common/uris/html/types.d.ts

@@ -5,6 +5,7 @@ type HtmlMethodOptions = {
   baseUrl?: string;
   linkSelectors: Array<LinkSelector>;
   anchorUris: Array<Pattern>;
+  anchorPathSegments?: Array<Pattern>;
   anchorIgnoredUris: Array<Pattern>;
   anchorLabels: Array<Pattern>;
 };

package/dist/feeds/defaults.d.ts

@@ -2,14 +2,14 @@ import { GuessMethodOptions } from "../common/uris/guess/types.js";
 import { HeadersMethodOptions } from "../common/uris/headers/types.js";
 import { HtmlMethodOptions } from "../common/uris/html/types.js";
 import { PlatformMethodOptions } from "../common/uris/platform/types.js";
-import { LinkSelector, UriEntry } from "../common/types.js";
+import { LinkSelector, Pattern, UriEntry } from "../common/types.js";
 
 //#region src/feeds/defaults.d.ts
 declare const mimeTypes: string[];
 declare const urisMinimal: string[];
 declare const urisBalanced: string[];
 declare const urisComprehensive: Array<UriEntry>;
-declare const ignoredUris: string[];
+declare const ignoredUris: Array<Pattern>;
 declare const anchorLabels: string[];
 declare const linkSelectors: Array<LinkSelector>;
 declare const anchorPathSegments: RegExp[];

package/dist/feeds/defaults.js

@@ -152,12 +152,15 @@ const urisComprehensive = [
 	"/feeds/posts/default?alt=rss",
 	"/feeds/comments/default"
 ];
-const ignoredUris = ["wp-json/oembed/", "wp-json/wp/"];
+const ignoredUris = [
+	"wp-json/oembed/",
+	"wp-json/wp/",
+	/[?&][^=&]*=(https?:|https?%3a|aHR0c)/i
+];
 const anchorLabels = [
 	"rss",
 	"feed",
 	"atom",
-	"subscribe",
 	"syndicate",
 	"syndication",
 	"json feed"
@@ -173,7 +176,8 @@ const anchorPathSegments = [
 ];
 const defaultHtmlOptions = {
 	linkSelectors,
-	anchorUris: [...urisComprehensive.flat(), ...anchorPathSegments],
+	anchorUris: urisComprehensive.flat(),
+	anchorPathSegments,
 	anchorIgnoredUris: ignoredUris,
 	anchorLabels
 };

package/dist/feeds/platform/handlers/discourse.js

@@ -22,12 +22,8 @@ const isDiscourseHtml = (content) => {
 };
 const discourseHandler = {
 	match: (url, content) => {
-		try {
-			if (!content || !isDiscourseHtml(content)) return false;
-			new URL(url);
-			return true;
-		} catch {}
-		return false;
+		if (!content || !isDiscourseHtml(content)) return false;
+		return URL.canParse(url);
 	},
 	resolve: (url) => {
 		try {

package/dist/feeds/platform/handlers/libsyn.js

@@ -8,7 +8,7 @@ const libsynHandler = {
 	resolve: (url) => {
 		const { origin, pathname } = new URL(url);
 		if (isHostOf(url, "feeds.libsyn.com")) {
-			const showId = pathname.split("/").filter(Boolean)[0];
+			const showId = pathname.split("/").find(Boolean);
 			if (showId && numericRegex.test(showId)) return [{
 				uri: `https://feeds.libsyn.com/${showId}/rss`,
 				hint: composeHint("libsyn:podcast")

package/package.json

@@ -68,14 +68,14 @@
     "htmlparser2": "^12.0.0"
   },
   "peerDependencies": {
-    "feedsmith": "^3.0.0-beta.4"
+    "feedsmith": "3.0.0-beta.5"
   },
   "devDependencies": {
     "@types/bun": "^1.3.13",
-    "feedsmith": "^3.0.0-beta.4",
-    "kvalita": "^1.13.0",
+    "feedsmith": "3.0.0-beta.5",
+    "kvalita": "^1.15.1",
     "tsdown": "^0.22.2",
     "vitepress": "^2.0.0-alpha.17"
   },
-  "version": "2.0.0-beta.1"
+  "version": "2.0.0-beta.2"
 }