@fedify/vocab-runtime 2.0.0-dev.1908 → 2.0.0-dev.196

package/dist/mod.cjs

@@ -22,6 +22,7 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
 
 //#endregion
 const __logtape_logtape = __toESM(require("@logtape/logtape"));
+const __opentelemetry_api = __toESM(require("@opentelemetry/api"));
 const node_process = __toESM(require("node:process"));
 const node_dns_promises = __toESM(require("node:dns/promises"));
 const node_net = __toESM(require("node:net"));
@@ -4178,6 +4179,45 @@ const preloadedContexts = {
 };
 var contexts_default = preloadedContexts;
 
+//#endregion
+//#region deno.json
+var name = "@fedify/vocab-runtime";
+var version = "2.0.0-dev.196+c3cfc0a9";
+var license = "MIT";
+var exports$1 = { ".": "./src/mod.ts" };
+var description = "Runtime library for @fedify/vocab";
+var author = {
+	"name": "Hong Minhee",
+	"email": "hong@minhee.org",
+	"url": "https://hongminhee.org/"
+};
+var imports = {
+	"@multiformats/base-x": "npm:@multiformats/base-x@^4.0.1",
+	"asn1js": "npm:asn1js@^3.0.6",
+	"byte-encodings": "npm:byte-encodings@^1.0.11",
+	"fetch-mock": "npm:fetch-mock@^12.5.4",
+	"multicodec": "npm:multicodec@^3.2.1",
+	"pkijs": "npm:pkijs@^3.2.5"
+};
+var exclude = ["dist", "node_modules"];
+var publish = { "exclude": ["**/*.test.ts"] };
+var tasks = {
+	"check": "deno fmt --check && deno lint && deno check src/*.ts",
+	"test": "deno test"
+};
+var deno_default = {
+	name,
+	version,
+	license,
+	exports: exports$1,
+	description,
+	author,
+	imports,
+	exclude,
+	publish,
+	tasks
+};
+
 //#endregion
 //#region src/link.ts
 const parametersNeedLowerCase = ["rel", "type"];
@@ -4360,42 +4400,6 @@ var HttpHeaderLink = class HttpHeaderLink {
 	}
 };
 
-//#endregion
-//#region deno.json
-var name = "@fedify/vocab-runtime";
-var version = "2.0.0-dev.1908+c31cc639";
-var license = "MIT";
-var exports$1 = { ".": "./src/mod.ts" };
-var description = "Runtime library for @fedify/vocab";
-var author = {
-	"name": "Hong Minhee",
-	"email": "hong@minhee.org",
-	"url": "https://hongminhee.org/"
-};
-var imports = {
-	"asn1js": "npm:asn1js@^3.0.6",
-	"byte-encodings": "npm:byte-encodings@^1.0.11",
-	"fetch-mock": "npm:fetch-mock@^12.5.4",
-	"multicodec": "npm:multicodec@^3.2.1",
-	"pkijs": "npm:pkijs@^3.2.5"
-};
-var exclude = ["dist", "node_modules"];
-var tasks = {
-	"check": "deno fmt --check && deno lint && deno check src/*.ts",
-	"test": "deno test"
-};
-var deno_default = {
-	name,
-	version,
-	license,
-	exports: exports$1,
-	description,
-	author,
-	imports,
-	exclude,
-	tasks
-};
-
 //#endregion
 //#region src/request.ts
 /**
@@ -4582,29 +4586,38 @@ async function getRemoteDocument(url, response, fetch$1) {
 	}
 	let document;
 	if (!jsonLd && (contentType === "text/html" || contentType?.startsWith("text/html;") || contentType === "application/xhtml+xml" || contentType?.startsWith("application/xhtml+xml;"))) {
-		const p = /<(a|link)((\s+[a-z][a-z:_-]*=("[^"]*"|'[^']*'|[^\s>]+))+)\s*\/?>/gi;
-		const p2 = /\s+([a-z][a-z:_-]*)=("([^"]*)"|'([^']*)'|([^\s>]+))/gi;
+		const MAX_HTML_SIZE = 1024 * 1024;
 		const html = await response.text();
-		let m;
-		const rawAttribs = [];
-		while ((m = p.exec(html)) !== null) rawAttribs.push(m[2]);
-		for (const rawAttrs of rawAttribs) {
-			let m2;
-			const attribs = {};
-			while ((m2 = p2.exec(rawAttrs)) !== null) {
-				const key = m2[1].toLowerCase();
-				const value = m2[3] ?? m2[4] ?? m2[5] ?? "";
-				attribs[key] = value;
-			}
-			if (attribs.rel === "alternate" && "type" in attribs && (attribs.type === "application/activity+json" || attribs.type === "application/ld+json" || attribs.type.startsWith("application/ld+json;")) && "href" in attribs && new URL(attribs.href, docUrl).href !== docUrl.href) {
-				logger.debug("Found alternate document: {alternateUrl} from {url}", {
-					alternateUrl: attribs.href,
-					url: documentUrl
-				});
-				return await fetch$1(new URL(attribs.href, docUrl).href);
+		if (html.length > MAX_HTML_SIZE) {
+			logger.warn("HTML response too large, skipping alternate link discovery: {url}", {
+				url: documentUrl,
+				size: html.length
+			});
+			document = JSON.parse(html);
+		} else {
+			const tagPattern = /<(a|link)\s+([^>]*?)\s*\/?>/gi;
+			const attrPattern = /([a-z][a-z:_-]*)=(?:"([^"]*)"|'([^']*)'|([^\s>]+))/gi;
+			let tagMatch;
+			while ((tagMatch = tagPattern.exec(html)) !== null) {
+				const tagContent = tagMatch[2];
+				let attrMatch;
+				const attribs = {};
+				attrPattern.lastIndex = 0;
+				while ((attrMatch = attrPattern.exec(tagContent)) !== null) {
+					const key = attrMatch[1].toLowerCase();
+					const value = attrMatch[2] ?? attrMatch[3] ?? attrMatch[4] ?? "";
+					attribs[key] = value;
+				}
+				if (attribs.rel === "alternate" && "type" in attribs && (attribs.type === "application/activity+json" || attribs.type === "application/ld+json" || attribs.type.startsWith("application/ld+json;")) && "href" in attribs && new URL(attribs.href, docUrl).href !== docUrl.href) {
+					logger.debug("Found alternate document: {alternateUrl} from {url}", {
+						alternateUrl: attribs.href,
+						url: documentUrl
+					});
+					return await fetch$1(new URL(attribs.href, docUrl).href);
+				}
 			}
+			document = JSON.parse(html);
 		}
-		document = JSON.parse(html);
 	} else document = await response.json();
 	logger.debug("Fetched document: {status} {url} {headers}", {
 		status: response.status,
@@ -4635,6 +4648,8 @@ async function getRemoteDocument(url, response, fetch$1) {
 * @since 1.3.0
 */
 function getDocumentLoader({ allowPrivateAddress, skipPreloadedContexts, userAgent } = {}) {
+	const tracerProvider = __opentelemetry_api.trace.getTracerProvider();
+	const tracer = tracerProvider.getTracer(deno_default.name, deno_default.version);
 	async function load(url, options) {
 		options?.signal?.throwIfAborted();
 		if (!skipPreloadedContexts && url in contexts_default) {
@@ -4654,14 +4669,38 @@ function getDocumentLoader({ allowPrivateAddress, skipPreloadedContexts, userAge
 			});
 			throw error;
 		}
-		const request = createActivityPubRequest(url, { userAgent });
-		logRequest(logger, request);
-		const response = await fetch(request, {
-			redirect: "manual",
-			signal: options?.signal
+		return await tracer.startActiveSpan("activitypub.fetch_document", {
+			kind: __opentelemetry_api.SpanKind.CLIENT,
+			attributes: { "url.full": url }
+		}, async (span) => {
+			try {
+				const request = createActivityPubRequest(url, { userAgent });
+				logRequest(logger, request);
+				const response = await fetch(request, {
+					redirect: "manual",
+					signal: options?.signal
+				});
+				span.setAttribute("http.response.status_code", response.status);
+				if (response.status >= 300 && response.status < 400 && response.headers.has("Location")) {
+					const redirectUrl = response.headers.get("Location");
+					span.setAttribute("http.redirect.url", redirectUrl);
+					return await load(redirectUrl, options);
+				}
+				const result = await getRemoteDocument(url, response, load);
+				span.setAttribute("docloader.document_url", result.documentUrl);
+				if (result.contextUrl != null) span.setAttribute("docloader.context_url", result.contextUrl);
+				return result;
+			} catch (error) {
+				span.recordException(error);
+				span.setStatus({
+					code: __opentelemetry_api.SpanStatusCode.ERROR,
+					message: String(error)
+				});
+				throw error;
+			} finally {
+				span.end();
+			}
 		});
-		if (response.status >= 300 && response.status < 400 && response.headers.has("Location")) return load(response.headers.get("Location"), options);
-		return getRemoteDocument(url, response, load);
 	}
 	return load;
 }

package/dist/mod.js

@@ -1,4 +1,5 @@
 import { getLogger } from "@logtape/logtape";
+import { SpanKind, SpanStatusCode, trace } from "@opentelemetry/api";
 import process from "node:process";
 import { lookup } from "node:dns/promises";
 import { isIP } from "node:net";
@@ -4155,6 +4156,45 @@ const preloadedContexts = {
 };
 var contexts_default = preloadedContexts;
 
+//#endregion
+//#region deno.json
+var name = "@fedify/vocab-runtime";
+var version = "2.0.0-dev.196+c3cfc0a9";
+var license = "MIT";
+var exports = { ".": "./src/mod.ts" };
+var description = "Runtime library for @fedify/vocab";
+var author = {
+	"name": "Hong Minhee",
+	"email": "hong@minhee.org",
+	"url": "https://hongminhee.org/"
+};
+var imports = {
+	"@multiformats/base-x": "npm:@multiformats/base-x@^4.0.1",
+	"asn1js": "npm:asn1js@^3.0.6",
+	"byte-encodings": "npm:byte-encodings@^1.0.11",
+	"fetch-mock": "npm:fetch-mock@^12.5.4",
+	"multicodec": "npm:multicodec@^3.2.1",
+	"pkijs": "npm:pkijs@^3.2.5"
+};
+var exclude = ["dist", "node_modules"];
+var publish = { "exclude": ["**/*.test.ts"] };
+var tasks = {
+	"check": "deno fmt --check && deno lint && deno check src/*.ts",
+	"test": "deno test"
+};
+var deno_default = {
+	name,
+	version,
+	license,
+	exports,
+	description,
+	author,
+	imports,
+	exclude,
+	publish,
+	tasks
+};
+
 //#endregion
 //#region src/link.ts
 const parametersNeedLowerCase = ["rel", "type"];
@@ -4337,42 +4377,6 @@ var HttpHeaderLink = class HttpHeaderLink {
 	}
 };
 
-//#endregion
-//#region deno.json
-var name = "@fedify/vocab-runtime";
-var version = "2.0.0-dev.1908+c31cc639";
-var license = "MIT";
-var exports = { ".": "./src/mod.ts" };
-var description = "Runtime library for @fedify/vocab";
-var author = {
-	"name": "Hong Minhee",
-	"email": "hong@minhee.org",
-	"url": "https://hongminhee.org/"
-};
-var imports = {
-	"asn1js": "npm:asn1js@^3.0.6",
-	"byte-encodings": "npm:byte-encodings@^1.0.11",
-	"fetch-mock": "npm:fetch-mock@^12.5.4",
-	"multicodec": "npm:multicodec@^3.2.1",
-	"pkijs": "npm:pkijs@^3.2.5"
-};
-var exclude = ["dist", "node_modules"];
-var tasks = {
-	"check": "deno fmt --check && deno lint && deno check src/*.ts",
-	"test": "deno test"
-};
-var deno_default = {
-	name,
-	version,
-	license,
-	exports,
-	description,
-	author,
-	imports,
-	exclude,
-	tasks
-};
-
 //#endregion
 //#region src/request.ts
 /**
@@ -4559,29 +4563,38 @@ async function getRemoteDocument(url, response, fetch$1) {
 	}
 	let document;
 	if (!jsonLd && (contentType === "text/html" || contentType?.startsWith("text/html;") || contentType === "application/xhtml+xml" || contentType?.startsWith("application/xhtml+xml;"))) {
-		const p = /<(a|link)((\s+[a-z][a-z:_-]*=("[^"]*"|'[^']*'|[^\s>]+))+)\s*\/?>/gi;
-		const p2 = /\s+([a-z][a-z:_-]*)=("([^"]*)"|'([^']*)'|([^\s>]+))/gi;
+		const MAX_HTML_SIZE = 1024 * 1024;
 		const html = await response.text();
-		let m;
-		const rawAttribs = [];
-		while ((m = p.exec(html)) !== null) rawAttribs.push(m[2]);
-		for (const rawAttrs of rawAttribs) {
-			let m2;
-			const attribs = {};
-			while ((m2 = p2.exec(rawAttrs)) !== null) {
-				const key = m2[1].toLowerCase();
-				const value = m2[3] ?? m2[4] ?? m2[5] ?? "";
-				attribs[key] = value;
-			}
-			if (attribs.rel === "alternate" && "type" in attribs && (attribs.type === "application/activity+json" || attribs.type === "application/ld+json" || attribs.type.startsWith("application/ld+json;")) && "href" in attribs && new URL(attribs.href, docUrl).href !== docUrl.href) {
-				logger.debug("Found alternate document: {alternateUrl} from {url}", {
-					alternateUrl: attribs.href,
-					url: documentUrl
-				});
-				return await fetch$1(new URL(attribs.href, docUrl).href);
+		if (html.length > MAX_HTML_SIZE) {
+			logger.warn("HTML response too large, skipping alternate link discovery: {url}", {
+				url: documentUrl,
+				size: html.length
+			});
+			document = JSON.parse(html);
+		} else {
+			const tagPattern = /<(a|link)\s+([^>]*?)\s*\/?>/gi;
+			const attrPattern = /([a-z][a-z:_-]*)=(?:"([^"]*)"|'([^']*)'|([^\s>]+))/gi;
+			let tagMatch;
+			while ((tagMatch = tagPattern.exec(html)) !== null) {
+				const tagContent = tagMatch[2];
+				let attrMatch;
+				const attribs = {};
+				attrPattern.lastIndex = 0;
+				while ((attrMatch = attrPattern.exec(tagContent)) !== null) {
+					const key = attrMatch[1].toLowerCase();
+					const value = attrMatch[2] ?? attrMatch[3] ?? attrMatch[4] ?? "";
+					attribs[key] = value;
+				}
+				if (attribs.rel === "alternate" && "type" in attribs && (attribs.type === "application/activity+json" || attribs.type === "application/ld+json" || attribs.type.startsWith("application/ld+json;")) && "href" in attribs && new URL(attribs.href, docUrl).href !== docUrl.href) {
+					logger.debug("Found alternate document: {alternateUrl} from {url}", {
+						alternateUrl: attribs.href,
+						url: documentUrl
+					});
+					return await fetch$1(new URL(attribs.href, docUrl).href);
+				}
 			}
+			document = JSON.parse(html);
 		}
-		document = JSON.parse(html);
 	} else document = await response.json();
 	logger.debug("Fetched document: {status} {url} {headers}", {
 		status: response.status,
@@ -4612,6 +4625,8 @@ async function getRemoteDocument(url, response, fetch$1) {
 * @since 1.3.0
 */
 function getDocumentLoader({ allowPrivateAddress, skipPreloadedContexts, userAgent } = {}) {
+	const tracerProvider = trace.getTracerProvider();
+	const tracer = tracerProvider.getTracer(deno_default.name, deno_default.version);
 	async function load(url, options) {
 		options?.signal?.throwIfAborted();
 		if (!skipPreloadedContexts && url in contexts_default) {
@@ -4631,14 +4646,38 @@ function getDocumentLoader({ allowPrivateAddress, skipPreloadedContexts, userAge
 			});
 			throw error;
 		}
-		const request = createActivityPubRequest(url, { userAgent });
-		logRequest(logger, request);
-		const response = await fetch(request, {
-			redirect: "manual",
-			signal: options?.signal
+		return await tracer.startActiveSpan("activitypub.fetch_document", {
+			kind: SpanKind.CLIENT,
+			attributes: { "url.full": url }
+		}, async (span) => {
+			try {
+				const request = createActivityPubRequest(url, { userAgent });
+				logRequest(logger, request);
+				const response = await fetch(request, {
+					redirect: "manual",
+					signal: options?.signal
+				});
+				span.setAttribute("http.response.status_code", response.status);
+				if (response.status >= 300 && response.status < 400 && response.headers.has("Location")) {
+					const redirectUrl = response.headers.get("Location");
+					span.setAttribute("http.redirect.url", redirectUrl);
+					return await load(redirectUrl, options);
+				}
+				const result = await getRemoteDocument(url, response, load);
+				span.setAttribute("docloader.document_url", result.documentUrl);
+				if (result.contextUrl != null) span.setAttribute("docloader.context_url", result.contextUrl);
+				return result;
+			} catch (error) {
+				span.recordException(error);
+				span.setStatus({
+					code: SpanStatusCode.ERROR,
+					message: String(error)
+				});
+				throw error;
+			} finally {
+				span.end();
+			}
 		});
-		if (response.status >= 300 && response.status < 400 && response.headers.has("Location")) return load(response.headers.get("Location"), options);
-		return getRemoteDocument(url, response, load);
 	}
 	return load;
 }