@jini-ai/platform 0.6.15-rc.1 → 0.6.15

package/dist/web/html-to-markdown.js

@@ -12,18 +12,25 @@ const MAX_DEPTH = 300;
 function isElement(node) {
     return "tagName" in node;
 }
+/** A node's children for {@link plainText}; skipped elements and leaves contribute none. */
+function scannableChildren(node) {
+    if (!("childNodes" in node) || (isElement(node) && SKIPPED.has(node.tagName)))
+        return [];
+    return node.childNodes;
+}
 /** Iterative so a degraded, very deep subtree cannot overflow the stack either. O(subtree nodes). */
 function plainText(root) {
     const pending = [root];
     const parts = [];
     while (pending.length) {
         const node = pending.pop();
-        if (node.nodeName === "#text" && "value" in node)
+        if (node.nodeName === "#text" && "value" in node) {
             parts.push(node.value);
-        else if ("childNodes" in node && !(isElement(node) && SKIPPED.has(node.tagName))) {
-            for (let i = node.childNodes.length - 1; i >= 0; i--)
-                pending.push(node.childNodes[i]);
+            continue;
         }
+        const children = scannableChildren(node);
+        for (let i = children.length - 1; i >= 0; i--)
+            pending.push(children[i]);
     }
     return parts.join("").replace(/\s+/g, " ");
 }
@@ -112,9 +119,11 @@ function takeEdgeWhitespace({ node, last }, _options = {}) {
     const edge = edgeText({ node, last });
     if (!edge)
         return "";
-    const whitespace = (last ? /\s+$/ : /^\s+/).exec(edge.value)?.[0] ?? "";
-    edge.value = last ? edge.value.slice(0, edge.value.length - whitespace.length) : edge.value.slice(whitespace.length);
-    return whitespace;
+    const value = edge.value;
+    // trimStart/trimEnd strip exactly the characters `\s` matches.
+    const kept = last ? value.trimEnd() : value.trimStart();
+    edge.value = kept;
+    return last ? value.slice(kept.length) : value.slice(0, value.length - kept.length);
 }
 /** The first/last leaf of the subtree when it is a text node. */
 function edgeText({ node, last }, _options = {}) {
@@ -141,9 +150,11 @@ function hastText(root) {
     while (pending.length) {
         const node = pending.pop();
         // HAST text nodes always carry a value.
-        if (node.type === "text")
+        if (node.type === "text") {
             parts.push(node.value);
-        else if (node.children)
+            continue;
+        }
+        if (node.children)
             for (let i = node.children.length - 1; i >= 0; i--)
                 pending.push(node.children[i]);
     }

package/dist/web/web-page.js

@@ -1,4 +1,5 @@
 import { boundHtml } from "./html-resource.js";
+import { firstDistinct } from "./first-distinct.js";
 import { ToolInputError } from "@jini-ai/core";
 import { EgressRefusedError } from "../http/guarded/errors.js";
 import { htmlToText, readHtmlDocument, stripActiveHtml, MAX_LINKS, MAX_IMAGES } from "./html-document.js";
@@ -145,28 +146,25 @@ const EMPTY_FACTS = { links: [], linksTruncated: false, images: [], imagesTrunca
 /** A sitemap's `<loc>` entries become links so an importer can enumerate pages from one call. */
 function sitemapFacts(xml, pageUrl) {
     const host = new URL(pageUrl).hostname.replace(/^www\./, "");
-    const seen = new Set();
-    const links = [];
-    let linksTruncated = false;
-    for (const match of xml.matchAll(/<loc>\s*([^<\s]+)\s*<\/loc>/gi)) {
-        let url;
-        try {
-            url = new URL(match[1].replace(/&amp;/g, "&"), pageUrl);
-        }
-        catch {
-            continue;
-        }
-        if ((url.protocol !== "http:" && url.protocol !== "https:") || seen.has(url.href))
-            continue;
-        seen.add(url.href);
-        if (links.length === MAX_LINKS) {
-            linksTruncated = true;
-            break;
-        }
-        links.push({ href: url.href, text: "", internal: url.hostname.replace(/^www\./, "") === host });
-    }
+    const { picked, truncated: linksTruncated } = firstDistinct({
+        items: [...xml.matchAll(/<loc>\s*([^<\s]+)\s*<\/loc>/gi)],
+        max: MAX_LINKS,
+        keyOf: (match) => sitemapHref(match[1], pageUrl),
+    });
+    const links = picked.map(([href]) => ({ href, text: "", internal: new URL(href).hostname.replace(/^www\./, "") === host }));
     return { ...EMPTY_FACTS, links, linksTruncated };
 }
+/** The absolute http(s) href of one `<loc>` value, or undefined when it is unparsable or another scheme. */
+function sitemapHref(loc, pageUrl) {
+    let url;
+    try {
+        url = new URL(loc.replace(/&amp;/g, "&"), pageUrl);
+    }
+    catch {
+        return undefined;
+    }
+    return url.protocol === "http:" || url.protocol === "https:" ? url.href : undefined;
+}
 function render(format, body, isHtml, pageUrl, ports) {
     if (!isHtml || format === "raw")
         return body;
@@ -364,9 +362,9 @@ function projectRendered({ input, ports, url, page }) {
 }
 /** Any clipped or dropped fact, image, data or text marks the whole rendered result truncated. */
 function renderedFlags({ page, result, images, factsTruncated, textTruncated }) {
-    const imagesTruncated = images.truncated || page.imagesTruncated || result.imagesTruncated || images.byUrl.size > MAX_IMAGES;
-    const truncated = imagesTruncated || factsTruncated || page.title.length > MAX_RENDERED_TITLE || result.truncated
-        || result.linksTruncated || page.dataTruncated || textTruncated;
+    const imagesTruncated = [images.truncated, page.imagesTruncated, result.imagesTruncated, images.byUrl.size > MAX_IMAGES].some(Boolean);
+    const truncated = [imagesTruncated, factsTruncated, page.title.length > MAX_RENDERED_TITLE, result.truncated,
+        result.linksTruncated, page.dataTruncated, textTruncated].some(Boolean);
     return { imagesTruncated, truncated };
 }
 /** Browser image facts come first, so their alt text wins over the parsed DOM's for the same src. */
@@ -391,8 +389,13 @@ function boundRenderedFacts({ result }, _optional = {}) {
     const stylesheets = result.stylesheets.filter(href => href.length <= MAX_URL_LENGTH);
     const meta = Object.fromEntries(Object.entries(result.meta).filter(([key]) => key.length <= MAX_RENDERED_META_KEY));
     const linksDropped = links.length < result.links.length;
-    const truncated = linksDropped || stylesheets.length < result.stylesheets.length || Object.keys(meta).length < Object.keys(result.meta).length
-        || (result.description?.length ?? 0) > MAX_RENDERED_DESCRIPTION || (result.lang?.length ?? 0) > MAX_RENDERED_LANG;
+    const truncated = [
+        linksDropped,
+        stylesheets.length < result.stylesheets.length,
+        Object.keys(meta).length < Object.keys(result.meta).length,
+        (result.description?.length ?? 0) > MAX_RENDERED_DESCRIPTION,
+        (result.lang?.length ?? 0) > MAX_RENDERED_LANG,
+    ].some(Boolean);
     return { links, stylesheets, meta, linksDropped, truncated };
 }
 /** `none` empties content; `summary` and text format use the browser's visible text instead of the projection. */

package/dist/windows-system-tools.d.ts

@@ -0,0 +1,15 @@
+/**
+ * Absolute path of a Windows system tool under `%SystemRoot%\System32`. A bare name such as
+ * `reg.exe` is resolved by the child-process search, which on Windows consults the current
+ * directory before PATH, so a same-named file in a project folder could run instead of the OS tool.
+ * Falls back to `C:\Windows` when the environment does not name the system root.
+ * @complexity O(1).
+ */
+export declare function windowsSystemTool({ relativePath }: {
+    relativePath: readonly string[];
+}, { env }?: {
+    env?: NodeJS.ProcessEnv;
+}): string;
+/** Windows PowerShell 5.1, which ships with every supported Windows release. */
+export declare function windowsPowerShell(): string;
+//# sourceMappingURL=windows-system-tools.d.ts.map

package/dist/windows-system-tools.js

@@ -0,0 +1,16 @@
+import { win32 } from "node:path";
+/**
+ * Absolute path of a Windows system tool under `%SystemRoot%\System32`. A bare name such as
+ * `reg.exe` is resolved by the child-process search, which on Windows consults the current
+ * directory before PATH, so a same-named file in a project folder could run instead of the OS tool.
+ * Falls back to `C:\Windows` when the environment does not name the system root.
+ * @complexity O(1).
+ */
+export function windowsSystemTool({ relativePath }, { env = process.env } = {}) {
+    return win32.join(env.SystemRoot ?? env.windir ?? "C:\\Windows", "System32", ...relativePath);
+}
+/** Windows PowerShell 5.1, which ships with every supported Windows release. */
+export function windowsPowerShell() {
+    return windowsSystemTool({ relativePath: ["WindowsPowerShell", "v1.0", "powershell.exe"] });
+}
+//# sourceMappingURL=windows-system-tools.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jini-ai/platform",
-  "version": "0.6.15-rc.1",
+  "version": "0.6.15",
   "description": "Generic OS/platform primitives: process exec and lifecycle, filesystem containment, HTTP readiness polling, toolchain discovery, blob storage, AWS signing, and an SSRF-safe asset cache.",
   "license": "Apache-2.0",
   "repository": {
@@ -206,7 +206,7 @@
     "!dist/**/*.map"
   ],
   "dependencies": {
-    "@jini-ai/core": "^0.6.15-rc.1"
+    "@jini-ai/core": "^0.6.13"
   },
   "publishConfig": {
     "access": "public",