@pagepocket/metadata-unit 0.13.0

package/dist/build-metadata.d.ts

@@ -0,0 +1,24 @@
+import type { EntryInfo, FileTree } from "@pagepocket/lib";
+export type PagePocketMetadata = {
+    version: 1;
+    url: string;
+    title: string;
+    snapshotType: string;
+    capturedAt: string;
+    generator: string;
+    resources: {
+        fileCount: number;
+    };
+};
+export type BuildMetadataInput = {
+    entry: EntryInfo;
+    html?: {
+        htmlString: string;
+        baseUrl: string;
+        url?: string;
+    };
+    files?: FileTree;
+    snapshotType?: string;
+    capturedAt?: Date;
+};
+export declare const buildMetadata: (input: BuildMetadataInput) => PagePocketMetadata;

package/dist/build-metadata.js

@@ -0,0 +1,40 @@
+const extractTitle = (htmlString) => {
+    const match = /<title[^>]*>([\s\S]*?)<\/title>/i.exec(htmlString);
+    const raw = match?.[1] ?? "";
+    return raw.replace(/\s+/g, " ").trim();
+};
+const resolveUrl = (entry, html) => {
+    if (entry.kind === "url") {
+        return entry.url;
+    }
+    if (entry.kind === "html-string" || entry.kind === "document") {
+        return entry.url ?? entry.baseUrl;
+    }
+    return html?.url ?? html?.baseUrl ?? "";
+};
+const countFiles = (dir) => {
+    let count = 0;
+    for (const entry of dir.entries) {
+        if (entry.kind === "file") {
+            count += 1;
+        }
+        else {
+            count += countFiles(entry);
+        }
+    }
+    return count;
+};
+export const buildMetadata = (input) => {
+    const url = resolveUrl(input.entry, input.html);
+    const title = input.html ? extractTitle(input.html.htmlString) : "";
+    const fileCount = input.files?.root ? countFiles(input.files.root) : 0;
+    return {
+        version: 1,
+        url,
+        title,
+        snapshotType: input.snapshotType ?? "unknown",
+        capturedAt: (input.capturedAt ?? new Date()).toISOString(),
+        generator: "pagepocket",
+        resources: { fileCount }
+    };
+};

package/dist/index.d.ts

@@ -0,0 +1,2 @@
+export { MetadataUnit, MetadataUnit as default } from "./metadata-unit.js";
+export type { PagePocketMetadata, BuildMetadataInput } from "./build-metadata.js";

package/dist/index.js

@@ -0,0 +1 @@
+export { MetadataUnit, MetadataUnit as default } from "./metadata-unit.js";

package/dist/metadata-unit.d.ts

@@ -0,0 +1,6 @@
+import { FileTreeUnit, type UnitContext, type UnitPatch, type UnitRuntime } from "@pagepocket/lib";
+export declare class MetadataUnit extends FileTreeUnit {
+    readonly id = "metadata";
+    readonly description = "Extracting metadata";
+    run(ctx: UnitContext, rt: UnitRuntime): Promise<UnitPatch>;
+}

package/dist/metadata-unit.js

@@ -0,0 +1,40 @@
+import { FileTreeUnit } from "@pagepocket/lib";
+import { buildMetadata } from "./build-metadata.js";
+const METADATA_PATH = "/.pagepocket.json";
+const injectFile = (root, path, data) => ({
+    ...root,
+    entries: [
+        ...root.entries,
+        { kind: "file", path, source: { kind: "bytes", data } }
+    ]
+});
+export class MetadataUnit extends FileTreeUnit {
+    constructor() {
+        super(...arguments);
+        this.id = "metadata";
+        this.description = "Extracting metadata";
+    }
+    async run(ctx, rt) {
+        const files = ctx.value.files;
+        const html = ctx.value.html;
+        const snapshotType = ctx.value.snapshotType;
+        const metadata = buildMetadata({
+            entry: rt.entry,
+            html,
+            files,
+            snapshotType
+        });
+        const bytes = new TextEncoder().encode(`${JSON.stringify(metadata, undefined, 2)}\n`);
+        if (!files?.root) {
+            const root = {
+                kind: "directory",
+                path: "",
+                entries: [{ kind: "file", path: METADATA_PATH, source: { kind: "bytes", data: bytes } }]
+            };
+            return { ...ctx.value, files: { root } };
+        }
+        const nextRoot = injectFile(files.root, METADATA_PATH, bytes);
+        const nextFiles = { ...files, root: nextRoot };
+        return { ...ctx.value, files: nextFiles };
+    }
+}

package/package.json

@@ -0,0 +1,23 @@
+{
+  "name": "@pagepocket/metadata-unit",
+  "version": "0.13.0",
+  "description": "PagePocket unit: emit .pagepocket.json metadata into the snapshot FileTree",
+  "type": "module",
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "files": [
+    "dist"
+  ],
+  "license": "ISC",
+  "dependencies": {
+    "@pagepocket/lib": "0.13.0"
+  },
+  "devDependencies": {
+    "tsx": "^4.19.3",
+    "typescript": "^5.4.5"
+  },
+  "scripts": {
+    "build": "tsc -p tsconfig.json",
+    "test": "tsx --test specs/*.spec.ts"
+  }
+}