@pagepocket/cli 0.4.0 → 0.4.2

package/README.md

@@ -0,0 +1,40 @@
+# @pagepocket/cli
+
+CLI for capturing offline snapshots of web pages. It fetches HTML, records network
+responses, downloads assets, rewrites links to local files, and injects a replay
+script so the snapshot works offline.
+
+## Install
+
+```bash
+npm i -g @pagepocket/cli
+```
+
+## Usage
+
+```bash
+pp https://example.com
+pp https://example.com -o ./snapshots
+```
+
+## Output
+
+Snapshots are written to the current directory by default:
+
+- `*.html`: offline snapshot page
+- `*.requests.json`: recorded requests/responses
+- `*_files/`: downloaded assets
+
+## Configuration
+
+Environment variables:
+
+- `PAGEPOCKET_FETCH_TIMEOUT_MS` (default: `60000`)
+- `PAGEPOCKET_FETCH_HEADERS` (JSON string of extra headers)
+
+## Development
+
+```bash
+pnpm install
+pnpm --filter @pagepocket/cli build
+```

package/dist/cli.js

@@ -5,14 +5,13 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 Object.defineProperty(exports, "__esModule", { value: true });
 const node_path_1 = __importDefault(require("node:path"));
 const core_1 = require("@oclif/core");
-const chalk_1 = __importDefault(require("chalk"));
 const lib_1 = require("@pagepocket/lib");
-const with_spinner_1 = require("./utils/with-spinner");
-const build_snapshot_data_1 = require("./stages/build-snapshot-data");
-const capture_network_1 = require("./stages/capture-network");
+const lighterceptor_adapter_1 = require("@pagepocket/lighterceptor-adapter");
+const chalk_1 = __importDefault(require("chalk"));
 const fetch_html_1 = require("./stages/fetch-html");
 const prepare_output_1 = require("./stages/prepare-output");
 const write_snapshot_1 = require("./stages/write-snapshot");
+const with_spinner_1 = require("./utils/with-spinner");
 class PagepocketCommand extends core_1.Command {
     async run() {
         const { args, flags } = await this.parse(PagepocketCommand);
@@ -39,48 +38,32 @@ class PagepocketCommand extends core_1.Command {
                 throw new Error("Invalid PAGEPOCKET_FETCH_HEADERS JSON.");
             }
         })();
-        const fetchXhrRecords = [];
         const fetched = await (0, with_spinner_1.withSpinner)(async () => (0, fetch_html_1.fetchHtml)(targetUrl, fetchTimeoutMs, headersOverride), "Fetching the target HTML");
-        const networkStage = await (async () => {
-            try {
-                return await (0, with_spinner_1.withSpinner)(async () => (0, capture_network_1.captureNetwork)(targetUrl, fetched.title), "Capturing network requests with lighterceptor");
-            }
-            catch {
-                return {
-                    networkRecords: [],
-                    lighterceptorNetworkRecords: [],
-                    capturedTitle: undefined,
-                    title: fetched.title
-                };
-            }
-        })();
-        const outputPaths = await (0, with_spinner_1.withSpinner)(async () => (0, prepare_output_1.prepareOutputPaths)(networkStage.title, outputFlag), "Preparing output paths");
-        const downloadStage = await (0, with_spinner_1.withSpinner)(async () => {
+        const { outputPaths, snapshotData, snapshotHtml } = await (0, with_spinner_1.withSpinner)(async () => {
             const originalCwd = process.cwd();
+            const interceptor = new lighterceptor_adapter_1.LighterceptorAdapter({ title: fetched.title });
+            const snapshotSeed = await interceptor.run(targetUrl);
+            const outputPaths = await (0, prepare_output_1.prepareOutputPaths)(snapshotSeed.title, outputFlag);
             const shouldRestoreCwd = outputPaths.baseDir !== originalCwd;
             try {
                 if (shouldRestoreCwd) {
                     process.chdir(outputPaths.baseDir);
                 }
-                const seedSnapshot = {
-                    url: targetUrl,
-                    title: networkStage.title,
-                    capturedAt: new Date().toISOString(),
-                    fetchXhrRecords,
-                    networkRecords: networkStage.lighterceptorNetworkRecords,
-                    resources: []
-                };
-                const pagepocket = new lib_1.PagePocket(fetched.html, seedSnapshot, {
-                    assetsDirName: outputPaths.assetsDirName,
+                const pagepocket = new lib_1.PagePocket(fetched.html, snapshotSeed, {
                     baseUrl: targetUrl,
+                    assetsDirName: outputPaths.assetsDirName,
                     requestsPath: node_path_1.default.basename(outputPaths.outputRequestsPath)
                 });
-                const snapshotHtml = await pagepocket.put();
+                const pageData = await pagepocket.put();
+                const snapshotData = {
+                    ...snapshotSeed,
+                    title: pageData.title,
+                    resources: pagepocket.resources
+                };
                 return {
-                    snapshotHtml,
-                    resourceMeta: pagepocket.resources,
-                    downloadedCount: pagepocket.downloadedCount,
-                    failedCount: pagepocket.failedCount
+                    outputPaths,
+                    snapshotData,
+                    snapshotHtml: pageData.content
                 };
             }
             finally {
@@ -94,19 +77,12 @@ class PagepocketCommand extends core_1.Command {
                 }
             }
         }, "Downloading resources");
-        const snapshotData = await (0, with_spinner_1.withSpinner)(async () => (0, build_snapshot_data_1.buildSnapshotData)({
-            targetUrl,
-            title: networkStage.title,
-            fetchXhrRecords,
-            lighterceptorNetworkRecords: networkStage.lighterceptorNetworkRecords,
-            resources: downloadStage.resourceMeta
-        }), "Preparing snapshot HTML");
         await (0, with_spinner_1.withSpinner)(async () => {
             await (0, write_snapshot_1.writeSnapshotFiles)({
                 outputRequestsPath: outputPaths.outputRequestsPath,
                 outputHtmlPath: outputPaths.outputHtmlPath,
                 snapshotData,
-                snapshotHtml: downloadStage.snapshotHtml
+                snapshotHtml: snapshotHtml
             });
         }, "Writing snapshot files");
         this.log(chalk_1.default.green("All done! Snapshot created."));

package/dist/stages/build-snapshot-data.js

@@ -7,7 +7,7 @@ const buildSnapshotData = (input) => {
         title: input.title,
         capturedAt: new Date().toISOString(),
         fetchXhrRecords: input.fetchXhrRecords,
-        networkRecords: input.lighterceptorNetworkRecords,
+        networkRecords: input.capturedNetworkRecords,
         resources: input.resources
     };
 };

package/dist/stages/capture-network.js

@@ -1,17 +1,17 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.captureNetwork = void 0;
-const lighterceptor_1 = require("lighterceptor");
 const lib_1 = require("@pagepocket/lib");
+const lighterceptor_1 = require("@pagepocket/lighterceptor");
 const captureNetwork = async (targetUrl, currentTitle) => {
     const result = await new lighterceptor_1.Lighterceptor(targetUrl, { recursion: true }).run();
-    const lighterceptorNetworkRecords = (result.networkRecords ?? []);
-    const networkRecords = (0, lib_1.mapLighterceptorRecords)(lighterceptorNetworkRecords);
+    const capturedNetworkRecords = (result.networkRecords ?? []);
+    const networkRecords = (0, lib_1.mapCapturedNetworkRecords)(capturedNetworkRecords);
     const capturedTitle = result.title;
     const title = currentTitle === "snapshot" && capturedTitle ? capturedTitle : currentTitle;
     return {
         networkRecords,
-        lighterceptorNetworkRecords,
+        capturedNetworkRecords,
         capturedTitle,
         title
     };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@pagepocket/cli",
-  "version": "0.4.0",
-  "description": "",
+  "version": "0.4.2",
+  "description": "CLI for capturing offline snapshots of web pages.",
   "main": "dist/index.js",
   "bin": {
     "pp": "dist/index.js"
@@ -14,16 +14,16 @@
   "license": "ISC",
   "dependencies": {
     "@oclif/core": "^4.0.9",
-    "cheerio": "^1.0.0-rc.12",
     "chalk": "^4.1.2",
+    "cheerio": "^1.0.0-rc.12",
     "got": "^11.8.6",
     "ora": "^9.0.0",
-    "@pagepocket/lib": "0.4.0",
-    "lighterceptor": "npm:@pagepocket/lighterceptor@0.4.0"
+    "@pagepocket/lighterceptor": "0.4.2",
+    "@pagepocket/lib": "0.4.2",
+    "@pagepocket/lighterceptor-adapter": "0.4.2"
   },
   "devDependencies": {
     "@types/node": "^20.11.30",
-    "prettier": "^3.7.4",
     "tsx": "^4.19.3",
     "typescript": "^5.4.5"
   },
@@ -36,8 +36,6 @@
   },
   "scripts": {
     "build": "tsc",
-    "format": "prettier --write .",
-    "format:check": "prettier --check .",
     "start": "node dist/index.js",
     "test": "tsx --test specs/**/*.test.ts"
   }