@pagepocket/cli 0.13.0 → 0.14.5

package/dist/commands/archive.js

@@ -6,6 +6,7 @@ import { ConfigService } from "../services/config-service.js";
 import { loadConfiguredPlugins } from "../services/load-configured-plugins.js";
 import { StrategyService } from "../services/strategy/strategy-service.js";
 import { loadStrategyUnits, resolveStrategy } from "../utils/archive-strategy.js";
+import { parseEnvVars, resolveStrategyVars } from "../utils/strategy-vars.js";
 export default class ArchiveCommand extends Command {
     static description = "Archive a web page as an offline snapshot.";
     static args = {
@@ -18,35 +19,29 @@ export default class ArchiveCommand extends Command {
         help: Flags.help({
             char: "h"
         }),
-        timeout: Flags.integer({
-            char: "t",
-            description: "Network idle duration in milliseconds before capture stops",
-            required: false
-        }),
-        maxDuration: Flags.integer({
-            description: "Hard max capture duration in milliseconds",
-            required: false
-        }),
         strategy: Flags.string({
             description: "Run an installed or built-in strategy (unit pipeline) by name"
+        }),
+        env: Flags.string({
+            char: "e",
+            description: "Strategy variable in key:value format (e.g. -e \"timeout:3000\")",
+            multiple: true
         })
     };
     async run() {
         const { args, flags } = await this.parse(ArchiveCommand);
         const targetUrl = args.url;
-        const timeoutMs = typeof flags.timeout === "number" ? flags.timeout : undefined;
-        const maxDurationMs = typeof flags.maxDuration === "number" ? flags.maxDuration : undefined;
         const strategyName = typeof flags.strategy === "string" ? flags.strategy.trim() : undefined;
         const configService = new ConfigService();
         const name = strategyName && strategyName.length > 0 ? strategyName : "default";
         const strategyService = new StrategyService(configService);
         strategyService.ensureConfigFileExists();
         const strategy = resolveStrategy({ strategyName: name, strategyService });
-        const units = await loadStrategyUnits({ strategy, configService });
-        const strategyFile = strategy.strategyFile;
-        const captureOptions = strategyFile.pipeline.captureOptions;
-        const effectiveTimeoutMs = typeof timeoutMs === "number" ? timeoutMs : captureOptions?.timeoutMs;
-        const effectiveMaxDurationMs = typeof maxDurationMs === "number" ? maxDurationMs : captureOptions?.maxDurationMs;
+        const envVars = parseEnvVars(flags.env ?? []);
+        const resolvedStrategyFile = resolveStrategyVars(strategy.strategyFile, envVars);
+        const resolvedStrategy = { ...strategy, strategyFile: resolvedStrategyFile };
+        const units = await loadStrategyUnits({ strategy: resolvedStrategy, configService });
+        const captureOptions = resolvedStrategy.strategyFile.pipeline.captureOptions;
         const pagepocket = PagePocket.fromURL(targetUrl);
         let spinner = ora();
         let activeUnitId = "";
@@ -69,13 +64,15 @@ export default class ArchiveCommand extends Command {
         let result;
         try {
             result = await pagepocket.capture({
-                ...(typeof effectiveTimeoutMs === "number" ? { timeoutMs: effectiveTimeoutMs } : {}),
-                ...(typeof effectiveMaxDurationMs === "number"
-                    ? { maxDurationMs: effectiveMaxDurationMs }
+                ...(typeof captureOptions?.timeoutMs === "number"
+                    ? { timeoutMs: captureOptions.timeoutMs }
+                    : {}),
+                ...(typeof captureOptions?.maxDurationMs === "number"
+                    ? { maxDurationMs: captureOptions.maxDurationMs }
                     : {}),
                 blacklist: [...ga, ...ns],
                 units,
-                plugins: await loadConfiguredPlugins().catch(() => [])
+                plugins: await loadConfiguredPlugins()
             });
         }
         catch (error) {

package/dist/index.js

@@ -1,3 +1,4 @@
+#!/usr/bin/env node
 import { execute } from "@oclif/core";
 import { normalizeArgv } from "./utils/normalize-argv.js";
 await execute({

package/dist/utils/strategy-vars.js

@@ -0,0 +1,117 @@
+const VAR_PATTERN = /\{\{([a-zA-Z_][a-zA-Z0-9_]*)(?::([^}]*))?\}\}/g;
+/**
+ * Parse `-e` flag values into a typed variable map.
+ *
+ * Each raw arg must be in `key:value` format. The key is the variable name,
+ * everything after the first colon is the raw value string.
+ *
+ * Value type inference:
+ * - Parseable as finite number → number
+ * - `"true"` / `"false"` → boolean
+ * - Otherwise → string
+ *
+ * Usage:
+ *   parseEnvVars(["timeout:3000", "overwrite:true", "outDir:./snapshots"])
+ *   // => { timeout: 3000, overwrite: true, outDir: "./snapshots" }
+ */
+export const parseEnvVars = (rawArgs) => {
+    const vars = {};
+    for (const raw of rawArgs) {
+        const colonIndex = raw.indexOf(":");
+        if (colonIndex === -1) {
+            throw new Error(`Invalid -e value: "${raw}". Expected format: key:value (e.g. -e "timeout:3000")`);
+        }
+        const key = raw.slice(0, colonIndex).trim();
+        const valueRaw = raw.slice(colonIndex + 1);
+        if (key.length === 0) {
+            throw new Error(`Invalid -e value: "${raw}". Key must not be empty.`);
+        }
+        if (!/^[a-zA-Z_][a-zA-Z0-9_]*$/.test(key)) {
+            throw new Error(`Invalid variable name: "${key}". Must be a valid identifier (letters, digits, underscores).`);
+        }
+        if (key in vars) {
+            throw new Error(`Duplicate variable: "${key}". Each variable can only be specified once.`);
+        }
+        vars[key] = inferValue(valueRaw);
+    }
+    return vars;
+};
+const inferValue = (raw) => {
+    const asNumber = Number(raw);
+    if (raw.length > 0 && Number.isFinite(asNumber)) {
+        return asNumber;
+    }
+    if (raw === "true") {
+        return true;
+    }
+    if (raw === "false") {
+        return false;
+    }
+    return raw;
+};
+/**
+ * Recursively resolve `{{varName}}` and `{{varName:default}}` placeholders in a JSON structure.
+ *
+ * - Only string values are scanned for placeholders.
+ * - `{{var:default}}` provides an inline default; the default is type-inferred the same way as `-e` values.
+ * - If the entire string is a single `{{var}}` or `{{var:default}}`, the value is replaced with the
+ *   variable's typed value (number, boolean, or string).
+ * - If the string contains mixed text and placeholders (e.g. `"path/{{name}}"`),
+ *   all placeholders are interpolated and the result remains a string.
+ * - Throws if any referenced variable has no `-e` value AND no inline default.
+ *
+ * Usage:
+ *   resolveStrategyVars({ timeoutMs: "{{timeout:5000}}" }, {})
+ *   // => { timeoutMs: 5000 }
+ */
+export const resolveStrategyVars = (json, vars) => {
+    const defaults = new Map();
+    const referenced = new Set();
+    const resolved = resolveNode(json, vars, defaults, referenced);
+    const missing = [...referenced].filter((name) => !(name in vars) && !defaults.has(name));
+    if (missing.length > 0) {
+        throw new Error(`Missing strategy variable(s): ${missing.join(", ")}. ` +
+            `Pass them with ${missing.map((name) => `-e "${name}:<value>"`).join(" ")}`);
+    }
+    return resolved;
+};
+const resolveNode = (node, vars, defaults, referenced) => {
+    if (typeof node === "string") {
+        return resolveString(node, vars, defaults, referenced);
+    }
+    if (Array.isArray(node)) {
+        return node.map((item) => resolveNode(item, vars, defaults, referenced));
+    }
+    if (isRecord(node)) {
+        const result = {};
+        for (const [key, value] of Object.entries(node)) {
+            result[key] = resolveNode(value, vars, defaults, referenced);
+        }
+        return result;
+    }
+    return node;
+};
+const resolveString = (value, vars, defaults, referenced) => {
+    const matches = [...value.matchAll(VAR_PATTERN)];
+    if (matches.length === 0) {
+        return value;
+    }
+    for (const match of matches) {
+        const varName = match[1];
+        const rawDefault = match[2];
+        referenced.add(varName);
+        if (rawDefault !== undefined && !defaults.has(varName)) {
+            defaults.set(varName, inferValue(rawDefault));
+        }
+    }
+    const isSingleWholeVar = matches.length === 1 && matches[0][0] === value;
+    if (isSingleWholeVar) {
+        const varName = matches[0][1];
+        return vars[varName] ?? defaults.get(varName);
+    }
+    return value.replace(VAR_PATTERN, (_match, varName) => {
+        const resolved = vars[varName] ?? defaults.get(varName);
+        return String(resolved ?? "");
+    });
+};
+const isRecord = (value) => Boolean(value) && typeof value === "object" && !Array.isArray(value);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pagepocket/cli",
-  "version": "0.13.0",
+  "version": "0.14.5",
   "description": "CLI for capturing offline snapshots of web pages.",
   "type": "module",
   "main": "dist/index.js",
@@ -14,33 +14,35 @@
   "author": "",
   "license": "ISC",
   "dependencies": {
+    "@hono/node-server": "^1.14.1",
     "@oclif/core": "^4.0.9",
     "chalk": "^4.1.2",
     "env-paths": "^2.2.1",
-    "@hono/node-server": "^1.14.1",
     "hono": "^4.7.6",
     "npm-package-arg": "^13.0.2",
     "ora": "^9.0.0",
-    "@pagepocket/main-content-unit": "0.13.0",
-    "@pagepocket/metadata-unit": "0.13.0",
-    "@pagepocket/capture-http-lighterceptor-unit": "0.13.0",
-    "@pagepocket/build-snapshot-unit": "0.13.0",
-    "@pagepocket/capture-http-puppeteer-unit": "0.13.0",
-    "@pagepocket/capture-http-cdp-unit": "0.13.0",
-    "@pagepocket/plugin-yt-dlp": "0.13.0",
-    "@pagepocket/contracts": "0.13.0",
-    "@pagepocket/shared": "0.13.0",
-    "@pagepocket/builtin-strategy": "0.13.0",
-    "@pagepocket/lib": "0.13.0",
-    "@pagepocket/single-file-unit": "0.13.0",
-    "@pagepocket/write-down-unit": "0.13.0"
+    "@pagepocket/build-snapshot-unit": "0.14.5",
+    "@pagepocket/builtin-strategy": "0.14.5",
+    "@pagepocket/capture-http-lighterceptor-unit": "0.14.5",
+    "@pagepocket/capture-http-cdp-unit": "0.14.5",
+    "@pagepocket/capture-http-puppeteer-unit": "0.14.5",
+    "@pagepocket/lib": "0.14.5",
+    "@pagepocket/contracts": "0.14.5",
+    "@pagepocket/metadata-unit": "0.14.5",
+    "@pagepocket/plugin-yt-dlp": "0.14.5",
+    "@pagepocket/shared": "0.14.5",
+    "@pagepocket/main-content-unit": "0.14.5",
+    "@pagepocket/single-file-unit": "0.14.5",
+    "@pagepocket/write-down-unit": "0.14.5"
   },
   "devDependencies": {
     "@types/node": "^20.11.30",
+    "cpx2": "^8.0.0",
+    "esbuild": "^0.25.12",
     "rimraf": "^6.0.1",
     "tsx": "^4.19.3",
     "typescript": "^5.4.5",
-    "@pagepocket/content-reader": "0.13.0"
+    "@pagepocket/content-reader": "0.14.5"
   },
   "oclif": {
     "bin": "pp",
@@ -52,8 +54,9 @@
   },
   "scripts": {
     "build": "rimraf dist && tsc && pnpm build:vendor",
-    "build:vendor": "mkdir -p dist/vendor && cp node_modules/@pagepocket/content-reader/dist/browser/content-reader.iife.js node_modules/@pagepocket/content-reader/dist/browser/content-reader.css dist/vendor/",
+    "build:vendor": "cpx \"node_modules/@pagepocket/content-reader/dist/browser/*\" dist/vendor",
     "start": "node dist/index.js",
-    "test": "pnpm build && tsx --test specs/*.spec.ts"
+    "test": "pnpm build && tsx --test specs/*.spec.ts",
+    "build:sea": "node --import tsx scripts/build-sea.ts"
   }
 }