nfunc-mcp 0.3.0 → 0.5.0

package/dist/utils/psiAuth.js

@@ -0,0 +1,36 @@
+/**
+ * PSI API key resolution.
+ *
+ * Precedence: explicit tool input → PAGESPEED_API_KEY → keyless.
+ *
+ * The preferred setup is the `env` block in the MCP client config, because the
+ * client launches this server and we do not control its flags — `--env-file`
+ * is not available to us. An explicit input is supported for people who want
+ * to paste a key once, but it is the worst option: it lands in the
+ * conversation transcript and in any client-side logging, so it is never
+ * echoed back and every error that touches a request URL is redacted.
+ */
+export const KEY_ENV_VAR = "PAGESPEED_API_KEY";
+/**
+ * Keyless PSI is rate-limited hard enough that a batch will 429 partway
+ * through, leaving a half-finished audit and a confusing error. Refusing up
+ * front with instructions is a better failure than discovering it at run 12.
+ */
+export const KEYLESS_RUN_CAP = 4;
+export function resolveApiKey(explicit) {
+    const trimmed = explicit?.trim();
+    if (trimmed)
+        return { key: trimmed, source: "input" };
+    const fromEnv = process.env[KEY_ENV_VAR]?.trim();
+    if (fromEnv)
+        return { key: fromEnv, source: "env" };
+    return { key: null, source: "none" };
+}
+export function keylessWarning(runs) {
+    return (`No ${KEY_ENV_VAR} set — running unauthenticated against a shared, ` +
+        `heavily rate-limited quota. This is fine for ${KEYLESS_RUN_CAP} runs or ` +
+        `fewer; ${runs} runs will almost certainly hit HTTP 429 partway through. ` +
+        `Get a key from the Google Cloud console (enable the PageSpeed Insights ` +
+        `API) and set ${KEY_ENV_VAR} in your MCP client config.`);
+}
+//# sourceMappingURL=psiAuth.js.map

package/dist/utils/psiAuth.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"psiAuth.js","sourceRoot":"","sources":["../../src/utils/psiAuth.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AASH,MAAM,CAAC,MAAM,WAAW,GAAG,mBAAmB,CAAC;AAE/C;;;;GAIG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC;AAEjC,MAAM,UAAU,aAAa,CAAC,QAAiB;IAC7C,MAAM,OAAO,GAAG,QAAQ,EAAE,IAAI,EAAE,CAAC;IACjC,IAAI,OAAO;QAAE,OAAO,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IAEtD,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,WAAW,CAAC,EAAE,IAAI,EAAE,CAAC;IACjD,IAAI,OAAO;QAAE,OAAO,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;IAEpD,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC;AACvC,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,IAAY;IACzC,OAAO,CACL,MAAM,WAAW,mDAAmD;QACpE,gDAAgD,eAAe,WAAW;QAC1E,UAAU,IAAI,4DAA4D;QAC1E,yEAAyE;QACzE,gBAAgB,WAAW,6BAA6B,CACzD,CAAC;AACJ,CAAC"}

package/dist/utils/psiParser.d.ts

@@ -0,0 +1,135 @@
+/**
+ * PageSpeed Insights response parsing.
+ *
+ * One PSI call returns two unrelated datasets: `lighthouseResult` (a lab run
+ * on Google's hardware) and `loadingExperience` (CrUX field data from real
+ * Chrome users). They are measured differently, they disagree routinely, and
+ * conflating them produces confidently wrong reports — so they stay separate
+ * all the way through this module and only meet in the comparator.
+ *
+ * The lab half is the same LHR schema `parseLighthouseJSON` already handles
+ * and is delegated to it unchanged. Everything here is about the field half
+ * and about pulling numeric lab metrics that survive aggregation.
+ */
+import { type ParsedLighthouse } from "./outputParsers.js";
+export type CruxCategory = "FAST" | "AVERAGE" | "SLOW" | "NONE";
+/** The five field metrics PSI reports. `fid` is deliberately absent — see below. */
+export type WebVital = "lcp" | "inp" | "cls" | "fcp" | "ttfb";
+export type FieldSource = "url" | "origin";
+export interface CruxMetric {
+    /**
+     * 75th percentile across the collection window. Milliseconds for every
+     * metric except `cls`, which is unitless.
+     */
+    p75: number;
+    category: CruxCategory;
+    /** Share of real users in each bucket. Sums to ~1. */
+    distribution: {
+        good: number;
+        needsImprovement: number;
+        poor: number;
+    };
+    /**
+     * Whether this specific metric came from the URL or from origin-wide data.
+     * Per metric, not per response — see the note on `parseCrux`.
+     */
+    source: FieldSource;
+}
+export interface ParsedCrux {
+    /**
+     * Always 28. PSI does not return the collection window, but CrUX in PSI is
+     * defined as a trailing 28-day aggregate, so this is a documented constant
+     * rather than a parsed value. Stated explicitly because reports must not
+     * describe field data as real-time.
+     */
+    collectionPeriodDays: 28;
+    overallCategory: CruxCategory | null;
+    /** Which source the majority of metrics came from, for a one-line summary. */
+    primarySource: FieldSource;
+    metrics: Partial<Record<WebVital, CruxMetric>>;
+}
+export interface LabMetrics {
+    lcpMs: number | null;
+    fcpMs: number | null;
+    cls: number | null;
+    tbtMs: number | null;
+    speedIndexMs: number | null;
+    ttiMs: number | null;
+    ttfbMs: number | null;
+}
+export interface ParsedPsi {
+    requestedUrl: string;
+    finalUrl: string;
+    strategy: string;
+    fetchTime: string | null;
+    lighthouseVersion: string | null;
+    scores: Record<string, number>;
+    lab: LabMetrics;
+    /** Null when the URL has too little real-user traffic and origin fallback is off or empty. */
+    field: ParsedCrux | null;
+    lighthouse: ParsedLighthouse;
+    /** Lighthouse audits that failed, for the existing defect formatter. */
+    runWarnings: string[];
+}
+/** PSI returned a 200 whose Lighthouse run did not actually complete. */
+export declare class PsiRuntimeError extends Error {
+    readonly code: string;
+    constructor(code: string, message: string);
+}
+interface RawCruxMetric {
+    percentile?: number;
+    category?: string;
+    distributions?: Array<{
+        min?: number;
+        max?: number;
+        proportion?: number;
+    }>;
+}
+interface RawLoadingExperience {
+    id?: string;
+    overall_category?: string;
+    metrics?: Record<string, RawCruxMetric>;
+}
+/**
+ * Merge URL-level and origin-level CrUX into one metric set.
+ *
+ * Field availability is **per metric, not per URL** — a page can have enough
+ * traffic for CrUX to report CLS but not LCP. A real response in the Five
+ * Below batch carried two of the five metrics at URL level while the origin
+ * carried all five, which is why the manual audit shows "No field data" for
+ * one metric and a category for another on the same row.
+ *
+ * So the fallback runs per metric, and each metric records where it came from.
+ * A single response-level `field_source` flag would have to either discard the
+ * URL-level metrics that do exist or silently label origin data as page data,
+ * and origin CrUX for a homepage says nothing about a checkout page.
+ */
+export declare function parseCrux(urlLevel: RawLoadingExperience | undefined, originLevel: RawLoadingExperience | undefined, originFallback: boolean): ParsedCrux | null;
+/**
+ * Lab metrics as numbers.
+ *
+ * `numericValue`, never `displayValue`. The Five Below CSV recorded TTFB as
+ * "Root document took 930 ms" — a localised string containing a non-breaking
+ * space — because it read displayValue, which made the whole column
+ * unaggregatable while `numericValue: 929` sat in the same audit object.
+ * Formatting is a report-time concern; nothing upstream of the report should
+ * hold a number as text.
+ */
+export declare function extractLabMetrics(auditsJson: string): LabMetrics;
+export interface ParsePsiOptions {
+    strategy: string;
+    /** Fall back to origin-wide CrUX per metric when URL-level data is absent. */
+    originFallback?: boolean;
+}
+/**
+ * Parse a full PSI v5 response.
+ *
+ * Throws `PsiRuntimeError` when Lighthouse itself failed inside a 200
+ * response. PSI does this routinely — a page that times out or refuses the
+ * fetch comes back as HTTP 200 with `lighthouseResult.runtimeError` set and
+ * every score null. Parsing it anyway records a real page as scoring zero
+ * across the board, which is worse than a failed run, because a zero looks
+ * like data. Callers should treat this as retryable.
+ */
+export declare function parsePsiResponse(rawJson: string, options: ParsePsiOptions): ParsedPsi;
+export {};

package/dist/utils/psiParser.js

@@ -0,0 +1,200 @@
+/**
+ * PageSpeed Insights response parsing.
+ *
+ * One PSI call returns two unrelated datasets: `lighthouseResult` (a lab run
+ * on Google's hardware) and `loadingExperience` (CrUX field data from real
+ * Chrome users). They are measured differently, they disagree routinely, and
+ * conflating them produces confidently wrong reports — so they stay separate
+ * all the way through this module and only meet in the comparator.
+ *
+ * The lab half is the same LHR schema `parseLighthouseJSON` already handles
+ * and is delegated to it unchanged. Everything here is about the field half
+ * and about pulling numeric lab metrics that survive aggregation.
+ */
+import { parseLighthouseJSON } from "./outputParsers.js";
+/** PSI returned a 200 whose Lighthouse run did not actually complete. */
+export class PsiRuntimeError extends Error {
+    code;
+    constructor(code, message) {
+        super(message);
+        this.code = code;
+        this.name = "PsiRuntimeError";
+    }
+}
+/**
+ * CrUX metric key → our vital name.
+ *
+ * FIRST_INPUT_DELAY_MS is intentionally not mapped. FID was retired in March
+ * 2024 and replaced by INP; it still appears in some responses, and treating
+ * it as a current metric would report a defect against a standard Google no
+ * longer measures. Read it only if a caller ever needs historical data.
+ */
+const CRUX_KEYS = {
+    LARGEST_CONTENTFUL_PAINT_MS: "lcp",
+    INTERACTION_TO_NEXT_PAINT: "inp",
+    CUMULATIVE_LAYOUT_SHIFT_SCORE: "cls",
+    FIRST_CONTENTFUL_PAINT_MS: "fcp",
+    EXPERIMENTAL_TIME_TO_FIRST_BYTE: "ttfb",
+};
+/**
+ * CLS is the one field metric PSI returns scaled by 100: a `percentile` of 55
+ * means a CLS of 0.55, and the distribution bucket boundaries (0-10, 10-25)
+ * are scaled the same way. Comparing the raw integer against the 0.25 "poor"
+ * threshold marks every site on earth as catastrophic; forgetting the scaling
+ * entirely turns a 0.55 into a 55. Verified against real responses.
+ */
+function scaleCruxValue(vital, percentile) {
+    return vital === "cls" ? percentile / 100 : percentile;
+}
+function toCategory(raw) {
+    return raw === "FAST" || raw === "AVERAGE" || raw === "SLOW" ? raw : "NONE";
+}
+function toDistribution(raw) {
+    const share = (i) => {
+        const p = raw?.[i]?.proportion;
+        return typeof p === "number" ? Number(p.toFixed(4)) : 0;
+    };
+    // PSI always emits exactly three buckets, ordered good → poor.
+    return { good: share(0), needsImprovement: share(1), poor: share(2) };
+}
+function readMetric(raw, vital, source) {
+    if (!raw || typeof raw.percentile !== "number")
+        return null;
+    return {
+        p75: scaleCruxValue(vital, raw.percentile),
+        category: toCategory(raw.category),
+        distribution: toDistribution(raw.distributions),
+        source,
+    };
+}
+/**
+ * Merge URL-level and origin-level CrUX into one metric set.
+ *
+ * Field availability is **per metric, not per URL** — a page can have enough
+ * traffic for CrUX to report CLS but not LCP. A real response in the Five
+ * Below batch carried two of the five metrics at URL level while the origin
+ * carried all five, which is why the manual audit shows "No field data" for
+ * one metric and a category for another on the same row.
+ *
+ * So the fallback runs per metric, and each metric records where it came from.
+ * A single response-level `field_source` flag would have to either discard the
+ * URL-level metrics that do exist or silently label origin data as page data,
+ * and origin CrUX for a homepage says nothing about a checkout page.
+ */
+export function parseCrux(urlLevel, originLevel, originFallback) {
+    const metrics = {};
+    let fromUrl = 0;
+    let fromOrigin = 0;
+    /**
+     * PSI performs its own origin fallback, and does it silently.
+     *
+     * When a URL has too little traffic, `loadingExperience` comes back populated
+     * with **origin-level** numbers and the only signal is its `id`, which holds
+     * the origin instead of the URL. Trusting the block's position in the
+     * response therefore labels origin data as page data: two different paginated
+     * URLs came back with byte-identical p75 values, both marked `source: "url"`.
+     *
+     * Comparing the two blocks' ids catches it without needing the requested URL.
+     */
+    const psiSubstitutedOrigin = Boolean(urlLevel?.id) && Boolean(originLevel?.id) && urlLevel?.id === originLevel?.id;
+    const urlLevelSource = psiSubstitutedOrigin ? "origin" : "url";
+    for (const [cruxKey, vital] of Object.entries(CRUX_KEYS)) {
+        const atUrl = readMetric(urlLevel?.metrics?.[cruxKey], vital, urlLevelSource);
+        if (atUrl) {
+            metrics[vital] = atUrl;
+            if (urlLevelSource === "url")
+                fromUrl++;
+            else
+                fromOrigin++;
+            continue;
+        }
+        if (!originFallback)
+            continue;
+        const atOrigin = readMetric(originLevel?.metrics?.[cruxKey], vital, "origin");
+        if (atOrigin) {
+            metrics[vital] = atOrigin;
+            fromOrigin++;
+        }
+    }
+    if (fromUrl === 0 && fromOrigin === 0)
+        return null;
+    // Only trust the URL-level verdict when the URL actually supplied data;
+    // otherwise the origin's verdict is the one describing these numbers.
+    const overallRaw = fromUrl > 0 ? urlLevel?.overall_category : originLevel?.overall_category;
+    return {
+        collectionPeriodDays: 28,
+        overallCategory: overallRaw ? toCategory(overallRaw) : null,
+        primarySource: fromUrl >= fromOrigin ? "url" : "origin",
+        metrics,
+    };
+}
+/**
+ * Lab metrics as numbers.
+ *
+ * `numericValue`, never `displayValue`. The Five Below CSV recorded TTFB as
+ * "Root document took 930 ms" — a localised string containing a non-breaking
+ * space — because it read displayValue, which made the whole column
+ * unaggregatable while `numericValue: 929` sat in the same audit object.
+ * Formatting is a report-time concern; nothing upstream of the report should
+ * hold a number as text.
+ */
+export function extractLabMetrics(auditsJson) {
+    const audits = JSON.parse(auditsJson);
+    const num = (id) => {
+        const v = audits.audits?.[id]?.numericValue;
+        return typeof v === "number" ? v : null;
+    };
+    const ms = (id) => {
+        const v = num(id);
+        return v === null ? null : Math.round(v);
+    };
+    return {
+        lcpMs: ms("largest-contentful-paint"),
+        fcpMs: ms("first-contentful-paint"),
+        // CLS is unitless and small; rounding it to an integer would destroy it.
+        cls: num("cumulative-layout-shift"),
+        tbtMs: ms("total-blocking-time"),
+        speedIndexMs: ms("speed-index"),
+        ttiMs: ms("interactive"),
+        ttfbMs: ms("server-response-time"),
+    };
+}
+/**
+ * Parse a full PSI v5 response.
+ *
+ * Throws `PsiRuntimeError` when Lighthouse itself failed inside a 200
+ * response. PSI does this routinely — a page that times out or refuses the
+ * fetch comes back as HTTP 200 with `lighthouseResult.runtimeError` set and
+ * every score null. Parsing it anyway records a real page as scoring zero
+ * across the board, which is worse than a failed run, because a zero looks
+ * like data. Callers should treat this as retryable.
+ */
+export function parsePsiResponse(rawJson, options) {
+    const { strategy, originFallback = true } = options;
+    const response = JSON.parse(rawJson);
+    const lhr = response.lighthouseResult;
+    if (!lhr) {
+        throw new PsiRuntimeError("NO_LIGHTHOUSE_RESULT", "PSI response contained no lighthouseResult.");
+    }
+    if (lhr.runtimeError?.code) {
+        throw new PsiRuntimeError(lhr.runtimeError.code, lhr.runtimeError.message ?? "Lighthouse reported a runtime error.");
+    }
+    // parseLighthouseJSON takes the serialised LHR. Re-serialising the object we
+    // already hold is a little wasteful, but it keeps the Phase 2-13 parser and
+    // its priority mapping untouched, which is worth more than the milliseconds.
+    const lhrJson = JSON.stringify(lhr);
+    const lighthouse = parseLighthouseJSON(lhrJson);
+    return {
+        requestedUrl: lhr.requestedUrl ?? "",
+        finalUrl: lhr.finalDisplayedUrl ?? lhr.finalUrl ?? lhr.requestedUrl ?? "",
+        strategy,
+        fetchTime: lhr.fetchTime ?? null,
+        lighthouseVersion: lhr.lighthouseVersion ?? null,
+        scores: lighthouse.categoryScores,
+        lab: extractLabMetrics(lhrJson),
+        field: parseCrux(response.loadingExperience, response.originLoadingExperience, originFallback),
+        lighthouse,
+        runWarnings: lhr.runWarnings ?? [],
+    };
+}
+//# sourceMappingURL=psiParser.js.map

package/dist/utils/psiParser.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"psiParser.js","sourceRoot":"","sources":["../../src/utils/psiParser.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,mBAAmB,EAAyB,MAAM,oBAAoB,CAAC;AAgEhF,yEAAyE;AACzE,MAAM,OAAO,eAAgB,SAAQ,KAAK;IAE7B;IADX,YACW,IAAY,EACrB,OAAe;QAEf,KAAK,CAAC,OAAO,CAAC,CAAC;QAHN,SAAI,GAAJ,IAAI,CAAQ;QAIrB,IAAI,CAAC,IAAI,GAAG,iBAAiB,CAAC;IAChC,CAAC;CACF;AA+BD;;;;;;;GAOG;AACH,MAAM,SAAS,GAA6B;IAC1C,2BAA2B,EAAE,KAAK;IAClC,yBAAyB,EAAE,KAAK;IAChC,6BAA6B,EAAE,KAAK;IACpC,yBAAyB,EAAE,KAAK;IAChC,+BAA+B,EAAE,MAAM;CACxC,CAAC;AAEF;;;;;;GAMG;AACH,SAAS,cAAc,CAAC,KAAe,EAAE,UAAkB;IACzD,OAAO,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,UAAU,GAAG,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC;AACzD,CAAC;AAED,SAAS,UAAU,CAAC,GAAuB;IACzC,OAAO,GAAG,KAAK,MAAM,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC;AAC9E,CAAC;AAED,SAAS,cAAc,CACrB,GAAmC;IAEnC,MAAM,KAAK,GAAG,CAAC,CAAS,EAAU,EAAE;QAClC,MAAM,CAAC,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC;QAC/B,OAAO,OAAO,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC1D,CAAC,CAAC;IACF,+DAA+D;IAC/D,OAAO,EAAE,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;AACxE,CAAC;AAED,SAAS,UAAU,CACjB,GAA8B,EAC9B,KAAe,EACf,MAAmB;IAEnB,IAAI,CAAC,GAAG,IAAI,OAAO,GAAG,CAAC,UAAU,KAAK,QAAQ;QAAE,OAAO,IAAI,CAAC;IAC5D,OAAO;QACL,GAAG,EAAE,cAAc,CAAC,KAAK,EAAE,GAAG,CAAC,UAAU,CAAC;QAC1C,QAAQ,EAAE,UAAU,CAAC,GAAG,CAAC,QAAQ,CAAC;QAClC,YAAY,EAAE,cAAc,CAAC,GAAG,CAAC,aAAa,CAAC;QAC/C,MAAM;KACP,CAAC;AACJ,CAAC;AAED;;;;;;;;;;;;;GAaG;AACH,MAAM,UAAU,SAAS,CACvB,QAA0C,EAC1C,WAA6C,EAC7C,cAAuB;IAEvB,MAAM,OAAO,GAA0C,EAAE,CAAC;IAC1D,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB;;;;;;;;;;OAUG;IACH,MAAM,oBAAoB,GACxB,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,IAAI,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,IAAI,QAAQ,EAAE,EAAE,KAAK,WAAW,EAAE,EAAE,CAAC;IACxF,MAAM,cAAc,GAAgB,oBAAoB,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC;IAE5E,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,CAAC;QACzD,MAAM,KAAK,GAAG,UAAU,CAAC,QAAQ,EAAE,OAAO,EAAE,CAAC,OAAO,CAAC,EAAE,KAAK,EAAE,cAAc,CAAC,CAAC;QAC9E,IAAI,KAAK,EAAE,CAAC;YACV,OAAO,CAAC,KAAK,CAAC,GAAG,KAAK,CAAC;YACvB,IAAI,cAAc,KAAK,KAAK;gBAAE,OAAO,EAAE,CAAC;;gBACnC,UAAU,EAAE,CAAC;YAClB,SAAS;QACX,CAAC;QACD,IAAI,CAAC,cAAc;YAAE,SAAS;QAC9B,MAAM,QAAQ,GAAG,UAAU,CAAC,WAAW,EAAE,OAAO,EAAE,CAAC,OAAO,CAAC,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAC;QAC9E,IAAI,QAAQ,EAAE,CAAC;YACb,OAAO,CAAC,KAAK,CAAC,GAAG,QAAQ,CAAC;YAC1B,UAAU,EAAE,CAAC;QACf,CAAC;IACH,CAAC;IAED,IAAI,OAAO,KAAK,CAAC,IAAI,UAAU,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAEnD,wEAAwE;IACxE,sEAAsE;IACtE,MAAM,UAAU,GACd,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,gBAAgB,CAAC,CAAC,CAAC,WAAW,EAAE,gBAAgB,CAAC;IAE3E,OAAO;QACL,oBAAoB,EAAE,EAAE;QACxB,eAAe,EAAE,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI;QAC3D,aAAa,EAAE,OAAO,IAAI,UAAU,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ;QACvD,OAAO;KACR,CAAC;AACJ,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,iBAAiB,CAAC,UAAkB;IAClD,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAEnC,CAAC;IACF,MAAM,GAAG,GAAG,CAAC,EAAU,EAAiB,EAAE;QACxC,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC,EAAE,YAAY,CAAC;QAC5C,OAAO,OAAO,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC1C,CAAC,CAAC;IACF,MAAM,EAAE,GAAG,CAAC,EAAU,EAAiB,EAAE;QACvC,MAAM,CAAC,GAAG,GAAG,CAAC,EAAE,CAAC,CAAC;QAClB,OAAO,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IAC3C,CAAC,CAAC;IAEF,OAAO;QACL,KAAK,EAAE,EAAE,CAAC,0BAA0B,CAAC;QACrC,KAAK,EAAE,EAAE,CAAC,wBAAwB,CAAC;QACnC,yEAAyE;QACzE,GAAG,EAAE,GAAG,CAAC,yBAAyB,CAAC;QACnC,KAAK,EAAE,EAAE,CAAC,qBAAqB,CAAC;QAChC,YAAY,EAAE,EAAE,CAAC,aAAa,CAAC;QAC/B,KAAK,EAAE,EAAE,CAAC,aAAa,CAAC;QACxB,MAAM,EAAE,EAAE,CAAC,sBAAsB,CAAC;KACnC,CAAC;AACJ,CAAC;AAQD;;;;;;;;;GASG;AACH,MAAM,UAAU,gBAAgB,CAC9B,OAAe,EACf,OAAwB;IAExB,MAAM,EAAE,QAAQ,EAAE,cAAc,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IACpD,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAmB,CAAC;IACvD,MAAM,GAAG,GAAG,QAAQ,CAAC,gBAAgB,CAAC;IAEtC,IAAI,CAAC,GAAG,EAAE,CAAC;QACT,MAAM,IAAI,eAAe,CACvB,sBAAsB,EACtB,6CAA6C,CAC9C,CAAC;IACJ,CAAC;IACD,IAAI,GAAG,CAAC,YAAY,EAAE,IAAI,EAAE,CAAC;QAC3B,MAAM,IAAI,eAAe,CACvB,GAAG,CAAC,YAAY,CAAC,IAAI,EACrB,GAAG,CAAC,YAAY,CAAC,OAAO,IAAI,sCAAsC,CACnE,CAAC;IACJ,CAAC;IAED,6EAA6E;IAC7E,4EAA4E;IAC5E,6EAA6E;IAC7E,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;IACpC,MAAM,UAAU,GAAG,mBAAmB,CAAC,OAAO,CAAC,CAAC;IAEhD,OAAO;QACL,YAAY,EAAE,GAAG,CAAC,YAAY,IAAI,EAAE;QACpC,QAAQ,EAAE,GAAG,CAAC,iBAAiB,IAAI,GAAG,CAAC,QAAQ,IAAI,GAAG,CAAC,YAAY,IAAI,EAAE;QACzE,QAAQ;QACR,SAAS,EAAE,GAAG,CAAC,SAAS,IAAI,IAAI;QAChC,iBAAiB,EAAE,GAAG,CAAC,iBAAiB,IAAI,IAAI;QAChD,MAAM,EAAE,UAAU,CAAC,cAAc;QACjC,GAAG,EAAE,iBAAiB,CAAC,OAAO,CAAC;QAC/B,KAAK,EAAE,SAAS,CACd,QAAQ,CAAC,iBAAiB,EAC1B,QAAQ,CAAC,uBAAuB,EAChC,cAAc,CACf;QACD,UAAU;QACV,WAAW,EAAE,GAAG,CAAC,WAAW,IAAI,EAAE;KACnC,CAAC;AACJ,CAAC"}

package/dist/utils/publicUrl.d.ts

@@ -0,0 +1,17 @@
+/**
+ * Whether PSI can reach a URL at all.
+ *
+ * PSI fetches the target from Google's own infrastructure, so "can I open this
+ * in my browser" is not the test — the page has to be reachable from the
+ * public internet. Localhost and RFC1918 addresses are not degraded cases that
+ * return partial data; they cannot be audited, full stop, and the only useful
+ * response is to say so before spending a request finding out.
+ */
+export type ReachVerdict = "public" | "loopback" | "private" | "non_http" | "malformed";
+export interface ReachResult {
+    verdict: ReachVerdict;
+    auditable: boolean;
+    reason?: string;
+}
+export declare function checkPublicReachability(rawUrl: string): ReachResult;
+export declare function isSessionGated(rawUrl: string): boolean;

package/dist/utils/publicUrl.js

@@ -0,0 +1,115 @@
+/**
+ * Whether PSI can reach a URL at all.
+ *
+ * PSI fetches the target from Google's own infrastructure, so "can I open this
+ * in my browser" is not the test — the page has to be reachable from the
+ * public internet. Localhost and RFC1918 addresses are not degraded cases that
+ * return partial data; they cannot be audited, full stop, and the only useful
+ * response is to say so before spending a request finding out.
+ */
+/** IPv4 in a private or link-local range, per RFC1918 / RFC3927. */
+function isPrivateIPv4(host) {
+    const parts = host.split(".");
+    if (parts.length !== 4)
+        return false;
+    const [a, b] = parts.map(Number);
+    if (parts.some((p) => !/^\d+$/.test(p)) || [a, b].some(Number.isNaN))
+        return false;
+    if (a === 10)
+        return true;
+    if (a === 172 && b >= 16 && b <= 31)
+        return true;
+    if (a === 192 && b === 168)
+        return true;
+    if (a === 169 && b === 254)
+        return true; // link-local
+    if (a === 100 && b >= 64 && b <= 127)
+        return true; // CGNAT
+    return false;
+}
+function isLoopback(host) {
+    if (host === "localhost" || host.endsWith(".localhost"))
+        return true;
+    if (host === "::1" || host === "[::1]")
+        return true;
+    return /^127\./.test(host);
+}
+/**
+ * Hostnames that never resolve on the public internet. `.local` is mDNS,
+ * `.internal` and `.test`/`.invalid` are reserved, and a bare hostname with no
+ * dot is a LAN name.
+ */
+function isNonPublicName(host) {
+    if (/\.(local|internal|localdomain|test|invalid|example)$/.test(host))
+        return true;
+    return !host.includes(".") && !/^\d+$/.test(host);
+}
+export function checkPublicReachability(rawUrl) {
+    let url;
+    try {
+        url = new URL(rawUrl);
+    }
+    catch {
+        return { verdict: "malformed", auditable: false, reason: `"${rawUrl}" is not a valid URL.` };
+    }
+    if (url.protocol !== "http:" && url.protocol !== "https:") {
+        return {
+            verdict: "non_http",
+            auditable: false,
+            reason: `PSI only audits http(s) URLs; got "${url.protocol}".`,
+        };
+    }
+    const host = url.hostname.toLowerCase();
+    if (isLoopback(host)) {
+        return {
+            verdict: "loopback",
+            auditable: false,
+            reason: "PSI fetches pages from Google's infrastructure, so it cannot reach " +
+                "localhost. Use run_lighthouse, which runs Chrome on this machine.",
+        };
+    }
+    if (isPrivateIPv4(host)) {
+        return {
+            verdict: "private",
+            auditable: false,
+            reason: `${host} is a private-network address and is not reachable from ` +
+                "Google's infrastructure. Use run_lighthouse for internal hosts.",
+        };
+    }
+    if (isNonPublicName(host)) {
+        return {
+            verdict: "private",
+            auditable: false,
+            reason: `"${host}" is not a publicly resolvable hostname. Use run_lighthouse ` +
+                "for hosts that only resolve on your network.",
+        };
+    }
+    return { verdict: "public", auditable: true };
+}
+/**
+ * Paths that PSI can fetch but cannot meaningfully audit, because an anonymous
+ * request does not see the real page.
+ *
+ * PSI has no session: it audits an empty cart, a redirect to a login form, or
+ * a search page with no query, and reports the result as if it were the page
+ * someone asked about. A confidently wrong number is worse than a gap, so
+ * these are flagged rather than run. `run_lighthouse` can carry cookies and is
+ * the right tool for them.
+ */
+const SESSION_GATED = [
+    /(^|\/)(cart|basket|bag)(\/|$)/,
+    /(^|\/)(checkout|payment|order-confirmation)(\/|$)/,
+    /(^|\/)(account|my-account|profile|dashboard|orders)(\/|$)/,
+    /(^|\/)(login|signin|sign-in|register|signup|sign-up|logout)(\/|$)/,
+    /(^|\/)(wishlist|favorites|favourites|saved)(\/|$)/,
+];
+export function isSessionGated(rawUrl) {
+    try {
+        const path = new URL(rawUrl).pathname.toLowerCase();
+        return SESSION_GATED.some((re) => re.test(path));
+    }
+    catch {
+        return false;
+    }
+}
+//# sourceMappingURL=publicUrl.js.map

package/dist/utils/publicUrl.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"publicUrl.js","sourceRoot":"","sources":["../../src/utils/publicUrl.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAUH,oEAAoE;AACpE,SAAS,aAAa,CAAC,IAAY;IACjC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC9B,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IACrC,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IACjC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IACnF,IAAI,CAAC,KAAK,EAAE;QAAE,OAAO,IAAI,CAAC;IAC1B,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE;QAAE,OAAO,IAAI,CAAC;IACjD,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,GAAG;QAAE,OAAO,IAAI,CAAC;IACxC,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,GAAG;QAAE,OAAO,IAAI,CAAC,CAAC,aAAa;IACtD,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,GAAG;QAAE,OAAO,IAAI,CAAC,CAAC,QAAQ;IAC3D,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,UAAU,CAAC,IAAY;IAC9B,IAAI,IAAI,KAAK,WAAW,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC;QAAE,OAAO,IAAI,CAAC;IACrE,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,OAAO;QAAE,OAAO,IAAI,CAAC;IACpD,OAAO,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC7B,CAAC;AAED;;;;GAIG;AACH,SAAS,eAAe,CAAC,IAAY;IACnC,IAAI,sDAAsD,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC;IACnF,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACpD,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,MAAc;IACpD,IAAI,GAAQ,CAAC;IACb,IAAI,CAAC;QACH,GAAG,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC;IACxB,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,SAAS,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,MAAM,uBAAuB,EAAE,CAAC;IAC/F,CAAC;IAED,IAAI,GAAG,CAAC,QAAQ,KAAK,OAAO,IAAI,GAAG,CAAC,QAAQ,KAAK,QAAQ,EAAE,CAAC;QAC1D,OAAO;YACL,OAAO,EAAE,UAAU;YACnB,SAAS,EAAE,KAAK;YAChB,MAAM,EAAE,sCAAsC,GAAG,CAAC,QAAQ,IAAI;SAC/D,CAAC;IACJ,CAAC;IAED,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;IAExC,IAAI,UAAU,CAAC,IAAI,CAAC,EAAE,CAAC;QACrB,OAAO;YACL,OAAO,EAAE,UAAU;YACnB,SAAS,EAAE,KAAK;YAChB,MAAM,EACJ,qEAAqE;gBACrE,mEAAmE;SACtE,CAAC;IACJ,CAAC;IACD,IAAI,aAAa,CAAC,IAAI,CAAC,EAAE,CAAC;QACxB,OAAO;YACL,OAAO,EAAE,SAAS;YAClB,SAAS,EAAE,KAAK;YAChB,MAAM,EACJ,GAAG,IAAI,0DAA0D;gBACjE,iEAAiE;SACpE,CAAC;IACJ,CAAC;IACD,IAAI,eAAe,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1B,OAAO;YACL,OAAO,EAAE,SAAS;YAClB,SAAS,EAAE,KAAK;YAChB,MAAM,EACJ,IAAI,IAAI,8DAA8D;gBACtE,8CAA8C;SACjD,CAAC;IACJ,CAAC;IAED,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC;AAChD,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,aAAa,GAAG;IACpB,+BAA+B;IAC/B,mDAAmD;IACnD,2DAA2D;IAC3D,mEAAmE;IACnE,mDAAmD;CACpD,CAAC;AAEF,MAAM,UAAU,cAAc,CAAC,MAAc;IAC3C,IAAI,CAAC;QACH,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;QACpD,OAAO,aAAa,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IACnD,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,KAAK,CAAC;IACf,CAAC;AACH,CAAC"}

package/dist/utils/sitemapReader.d.ts

@@ -0,0 +1,27 @@
+/**
+ * Sitemap discovery.
+ *
+ * The cheapest honest way to learn which pages a site has. A sitemap is
+ * authoritative (the site published it), instant (one or two requests), and
+ * needs no crawling, no robots.txt politeness budget and no HTML parsing. A
+ * crawler is the fallback for sites without one, and is deliberately not built
+ * yet — everything downstream works without it.
+ *
+ * Sitemaps are parsed with regular expressions rather than an XML library.
+ * That is normally a mistake, but a sitemap is a machine-generated document
+ * with a fixed two-element vocabulary, and the alternative is a dependency
+ * carried by every install of this server for one code path. The parser reads
+ * <loc> only and ignores everything else, so malformed markup degrades to
+ * fewer URLs rather than to wrong ones.
+ */
+export interface SitemapResult {
+    urls: string[];
+    /** Every sitemap document actually fetched, in order. */
+    sitemapsRead: string[];
+    /** True when a cap stopped the walk early — the URL list is a subset. */
+    truncated: boolean;
+    /** Every location tried, so a failure can say what was ruled out. */
+    attempted: string[];
+    warnings: string[];
+}
+export declare function readSitemap(origin: string): Promise<SitemapResult>;