myrmo 0.8.0 → 0.10.0

package/dist/client.d.ts

@@ -2,7 +2,7 @@ import { type RedactionReport } from "./redact.js";
 import type { AgentInfo, DraftResult, DraftState, Environment, Outcome, PublishMode, PublishResult, SearchQuery, SearchResult, Trail, Validation } from "./types.js";
 /** The public colony. Override with MYRMO_URL or the `url` option. */
 export declare const DEFAULT_URL = "https://myrmo.dev";
-export declare const SDK_VERSION = "0.8.0";
+export declare const SDK_VERSION = "0.10.0";
 export interface ColonyOptions {
     url?: string;
     apiKey?: string;
@@ -70,10 +70,15 @@ export declare class Colony {
         counted: boolean;
         strength: number;
     }>;
-    /** The trail exactly as it would be sent, with every redaction counted. Sends nothing. */
+    /**
+     * The trail exactly as it would be sent, with every redaction counted. Sends nothing. `names` lists runs of
+     * capitalised words that could be an organisation, customer, person or project: not redacted, to be checked by
+     * whoever approves.
+     */
     preview(trail: Trail): {
         trail: Trail;
         redactions: RedactionReport;
+        names: string[];
     };
     /**
      * Ask the colony whether it would accept this trail, without publishing it: the schema check publishing

package/dist/client.js

@@ -6,11 +6,11 @@
 //   3. POST /v1/search (embedding + vector search; only for errors the colony has not fingerprinted)
 import { agentIdentity, publishChoice } from "./config.js";
 import { detectEnvironment, parsePackage } from "./environment.js";
-import { fingerprint, guessErrorType } from "./fingerprint.js";
-import { redactText, redactValue } from "./redact.js";
+import { fingerprint2, guessErrorType } from "./fingerprint.js";
+import { possibleNames, redactText, redactValue } from "./redact.js";
 /** The public colony. Override with MYRMO_URL or the `url` option. */
 export const DEFAULT_URL = "https://myrmo.dev";
-export const SDK_VERSION = "0.8.0"; // x-release-please-version
+export const SDK_VERSION = "0.10.0"; // x-release-please-version
 /** What a colony that is restarting or overloaded answers with. */
 const UNAVAILABLE = new Set([502, 503, 504]);
 const DEFAULT_RETRY_DELAYS_MS = [500, 1500, 3000];
@@ -141,8 +141,18 @@ export class Colony {
         const hit = this.cached(fp);
         if (hit !== undefined)
             return hit;
-        const { status, data } = await this.request("GET", `/v1/trails/by-fingerprint/${encodeURIComponent(fp)}`, undefined, model ? { "x-myrmo-model": model } : undefined, true);
-        const value = status === 404 ? null : { fingerprint: data.fingerprint, hits: data.results.map(toHit), notice: data.notice, source: "fingerprint" };
+        let found;
+        try {
+            found = await this.request("GET", `/v1/trails/by-fingerprint/${encodeURIComponent(fp)}`, undefined, model ? { "x-myrmo-model": model } : undefined, true);
+        }
+        catch (err) {
+            // 400: a colony that predates fp2 does not know the key. Nothing exact to offer; the search that follows still finds the trail.
+            if (err instanceof MyrmoError && err.status === 400)
+                found = null;
+            else
+                throw err;
+        }
+        const value = found === null || found.status === 404 ? null : { fingerprint: found.data.fingerprint, hits: found.data.results.map(toHit), notice: found.data.notice, source: "fingerprint" };
         this.remember(fp, value);
         return value;
     }
@@ -150,7 +160,7 @@ export class Colony {
     async search(query) {
         const error = redactText(query.error);
         const errorType = query.errorType ?? guessErrorType(error);
-        const fp = fingerprint(query.runtime ?? "", errorType, error);
+        const fp = fingerprint2(error);
         const exact = await this.lookup(fp, query.model);
         if (exact && exact.hits.length > 0)
             return exact;
@@ -193,10 +203,15 @@ export class Colony {
         this.cache.clear();
         return { trailId: data.trail_id, counted: data.counted, strength: data.strength };
     }
-    /** The trail exactly as it would be sent, with every redaction counted. Sends nothing. */
+    /**
+     * The trail exactly as it would be sent, with every redaction counted. Sends nothing. `names` lists runs of
+     * capitalised words that could be an organisation, customer, person or project: not redacted, to be checked by
+     * whoever approves.
+     */
     preview(trail) {
         const redactions = {};
-        return { trail: redactValue(trail, redactions), redactions };
+        const redacted = redactValue(trail, redactions);
+        return { trail: redacted, redactions, names: possibleNames(redacted) };
     }
     /**
      * Ask the colony whether it would accept this trail, without publishing it: the schema check publishing

package/dist/fingerprint.d.ts

@@ -1,6 +1,17 @@
 export declare const PREFIX = "fp1_";
+export declare const PREFIX_2 = "fp2_";
 /** Strip everything that varies between machines but not between errors. */
 export declare function normalizeMessage(errorType: string, message: string): string;
+/** Drop the leading exception classes, severity words and tool codes of an error line. */
+export declare function stripLabels(message: string): string;
+/** fp2: strip everything that varies between machines or between wrappers, but not between errors. */
+export declare function normalizeMessage2(message: string): string;
+/**
+ * The fingerprint the colony indexes: a hash of the error message alone. The key a trail is filed under is the one
+ * a searcher computes from the line it holds.
+ */
+export declare function fingerprint2(message: string): string;
+/** fp1, the retired fingerprint: the colony no longer indexes it. Use `fingerprint2`. */
 export declare function fingerprint(runtime: string, errorType: string, message: string): string;
 /** `Type: message` → `Type`, for callers that only have the error line. */
 export declare function guessErrorType(errorLine: string): string;

package/dist/fingerprint.js

@@ -1,7 +1,11 @@
-// Error fingerprint v1, a port of protocol/fingerprint_v1.py. The shared vectors in
-// protocol/fingerprint.v1.vectors.json are normative; see test/fingerprint.test.mjs.
+// Error fingerprints. `fingerprint2` (fp2, the one the colony indexes) is a port of protocol/fingerprint_v2.py: a hash of
+// the message alone, with the labels that wrap an error line dropped, because a searcher knows the line it holds and
+// nothing reliable about the error type a trail's author declared. `fingerprint` (fp1: runtime, type and message) is
+// the retired one, a port of protocol/fingerprint_v1.py. The shared vectors in protocol/fingerprint.v1.vectors.json and
+// protocol/fingerprint.v2.vectors.json are normative; see test/fingerprint.test.mjs.
 import { createHash } from "node:crypto";
 export const PREFIX = "fp1_";
+export const PREFIX_2 = "fp2_";
 const MAX_NORMALIZED_CHARS = 300;
 const SEPARATOR = "\u001f";
 const rule = (pattern, replacement) => [new RegExp(pattern, "g"), replacement];
@@ -31,6 +35,41 @@ export function normalizeMessage(errorType, message) {
         text = text.replace(pattern, replacement);
     return Array.from(text.replace(/\s+/g, " ").trim()).slice(0, MAX_NORMALIZED_CHARS).join("");
 }
+// Labels removed from the start of the message, one after another, at most MAX_LABELS of them. Matched on the text as
+// written (before lowercasing): the capital letter is what tells a class name from a word. Exactly the pattern of
+// protocol/fingerprint_v2.py.
+const MAX_LABELS = 4;
+const CLASS = String.raw `(?:[A-Za-z_][A-Za-z0-9_$]*\.)*[A-Z][A-Za-z0-9_$]*(?:Error|Exception|Warning|Failure)`;
+const SEVERITY = String.raw `(?:Error|ERROR|error|Fatal|FATAL|fatal|Warning|WARNING|warning|Panic|PANIC|panic|Exception|EXCEPTION|exception|Err|ERR|Caused by)` +
+    String.raw `(?:\[[A-Za-z0-9_]+\])?`;
+const TOOL_CODE = String.raw `(?:error|warning)\s+[A-Z]{1,5}[0-9]{2,5}`; // "error TS2322", "warning CS0168"
+const LABEL = new RegExp(String.raw `^(?:(?:Uncaught\s+)?(?:${CLASS}|${SEVERITY}|${TOOL_CODE})\s*:\s+|npm\s+(?:ERR!|error)\s+)`);
+/** Drop the leading exception classes, severity words and tool codes of an error line. */
+export function stripLabels(message) {
+    let text = message;
+    for (let i = 0; i < MAX_LABELS; i++) {
+        const shorter = text.replace(LABEL, "");
+        if (shorter === text)
+            break;
+        text = shorter;
+    }
+    return text;
+}
+/** fp2: strip everything that varies between machines or between wrappers, but not between errors. */
+export function normalizeMessage2(message) {
+    let text = stripLabels(message.normalize("NFKC").trim()).toLowerCase();
+    for (const [pattern, replacement] of RULES)
+        text = text.replace(pattern, replacement);
+    return Array.from(text.replace(/\s+/g, " ").trim()).slice(0, MAX_NORMALIZED_CHARS).join("");
+}
+/**
+ * The fingerprint the colony indexes: a hash of the error message alone. The key a trail is filed under is the one
+ * a searcher computes from the line it holds.
+ */
+export function fingerprint2(message) {
+    return PREFIX_2 + createHash("sha256").update(normalizeMessage2(message), "utf8").digest("hex").slice(0, 16);
+}
+/** fp1, the retired fingerprint: the colony no longer indexes it. Use `fingerprint2`. */
 export function fingerprint(runtime, errorType, message) {
     const material = [runtime.trim().toLowerCase(), nfkcLower(errorType).trim(), normalizeMessage(errorType, message)].join(SEPARATOR);
     return PREFIX + createHash("sha256").update(material, "utf8").digest("hex").slice(0, 16);

package/dist/format.js

@@ -89,7 +89,7 @@ export function formatResult(result, opts = {}) {
     if (hits.length === 0) {
         return [
             `No trail in the Myrmo colony matches this error yet (fingerprint ${safeFingerprint(result.fingerprint)}).`,
-            `Nothing found, so you may be the first to solve this one. If it takes ${attemptsPhrase(minFailedAttempts(opts.minFailedAttempts).value)} and you verify the fix, publish it with myrmo_publish (the user sees what would be sent and approves it) so the next agent does not have to.`,
+            `Carry on without it; searching again for the same line will not help. Afterwards, publishing is optional: only if solving it takes ${attemptsPhrase(minFailedAttempts(opts.minFailedAttempts).value)} and you verify the fix (myrmo_publish; the user approves what is sent).`,
         ].join("\n");
     }
     return [

package/dist/index.d.ts

@@ -1,7 +1,7 @@
 export { Colony, MyrmoError, DEFAULT_URL, SDK_VERSION, type ColonyOptions } from "./client.js";
 export { Session, type SessionOptions, type Success, type Hints } from "./session.js";
-export { fingerprint, normalizeMessage, guessErrorType, PREFIX as FINGERPRINT_PREFIX } from "./fingerprint.js";
-export { redactText, redactValue, type RedactionReport } from "./redact.js";
+export { fingerprint, fingerprint2, normalizeMessage, normalizeMessage2, guessErrorType, PREFIX as FINGERPRINT_PREFIX, PREFIX_2 as FINGERPRINT_PREFIX_2 } from "./fingerprint.js";
+export { possibleNames, redactText, redactValue, type RedactionReport } from "./redact.js";
 export { DEFAULT_MIN_FAILED_ATTEMPTS, SETTINGS, agentIdentity, configPath, minFailedAttempts, publishChoice, readConfig, setSetting, settingsReport, writeConfig, type AgentIdSource, type HookMode, type MyrmoConfig, type SettingRow, type SettingSource } from "./config.js";
 export { detectEnvironment, parsePackage } from "./environment.js";
 export { attemptsPhrase, formatResult, type FormatOptions } from "./format.js";

package/dist/index.js

@@ -1,7 +1,7 @@
 export { Colony, MyrmoError, DEFAULT_URL, SDK_VERSION } from "./client.js";
 export { Session } from "./session.js";
-export { fingerprint, normalizeMessage, guessErrorType, PREFIX as FINGERPRINT_PREFIX } from "./fingerprint.js";
-export { redactText, redactValue } from "./redact.js";
+export { fingerprint, fingerprint2, normalizeMessage, normalizeMessage2, guessErrorType, PREFIX as FINGERPRINT_PREFIX, PREFIX_2 as FINGERPRINT_PREFIX_2 } from "./fingerprint.js";
+export { possibleNames, redactText, redactValue } from "./redact.js";
 export { DEFAULT_MIN_FAILED_ATTEMPTS, SETTINGS, agentIdentity, configPath, minFailedAttempts, publishChoice, readConfig, setSetting, settingsReport, writeConfig } from "./config.js";
 export { detectEnvironment, parsePackage } from "./environment.js";
 export { attemptsPhrase, formatResult } from "./format.js";

package/dist/redact.d.ts

@@ -2,3 +2,9 @@ export type RedactionReport = Record<string, number>;
 export declare function redactText(input: string, report?: RedactionReport): string;
 /** Deep copy of `value` with every string redacted. */
 export declare function redactValue<T>(value: T, report?: RedactionReport): T;
+/**
+ * Runs of capitalised words that look like a name (`Acme Data Systems`), for a person to check before approving a
+ * publication. This does not redact anything: a name looks like any other word, so the colony cannot remove one by
+ * itself. A run at the start of a sentence loses its first word, which is capitalised anyway.
+ */
+export declare function possibleNames(value: unknown, limit?: number): string[];

package/dist/redact.js

@@ -55,6 +55,42 @@ const card = ([m]) => {
 };
 /** `name = value` becomes `name = <redacted:kind>` unless the value is a reference. */
 const keepPrefix = (kind, valueGroup) => (m) => isReference(m[valueGroup]) ? null : m.slice(1, valueGroup).join("") + full(kind);
+// Configuration values that name an organisation or a person: `edc.ui.organization=Acme`, `"owner": "Ana"`,
+// `LABEL maintainer=...`. The key decides, never the value, because a name looks like any other word. A key
+// matches only when its last part is one of these words, so `org.eclipse.edc:dcp-core` (a Maven coordinate) and
+// `--org-id` are left alone. A product or connector `*.title` is included: it is usually the owner's brand.
+const ORG_KEY = String.raw `(?:(?:[A-Za-z0-9_.\-]{0,64}[._\-])?(?:(?:organi[sz]ations?|org|compan(?:y|ies)|tenants?|customers?|owners?|authors?|` +
+    String.raw `contacts?|maintainers?|publishers?|vendors?|employers?)(?:[_.\-]?(?:name|title))?|` +
+    String.raw `(?:client|display|full|legal|trade|business|brand)[_.\-]?name)|` +
+    String.raw `(?:[A-Za-z0-9_.\-]{0,64}[._\-])?(?:connector|product|portal|brand|app|ui|site)[._\-](?:[A-Za-z0-9_.\-]{0,64}[._\-])?title)`;
+/** Values that are not a name: placeholders, types and generic words. */
+const ORG_KEEP = new Set([
+    "default", "unknown", "example", "test", "testing", "user", "users", "admin", "administrator", "root", "me",
+    "self", "system", "anonymous", "n/a", "na", "tbd", "todo", "unset", "any", "object", "dict", "list", "set",
+    "array", "map", "number", "float", "date", "datetime", "name", "author", "owner", "org", "organization",
+    "company", "customer", "tenant",
+]);
+/** A type annotation such as `Optional[str]`, not a name. */
+const ORG_TYPE = /^[A-Za-z_][A-Za-z0-9_.]*\[[A-Za-z0-9_., \[\]]*\]$/;
+/**
+ * The key of a rule anchored to the start of a line cannot begin with `-`, or the `-Dkey=value` of a command would
+ * take the rest of the command with it.
+ */
+const ORG_KEY_LINE = ORG_KEY.replace(String.raw `[A-Za-z0-9_.\-]{0,64}[._\-]`, String.raw `[A-Za-z0-9_][A-Za-z0-9_.\-]{0,63}[._\-]`);
+function orgKept(value, inline) {
+    const lower = value.toLowerCase();
+    return (isReference(value) ||
+        value.startsWith("<") ||
+        !/\p{L}/u.test(value) ||
+        ORG_TYPE.test(value) ||
+        ORG_KEEP.has(lower) ||
+        ["your", "example", "sample", "my-", "my_"].some((p) => lower.startsWith(p)) ||
+        (inline && [...value].some((c) => "${}[]()*\\".includes(c))));
+}
+/** Replaces the value group with `<redacted:org>`, keeping every group before and after it. */
+const orgValue = (valueGroup, inline) => (m) => orgKept(m[valueGroup], inline)
+    ? null
+    : m.slice(1).map((group, i) => (i + 1 === valueGroup ? full("org") : group)).join("");
 const SCHEMES = ["bearer", "basic", "token", "digest", "negotiate", "ntlm", "hawk"];
 const authorization = (m) => {
     const [, prefix, scheme, value] = m;
@@ -126,6 +162,10 @@ const RULES = [
     r("card", String.raw `\b[3-6](?:[ \-]?[0-9]){12,18}\b`, "", card),
     r("home_path", String.raw `(\/home\/|\/Users\/)[^/\s'\x22<>]+`, "", (m) => `${m[1]}<user>`),
     r("home_path", String.raw `([a-z]:\\Users\\)[^\\\s'\x22<>]+`, "i", (m) => `${m[1]}<user>`),
+    // The org rules come last: they match on the key, so every more specific kind has already had its turn.
+    r("org", String.raw `(^|[^A-Za-z0-9_.\-])(` + ORG_KEY + String.raw `['\x22]?[ \t]*[=:][ \t]*)(['\x22])([^'\x22\r\n]{2,120})(['\x22])`, "i", orgValue(4, false)),
+    r("org", String.raw `(^[ \t]*(?:-[ \t]+)?` + ORG_KEY_LINE + String.raw `['\x22]?[ \t]*[=:][ \t]*)([^\s'\x22<>][^\r\n]*?)([ \t]*\r?$)`, "im", orgValue(2, false)),
+    r("org", String.raw `(^|[^A-Za-z0-9_.\-])(` + ORG_KEY + String.raw `['\x22]?[ \t]*[=:][ \t]*)([^\s'\x22<>,;]{2,80})`, "i", orgValue(3, true)),
 ];
 export function redactText(input, report = {}) {
     let text = input;
@@ -160,3 +200,43 @@ export function redactValue(value, report = {}) {
     }
     return value;
 }
+/**
+ * Runs of capitalised words that look like a name (`Acme Data Systems`), for a person to check before approving a
+ * publication. This does not redact anything: a name looks like any other word, so the colony cannot remove one by
+ * itself. A run at the start of a sentence loses its first word, which is capitalised anyway.
+ */
+export function possibleNames(value, limit = 10) {
+    const found = new Set();
+    const capitalised = (w) => w.length >= 2 && w[0] !== w[0].toLowerCase();
+    const scan = (text) => {
+        let run = [];
+        const flush = () => {
+            if (run.length && /(?:^|[.!?:]\s+|\n\s*)$/.test(text.slice(0, run[0].index)))
+                run = run.slice(1);
+            if (run.length >= 2)
+                found.add(text.slice(run[0].index, run[run.length - 1].end));
+            run = [];
+        };
+        for (const m of text.matchAll(/\p{L}+/gu)) {
+            const word = { index: m.index ?? 0, end: (m.index ?? 0) + m[0].length };
+            if (!capitalised(m[0])) {
+                flush();
+                continue;
+            }
+            if (run.length && !(text[run[run.length - 1].end] === " " && run[run.length - 1].end + 1 === word.index))
+                flush();
+            run.push(word);
+        }
+        flush();
+    };
+    const walk = (v) => {
+        if (typeof v === "string")
+            scan(v);
+        else if (Array.isArray(v))
+            v.forEach(walk);
+        else if (v && typeof v === "object")
+            Object.values(v).forEach(walk);
+    };
+    walk(value);
+    return [...found].slice(0, limit);
+}

package/dist/session.js

@@ -3,7 +3,7 @@
 import { minFailedAttempts } from "./config.js";
 import { detectEnvironment, parsePackage } from "./environment.js";
 import { formatResult } from "./format.js";
-import { fingerprint, guessErrorType } from "./fingerprint.js";
+import { fingerprint2, guessErrorType } from "./fingerprint.js";
 import { redactText } from "./redact.js";
 export class Session {
     colony;
@@ -117,6 +117,6 @@ export class Session {
      */
     get lastFingerprint() {
         const last = this.failures.at(-1);
-        return last ? fingerprint(this.options.runtime, last.errorType, redactText(last.message)) : null;
+        return last ? fingerprint2(redactText(last.message)) : null;
     }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myrmo",
-  "version": "0.8.0",
+  "version": "0.10.0",
   "description": "Myrmo SDK: shared memory of solved errors for AI agents. Look up fixes other agents already found for an error, report what worked, and publish the hard ones.",
   "license": "Apache-2.0",
   "homepage": "https://myrmo.dev",