@forwardimpact/libsyntheticgen 0.1.1 → 0.1.2

package/dsl/parser.js

@@ -173,6 +173,15 @@ export function parse(tokens) {
           people.disciplines[disc] = pct;
         }
         expect("RBRACE");
+      } else if (kw.value === "archetypes") {
+        expect("LBRACE");
+        people.archetypes = {};
+        while (peek().type !== "RBRACE") {
+          const archetype = parseStringOrIdent();
+          const pct = parseNumberValue();
+          people.archetypes[archetype] = pct;
+        }
+        expect("RBRACE");
       } else
         throw new Error(
           `Unexpected '${kw.value}' in people at line ${kw.line}`,
@@ -199,6 +208,10 @@ export function parse(tokens) {
       else if (kw.value === "prose_topic")
         proj.prose_topic = parseStringValue();
       else if (kw.value === "prose_tone") proj.prose_tone = parseStringValue();
+      else if (kw.value === "milestones") proj.milestones = parseArray();
+      else if (kw.value === "risks") proj.risks = parseArray();
+      else if (kw.value === "technical_choices")
+        proj.technical_choices = parseArray();
       else
         throw new Error(
           `Unexpected '${kw.value}' in project at line ${kw.line}`,
@@ -263,6 +276,8 @@ export function parse(tokens) {
     while (peek().type !== "RBRACE") {
       const kw = advance();
       if (kw.value === "name") scenario.name = parseStringValue();
+      else if (kw.value === "narrative")
+        scenario.narrative = parseStringValue();
       else if (kw.value === "timerange_start")
         scenario.timerange_start = parseDateValue();
       else if (kw.value === "timerange_end")
@@ -634,7 +649,16 @@ export function parse(tokens) {
         content.briefings_per_persona = parseNumberValue();
       else if (kw.value === "notes_per_persona")
         content.notes_per_persona = parseNumberValue();
-      else
+      else if (kw.value === "blog_topics") {
+        expect("LBRACE");
+        content.blog_topics = {};
+        while (peek().type !== "RBRACE") {
+          const topic = parseStringOrIdent();
+          const pct = parseNumberValue();
+          content.blog_topics[topic] = pct;
+        }
+        expect("RBRACE");
+      } else
         throw new Error(
           `Unexpected '${kw.value}' in content at line ${kw.line}`,
         );

package/engine/entities.js

@@ -15,39 +15,42 @@ import {
  * Build all entities from AST and RNG.
  * @param {import('../dsl/parser.js').UniverseAST} ast
  * @param {import('./rng.js').SeededRNG} rng
+ * @param {object} [logger] - Logger instance for warnings
  * @returns {{ orgs: object[], departments: object[], teams: object[], people: object[], projects: object[] }}
  */
-export function buildEntities(ast, rng) {
+export function buildEntities(ast, rng, logger) {
   const domain = ast.domain;
   const orgs = ast.orgs.map((o) => ({
     ...o,
-    iri: `https://${domain}/org/${o.id}`,
+    iri: `https://${domain}/id/org/${o.id}`,
   }));
   const departments = ast.departments.map((d) => ({
     ...d,
-    iri: `https://${domain}/department/${d.id}`,
+    iri: `https://${domain}/id/department/${d.id}`,
   }));
   const teams = ast.teams.map((t) => ({
     ...t,
     repos: t.repos || [],
-    iri: `https://${domain}/team/${t.id}`,
+    iri: `https://${domain}/id/team/${t.id}`,
     getdx_team_id: `gdx_team_${t.id}`,
   }));
-  const people = generatePeople(ast, rng, teams, domain);
+  const people = generatePeople(ast, rng, teams, domain, logger);
   const projects = ast.projects.map((p) => ({
     ...p,
     teams: p.teams || [],
     phase: p.phase || null,
     prose_topic: p.prose_topic || null,
     prose_tone: p.prose_tone || null,
-    iri: `https://${domain}/project/${p.id}`,
+    iri: `https://${domain}/id/project/${p.id}`,
   }));
 
   return { orgs, departments, teams, people, projects };
 }
 
-function generatePeople(ast, rng, teams, domain) {
-  const { count, distribution, disciplines } = ast.people;
+function generatePeople(ast, rng, teams, domain, logger) {
+  const { count, distribution, disciplines, archetypes } = ast.people;
+  const archetypeKeys = archetypes ? Object.keys(archetypes) : [];
+  const archetypeWeights = archetypes ? Object.values(archetypes) : [];
   const people = [];
   const usedNames = new Set();
 
@@ -71,6 +74,9 @@ function generatePeople(ast, rng, teams, domain) {
   for (const team of teams) {
     if (!team.manager) continue;
     const name = managerAssignments.get(team.id);
+    const archetype = archetypeKeys.length
+      ? archetypeKeys[rng.weightedPick(archetypeWeights)]
+      : "steady_contributor";
     people.push(
       makePerson(
         name,
@@ -80,6 +86,8 @@ function generatePeople(ast, rng, teams, domain) {
         domain,
         true,
         null,
+        undefined,
+        archetype,
       ),
     );
   }
@@ -92,6 +100,9 @@ function generatePeople(ast, rng, teams, domain) {
     const disc = discKeys[rng.weightedPick(discWeights)];
     const team = rng.pick(teams);
     const mgr = people.find((p) => p.is_manager && p.team_id === team.id);
+    const archetype = archetypeKeys.length
+      ? archetypeKeys[rng.weightedPick(archetypeWeights)]
+      : "steady_contributor";
     people.push(
       makePerson(
         name,
@@ -102,10 +113,17 @@ function generatePeople(ast, rng, teams, domain) {
         false,
         mgr?.email || null,
         `2023-${pad2(rng.randomInt(1, 12))}-${pad2(rng.randomInt(1, 28))}`,
+        archetype,
       ),
     );
   }
 
+  if (people.length < count && logger) {
+    logger.warn(
+      `People shortfall: requested ${count}, generated ${people.length} (name pool exhausted)`,
+    );
+  }
+
   return people;
 }
 
@@ -118,6 +136,7 @@ function makePerson(
   isManager,
   managerEmail,
   hireDate = "2023-01-15",
+  archetype = "steady_contributor",
 ) {
   const id = name.toLowerCase().replace(/\s+/g, "-");
   return {
@@ -134,7 +153,8 @@ function makePerson(
     is_manager: isManager,
     manager_email: managerEmail,
     hire_date: hireDate,
-    iri: `https://${domain}/person/${id}`,
+    archetype,
+    iri: `https://${domain}/id/person/${id}`,
   };
 }
 

package/engine/prose-keys.js

@@ -11,13 +11,15 @@
  */
 export function collectProseKeys(entities) {
   const keys = new Map();
+  const orgName = entities.orgs[0]?.name || "BioNova";
 
   // Organization README prose
   keys.set("org_readme", {
-    topic: `${entities.orgs[0]?.name || "BioNova"} company overview`,
+    topic: `${orgName} company overview`,
     tone: "corporate, informative",
     length: "3-4 paragraphs",
     domain: entities.domain,
+    orgName,
   });
 
   // Project descriptions
@@ -28,6 +30,7 @@ export function collectProseKeys(entities) {
         tone: proj.prose_tone || "technical",
         length: "2-3 paragraphs",
         domain: entities.domain,
+        orgName,
       });
     }
   }
@@ -41,6 +44,7 @@ export function collectProseKeys(entities) {
         tone: "technical, informative",
         length: "6-8 paragraphs",
         domain: entities.domain,
+        orgName,
       });
     }
 
@@ -51,6 +55,7 @@ export function collectProseKeys(entities) {
         tone: "conversational, technical",
         length: "4-5 paragraphs",
         domain: entities.domain,
+        orgName,
       });
     }
 
@@ -61,6 +66,7 @@ export function collectProseKeys(entities) {
         tone: "helpful, concise",
         length: "1 paragraph",
         domain: entities.domain,
+        orgName,
       });
     }
 
@@ -71,6 +77,7 @@ export function collectProseKeys(entities) {
         tone: "instructional",
         length: "5-6 paragraphs",
         domain: entities.domain,
+        orgName,
       });
     }
 
@@ -82,6 +89,7 @@ export function collectProseKeys(entities) {
         length: "1-2 sentences",
         maxTokens: 100,
         domain: entities.domain,
+        orgName,
       });
     }
 
@@ -93,6 +101,7 @@ export function collectProseKeys(entities) {
         length: "1-2 sentences",
         maxTokens: 80,
         domain: entities.domain,
+        orgName,
       });
     }
   }
@@ -111,6 +120,7 @@ export function collectProseKeys(entities) {
           tone: "professional, concise",
           length: "2-3 paragraphs",
           domain: entities.domain,
+          orgName,
           role: `${persona.level} ${persona.discipline}`,
         });
       }
@@ -122,6 +132,7 @@ export function collectProseKeys(entities) {
           tone: "personal, technical",
           length: "1-2 paragraphs",
           domain: entities.domain,
+          orgName,
           role: `${persona.level} ${persona.discipline}`,
         });
       }
@@ -141,6 +152,7 @@ export function collectProseKeys(entities) {
           length: "1-2 sentences",
           maxTokens: 80,
           domain: entities.domain,
+          orgName,
           role: `${ck.person_level} ${ck.person_discipline.replace(/_/g, " ")} on the ${ck.team_name}`,
           scenario: ck.scenario_name,
           driver: ck.driver_name,

package/engine/tier0.js

@@ -33,6 +33,7 @@ export class EntityGenerator {
     const { orgs, departments, teams, people, projects } = buildEntities(
       ast,
       rng,
+      this.logger,
     );
     const activity = generateActivity(ast, rng, people, teams);
 

package/index.js

@@ -2,6 +2,11 @@ export { DslParser, createDslParser } from "./dsl/index.js";
 export { EntityGenerator, createEntityGenerator } from "./engine/tier0.js";
 export { createSeededRNG } from "./engine/rng.js";
 export { collectProseKeys } from "./engine/prose-keys.js";
+export {
+  PROFICIENCY_LEVELS,
+  MATURITY_LEVELS,
+  STAGE_NAMES,
+} from "./vocabulary.js";
 export { FakerTool, createFakerTool } from "./tools/faker.js";
 export { SyntheaTool, createSyntheaTool } from "./tools/synthea.js";
 export { SdvTool, createSdvTool } from "./tools/sdv.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@forwardimpact/libsyntheticgen",
-  "version": "0.1.1",
+  "version": "0.1.2",
   "description": "DSL parsing and deterministic entity generation for synthetic data",
   "license": "Apache-2.0",
   "repository": {
@@ -16,6 +16,8 @@
     "./engine": "./engine/tier0.js",
     "./engine/entities": "./engine/entities.js",
     "./engine/activity": "./engine/activity.js",
+    "./vocabulary": "./vocabulary.js",
+    "./vocabulary.js": "./vocabulary.js",
     "./rng": "./engine/rng.js",
     "./tools/faker": "./tools/faker.js",
     "./tools/synthea": "./tools/synthea.js",

package/vocabulary.js

@@ -0,0 +1,65 @@
+/**
+ * Shared vocabulary constants for synthetic data generation.
+ *
+ * Single source of truth for proficiency levels, maturity levels,
+ * and stage names used across libsyntheticgen, libsyntheticprose,
+ * and libsyntheticrender.
+ */
+
+/** @type {string[]} */
+export const PROFICIENCY_LEVELS = [
+  "awareness",
+  "foundational",
+  "working",
+  "practitioner",
+  "expert",
+];
+
+/** @type {string[]} */
+export const MATURITY_LEVELS = [
+  "emerging",
+  "developing",
+  "practicing",
+  "role_modeling",
+  "exemplifying",
+];
+
+/** @type {string[]} */
+export const STAGE_NAMES = [
+  "specify",
+  "plan",
+  "scaffold",
+  "code",
+  "review",
+  "deploy",
+];
+
+/** @type {Record<string, number>} */
+const PROFICIENCY_INDEX = Object.fromEntries(
+  PROFICIENCY_LEVELS.map((p, i) => [p, i]),
+);
+
+/** @type {Record<string, number>} */
+const ARCHETYPE_OFFSET = {
+  high_performer: 1,
+  steady_contributor: 0,
+  new_hire: -1,
+  struggling: -2,
+};
+
+/**
+ * Adjust a proficiency level based on a person's archetype.
+ * @param {string} expected - Base proficiency level
+ * @param {string} archetype - Person archetype
+ * @returns {string} Adjusted proficiency level
+ */
+export function adjustProficiency(expected, archetype) {
+  const base = PROFICIENCY_INDEX[expected];
+  if (base === undefined) return expected;
+  const offset = ARCHETYPE_OFFSET[archetype] || 0;
+  const clamped = Math.max(
+    0,
+    Math.min(PROFICIENCY_LEVELS.length - 1, base + offset),
+  );
+  return PROFICIENCY_LEVELS[clamped];
+}