@unified-product-graph/core 0.11.0 → 0.11.2

package/CHANGELOG.md

@@ -7,6 +7,30 @@ This project uses [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ---
 
+## [0.11.2] - 2026-06-14
+
+**Two analysis reads over the classification layer: compare two rivals, and digest a property's distribution.** The competitive tier could render the landscape (`get_portfolio_tree`), audit its completeness (`audit_property_coverage`), and diff its history (`diff_classification`), but two by-hand motions from the backfill stayed manual: comparing two competitors axis-by-axis, and counting a property's distribution over the `jq` dump. This release lands both as read tools (Data's tooling-gaps brief #5/#6). Current-state reads, no history substrate, no schema or data migration. New tools are a patch under spec policy.
+
+### Added
+- **`compare_classifications` (new tool, 129 of 130; local-only).** `compare_classifications({ a, b, axis? })` joins two classified nodes axis-by-axis: where they sit at the same value (`agree`), at different values (`diverge`), or where only one has been graded (`a_only` / `b_only`). Divergences are ordered first (the actionable rows). It reuses the same per-node profile assembly as `get_portfolio_tree` competitor_profile, so axis / value / confidence resolution is identical. This is the bridge from the classification layer to the parity layer: `create_parity_edge` is the writer; this derives which axes warrant one.
+- **`aggregate_edge_properties` (new tool, 130 of 130; local-only).** `aggregate_edge_properties({ edge_type, group_by?, property? })` returns the distribution of one edge property across every portfolio cross-edge of a type — overall, or grouped by `axis`, `competitor`, or `value`. `property` defaults to `confidence`, and an assessment-object property buckets by its label, so "165 Confident / 53 Some evidence, the uncertain ones cluster on `ext_api_sdk`" is one call instead of a count over a dump. The digest of the property layer.
+
+---
+
+## [0.11.1] - 2026-06-14
+
+**Bug fix: the typed classification writer misrouted competitor sources, duplicated instead of upserting, and disagreed with the confidence scale.** The generic cross-edge writers got their in-place upsert in 0.10.6; the typed convenience writer `create_classification_edge` never inherited it, and on a qualified `{product}/{node}` competitor source it wrote the wrong edge type, created a duplicate, and expanded `high` to a different `confidence_5` value than the rest of the graph. Three defects in one call (Data's dogfood brief), all on a path the generic writers already do correctly. No data migration.
+
+### Fixed
+- **`create_classification_edge` now upserts instead of duplicating.** Root cause was the misroute below: a qualified competitor source was mis-typed as the polymorphic `node_classified_as_classification_value`, so the `(source, target, type)` dedup never matched the existing `competitor_…` edge and a second edge was created. Correct routing makes the write land on the existing cell in place (`status: "updated"`).
+- **Qualified-source type routing.** A `{product}/{node}` source whose node is a `competitor` now resolves to `competitor` — via the owning product file, and failing that the portfolio's `instance_of` index (which maps `{pid}/{nid}` to its canonical type without a local product file) — and writes `competitor_classified_as_classification_value`. Bare local sources keep the polymorphic type.
+- **Confidence-scale agreement.** `high` now expands to `confidence_5` value **4** (canonical label `Confident`), not 5 — matching the generic writers and the existing classify-edge population, so two writers can no longer populate the same axis with different numbers for the same word.
+
+### Added
+- **Pinned friendly confidence mapping on the scale (`confidence_5.friendly_aliases = { low: 2, medium: 3, high: 4 }`).** The single, introspectable source of truth — surfaced via `get_scale` — that `create_classification_edge` (and any future friendly-confidence writer) resolves through `friendlyToAssessment(scaleId, word)`. The expansion carries the canonical point label (`Confident`), not the input word.
+
+---
+
 ## [0.11.0] - 2026-06-14
 
 **Self-documenting competitive history: a competitor's classification change records itself.** The 0.10.x tier classifies competitors against axes and values and carries `confidence` / `assessed_on`, but a re-assessment overwrites the prior value — there was no way to ask *what moved*. This release makes the classification landscape remember its own changes. A minor bump: it widens the `competitor_signal.signal_type` enum (union widening), which spec policy treats as a minor; additive, so old graphs still parse.

package/dist/index.cjs

@@ -119,6 +119,7 @@ __export(index_exports, {
   evaluateAntiPatterns: () => evaluateAntiPatterns,
   findInvalidStatusMigrationTargets: () => findInvalidStatusMigrationTargets,
   formatUpgText: () => formatUpgText,
+  friendlyToAssessment: () => friendlyToAssessment,
   generateSlug: () => generateSlug,
   getAnchorEntity: () => getAnchorEntity,
   getAntiPatternById: () => getAntiPatternById,
@@ -6801,7 +6802,11 @@ var UPG_SCALES = {
       { value: 3, label: "Some evidence", description: "A few data points" },
       { value: 4, label: "Confident", description: "Multiple data sources" },
       { value: 5, label: "Data-backed", description: "Strong quantitative evidence" }
-    ]
+    ],
+    // The pinned friendly mapping (0.11.1). `high` is Confident (4), not
+    // Data-backed (5) — 5 is reserved for genuinely quantified claims. Matches
+    // the existing classify-edge population, so nothing needs re-stamping.
+    friendly_aliases: { low: 2, medium: 3, high: 4 }
   },
   // ── Effort ─────────────────────────────────────────────────────────────────
   effort_5: {
@@ -6823,6 +6828,13 @@ var UPG_SCALES = {
 function getScale(scaleId) {
   return UPG_SCALES[scaleId];
 }
+function friendlyToAssessment(scaleId, word) {
+  const scale = UPG_SCALES[scaleId];
+  const value = scale?.friendly_aliases?.[word];
+  if (value === void 0) return null;
+  const point = scale.points.find((p) => p.value === value);
+  return { value, label: point?.label ?? word, scale_id: scaleId };
+}
 var PROPERTY_SCALE_MAP = {
   // ── Reach ────────────────────────────────────────────────────────────────
   reach: "reach_5",
@@ -28755,7 +28767,7 @@ function isCanonical(text) {
 }
 
 // src/index.ts
-var UPG_VERSION = "0.11.0";
+var UPG_VERSION = "0.11.2";
 var UPG_FORMAT_VERSION = "0.4.0";
 var MARKDOWN_FORMAT_VERSION = "0.1";
 var UPG_TYPES = getTypes();
@@ -28919,6 +28931,7 @@ var UPG_META_COUNT = UPG_ENTITY_META.length;
   evaluateAntiPatterns,
   findInvalidStatusMigrationTargets,
   formatUpgText,
+  friendlyToAssessment,
   generateSlug,
   getAnchorEntity,
   getAntiPatternById,