nodality 1.0.221 → 1.1.0

package/dist/index.esm.js.LICENSE.txt

@@ -1,7 +1,164 @@
 /*!
- * nodality v1.0.220
+ * codegen.js — options object -> JavaScript source, for toCode().
+ *
+ * Every toCode() in the library used to hand-roll the same two steps:
+ *
+ *     JSON.stringify(obj, null, 4).replace(/"([^"]+)":/g, '$1:')
+ *
+ * The replace unquotes EVERY key. That is correct for `radius` and wrong
+ * for any key that is not a bare identifier — and the library emits those
+ * now: `vars: { --nod-split: 0.9 }` from the morph expansion, and content
+ * slots like `"main.hero.title"`. Neither parses as JavaScript, so the
+ * emitted code could not be run, pasted, or ejected. Fourteen copies of
+ * the bad spelling had drifted across twelve files, while the correct one
+ * (`/"(\w+)"\s*:/g`) already sat in image.js, table.js, button.js and
+ * side-nav-bar.js — the same rule, written both ways, in one codebase.
+ *
+ * This module is the one place that rule lives now. It is pure: no DOM, no
+ * imports, safe to load anywhere (see __tests__/unit/import-inert).
+ *
+ *   toObjectSource(obj, 4)   // an object -> "{\n    radius: \"2px\"\n}"
+ *   unquoteKeys(jsonString)  // already have the JSON? just fix the keys
+ *   IDENT_KEY                // the pattern, for call sites that chain
+ *                            // further replaces onto the same string
+ */
+
+/*!
+ * morph-node.js — `{ op: "morph" }`, the (E,N) form of a transition.
+ *
+ * Everything else in N describes how an element LOOKS. A morph node
+ * describes which element BECOMES which, and on what interaction:
+ *
+ *     { op: "morph", from: "topnav", to: ["work", "about", "contact"],
+ *       effect: "t-vhs", duration: 900, back: true }
+ *
+ * Before this, a transition was the one thing the two arrays could not
+ * express. A page wanting a morph had to capture both sides itself,
+ * measure two rects, call applyRasterPipeline with a `transition` block,
+ * drive `t` through a timeline, and wire the clicks — a few hundred lines
+ * sitting *around* the arrays rather than in them.
+ *
+ * What this owns, so a caller does not have to:
+ *
+ *   - the destinations start hidden, so only the source is on screen
+ *   - link i in the source maps to destination i
+ *   - both sides are captured as stills and travel one converging box
+ *   - `back: true` wires the destination's own button to return
+ *
+ * Runs AFTER Des has mounted, because it needs rendered geometry: rects
+ * come from real layout, not from the descriptors.
+ */
+
+/*!
+ * nodality v1.0.224
  * (c) 2026 Filip Vabrousek
  * License: MIT
  */
 
+/*!
+ * nodality/data — runtime/SSG-aware data loader
+ *
+ * `loadJson(name, opts?)` resolves JSON in whichever environment the
+ * page entry is currently executing:
+ *
+ *   • Browser  → relative `fetch("./<name>")` from the page's URL.
+ *   • Node     → ESM JSON import from `../<name>` relative to this
+ *                module (i.e. from the project's `upload/` directory).
+ *
+ * The two paths differ because under SSG (nodality prerender + jsdom)
+ * there is no dev server to fetch from — fetches silently fail and
+ * the page ships an empty mount. The ESM import path reads the JSON
+ * straight off disk, embedding the data into the prerendered HTML so
+ * crawlers see real content.
+ *
+ * Usage in a page entry:
+ *
+ *   import { loadJson } from "nodality/data";
+ *   const products = await loadJson("products.json");
+ *
+ * Conventions:
+ *   • At runtime the file must live next to the page HTML (so
+ *     `./products.json` resolves under the deployed origin).
+ *   • At build time the file must live in `upload/` (the project's
+ *     `uploadDir`) — which is the same place. Single source of truth.
+ *
+ * `fallback` is an opt-in escape hatch: pass `{ fallback: [] }` to
+ * return the supplied value instead of throwing when the file is
+ * missing. Useful for "team.json doesn't exist yet" stub pages.
+ *
+ * Add `{ quiet: true }` when that missing file is the EXPECTED state, so
+ * the build stays silent for a stub and still warns when a file that is
+ * supposed to be there fails to load.
+ */
+
+/*!
+ * nodality/seo — SEO metadata helpers
+ *
+ * Exposes a single `applySeoMeta()` call that idempotently writes a
+ * full per-page metadata bundle into <head>: description, canonical,
+ * Open Graph (Facebook / iMessage / WhatsApp / Slack / LinkedIn),
+ * Twitter / X card, and an optional schema.org JSON-LD payload.
+ *
+ * Also exports a handful of pre-built JSON-LD builders for the
+ * common schema types (Organization, WebSite, Product, Article,
+ * BreadcrumbList) so projects don't have to memorise the schema.org
+ * field names.
+ *
+ * Works in both directions:
+ *   • At SSG time (nodality prerender) the tags land in the static
+ *     HTML response — crawlers and social-card scrapers index them.
+ *   • At browser hydration time the same call re-applies the tags
+ *     against the live <head>; idempotent via `data-seo="1"` markers,
+ *     so re-rendering the page doesn't pile up duplicates.
+ *
+ * Origin handling: pass `path` as a leading-slash route ("/" or
+ * "/blog/post-1") and configure the origin globally via
+ * `setSeoOrigin("https://example.com")` once at app boot. Falls back
+ * to `window.location.origin` in the browser if no origin is set;
+ * falls back to "" under SSG (canonical-relative).
+ */
+
+/*!
+ * suggest.js — "did you mean …?" over a closed vocabulary.
+ *
+ * The library's most expensive historical bug class is the silently
+ * ignored key: a typo'd option or element type that does nothing, reports
+ * nothing, and is found minutes later by eye. Everywhere a vocabulary is
+ * closed — morph's tokens/kinds/axes, ElementMapper's element types — the
+ * rule is the same: throw, and name the nearest valid word.
+ *
+ * Extracted from layout/morph.js (phase M1) so the mapper can use the same
+ * matching, and so the ~15-line DP exists once. No dependency: writing it
+ * is cheaper than owning one, which is the same call the allocator made.
+ *
+ * Pure — no DOM, safe to import anywhere.
+ */
+
+/*!
+ * transition.js — driving transition progress. Phase T4.
+ *
+ * The pipeline already renders any `t` you hand it (phases T1–T3). This
+ * module is only about WHERE t comes from: a timeline, a scroll range, a
+ * drag, a test. It owns no pixels.
+ *
+ *     import { progressTimeline } from "nodality/transition";
+ *
+ *     const tl = progressTimeline(pipeline, { duration: 600 });
+ *     tl.to(1);            // play forward
+ *     tl.to(0);            // reverse — from wherever it is NOW
+ *     tl.bindScroll("#section");
+ *     tl.set(0.37);        // deterministic, for tests and gestures
+ *
+ * The one property everything rests on: **retargeting starts from the
+ * current value, never from zero.** Interrupting a half-finished morph
+ * and sending it back is `to(0)`, and it continues from 0.5 rather than
+ * snapping. That falls out of `t` being an input rather than a timer
+ * (P-1) — there is no timeline state to unwind, only a number to move.
+ *
+ * Duration scales with the distance actually travelled, so a reversal
+ * from the midpoint takes half as long as a full play. A fixed duration
+ * makes short corrections feel sluggish and is the usual reason
+ * interruptible animation feels wrong.
+ */
+
 /*! regenerator-runtime -- Copyright (c) 2014-present, Facebook, Inc. -- license (MIT): https://github.com/babel/babel/blob/main/packages/babel-helpers/LICENSE */

package/dist/keyframe-animation.cjs.js.LICENSE.txt

@@ -1,5 +1,5 @@
 /*!
- * nodality v1.0.220
+ * nodality v1.0.224
  * (c) 2026 Filip Vabrousek
  * License: MIT
  */

package/dist/keyframe-animation.esm.js.LICENSE.txt

@@ -1,5 +1,5 @@
 /*!
- * nodality v1.0.220
+ * nodality v1.0.224
  * (c) 2026 Filip Vabrousek
  * License: MIT
  */