ucode 0.1.0

data/lib/ucode/site/config_emitter.rb

@@ -0,0 +1,124 @@
+# frozen_string_literal: true
+
+require "json"
+require "pathname"
+
+require "ucode/repo/atomic_writes"
+
+module Ucode
+  module Site
+    # Writes `site/.vitepress/config.ts` from the JSON output tree.
+    #
+    # Reads:
+    #   output/planes/<n>.json     → top nav (17 planes)
+    #   output/blocks/index.json   → sidebar groups (one per plane)
+    #
+    # Writes:
+    #   site/.vitepress/config.ts  — TypeScript module that exports the
+    #   Vitepress `defineConfig` payload. Idempotent via AtomicWrites.
+    #
+    # **Pure**: reads JSON, writes TS. No Vitepress dependency on the
+    # Ruby side; the site is a thin shell that imports this generated
+    # config and re-exports `defineConfig(config)`.
+    class ConfigEmitter
+      include Repo::AtomicWrites
+
+      # @param output_root [String, Pathname] dataset root
+      # @param site_root [String, Pathname] Vitepress project root
+      def initialize(output_root:, site_root:)
+        @output_root = Pathname.new(output_root)
+        @site_root = Pathname.new(site_root)
+      end
+
+      # @return [Pathname] the config file path that #emit writes
+      def target_path
+        @site_root.join(".vitepress", "config.ts")
+      end
+
+      # Read the output tree, render config.ts, write atomically.
+      # @return [Boolean] true if the file was (re)written, false if skipped
+      def emit
+        write_atomic(target_path, render(load_planes, load_blocks))
+      end
+
+      private
+
+      def load_planes
+        (0..16).map do |n|
+          path = @output_root.join("planes", "#{n}.json")
+          next nil unless path.exist?
+
+          JSON.parse(path.read).tap do |plane|
+            plane["number"] = n
+          end
+        end.compact
+      end
+
+      def load_blocks
+        path = @output_root.join("blocks", "index.json")
+        return [] unless path.exist?
+
+        JSON.parse(path.read)
+      end
+
+      def render(planes, blocks)
+        <<~TS
+          // AUTO-GENERATED by Ucode::Site::ConfigEmitter. Do not edit.
+          import { defineConfig } from "vitepress";
+
+          export const planes = #{planes_to_ts(planes)};
+
+          export const blocks = #{blocks_to_ts(blocks)};
+
+          export default defineConfig({
+            title: "ucode",
+            description: "Unicode Character Database — code charts, properties, relationships",
+            cleanUrls: true,
+            themeConfig: {
+              siteTitle: "ucode",
+              nav: [
+                { text: "Planes", link: "/plane/0" },
+                { text: "Search", link: "/search" },
+              ],
+              sidebar: sidebar(planes, blocks),
+              socialLinks: [
+                { icon: "github", link: "https://github.com/fontist/ucode" },
+              ],
+              search: { provider: "local" },
+            },
+          });
+
+          function sidebar(planes, blocks) {
+            const groups = {};
+            for (const plane of planes) {
+              groups[plane.number] = {
+                text: plane.abbrev + " — " + plane.name,
+                items: blocks
+                  .filter((b) => b.plane_number === plane.number)
+                  .map((b) => ({ text: b.name, link: "/block/" + b.id })),
+              };
+            }
+            // Sidebar per top-level section: plane pages share one sidebar,
+            // block pages get their plane's sidebar.
+            return {
+              "/plane/": Object.values(groups),
+              "/block/": groups,
+            };
+          }
+        TS
+      end
+
+      def planes_to_ts(planes)
+        JSON.pretty_generate(planes.map { |p|
+          { number: p["number"], name: p["name"], abbrev: p["abbrev"] }
+        })
+      end
+
+      def blocks_to_ts(blocks)
+        JSON.pretty_generate(blocks.map { |b|
+          { id: b["id"], name: b["name"], plane_number: b["plane_number"] }
+        })
+      end
+    end
+  end
+end

data/lib/ucode/site/generator.rb

@@ -0,0 +1,178 @@
+# frozen_string_literal: true
+
+require "fileutils"
+require "find"
+require "json"
+require "pathname"
+
+require "ucode/repo/atomic_writes"
+require "ucode/site/config_emitter"
+require "ucode/site/search_index"
+
+module Ucode
+  module Site
+    # Orchestrates `ucode site init` and `ucode site build`.
+    #
+    # **init** copies the static Vitepress template from
+    # `lib/ucode/site/template/` into the user's `site/`. The template
+    # ships package.json, theme, and the dynamic route components
+    # (`char/[codepoint].vue`, `block/[id].vue`, `plane/[n].md` stub).
+    #
+    # **build** regenerates only the parts that depend on the dataset:
+    #   - `.vitepress/config.ts`           — ConfigEmitter
+    #   - `public/data/`                   — symlinked or copied from `output/`
+    #   - `public/data/index/search.json`  — SearchIndex
+    #   - `plane/<n>.md`                   — one thin stub per plane (frontmatter)
+    #   - `block/<id>.md`                  — one thin stub per block
+    #
+    # Static pages are markdown stubs that mount a Vue component; the
+    # component fetches the JSON for that plane/block at runtime. This
+    # keeps the generator cheap (~363 small writes) and the per-character
+    # route dynamic (~160k static pages is infeasible).
+    #
+    # **Idempotent**: every write goes through AtomicWrites.
+    class Generator
+      include Repo::AtomicWrites
+
+      TemplateDir = File.expand_path("template", __dir__).freeze
+      private_constant :TemplateDir
+
+      # @param output_root [String, Pathname] dataset root (read)
+      # @param site_root [String, Pathname] Vitepress project root (write)
+      def initialize(output_root:, site_root:)
+        @output_root = Pathname.new(output_root)
+        @site_root = Pathname.new(site_root)
+      end
+
+      # Copy the static template into `site_root`. No-op for any file
+      # that already exists with identical content (AtomicWrites).
+      # @return [Integer] number of files written
+      def init
+        count = 0
+        each_template_file do |src, rel|
+          dst = @site_root.join(rel)
+          count += 1 if write_atomic(dst, src.read)
+        end
+        count
+      end
+
+      # Regenerate config + pages + search index from the current
+      # `output/` tree. Returns a tally of what changed.
+      # @return [Hash{Symbol => Integer}]
+      def build
+        tally = { config: 0, pages: 0, search: 0, data_link: 0 }
+
+        tally[:config] = config_emitter.emit ? 1 : 0
+        tally[:pages] = write_pages
+        tally[:search] = search_index.build ? 1 : 0
+        tally[:data_link] = link_data_dir ? 1 : 0
+
+        tally
+      end
+
+      private
+
+      def config_emitter
+        ConfigEmitter.new(output_root: @output_root, site_root: @site_root)
+      end
+
+      def search_index
+        SearchIndex.new(@output_root)
+      end
+
+      # Walk TemplateDir, yielding (src_path, relative_path) for each file.
+      def each_template_file(&block)
+        return unless Dir.exist?(TemplateDir)
+
+        Find.find(TemplateDir) do |src|
+          next if File.directory?(src)
+
+          rel = Pathname.new(src).relative_path_from(Pathname.new(TemplateDir))
+          yield Pathname.new(src), rel
+        end
+      end
+
+      # Write plane and block markdown stubs from the output tree.
+      def write_pages
+        count  = 0
+        count += write_plane_pages
+        count += write_block_pages
+        count
+      end
+
+      def write_plane_pages
+        planes_dir = @output_root.join("planes")
+        return 0 unless planes_dir.directory?
+
+        planes_dir.children.sort.sum do |path|
+          next 0 unless path.file? && path.extname == ".json"
+
+          plane = JSON.parse(path.read)
+          n = plane["number"]
+          next 0 unless n
+
+          payload = plane_md(n)
+          write_atomic(@site_root.join("plane", "#{n}.md"), payload) ? 1 : 0
+        end
+      end
+
+      def write_block_pages
+        blocks = read_json_list(@output_root.join("blocks", "index.json"))
+        blocks.sum do |block|
+          id = block["id"]
+          next 0 unless id
+
+          payload = block_md(id)
+          write_atomic(@site_root.join("block", "#{id}.md"), payload) ? 1 : 0
+        end
+      end
+
+      def read_json_list(path)
+        return [] unless path&.exist?
+
+        JSON.parse(path.read)
+      end
+
+      def plane_md(plane_number)
+        <<~MD
+          ---
+          layout: plane
+          title: "Plane #{plane_number}"
+          plane: #{plane_number}
+          ---
+
+          <PlaneView plane="#{plane_number}" />
+        MD
+      end
+
+      def block_md(block_id)
+        <<~MD
+          ---
+          layout: block
+          title: "#{block_id}"
+          block: "#{block_id}"
+          ---
+
+          <BlockView block="#{block_id}" />
+        MD
+      end
+
+      # Vitepress serves `site/public/` at the site root. Symlink the
+      # dataset into `public/data/` so the Vue components can fetch
+      # `/data/...` URLs. Falls back to a recursive copy on filesystems
+      # that don't support symlinks.
+      def link_data_dir
+        link = @site_root.join("public", "data")
+        return 0 if link.symlink? && link.dirname.exist?
+
+        link.dirname.mkpath
+        begin
+          File.symlink(@output_root.relative_path_from(link.dirname).to_s, link.to_s)
+        rescue SystemCallError
+          FileUtils.cp_r(@output_root, link)
+        end
+        1
+      end
+    end
+  end
+end

data/lib/ucode/site/search_index.rb

@@ -0,0 +1,68 @@
+# frozen_string_literal: true
+
+require "json"
+require "pathname"
+
+require "ucode/repo/atomic_writes"
+require "ucode/repo/paths"
+
+module Ucode
+  module Site
+    # Builds the client-side search payload consumed by MiniSearch.
+    #
+    # Input: `output/index/labels.json` written by `Repo::AggregateWriter`
+    #   — a flat `{ "U+XXXX" => { name, gc, sc } }` map (~160k entries,
+    #   ~5 MB raw).
+    #
+    # Output: `output/index/search.json` — an array of `{ id, name, gc, sc }`
+    #   objects, ready to feed `new MiniSearch(payload, ...)`.
+    #
+    # **Streaming**: labels.json is parsed incrementally via the stdlib
+    # JSON parser; the entire payload is materialised once for atomic
+    # write. For ~160k codepoints this peaks around ~30 MB — acceptable
+    # for a build-time tool.
+    #
+    # **Idempotent**: re-runs are byte-compared no-ops via AtomicWrites.
+    class SearchIndex
+      include Repo::AtomicWrites
+
+      # @param output_root [String, Pathname]
+      def initialize(output_root)
+        @output_root = Pathname.new(output_root)
+      end
+
+      # Build and write `search.json`. Returns the entry count, or nil
+      # if labels.json is absent (nothing to index).
+      # @return [Integer, nil]
+      def build
+        labels = load_labels
+        return nil unless labels
+
+        entries = labels.map { |cp_id, meta| entry_for(cp_id, meta) }
+        payload = JSON.generate(entries)
+        write_atomic(target_path, payload)
+        entries.size
+      end
+
+      # The path that #build writes to. Exposed so specs and the site
+      # generator can reference it without duplicating the convention.
+      # @return [Pathname]
+      def target_path
+        Pathname(@output_root).join("index", "search.json")
+      end
+
+      private
+
+      def load_labels
+        path = Repo::Paths.labels_index_path(@output_root)
+        return nil unless path.exist?
+
+        JSON.parse(path.read)
+      end
+
+      def entry_for(cp_id, meta)
+        { id: cp_id, name: meta["name"], gc: meta["gc"], sc: meta["sc"] }
+      end
+    end
+  end
+end

data/lib/ucode/site/template/.gitignore

@@ -0,0 +1,4 @@
+node_modules/
+.vitepress/dist/
+.vitepress/cache/
+public/data/

data/lib/ucode/site/template/.vitepress/config.ts

@@ -0,0 +1,8 @@
+// Placeholder config; replaced by `ucode site build` (which writes
+// config.ts from the current output/ tree). Safe to delete.
+import { defineConfig } from "vitepress";
+
+export default defineConfig({
+  title: "ucode",
+  description: "Run `ucode site build` to regenerate this config from output/.",
+});

data/lib/ucode/site/template/.vitepress/theme/index.js

@@ -0,0 +1,20 @@
+import DefaultTheme from "vitepress/theme";
+import { defineComponent, h } from "vue";
+import PlaneView from "../components/PlaneView.vue";
+import BlockView from "../components/BlockView.vue";
+import CharView from "../components/CharView.vue";
+
+export default {
+  ...DefaultTheme,
+  Layout: defineComponent({
+    name: "UcodeLayout",
+    setup() {
+      return () => h(DefaultTheme.Layout, null, {});
+    },
+  }),
+  enhanceApp({ app }) {
+    app.component("PlaneView", PlaneView);
+    app.component("BlockView", BlockView);
+    app.component("CharView", CharView);
+  },
+};

data/lib/ucode/site/template/char/[codepoint].md

@@ -0,0 +1,13 @@
+---
+layout: doc
+---
+
+<script setup>
+import { useRoute } from "vitepress";
+import CharView from "../components/CharView.vue";
+
+const route = useRoute();
+const codepoint = route.params.codepoint;
+</script>
+
+<CharView :codepoint="codepoint" />

data/lib/ucode/site/template/components/BlockView.vue

@@ -0,0 +1,57 @@
+<script setup>
+import { ref, onMounted, computed } from "vue";
+
+const props = defineProps({ block: { type: String, required: true } });
+
+const blockMeta = ref(null);
+const cpToBlock = ref({});
+const labels = ref({});
+
+const cells = computed(() => {
+  if (!blockMeta.value) return [];
+  return blockMeta.value.codepoint_ids.map((cpId) => ({
+    id: cpId,
+    label: labels.value[cpId]?.name,
+    glyphUrl: `/data/blocks/${props.block}/${cpId}/glyph.svg`,
+    detailUrl: `/char/${cpId.slice(2)}`,
+  }));
+});
+
+onMounted(async () => {
+  const [blockRes, labelsRes] = await Promise.all([
+    fetch(`/data/blocks/${props.block}.json`),
+    fetch(`/data/index/labels.json`),
+  ]);
+  blockMeta.value = await blockRes.json();
+  labels.value = await labelsRes.json();
+});
+</script>
+
+<template>
+  <div class="block">
+    <header v-if="blockMeta">
+      <h1>{{ blockMeta.name }}</h1>
+      <p class="lead">
+        <code>{{ blockMeta.id }}</code>
+        — U+{{ blockMeta.range_first.toString(16).toUpperCase().padStart(4, "0") }}
+        – U+{{ blockMeta.range_last.toString(16).toUpperCase().padStart(4, "0") }}
+        ({{ cells.length }} codepoints)
+      </p>
+    </header>
+
+    <section class="grid">
+      <a v-for="cell in cells" :key="cell.id" :href="cell.detailUrl" class="cell">
+        <img :src="cell.glyphUrl" :alt="cell.label" loading="lazy" />
+        <span class="cp-id">{{ cell.id }}</span>
+      </a>
+    </section>
+  </div>
+</template>
+
+<style scoped>
+.lead { color: var(--vp-c-text-2); font-family: monospace; }
+.grid { display: grid; grid-template-columns: repeat(auto-fill, minmax(96px, 1fr)); gap: 1px; background: var(--vp-c-divider); border: 1px solid var(--vp-c-divider); }
+.cell { display: flex; flex-direction: column; align-items: center; padding: 0.5rem; background: var(--vp-c-bg); text-decoration: none; }
+.cell img { width: 64px; height: 64px; object-fit: contain; }
+.cp-id { font-family: monospace; font-size: 0.7rem; color: var(--vp-c-text-3); margin-top: 0.25rem; }
+</style>

data/lib/ucode/site/template/components/CharView.vue

@@ -0,0 +1,85 @@
+<script setup>
+import { ref, onMounted, computed } from "vue";
+
+const props = defineProps({ codepoint: { type: [String, Number], required: true } });
+
+const cpData = ref(null);
+const cpToBlock = ref({});
+const error = ref(null);
+
+const cpId = computed(() => {
+  const n = Number(props.codepoint);
+  return `U+${n.toString(16).toUpperCase().padStart(4, "0")}`;
+});
+
+const blockId = computed(() => cpToBlock.value[cpId.value]);
+const glyphUrl = computed(() =>
+  blockId.value ? `/data/blocks/${blockId.value}/${cpId.value}/glyph.svg` : null
+);
+
+onMounted(async () => {
+  try {
+    const [mappingRes] = await Promise.all([
+      fetch(`/data/index/codepoint_to_block.json`),
+    ]);
+    cpToBlock.value = await mappingRes.json();
+    if (!blockId.value) {
+      error.value = `No block found for ${cpId.value}`;
+      return;
+    }
+    const dataRes = await fetch(`/data/blocks/${blockId.value}/${cpId.value}/index.json`);
+    if (!dataRes.ok) {
+      error.value = `No data file for ${cpId.value} (HTTP ${dataRes.status})`;
+      return;
+    }
+    cpData.value = await dataRes.json();
+  } catch (e) {
+    error.value = e.message;
+  }
+});
+
+const propertyRows = computed(() => {
+  if (!cpData.value) return [];
+  return Object.entries(cpData.value)
+    .filter(([_, v]) => v !== null && v !== undefined && v !== "")
+    .map(([k, v]) => ({ key: k, value: JSON.stringify(v) }));
+});
+</script>
+
+<template>
+  <div class="char">
+    <header>
+      <h1><code>{{ cpId }}</code></h1>
+      <p v-if="cpData" class="name">{{ cpData.name }}</p>
+      <p v-if="error" class="error">{{ error }}</p>
+    </header>
+
+    <div v-if="cpData" class="body">
+      <aside v-if="glyphUrl" class="glyph">
+        <img :src="glyphUrl" :alt="cpData.name" />
+      </aside>
+
+      <section class="props">
+        <h2>Properties</h2>
+        <table>
+          <tbody>
+            <tr v-for="row in propertyRows" :key="row.key">
+              <th>{{ row.key }}</th>
+              <td><code>{{ row.value }}</code></td>
+            </tr>
+          </tbody>
+        </table>
+      </section>
+    </div>
+  </div>
+</template>
+
+<style scoped>
+.name { font-family: monospace; color: var(--vp-c-text-2); }
+.error { color: var(--vp-c-danger-1); }
+.body { display: grid; grid-template-columns: 200px 1fr; gap: 2rem; }
+.glyph img { width: 100%; height: auto; border: 1px solid var(--vp-c-divider); background: var(--vp-c-bg-alt); }
+.props table { border-collapse: collapse; }
+.props th, .props td { text-align: left; padding: 0.25rem 0.75rem; border-bottom: 1px solid var(--vp-c-divider); vertical-align: top; }
+.props th { font-weight: 500; color: var(--vp-c-text-2); white-space: nowrap; }
+</style>

data/lib/ucode/site/template/components/PlaneView.vue

@@ -0,0 +1,56 @@
+<script setup>
+import { ref, onMounted, computed } from "vue";
+
+const props = defineProps({ plane: { type: [String, Number], required: true } });
+
+const planeMeta = ref(null);
+const blocksIndex = ref([]);
+
+const planeNumber = computed(() => Number(props.plane));
+const blocksForPlane = computed(() =>
+  blocksIndex.value.filter((b) => b.plane_number === planeNumber.value)
+);
+
+onMounted(async () => {
+  const [planeRes, blocksRes] = await Promise.all([
+    fetch(`/data/planes/${planeNumber.value}.json`),
+    fetch(`/data/blocks/index.json`),
+  ]);
+  planeMeta.value = await planeRes.json();
+  blocksIndex.value = await blocksRes.json();
+});
+</script>
+
+<template>
+  <div class="plane">
+    <header v-if="planeMeta">
+      <h1>
+        <code>U+{{ planeNumber.toString(16).toUpperCase().padStart(2, "0") }}0000</code>
+        –
+        <code>U+{{ planeNumber.toString(16).toUpperCase().padStart(2, "0") }}FFFF</code>
+      </h1>
+      <p class="lead">{{ planeMeta.name }} ({{ planeMeta.abbrev }})</p>
+    </header>
+
+    <section>
+      <h2>Blocks ({{ blocksForPlane.length }})</h2>
+      <ul class="block-list">
+        <li v-for="b in blocksForPlane" :key="b.id">
+          <a :href="`/block/${b.id}`">{{ b.name }}</a>
+          <small>
+            U+{{ b.first_cp.toString(16).toUpperCase().padStart(4, "0") }}
+            –
+            U+{{ b.last_cp.toString(16).toUpperCase().padStart(4, "0") }}
+          </small>
+        </li>
+      </ul>
+    </section>
+  </div>
+</template>
+
+<style scoped>
+.lead { color: var(--vp-c-text-2); }
+.block-list { list-style: none; padding: 0; display: grid; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); gap: 0.5rem; }
+.block-list li { padding: 0.5rem 0.75rem; border: 1px solid var(--vp-c-divider); border-radius: 4px; }
+.block-list small { display: block; color: var(--vp-c-text-3); font-family: monospace; }
+</style>

data/lib/ucode/site/template/components/SearchView.vue

@@ -0,0 +1,66 @@
+<script setup>
+import { ref, onMounted, computed } from "vue";
+import MiniSearch from "minisearch";
+
+const query = ref("");
+const results = ref([]);
+const status = ref("loading");
+let index = null;
+
+onMounted(async () => {
+  try {
+    const res = await fetch("/data/index/search.json");
+    const docs = await res.json();
+    index = new MiniSearch({
+      fields: ["name", "id"],
+      storeFields: ["name", "gc", "sc"],
+      idField: "id",
+    });
+    await index.addAllAsync(docs);
+    status.value = "ready";
+  } catch (e) {
+    status.value = "error";
+  }
+});
+
+const run = () => {
+  if (!index || query.value.trim().length < 2) {
+    results.value = [];
+    return;
+  }
+  results.value = index.search(query.value, { prefix: true, fuzzy: 0.2, limit: 50 });
+};
+</script>
+
+<template>
+  <div class="search">
+    <h1>Search</h1>
+    <p v-if="status === 'loading'">Loading index…</p>
+    <p v-else-if="status === 'error'">Failed to load search index.</p>
+    <div v-else>
+      <input
+        v-model="query"
+        @input="run"
+        placeholder="Search by name (e.g. LATIN CAPITAL LETTER A) or U+0041"
+        autofocus
+      />
+      <p class="meta">{{ results.length }} result(s)</p>
+      <ul class="results">
+        <li v-for="r in results" :key="r.id">
+          <a :href="`/char/${r.id.slice(2)}`">
+            <code>{{ r.id }}</code> — {{ r.name }}
+          </a>
+          <small v-if="r.gc || r.sc">({{ [r.gc, r.sc].filter(Boolean).join(", ") }})</small>
+        </li>
+      </ul>
+    </div>
+  </div>
+</template>
+
+<style scoped>
+input { width: 100%; padding: 0.5rem 0.75rem; font-size: 1rem; border: 1px solid var(--vp-c-divider); border-radius: 4px; background: var(--vp-c-bg); color: var(--vp-c-text-1); }
+.meta { color: var(--vp-c-text-3); font-size: 0.85rem; }
+.results { list-style: none; padding: 0; }
+.results li { padding: 0.5rem 0; border-bottom: 1px solid var(--vp-c-divider); }
+.results small { color: var(--vp-c-text-3); margin-left: 0.5rem; }
+</style>