@bettercms-ai/convert 0.9.0 → 0.10.0

package/dist/index.d.ts

@@ -37,6 +37,18 @@ interface BriefPath {
     kind: string;
     /** The declared field type, which `kind` deliberately loses (richtext vs document). */
     type: string;
+    /**
+     * The richtext value IS the element's whole inline subtree — the field's `config.inline`.
+     *
+     * A heading with a styled run (`<h1>Notes from a <span class="serif">web studio</span></h1>`)
+     * derives as ONE richtext field whose value is that innerHTML, not as the text node beside the
+     * span. Without this the codemod cannot tell it from the icon+text shape, whose value really is
+     * one text node — and it wraps the text node in a `<span>`, leaving the styled run rendered
+     * twice. Optional on purpose: a brief written before the distinction existed carries none, and
+     * the old wrap-the-text-node reading is still the right answer for it.
+     * @see derive-schema.ts textType, which mints `config: { inline: true }` for exactly this shape.
+     */
+    inline?: true;
     label: string;
     /** The copy the REPO renders today. Null when the import captured no default. */
     original: string | null;
@@ -142,8 +154,13 @@ interface UnlocatedPath {
  * correctly formatted file, which is every file.
  */
 declare const flat: (value: string) => string;
-/** The same text with tags removed — what a reader sees, not what the file says. */
-declare const stripTags: (value: string) => string;
+/**
+ * The same text with tags removed — what a reader sees, not what the file says.
+ *
+ * `br` is what a line break becomes. A SPACE is what a reader sees; the empty string is what
+ * `textContent` sees, and the matcher works on the latter. @see locate, which searches both.
+ */
+declare const stripTags: (value: string, br?: string) => string;
 interface BriefLike {
     pages: {
         route: string;
@@ -192,7 +209,9 @@ declare function locate(brief: BriefLike, sources: SourceFile[], options?: Locat
  * sentence appears in three files" and "this field is bound" are different questions — and a path
  * is only `rewritten` when EVERY located occurrence of it was.
  */
-type PendingReason = "NO_ORIGINAL" | "NOT_IN_SOURCE" | "DIALECT_UNSUPPORTED" | "IN_EXPRESSION" | "IN_SCRIPT_OR_COMMENT" | "IN_DATA_FILE" | "AMBIGUOUS_LITERAL" | "KIND_MISMATCH" | "SUBSTRING_ONLY" | "PROP_TARGET_NOT_FOUND" | "PROP_DRILLED_DEEP" | "REPEATER_FIXED_LENGTH"
+type PendingReason = "NO_ORIGINAL" | "NOT_IN_SOURCE" | "DIALECT_UNSUPPORTED" | "IN_EXPRESSION" | "IN_SCRIPT_OR_COMMENT" | "IN_DATA_FILE" | "AMBIGUOUS_LITERAL" | "KIND_MISMATCH" | "SUBSTRING_ONLY"
+/** An INLINE richtext value whose element holds a component, a `<slot>` or an expression. */
+ | "SUBTREE_NOT_HTML" | "PROP_TARGET_NOT_FOUND" | "PROP_DRILLED_DEEP" | "REPEATER_FIXED_LENGTH"
 /** One array declaration, two loops over it: the row this leaf belongs to has two homes. */
  | "REPEATER_AMBIGUOUS"
 /** An image whose built url maps to no repository asset, or to more than one. */
@@ -515,6 +534,12 @@ interface Site {
     literal: string;
     /** The literal is only PART of what this element renders; rewriting it would eat the rest. */
     partial: boolean;
+    /**
+     * The subtree holds markup only the FRAMEWORK renders — a component, a `<slot>`, an expression.
+     *
+     * Absent on every site whose children are plain html, which is the common case. @see ownsSubtree
+     */
+    nonHtml?: true;
     /** Tag chain from the document root down to the element's parent, outermost first. */
     ancestors: string[];
     /** Index among the element's ELEMENT siblings. */
@@ -629,6 +654,8 @@ interface Rewrite {
     path: string;
     /** The brief's kind: "text" | "richtext" | "image". */
     kind: string;
+    /** The richtext value is the element's whole inline subtree. @see BriefPath.inline */
+    inline?: true;
     /** The copy the repo renders today, kept as the in-code fallback. */
     fallback: string;
     /** The identifier this page's snapshot is imported under. */
@@ -1132,6 +1159,41 @@ declare const SOURCE_EXTENSIONS: string[];
 /** Is this repository-relative path a file worth opening? */
 declare function isSourceCandidate(path: string): boolean;
 
+/**
+ * Text that is NOT copy, and is therefore outside both halves of the ratio.
+ *
+ * 🔴 IN THE DENOMINATOR TOO. Counting a `✓`, a `01`, the platform badge or a skip link as
+ * unaddressable text would put a floor under every route's percentage that no amount of binding
+ * could lift — and the target is a number someone has to be able to reach. They are not copy an
+ * author edits, so they are not copy this measures.
+ *
+ *  - a decorative glyph: two characters or fewer, a single non-letter symbol, or an ordinal
+ *    (`01`, `2.`, `3 /`) that numbers a list rather than saying anything;
+ *  - the platform badge every free site renders;
+ *  - a skip link, which exists for keyboards and is never edited.
+ */
+declare const isDecorativeGlyph: (text: string) => boolean;
+interface XrayBucket {
+    /** The element that renders it. */
+    tag: string;
+    /** Its landmark chain, outermost first — `main>ul`. */
+    context: string;
+    chars: number;
+    nodes: number;
+    /** Up to two examples, ≤ 50 characters each. */
+    samples: string[];
+}
+interface Xray {
+    /** Characters of visible copy, decorative glyphs excluded. */
+    visible: number;
+    /** Of those, the characters no binding owns. */
+    unaddressable: number;
+    /** The biggest buckets, largest first — at most 12. */
+    buckets: XrayBucket[];
+}
+/** What a reader sees on this page, and how much of it no field owns. */
+declare function measureUnaddressable(html: string): Xray;
+
 /**
  * The project's own `compilerOptions.paths`, when a tsconfig is among the files.
  *
@@ -1644,4 +1706,4 @@ declare function convertSources(briefIn: Brief, sources: SourceFile[], options?:
     receipt: ConversionReceipt;
 }>;
 
-export { ASTRO_LANE_PENDING, type AstNode, type AttrBinding, type Brief, type BriefForm, type BriefPage, type BriefPath, COMPONENT_DIR, CONTENT_DIR, type CanvasResult, type CanvasSource, type Carried, type ComponentizeOptions, type ComponentizePlan, type ComponentizePlanComponent, type ComponentizePlanField, type ComponentizePlanPage, type ComponentizePlanSection, type ComponentizeReceipt, type ConversionReceipt, ConvertError, type ConvertErrorCode, type ConvertOptions, DIALECT_RULES, type DataMember, type DataResult, type DeclaredPath, type Dialect, type DynamicBinding, FORM_SCRIPT, type FileDeclaration, type FindSitesResult, FormInvariantError, type FormPendingReason, type FormsReceipt, HELPER_PATH, HELPER_VERSION, type ImageResult, type ImageTarget, type ImportedFrom, type LlmFallback, type LocatedPath, type Node, PARSE_FILE, type ParsedFile, type ParserError, type PathLocator, type PendingForm, type PendingPath, type PendingReason, type PendingSection, type PlanFile, REGISTRY_MARKER, type ReceiptFile, ReceiptInvariantError, type ReceiptNote, type Rewrite, SECTIONS_LIB, SECTION_MARKER, SKIP_DIRS, SKIP_FILES, SOURCE_EXTENSIONS, SSR_DRAFT_RECIPE, STUB_CONTENT, SectionInvariantError, type SectionPendingReason, type SectionsReceipt, type Site, type SiteWhere, type SourceFile, type Splice, type TargetIdentity, type UnlocatedPath, type ValidationCode, ValidationError, aliasesFrom, assertForms, assertReceipt, assertSections, assetCandidates, assetImport, attrsFor, bindDataLiterals, bindImages, briefDigest, canvasBridge, carryFor, componentizeSources, convertSources, convertedHere, coverageOf, declaringElements, dialectOf, exportedFunction, extendedConfigs, findSites, findSitesTolerant, flat, helperSource, importsIn, isDynamicRoute, isKnownHelper, isSourceCandidate, locate, overlapping, pageFilesFor, parseFile, propsAttribute, readBrief, readComponentizePlan, readDeclarations, readExpr, readPlan, relativeImport, resolveSpecifier, rewriteFile, routeOfFile, stripTags, unreadableConfigs, walkAst, wireForms };
+export { ASTRO_LANE_PENDING, type AstNode, type AttrBinding, type Brief, type BriefForm, type BriefPage, type BriefPath, COMPONENT_DIR, CONTENT_DIR, type CanvasResult, type CanvasSource, type Carried, type ComponentizeOptions, type ComponentizePlan, type ComponentizePlanComponent, type ComponentizePlanField, type ComponentizePlanPage, type ComponentizePlanSection, type ComponentizeReceipt, type ConversionReceipt, ConvertError, type ConvertErrorCode, type ConvertOptions, DIALECT_RULES, type DataMember, type DataResult, type DeclaredPath, type Dialect, type DynamicBinding, FORM_SCRIPT, type FileDeclaration, type FindSitesResult, FormInvariantError, type FormPendingReason, type FormsReceipt, HELPER_PATH, HELPER_VERSION, type ImageResult, type ImageTarget, type ImportedFrom, type LlmFallback, type LocatedPath, type Node, PARSE_FILE, type ParsedFile, type ParserError, type PathLocator, type PendingForm, type PendingPath, type PendingReason, type PendingSection, type PlanFile, REGISTRY_MARKER, type ReceiptFile, ReceiptInvariantError, type ReceiptNote, type Rewrite, SECTIONS_LIB, SECTION_MARKER, SKIP_DIRS, SKIP_FILES, SOURCE_EXTENSIONS, SSR_DRAFT_RECIPE, STUB_CONTENT, SectionInvariantError, type SectionPendingReason, type SectionsReceipt, type Site, type SiteWhere, type SourceFile, type Splice, type TargetIdentity, type UnlocatedPath, type ValidationCode, ValidationError, type Xray, type XrayBucket, aliasesFrom, assertForms, assertReceipt, assertSections, assetCandidates, assetImport, attrsFor, bindDataLiterals, bindImages, briefDigest, canvasBridge, carryFor, componentizeSources, convertSources, convertedHere, coverageOf, declaringElements, dialectOf, exportedFunction, extendedConfigs, findSites, findSitesTolerant, flat, helperSource, importsIn, isDecorativeGlyph, isDynamicRoute, isKnownHelper, isSourceCandidate, locate, measureUnaddressable, overlapping, pageFilesFor, parseFile, propsAttribute, readBrief, readComponentizePlan, readDeclarations, readExpr, readPlan, relativeImport, resolveSpecifier, rewriteFile, routeOfFile, stripTags, unreadableConfigs, walkAst, wireForms };

package/dist/index.js

@@ -2560,6 +2560,7 @@ var IMAGE_SOURCES = {
   video: ["poster"]
 };
 var isImageSource = (site) => site.where === "attr" && (IMAGE_SOURCES[site.tag.toLowerCase()] ?? []).includes(site.attr ?? "");
+var ownsSubtree = (kind, site, inline) => kind === "richtext" && (inline === true ? !site.nonHtml : !(site.mixed && site.textRange !== null));
 var norm = (value) => decodeEntities(value.normalize("NFC")).replace(/\s+/g, " ").trim();
 var BCMS_ATTRS = [
   "data-bcms-field",
@@ -2592,6 +2593,7 @@ function openTagEnd(content, start) {
   }
   return content.length;
 }
+var OPAQUE_TAGS = /* @__PURE__ */ new Set(["pre"]);
 var isBlank = (value) => value.trim() === "";
 function renderedText(node) {
   if (node.type === "text") return node.value;
@@ -2599,6 +2601,11 @@ function renderedText(node) {
   return "";
 }
 var attrsOf = (node) => Object.fromEntries(node.attrs.map((a) => [a.name, a.value]));
+function nonHtmlSubtree(node) {
+  return node.children.some(
+    (c) => c.type === "expr" || c.type === "element" && (!isIntrinsic(c.tag) || c.tag === "slot" || nonHtmlSubtree(c))
+  );
+}
 function soleTextChild(node) {
   const texts = node.children.filter((c) => c.type === "text" && !isBlank(c.value));
   return texts.length === 1 ? texts[0] : null;
@@ -2635,10 +2642,12 @@ function collectSites(roots, literals) {
     if (node.type !== "element") return matched;
     const attrs = attrsOf(node);
     let index = 0;
-    for (const child of node.children) {
-      const childWalk = { ancestors: [...walk.ancestors, node.tag], siblingIndex: index };
-      if (child.type === "element") index++;
-      for (const hit of visit(child, childWalk)) matched.add(hit);
+    if (!OPAQUE_TAGS.has(node.tag)) {
+      for (const child of node.children) {
+        const childWalk = { ancestors: [...walk.ancestors, node.tag], siblingIndex: index };
+        if (child.type === "element") index++;
+        for (const hit of visit(child, childWalk)) matched.add(hit);
+      }
     }
     if (node.raw) {
       const text2 = norm(renderedTextRaw(node));
@@ -2664,6 +2673,7 @@ function collectSites(roots, literals) {
         textRange: soleTextChild(node)?.range ?? null,
         literal,
         partial: false,
+        ...nonHtmlSubtree(node) ? { nonHtml: true } : {},
         ancestors: walk.ancestors,
         siblingIndex: walk.siblingIndex
       });
@@ -2689,6 +2699,7 @@ function collectSites(roots, literals) {
           textRange: own.range,
           literal,
           partial: false,
+          ...nonHtmlSubtree(node) ? { nonHtml: true } : {},
           ancestors: walk.ancestors,
           siblingIndex: walk.siblingIndex
         });
@@ -2739,10 +2750,12 @@ function partials(node, literals, walk, out) {
   if (node.type !== "element" || node.raw) return false;
   let deeper = false;
   let index = 0;
-  for (const child of node.children) {
-    const childWalk = { ancestors: [...walk.ancestors, node.tag], siblingIndex: index };
-    if (child.type === "element") index++;
-    if (partials(child, literals, childWalk, out)) deeper = true;
+  if (!OPAQUE_TAGS.has(node.tag)) {
+    for (const child of node.children) {
+      const childWalk = { ancestors: [...walk.ancestors, node.tag], siblingIndex: index };
+      if (child.type === "element") index++;
+      if (partials(child, literals, childWalk, out)) deeper = true;
+    }
   }
   if (deeper) return true;
   const text = norm(renderedText(node));
@@ -4382,14 +4395,23 @@ function isKnownHelper(content, dialect) {
 
 // src/locate.ts
 var flat = (value) => value.replace(/\s+/g, " ").trim();
-var stripTags = (value) => value.replace(/<br\s*\/?>/gi, " ").replace(/<[^>]*>/g, "");
+var stripTags = (value, br = " ") => value.replace(/<br\s*\/?>/gi, br).replace(/<[^>]*>/g, "");
 function locate(brief, sources, options) {
   const flattened = sources.map((s) => {
     const readable = unwrapJsxStrings(decodeEntities(s.content));
     return {
       path: s.path,
       content: flat(readable),
-      text: options?.stripTags ? flat(stripTags(readable)) : null
+      text: options?.stripTags ? flat(stripTags(readable)) : null,
+      /**
+       * 🔴 A `<br>` IS A SPACE TO A READER AND NOTHING TO `textContent`, and the derive lane reads
+       * the DOM. `<h1>Launch your site<br /><span>fast.</span></h1>` is one INLINE richtext field
+       * whose value is `Launch your sitefast.` — no space, because no text node holds one — while
+       * this prefilter's only reading inserted one, so the file was never opened and every split
+       * heading on every imported site came back NOT_IN_SOURCE. Both readings are searched: this
+       * decides which FILES to open, and `collectSites` still has to match an element exactly.
+       */
+      tight: options?.stripTags ? flat(stripTags(readable, "")) : null
     };
   });
   const located = [];
@@ -4401,7 +4423,9 @@ function locate(brief, sources, options) {
         unlocated.push({ route: page.route, scope: p.scope, path: p.path, reason: "no-original" });
         continue;
       }
-      const files = flattened.filter((f) => f.content.includes(original) || f.text !== null && f.text.includes(original)).map((f) => f.path);
+      const files = flattened.filter(
+        (f) => f.content.includes(original) || f.text !== null && f.text.includes(original) || f.tight !== null && f.tight.includes(original)
+      ).map((f) => f.path);
       if (files.length === 0) {
         unlocated.push({ route: page.route, scope: p.scope, path: p.path, reason: "not-in-source" });
         continue;
@@ -5100,7 +5124,7 @@ function rewriteFile(content, dialect, rewrites, imports, extra = []) {
       addDeclaration(site.attrInsertAt, declaration(rewrite));
       continue;
     }
-    if (kind === "richtext" && !(site.mixed && site.textRange)) {
+    if (ownsSubtree(kind, site, rewrite.inline)) {
       if (!isIntrinsic(site.tag)) continue;
       if (rule.richtext && site.inner) {
         source.remove(site.inner.start, site.inner.end);
@@ -6386,6 +6410,66 @@ function isSourceCandidate(path) {
   return lower.split("/").slice(0, -1).some((segment) => TEMPLATE_DIRS.has(segment));
 }
 
+// src/xray.ts
+var SKIP_TAGS = /* @__PURE__ */ new Set(["svg", "iframe", "head", "title"]);
+var CONTEXT_TAGS = /* @__PURE__ */ new Set([
+  "nav",
+  "header",
+  "footer",
+  "main",
+  "aside",
+  "form",
+  "button",
+  "a",
+  "ul",
+  "ol",
+  "table",
+  "dialog",
+  "details",
+  "figure",
+  "blockquote"
+]);
+var BOUND_ATTRS = ["data-bcms-field", "data-bcms-layout-field", "data-bcms-path", "data-bcms-block"];
+var isDecorativeGlyph = (text) => text.length <= 2 || /^\d{1,2}\s*[/.)\-–]?$/.test(text) || !new RegExp("\\p{L}", "u").test(text) || /^made with\b/i.test(text) || /^skip to\b/i.test(text);
+var MAX_BUCKETS = 12;
+var flat2 = (value) => value.replace(/\s+/g, " ").trim();
+function measureUnaddressable(html) {
+  const parsed = parse2(html);
+  let visible = 0;
+  let unaddressable = 0;
+  const buckets = /* @__PURE__ */ new Map();
+  const visit = (node, tag, context, bound) => {
+    if (node.type === "text") {
+      const text = flat2(node.value);
+      if (!text || isDecorativeGlyph(text)) return;
+      visible += text.length;
+      if (bound) return;
+      unaddressable += text.length;
+      const key = `${tag} @ ${context}`;
+      const held = buckets.get(key) ?? { tag, context, chars: 0, nodes: 0, samples: [] };
+      held.chars += text.length;
+      held.nodes += 1;
+      if (held.samples.length < 2) held.samples.push(text.slice(0, 50));
+      buckets.set(key, held);
+      return;
+    }
+    if (node.type !== "element" || node.raw || SKIP_TAGS.has(node.tag)) return;
+    const attrs = new Map(node.attrs.map((a) => [a.name, a.value]));
+    if (attrs.has("hidden") || attrs.get("aria-hidden") === "true" || /display\s*:\s*none/.test(attrs.get("style") ?? "")) {
+      return;
+    }
+    const owned = bound || BOUND_ATTRS.some((name) => attrs.has(name));
+    const chain = CONTEXT_TAGS.has(node.tag) ? context === "body" ? node.tag : `${context}>${node.tag}` : context;
+    for (const child of node.children) visit(child, node.tag, chain, owned);
+  };
+  for (const root of parsed.roots) visit(root, "body", "body", false);
+  return {
+    visible,
+    unaddressable,
+    buckets: [...buckets.values()].sort((a, b) => b.chars - a.chars).slice(0, MAX_BUCKETS)
+  };
+}
+
 // src/declared.ts
 async function readDeclarations(file, content) {
   const dialect = dialectOf(file, content);
@@ -8274,7 +8358,8 @@ async function convertSources(briefIn, sources, options) {
         kind: p.kind,
         fallback: p.original ?? "",
         literal: norm(p.original ?? ""),
-        locator: p.locator
+        locator: p.locator,
+        ...p.inline ? { inline: true } : {}
       });
     }
   }
@@ -8499,6 +8584,16 @@ async function convertSources(briefIn, sources, options) {
       declaredRow(target, witness !== null, witness ?? byHand);
     } else rewritten++;
   }
+  const templated = dynamicBindings.filter((d) => d.prop === null);
+  for (let i = pending.length - 1; i >= 0; i--) {
+    const row = pending[i];
+    const target = targets.get(keyOf(row.route, row.scope, row.path));
+    if (row.reason !== "NO_ORIGINAL" || !target || target.fallback !== "") continue;
+    const shape = shapeOfPath(target.path);
+    if (!templated.some((d) => d.path === shape && attributable(target, d.file))) continue;
+    pending.splice(i, 1);
+    rewritten++;
+  }
   for (const [file, splices] of [...componentEdits].sort(([a], [b]) => a.localeCompare(b))) {
     const source = new MagicString6(contents.get(file));
     const seen = /* @__PURE__ */ new Set();
@@ -8811,6 +8906,10 @@ async function plan(file, content, dialect, targets, sites, importAt, outcomes,
       }
       let refusal = null;
       for (const site of fresh) {
+        if (target.inline && site.where === "text" && !ownsSubtree(target.kind, site, true)) {
+          refusal = "SUBTREE_NOT_HTML";
+          continue;
+        }
         let bindings;
         let bindingsMerge;
         if (site.where === "prop" && site.attr) {
@@ -8845,7 +8944,8 @@ async function plan(file, content, dialect, targets, sites, importAt, outcomes,
           // 🔴 AND ONLY WHEN THE ELEMENT'S SUBTREE IS THE VALUE. A value that is one TEXT NODE of
           // a mixed element gets a wrapper of its own, so taking the element's inner range as its
           // fallback swallowed the `<em>` beside it — the wrapper then rendered the sibling twice.
-          fallback: target.kind === "richtext" && site.inner && !(site.mixed && site.textRange) ? content.slice(site.inner.start, site.inner.end).trim() : target.fallback,
+          ...target.inline ? { inline: true } : {},
+          fallback: site.inner && ownsSubtree(target.kind, site, target.inline) ? content.slice(site.inner.start, site.inner.end).trim() : target.fallback,
           snapshot: snapshotName(target.slug),
           slug: target.slug,
           scope: target.scope,
@@ -9249,10 +9349,12 @@ export {
   flat,
   helperSource,
   importsIn,
+  isDecorativeGlyph,
   isDynamicRoute,
   isKnownHelper,
   isSourceCandidate,
   locate,
+  measureUnaddressable,
   overlapping,
   pageFilesFor,
   parseFile4 as parseFile,