@particle-academy/last-word 0.1.0 → 0.2.0

package/dist/index.d.cts

@@ -105,7 +105,7 @@ interface WriteResult {
  */
 
 /** Feature-parity baseline with PHP last-word; bumped independently on npm. */
-declare const VERSION = "0.1.0";
+declare const VERSION = "0.2.0";
 type Any$2 = any;
 declare const Agent: {
     /** Validate a doc without writing. Empty array = valid. */
@@ -239,7 +239,9 @@ declare function resolveImageSize(block: {
 
 /**
  * DocxReader — .docx bytes → Doc JSON model. Handles this package's own
- * writer output losslessly (round-trip) and tolerates Word-authored files:
+ * writer output losslessly (round-trip), the PHP mirror's output (same
+ * metadata slots since 0.2.0, plus the legacy `LastWordCode_{lang}` bookmark
+ * fallback for PHP ≤0.1.x files) and tolerates Word-authored files:
  * headings via pStyle Heading1-9 OR outlineLvl, numPr lists with ilvl nesting,
  * hyperlinks via rels, images via blip r:embed, page breaks, bottom-border-only
  * paragraphs → hr. Unknown constructs degrade to plain paragraphs, never throw.
@@ -258,6 +260,12 @@ declare class DocxReader {
     private walkBody;
     private parseSdt;
     private classifyParagraph;
+    /**
+     * Back-compat: the code language from a PHP last-word ≤0.1.x
+     * `LastWordCode_{lang}` bookmark on this paragraph, if present. The
+     * canonical carrier is the `lastword:code:{lang}` sdt tag.
+     */
+    private legacyBookmarkLanguage;
     /** Concatenated visible text of a paragraph (tabs and soft breaks included). */
     private plainText;
     private parseRuns;

package/dist/index.d.ts

@@ -105,7 +105,7 @@ interface WriteResult {
  */
 
 /** Feature-parity baseline with PHP last-word; bumped independently on npm. */
-declare const VERSION = "0.1.0";
+declare const VERSION = "0.2.0";
 type Any$2 = any;
 declare const Agent: {
     /** Validate a doc without writing. Empty array = valid. */
@@ -239,7 +239,9 @@ declare function resolveImageSize(block: {
 
 /**
  * DocxReader — .docx bytes → Doc JSON model. Handles this package's own
- * writer output losslessly (round-trip) and tolerates Word-authored files:
+ * writer output losslessly (round-trip), the PHP mirror's output (same
+ * metadata slots since 0.2.0, plus the legacy `LastWordCode_{lang}` bookmark
+ * fallback for PHP ≤0.1.x files) and tolerates Word-authored files:
  * headings via pStyle Heading1-9 OR outlineLvl, numPr lists with ilvl nesting,
  * hyperlinks via rels, images via blip r:embed, page breaks, bottom-border-only
  * paragraphs → hr. Unknown constructs degrade to plain paragraphs, never throw.
@@ -258,6 +260,12 @@ declare class DocxReader {
     private walkBody;
     private parseSdt;
     private classifyParagraph;
+    /**
+     * Back-compat: the code language from a PHP last-word ≤0.1.x
+     * `LastWordCode_{lang}` bookmark on this paragraph, if present. The
+     * canonical carrier is the `lastword:code:{lang}` sdt tag.
+     */
+    private legacyBookmarkLanguage;
     /** Concatenated visible text of a paragraph (tabs and soft breaks included). */
     private plainText;
     private parseRuns;

package/dist/index.js

@@ -1095,6 +1095,7 @@ var DocxReader = class {
     const blocks = [];
     let listBuf = [];
     let codeBuf = null;
+    let codeLang = null;
     let quoteBuf = null;
     const flushList = () => {
       if (listBuf.length > 0) {
@@ -1104,8 +1105,11 @@ var DocxReader = class {
     };
     const flushCode = () => {
       if (codeBuf !== null) {
-        blocks.push({ type: "code", text: codeBuf.join("\n") });
+        const block = { type: "code", text: codeBuf.join("\n") };
+        if (codeLang !== null) block.language = codeLang;
+        blocks.push(orderCodeKeys(block));
         codeBuf = null;
+        codeLang = null;
       }
     };
     const flushQuote = () => {
@@ -1130,7 +1134,10 @@ var DocxReader = class {
           } else if (kind.kind === "codeLine") {
             flushList();
             flushQuote();
-            if (codeBuf === null) codeBuf = [];
+            if (codeBuf === null) {
+              codeBuf = [];
+              codeLang = kind.language ?? null;
+            }
             codeBuf.push(kind.text);
           } else if (kind.kind === "quoteParagraph" && !ctx.insideQuote) {
             flushList();
@@ -1199,7 +1206,8 @@ var DocxReader = class {
       return { kind: "listItem", item: { numId, ilvl, runs: this.parseRuns(p, void 0) } };
     }
     if (/^(CodeBlock|SourceCode|HTMLPreformatted|Code)$/i.test(styleId)) {
-      return { kind: "codeLine", text: this.plainText(p) };
+      const language = this.legacyBookmarkLanguage(p);
+      return language !== void 0 ? { kind: "codeLine", text: this.plainText(p), language } : { kind: "codeLine", text: this.plainText(p) };
     }
     const blocks = [];
     const drawings = findAll(p, "drawing");
@@ -1235,6 +1243,21 @@ var DocxReader = class {
     }
     return { kind: "blocks", blocks };
   }
+  /**
+   * Back-compat: the code language from a PHP last-word ≤0.1.x
+   * `LastWordCode_{lang}` bookmark on this paragraph, if present. The
+   * canonical carrier is the `lastword:code:{lang}` sdt tag.
+   */
+  legacyBookmarkLanguage(p) {
+    for (const child of p.children) {
+      if (child.name !== "bookmarkStart") continue;
+      const name = at(child, "name") ?? "";
+      if (name.startsWith("LastWordCode_") && name.length > "LastWordCode_".length) {
+        return name.slice("LastWordCode_".length);
+      }
+    }
+    return void 0;
+  }
   /** Concatenated visible text of a paragraph (tabs and soft breaks included). */
   plainText(p) {
     return this.parseRuns(p, void 0).map((r) => r.text).join("");
@@ -2283,7 +2306,7 @@ var Validator = class {
 };
 
 // src/agent.ts
-var VERSION = "0.1.0";
+var VERSION = "0.2.0";
 function toU8(input) {
   return input instanceof Uint8Array ? input : new Uint8Array(input);
 }