@muhammara/native-core 7.0.0-beta.2 → 7.0.0-beta.4

package/lib/Recipe.js

@@ -263,8 +263,9 @@ class Recipe {
   /**
    * End the pdfDoc. Finalization happens once; later calls do not rewrite the
    * PDF and invoke the callback with the completed output when applicable.
-   * A failed finalization retires the Recipe and later calls rethrow the
-   * original error.
+   * An active page is finished first, so a forgotten endPage() does not cost
+   * that page. A failed finalization retires the Recipe and later calls
+   * rethrow the original error.
    * @function
    * @memberof Recipe
    * @param {function} [callback] - The callback function.
@@ -292,6 +293,10 @@ class Recipe {
     if (deletingPages && this.page) {
       throw new Error("Finish the current page before endPDF");
     }
+    // The writer cannot finalize around an open content stream: it drops the
+    // page and emits a catalog with no page tree. Finish the page for the
+    // caller instead, before finalization retires the Recipe.
+    if (this.page) this.endPage();
     try {
       this._deletePages();
       this._writeInfo();

package/lib/recipe/annotation.js

@@ -1,3 +1,17 @@
+/**
+ * Encodes annotation text as a PDF text string, so characters outside
+ * PDFDocEncoding are written as UTF-16BE instead of raw UTF-8 bytes.
+ * @private
+ * @param {Object} writer - The PDF writer.
+ * @param {string} [value] - The text to encode.
+ * @returns {number[]} The encoded string bytes.
+ */
+function textString(writer, value) {
+  var text = writer.createPDFTextString();
+  text.fromString(String(value ?? ""));
+  return text.toBytesArray();
+}
+
 /**
  * Create a comment annotation
  * @name comment
@@ -171,9 +185,9 @@ exports._annot = function _annot(subtype, args = {}, pageNumber, ref) {
     .writeKey("Rect")
     .writeRectangleValue(position)
     .writeKey("Subj")
-    .writeLiteralStringValue(params.subject)
+    .writeLiteralStringValue(textString(this.writer, params.subject))
     .writeKey("T")
-    .writeLiteralStringValue(params.title || "")
+    .writeLiteralStringValue(textString(this.writer, params.title))
     .writeKey("M")
     .writeLiteralStringValue(
       this.writer.createPDFDate(new Date(params.date)).toString(),
@@ -194,16 +208,16 @@ exports._annot = function _annot(subtype, args = {}, pageNumber, ref) {
    */
   if (text && params.richText) {
     const richText =
-      text.substring(0, 5) !== "<?xml" ? contentToRC(text) : params.richText;
+      text.substring(0, 5) !== "<?xml" ? contentToRC(text) : text;
     const richTextContent = richText;
     this.dictionaryContext
       .writeKey("RC")
-      .writeLiteralStringValue(richTextContent);
+      .writeLiteralStringValue(textString(this.writer, richTextContent));
   } else if (text) {
     const textContent = text;
     this.dictionaryContext
       .writeKey("Contents")
-      .writeLiteralStringValue(textContent);
+      .writeLiteralStringValue(textString(this.writer, textContent));
   }
 
   if (reply && ref) {

package/lib/recipe/appendPage.js

@@ -2,7 +2,8 @@ const muhammara = require("../muhammara");
 const utils = require("./utils");
 
 /**
- * Append pages from the other pdf to the current pdf
+ * Append pages from the other pdf to the current pdf. An active page is
+ * finished first, so appended pages follow it in the output.
  * @name appendPage
  * @function
  * @memberof Recipe#
@@ -55,6 +56,10 @@ exports.appendPage = function appendPage(pdfSrc, pages = []) {
     }
     return range.map(transformPageNumber);
   });
+  // Appending writes whole pages, which the writer cannot do around an open
+  // content stream. Close the active page only once the selection is known to
+  // be valid, so a rejected selection leaves the page exactly as it was.
+  if (this.page) this.endPage();
   if (pages.length > 0) {
     utils.appendPDFPagesFromPDFWithAnnotations(this.writer, pdfSrc, {
       specificRanges: pages,

package/lib/recipe/htmlToTextObjects.js

@@ -10,7 +10,6 @@ const DOMParser = require("@xmldom/xmldom").DOMParser;
  * @returns {Object[]} The parsed text layout objects.
  */
 exports.htmlToTextObjects = function (htmlCodes, options = {}) {
-  htmlCodes = htmlCodes.replace(/<br\/?>/g, "<p>[@@DONOT_RENDER_THIS@@]</p>");
   const nodes = new DOMParser().parseFromString(
     `<html>${htmlCodes}</html>`,
     "text/html",
@@ -39,6 +38,33 @@ function needsLineBreaker(tagName = "") {
   return lineBreakers.includes(tagName);
 }
 
+/**
+ * Reports whether text in `node` begins a visual line: after a line break or
+ * block element, or at the start of a block. Only such text drops its
+ * leading whitespace; text that follows inline content keeps one space.
+ * @private
+ */
+function startsLine(node) {
+  const previous = node.previousSibling;
+  if (previous) {
+    const tag = (previous.tagName || "").toLowerCase();
+    return tag === "br" || needsLineBreaker(tag) || ["ul", "ol"].includes(tag);
+  }
+  const parent = node.parentNode;
+  const parentTag = (
+    parent && parent.tagName ? parent.tagName : ""
+  ).toLowerCase();
+  if (
+    !parent ||
+    !parentTag ||
+    parentTag === "html" ||
+    needsLineBreaker(parentTag)
+  ) {
+    return true;
+  }
+  return startsLine(parent);
+}
+
 function isBoldTag(tagName = "") {
   const boldTags = ["b", "strong"];
   return boldTags.includes(tagName);
@@ -84,6 +110,14 @@ function parseNode(node, options) {
     }
   }
   let value = node.data ? node.data.replace(/^\s*/gm, "") : null;
+  if (value !== null && /^\s/.test(node.data) && !startsLine(node)) {
+    value = " " + value;
+  }
+  // Whitespace before a line break would only pad the end of the line.
+  const next = node.nextSibling;
+  if (value !== null && next && /^br$/i.test(next.tagName || "")) {
+    value = value.replace(/\s+$/, "");
+  }
   if (value && value.charCodeAt(0) == 8203) {
     // zero width space
     value = value.substring(1);
@@ -99,6 +133,8 @@ function parseNode(node, options) {
     attributes,
     styles,
     needsLineBreaker: needsLineBreaker(node.tagName),
+    // An explicit line break; text layout ends the current line here.
+    lineBreak: /^br$/i.test(node.tagName || ""),
     size: options.size,
     sizeRatio: getFontSizeRatio(node.tagName),
     sizeRatios: [getFontSizeRatio(node.tagName)],

package/lib/recipe/page.js

@@ -1,4 +1,5 @@
 const muhammara = require("../muhammara");
+const { PAGE_CONTEXT_STATE } = require("./utils");
 
 /** Builds the retained page tree while marking deleted leaf pages. @private */
 function readPageTree(
@@ -615,7 +616,8 @@ exports.createPage = function createPage(pageWidth, pageHeight, margins) {
     }
   }
   // from 0
-  this.metadata.pageCount += 1;
+  this.metadata.pageCount =
+    (this.metadata.pageCount ?? this.metadata.pages ?? 0) + 1;
   const pageNumber = this.metadata.pageCount;
   const dimensions = [0, 0, pageWidth, pageHeight];
   const layout = pageWidth > pageHeight ? "landscape" : "portrait";
@@ -635,7 +637,7 @@ exports.createPage = function createPage(pageWidth, pageHeight, margins) {
   this.pageNumber = pageNumber;
   this.pageContext = this.writer.startPageContentContext(this.page);
   this.editingPage = false;
-  this.contextState = "active-new";
+  this.contextState = PAGE_CONTEXT_STATE.ACTIVE_NEW;
   if (!this.isNewPDF) this.pagesCreated = true;
 
   if (margins) {
@@ -714,7 +716,8 @@ exports.endPage = function endPage() {
   }
 
   if (this.page.endContext) {
-    if (this.contextState === "active-edit") this.page.endContext();
+    if (this.contextState === PAGE_CONTEXT_STATE.ACTIVE_EDIT)
+      this.page.endContext();
     this.page.writePage();
   } else {
     this.writer.writePage(this.page);
@@ -723,7 +726,7 @@ exports.endPage = function endPage() {
   this.pageContext = null;
   this.pageNumber = 0;
   this.editingPage = false;
-  this.contextState = "idle";
+  this.contextState = PAGE_CONTEXT_STATE.IDLE;
 
   return this;
 };
@@ -748,7 +751,7 @@ exports.editPage = function editPage(pageNumber) {
   this.pageNumber = pageNumber;
   this.pageContext = pageModifier.startContext().getContext();
   this.editingPage = true;
-  this.contextState = "active-edit";
+  this.contextState = PAGE_CONTEXT_STATE.ACTIVE_EDIT;
   this.modifiedSourcePages.add(pageNumber);
 
   this._resumePageRotation(pageNumber);
@@ -1016,13 +1019,13 @@ exports.getCurrentPageInfo = function getCurrentPageInfo() {
  * @throws {Error} If there is no active page content context.
  */
 exports.pauseContext = function pauseContext() {
-  if (this.contextState === "active-edit") {
+  if (this.contextState === PAGE_CONTEXT_STATE.ACTIVE_EDIT) {
     this.page.endContext();
     this.pageContext = null;
-    this.contextState = "paused-edit";
-  } else if (this.contextState === "active-new") {
+    this.contextState = PAGE_CONTEXT_STATE.PAUSED_EDIT;
+  } else if (this.contextState === PAGE_CONTEXT_STATE.ACTIVE_NEW) {
     this.writer.pausePageContentContext(this.pageContext);
-    this.contextState = "paused-new";
+    this.contextState = PAGE_CONTEXT_STATE.PAUSED_NEW;
   } else {
     throw new Error("No active page content context to pause");
   }
@@ -1038,12 +1041,12 @@ exports.pauseContext = function pauseContext() {
  * @throws {Error} If there is no paused page content context.
  */
 exports.resumeContext = function resumeContext() {
-  if (this.contextState === "paused-edit") {
+  if (this.contextState === PAGE_CONTEXT_STATE.PAUSED_EDIT) {
     this.pageContext = this.page.startContext().getContext();
     this._resumePageRotation();
-    this.contextState = "active-edit";
-  } else if (this.contextState === "paused-new") {
-    this.contextState = "active-new";
+    this.contextState = PAGE_CONTEXT_STATE.ACTIVE_EDIT;
+  } else if (this.contextState === PAGE_CONTEXT_STATE.PAUSED_NEW) {
+    this.contextState = PAGE_CONTEXT_STATE.ACTIVE_NEW;
   } else {
     throw new Error("No paused page content context to resume");
   }

package/lib/recipe/parameters.js

@@ -1,3 +1,5 @@
+const { PAGE_CONTEXT_STATE } = require("./utils");
+
 this.mediumSizes = {
   executive: [521.86, 756.0], // north american
   folio: [612.0, 936.0],
@@ -81,7 +83,7 @@ exports._setParameters = function _setParameters() {
   };
 
   this._margin = Object.assign({}, this.default.pageMargin);
-  this.contextState = "idle";
+  this.contextState = PAGE_CONTEXT_STATE.IDLE;
   this.modifiedSourcePages = new Set();
 
   // Object.assign(this, options);