Support for line breaking on hyphens (#451)

* support for line breaking on hyphens

* update test pdfs

packages/schemas/__tests__/text.test.ts

@@ -74,23 +74,29 @@ describe('getSplitPosition test with mocked font width calculations', () => {
     expect(getSplittedLines('aaaa', mockCalcValues)).toEqual(['aaaa']);
   });
 
-  it('splits a line to the nearest previous space', () => {
+  it('splits a line to the nearest previous breakable char', () => {
     expect(getSplittedLines('aaa bbb', mockCalcValues)).toEqual(['aaa', 'bbb']);
+    expect(getSplittedLines('top-hat', mockCalcValues)).toEqual(['top-', 'hat']);
+    expect(getSplittedLines('top—hat', mockCalcValues)).toEqual(['top—', 'hat']); // em dash
+    expect(getSplittedLines('top–hat', mockCalcValues)).toEqual(['top–', 'hat']); // en dash
   });
 
-  it('splits a line where the split point is on a space', () => {
+  it('splits a line where the split point is on a breakable char', () => {
     expect(getSplittedLines('aaaaa bbbbb', mockCalcValues)).toEqual(['aaaaa', 'bbbbb']);
+    expect(getSplittedLines('left-hand', mockCalcValues)).toEqual(['left-', 'hand']);
   });
 
   it('splits a long line in the middle of a word if too long', () => {
     expect(getSplittedLines('aaaaaa bbb', mockCalcValues)).toEqual(['aaaaa', 'a bbb']);
+    expect(getSplittedLines('aaaaaa-a b', mockCalcValues)).toEqual(['aaaaa', 'a-a b']);
+    expect(getSplittedLines('aaaaa-aa b', mockCalcValues)).toEqual(['aaaaa', '-aa b']);
   });
 
-  it('splits a long line without spaces at exactly 5 chars', () => {
+  it('splits a long line without breakable chars at exactly 5 chars', () => {
     expect(getSplittedLines('abcdef', mockCalcValues)).toEqual(['abcde', 'f']);
   });
 
-  it('splits a very long line without spaces at exactly 5 chars', () => {
+  it('splits a very long line without breakable chars at exactly 5 chars', () => {
     expect(getSplittedLines('abcdefghijklmn', mockCalcValues)).toEqual(['abcde', 'fghij', 'klmn']);
   });
 

packages/schemas/src/text/helper.ts

@@ -152,21 +152,41 @@ const getOverPosition = (textLine: string, calcValues: FontWidthCalcValues) => {
   return null;
 };
 
+/**
+ * Line breakable chars depend on the language and writing system.
+ * Western writing systems typically use spaces and hyphens as line breakable chars.
+ * Other writing systems often break on word boundaries so the following
+ * does not negatively impact them.
+ * However, this might need to be revisited for broader language support.
+ */
+const isLineBreakableChar = (char: string) => {
+  const lineBreakableChars = [' ', '-', "\u2014", "\u2013"];
+  return lineBreakableChars.includes(char);
+}
+
 /**
  * Gets the position of the split. Splits the exceeding line at
- * the last whitespace prior to it exceeding the bounding box width.
+ * the last breakable char prior to it exceeding the bounding box width.
  */
 const getSplitPosition = (textLine: string, calcValues: FontWidthCalcValues) => {
   const overPos = getOverPosition(textLine, calcValues);
   if (overPos === null) return textLine.length; // input line is shorter than the available space
 
-  let overPosTmp = overPos;
-  while (textLine[overPosTmp] !== ' ' && overPosTmp >= 0) {
+  if (textLine[overPos] === ' ') {
+    // if the character immediately beyond the boundary is a space, split
+    return overPos;
+  }
+
+  let overPosTmp = overPos - 1;
+  while (overPosTmp >= 0) {
+    if (isLineBreakableChar(textLine[overPosTmp])) {
+      return overPosTmp+1;
+    }
     overPosTmp--;
   }
 
-  // For very long lines with no whitespace use the original overPos
-  return overPosTmp > 0 ? overPosTmp : overPos;
+  // For very long lines with no breakable chars use the original overPos
+  return overPos;
 };
 
 /**
@@ -175,7 +195,7 @@ const getSplitPosition = (textLine: string, calcValues: FontWidthCalcValues) =>
  */
 export const getSplittedLines = (textLine: string, calcValues: FontWidthCalcValues): string[] => {
   const splitPos = getSplitPosition(textLine, calcValues);
-  const splittedLine = textLine.substring(0, splitPos);
+  const splittedLine = textLine.substring(0, splitPos).trimEnd();
   const rest = textLine.substring(splitPos).trimStart();
 
   if (rest === textLine) {