Fix extraneous newlines (#232)

Author: KillyMXI

lib/block-text-builder.js

@@ -135,19 +135,31 @@ class BlockTextBuilder {
     }
 
     if (
-      this.whitespaceProcessor.testContainsWords(str) || // There are words to add;
-      (str.length && !this._stackItem.stashedLineBreaks) // or at least spaces to take into account.
-    ) {
-      if (this._stackItem.stashedLineBreaks) {
-        this._stackItem.inlineTextBuilder.startNewLine(this._stackItem.stashedLineBreaks);
+      str.length === 0 || // empty string
+      (
+        this._stackItem.stashedLineBreaks && // stashed linebreaks make whitespace irrelevant
+        !this.whitespaceProcessor.testContainsWords(str) // no words to add
+      )
+    ) { return; }
+
+    if (this.options.preserveNewlines) {
+      const newlinesNumber = this.whitespaceProcessor.countNewlinesNoWords(str);
+      if (newlinesNumber > 0) {
+        this._stackItem.inlineTextBuilder.startNewLine(newlinesNumber);
+        // keep stashedLineBreaks unchanged
+        return;
       }
-      this.whitespaceProcessor.shrinkWrapAdd(
-        str,
-        this._stackItem.inlineTextBuilder,
-        (this._wordTransformer && !noWordTransform) ? this._getCombinedWordTransformer() : undefined
-      );
-      this._stackItem.stashedLineBreaks = 0; // inline text doesn't introduce line breaks
     }
+
+    if (this._stackItem.stashedLineBreaks) {
+      this._stackItem.inlineTextBuilder.startNewLine(this._stackItem.stashedLineBreaks);
+    }
+    this.whitespaceProcessor.shrinkWrapAdd(
+      str,
+      this._stackItem.inlineTextBuilder,
+      (this._wordTransformer && !noWordTransform) ? this._getCombinedWordTransformer() : undefined
+    );
+    this._stackItem.stashedLineBreaks = 0; // inline text doesn't introduce line breaks
   }
 
   /**

lib/whitespace-processor.js

@@ -33,10 +33,11 @@ class WhitespaceProcessor {
     this.leadingWhitespaceRe = new RegExp(`^[${whitespaceCodes}]`);
     this.trailingWhitespaceRe = new RegExp(`[${whitespaceCodes}]$`);
     this.allWhitespaceOrEmptyRe = new RegExp(`^[${whitespaceCodes}]*$`);
+    this.newlineOrNonWhitespaceRe = new RegExp(`(\\n|[^\\n${whitespaceCodes}])`, 'g');
 
     if (options.preserveNewlines) {
 
-      const wordOrNewlineRe = new RegExp(`\n|[^\n${whitespaceCodes}]+`, 'gm');
+      const wordOrNewlineRe = new RegExp(`\\n|[^\\n${whitespaceCodes}]+`, 'gm');
 
       /**
        * Shrink whitespaces and wrap text, add to the builder.
@@ -128,6 +129,28 @@ class WhitespaceProcessor {
     return !this.allWhitespaceOrEmptyRe.test(text);
   }
 
+  /**
+   * Return the number of newlines if there are no words.
+   *
+   * If any word is found then return zero regardless of the actual number of newlines.
+   *
+   * @param   { string }  text  Input string.
+   * @returns { number }
+   */
+  countNewlinesNoWords (text) {
+    this.newlineOrNonWhitespaceRe.lastIndex = 0;
+    let counter = 0;
+    let match;
+    while ((match = this.newlineOrNonWhitespaceRe.exec(text)) !== null) {
+      if (match[0] === '\n') {
+        counter++;
+      } else {
+        return 0;
+      }
+    }
+    return counter;
+  }
+
 }
 
 module.exports = { WhitespaceProcessor: WhitespaceProcessor };

test/html-to-text.js

@@ -182,6 +182,40 @@ describe('html-to-text', function () {
       expect(convert(html, { preserveNewlines: true })).to.equal(expected);
     });
 
+    it('should produce equal results regardless of newline position between blocks', function () {
+      const newlineOutside = '<p>A</p>\n<p>B</p>';
+      const newlineInside = '<p>A</p><p>\nB</p>';
+      const r1 = convert(newlineOutside, { preserveNewlines: true });
+      const r2 = convert(newlineInside, { preserveNewlines: true });
+      expect(r1).to.equal(r2);
+    });
+
+    it('should produce equal results for preserved newlines and BR tags', function () {
+      const nlHtml = '<p>A</p>\n<p>B</p><p>\nC</p>';
+      const brHtml = '<p>A</p><br/><p>B</p><p><br/>C</p>';
+      const nlResult = convert(nlHtml, { preserveNewlines: true });
+      const brResult = convert(brHtml);
+      expect(nlResult).to.equal(brResult);
+    });
+
+    it('should account for trailing/leading linebreaks of adjacent blocks equally', function () {
+      const html = '<p>A</p>\n<div>B</div>\n<div>C</div>\n<p>D</p>';
+      const newlineInside = 'A\n\n\nB\n\nC\n\n\nD';
+      expect(convert(html, { preserveNewlines: true })).to.equal(newlineInside);
+    });
+
+    it('should work with multiple linebreaks and in presence of whitespaces', function () {
+      const html = '<p>A</p> \n \n <p>B</p>';
+      const newlineInside = 'A\n\n\n\nB';
+      expect(convert(html, { preserveNewlines: true })).to.equal(newlineInside);
+    });
+
+    it('should have no special behavior in presence of words among linebreaks', function () {
+      const html = '<p>A</p> \n B \n <p>C</p>';
+      const newlineInside = 'A\n\n\nB\n\n\nC';
+      expect(convert(html, { preserveNewlines: true })).to.equal(newlineInside);
+    });
+
   });
 
   describe('unicode and html entities', function () {