iText7.1.3 adding SVG into PdfDocument

Question

Using iText 7.1.3 and trying to add a SVG file into PdfDocument gives an output where texts with length 1 are not rendered. I found where it might be the problem. I please the iText team members to check it.

try {
      SvgConverter.drawOnCanvas(svgUrl.openStream(), pdfCanvas_, imageLlx, imageLly);
} catch (IOException e) {
      e.printStackTrace();
}

Debugger callings:

processText:255, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
visit:212, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
visit:204, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
visit:204, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
executeDepthFirstTraversal:153, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
process:106, DefaultSvgProcessor (com.itextpdf.svg.processors.impl)
process:768, SvgConverter (com.itextpdf.svg.converter)
convertToXObject:555, SvgConverter (com.itextpdf.svg.converter)
convertToXObject:590, SvgConverter (com.itextpdf.svg.converter)
drawOnCanvas:380, SvgConverter (com.itextpdf.svg.converter)

In function processText, in the line for Trim trailing whitespace

trimmedText = SvgTextUtil.trimTrailingWhitespace("A");

for the trimmedText = A (length = 1) returns empty String

/**
* Process the text contained in the text-node
*
* @param textNode node containing text to process
*/
private void processText(ITextNode textNode) {
    ISvgNodeRenderer parentRenderer = this.processorState.top();

    if (parentRenderer instanceof TextSvgNodeRenderer) {
        // when svg is parsed by jsoup it leaves all whitespace in text element as is. Meaning that
        // tab/space indented xml files will retain their tabs and spaces.
        // The following regex replaces all whitespace with a single space.
        //TODO(RND-906) evaluate regex and trim methods
        String trimmedText = textNode.wholeText().replaceAll("\\s+", " ");
        //Trim leading whitespace
        trimmedText = SvgTextUtil.trimLeadingWhitespace(trimmedText);
        //Trim trailing whitespace
        trimmedText = SvgTextUtil.trimTrailingWhitespace(trimmedText);
        parentRenderer.setAttribute(SvgConstants.Attributes.TEXT_CONTENT, trimmedText);
    }
}

Answer 1

There indeed is a bug in the trimTrailingWhitespace you pointed out

public static String trimTrailingWhitespace(String toTrim) {
    if(toTrim == null){
        return "";
    }
    int end = toTrim.length();
    if (end > 0) {
        int current = end - 1;
        while (current > 0) {
            char currentChar = toTrim.charAt(current);
            if (Character.isWhitespace(currentChar) && !(currentChar == '\n' || currentChar == '\r')) {
                //if the character is whitespace and not a newline, increase current
                current--;
            } else {
                break;
            }
        }
        if(current == 0){
            return "";
        }else {
            return toTrim.substring(0, current + 1);
        }
    }else{
        return toTrim;
    }
}

As the comment //if the character is whitespace and not a newline, increase current followed by current--; already indicates, this method is a copy of trimLeadingWhitespace (where the line after the identical comment indeed increases current) modified to work at the other end of the String parameter. Unfortunately, though, the modification was incorrect: If the string has a non-whitespace character at position 0 and thereafter only whitespaces, it erroneously is considered empty.

The fix would be to replace

while (current > 0)

by

while (current >= 0)

and

if(current == 0)

by

if(current < 0)

With that fix the

if (end > 0) {
    [...]
}else{
    return toTrim;
}

frame around [...] furthermore becomes unnecessary. And the while loop could more compactly have been formulated as a for loop, e.g. like this:

public static String trimTrailingWhitespace(String toTrim) {
    if (toTrim == null) {
        return "";
    }
    int current = toTrim.length() - 1;
    for ( ; current >= 0; current--) {
        char currentChar = toTrim.charAt(current);
        if (!(Character.isWhitespace(currentChar) && !(currentChar == '\n' || currentChar == '\r'))) {
            break;
        }
    }
    return current < 0 ? "" : toTrim.substring(0, current + 1);
}