PreprocessorReader being manipulated is not acting like a Preprocessor

[Systemdir]

When registering a Preprocessor extension and pushing ifdef or include onto the PreprocessorReader, the directives are not being processed.

As @mojavelinux wrote in asciidoctor/asciidoctor-maven-plugin#210:

I tried this with a Ruby extension and it worked. I wonder if the delegation in AsciidoctorJ is causing the PreprocessorReader to not act like a preprocessor (instead just a regular reader).

...

In a quick test with AsciidoctorJ, but assumption seems to hold. The PreprocessorReader being manipulated is not acting like a Preprocessor. Very likely AsciidoctorJ would fail such a test. Therefore, this should be opened in AsciidoctorJ.

Until then, you'll have to push preprocessed lines onto the reader (in other words, you have to do the preprocessing yourself).

[robertpanzer]

Hi systemdir!

Hmm, kinda works for me. I made the following test with the current master of asciidoctorj which is a few commits from 1.5.4, but nothing that could have any impact on this. (I guess :wink:)

I have the following asciidoctor files:

• test.adoc:

== Test

A test

XXX

• included.adoc:

This is the included content

The Preprocessor should add an include of included.doc if it sees the content XXX. My Preprocessor looks like this:

public class IncluderPreprocessor extends Preprocessor {

    public IncluderPreprocessor(Map<String, Object> config) {
        super(config);
    }

    @Override
    public PreprocessorReader process(Document document, PreprocessorReader reader) {
        // Read the full document
        List<String> lines = reader.readLines();
        // Copy all lines into a new list, replacing a line 'XXX' with an include directive
        List<String> newLines = new ArrayList<String>();
        for (String line: lines) {
            if (line.equals("XXX")) {
                newLines.add("include::included.adoc[]");
            } else if (line.trim().equals("")) {
                newLines.add(" "); // That's a bad one: If we push empty lines they somehow get lost for whatever reason and we have no separate block, but that's another issue
            } else {
                newLines.add(line);
            }
        }
        // For reuse. We always pass empty attributes
        Map<String, Object> attributes = new HashMap<String, Object>();
        // Finally push line by line in reverse order, it's a stack :)
        for (int i = newLines.size() - 1; i >= 0; i--) {
            reader.push_include(newLines.get(i), null, null, 1, attributes);
        }
        return reader;
    }
}

And my test passes:

    @Test
    public void test() {
        File f = classpath.getResource("includetest/test.adoc");

        asciidoctor.javaExtensionRegistry().preprocessor(IncluderPreprocessor.class);

        org.jsoup.nodes.Document document = 
            Jsoup.parse(
                asciidoctor.convertFile(f, OptionsBuilder.options().toFile(false).headerFooter(false).safe(SafeMode.UNSAFE)));
        assertEquals("This is the included content", document.select("p").get(1).text());
    }

Cheers Robert

[milefork]

@robertpanzer Yes your test case works. But if included.adoc would include another .adoc then it will not be included and instead just the bare include::somefile.adoc will be printed out.