jyapayne/swig
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Rewrite Doxygen unit tests for Java using Java 9 API

Browse source 
In particular, do not use com.sun.javadoc deprecated since Java 9 and
finally removed in Java 13, to allow the tests to run under modern JRE.

They don't run under Java 8 and earlier any more, but this shouldn't be
a huge problem nowadays and as SWIG output is independent from the Java
version used, it's enough to test it with modern Java versions.

Note that the tests themselves were changed only in the most minimal
way, to adapt them to the new way of running javadoc (which is now also
integrated into CommentParser itself instead of being duplicated in
every test).
This commit is contained in:
Vadim Zeitlin 2019-12-18 02:42:12 +01:00
commit 66a7826192
18 changed files with 140 additions and 180 deletions
Download patch file Download diff file Expand all files Collapse all files

134
Examples/test-suite/java/CommentParser.java
View file

@ -1,5 +1,6 @@
import com.sun.javadoc.*;
import com.sun.source.doctree.*;
import com.sun.source.util.DocTrees;
import java.util.HashMap;
import java.util.Map.Entry;
import java.util.Map;
@ -9,45 +10,120 @@ import java.io.BufferedWriter;
import java.io.OutputStreamWriter;
import java.io.FileOutputStream;
import java.io.IOException;
import java.util.*;
import java.util.spi.ToolProvider;
import javax.lang.model.*;
import javax.lang.model.element.*;
import javax.lang.model.util.*;
import jdk.javadoc.doclet.*;
public class CommentParser {
public class CommentParser implements Doclet {
private static Map<String, String> m_parsedComments = new HashMap<String, String>();
public static boolean start(RootDoc root) {
// We need to implement these base class pure virtual methods.
@Override
public void init(Locale locale, Reporter reporter) {
}
@Override
public Set<? extends Option> getSupportedOptions() {
return new HashSet<>();
}
@Override
public SourceVersion getSupportedSourceVersion() {
return SourceVersion.latest();
}
@Override
public String getName() {
return "CommentParser";
}
// Element name must be the fully qualified name of the element.
//
// If there is no comment associated with this element, simply do nothing.
private void storeCommentFor(DocTrees docTrees, String fullName, Element e) {
DocCommentTree docCommentTree = docTrees.getDocCommentTree(e);
if (docCommentTree == null)
return;
StringBuilder name = new StringBuilder(fullName);
// We must use signature in the key for methods for compatibility with
// the existing tests and to allow distinguishing between overloaded
// methods.
if (e instanceof ExecutableElement) {
ExecutableElement ex = (ExecutableElement)e;
name.append("(");
boolean firstParam = true;
for (VariableElement p : ex.getParameters()) {
if (firstParam) {
firstParam = false;
} else {
name.append(", ");
}
name.append(p.asType().toString());
}
name.append(")");
}
// For some reason the comment in the source is split into "body" and
// "block tags" parts, so we need to concatenate them back together.
StringBuilder comment = new StringBuilder();
for (DocTree d : docCommentTree.getFullBody()) {
comment.append(d.toString());
comment.append("\n");
}
boolean firstBlockTag = true;
for (DocTree d : docCommentTree.getBlockTags()) {
if (firstBlockTag) {
firstBlockTag = false;
comment.append("\n");
}
comment.append(d.toString());
comment.append("\n");
}
m_parsedComments.put(name.toString(), comment.toString());
}
@Override
public boolean run(DocletEnvironment docEnv) {
/*
* This method is called by 'javadoc' and gets the whole parsed java
* file, we get comments and store them
*/
DocTrees docTrees = docEnv.getDocTrees();
for (TypeElement t : ElementFilter.typesIn(docEnv.getIncludedElements())) {
String typeName = t.getQualifiedName().toString();
for (ClassDoc classDoc : root.classes()) {
storeCommentFor(docTrees, typeName, t);
if (classDoc.getRawCommentText().length() > 0)
m_parsedComments.put(classDoc.qualifiedName(), classDoc.getRawCommentText());
for (Element e : t.getEnclosedElements()) {
// Omit the method name for ctors: this is a bit weird, but
// this is what the existing tests expect.
String fullName = typeName;
if (e.getKind() != ElementKind.CONSTRUCTOR) {
fullName = fullName + "." + e.getSimpleName();
}
for (FieldDoc f : classDoc.enumConstants()) {
if (f.getRawCommentText().length() > 0)
m_parsedComments.put(f.qualifiedName(), f.getRawCommentText());
}
for (FieldDoc f : classDoc.fields()) {
if (f.getRawCommentText().length() > 0)
m_parsedComments.put(f.qualifiedName(), f.getRawCommentText());
}
for (ConstructorDoc c : classDoc.constructors()) {
if (c.getRawCommentText().length() > 0)
m_parsedComments.put(c.toString(), c.getRawCommentText());
}
for (MethodDoc m : classDoc.methods()) {
if (m.getRawCommentText().length() > 0)
m_parsedComments.put(m.toString(), m.getRawCommentText());
storeCommentFor(docTrees, fullName, e);
}
}
return true;
}
public int check(Map<String, String> wantedComments) {
public static int check(Map<String, String> wantedComments) {
int errorCount=0;
Iterator<Entry<String, String>> it = m_parsedComments.entrySet().iterator();
@ -131,7 +207,7 @@ public class CommentParser {
}
private void printKeys(Map<String, String> map) {
private static void printKeys(Map<String, String> map) {
Set<String> keys = map.keySet();
for (String key : keys) {
@ -155,6 +231,15 @@ public class CommentParser {
}
}
public static void parse(String sourcefile) {
ToolProvider javadoc = ToolProvider.findFirst("javadoc").orElseThrow();
int result = javadoc.run(System.out, System.err, new String[]{"-quiet", "-doclet", "CommentParser", sourcefile});
if (result != 0) {
System.err.println("Executing javadoc failed.");
System.exit(result);
}
}
public static void main(String argv[]) {
@ -163,8 +248,7 @@ public class CommentParser {
System.exit(1);
}
com.sun.tools.javadoc.Main.execute("The comment parser program",
"CommentParser", new String[]{"-quiet", argv[0]});
parse(argv[0]);
// if we are run as standalone app, print the list of found comments as it would appear in java source