Fix issue with garbage after the diagram and created better tests

src/net/sourceforge/plantuml/Pipe.java

@@ -35,11 +35,15 @@
  */
 package net.sourceforge.plantuml;
 
+import java.io.BufferedReader;
 import java.io.ByteArrayOutputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.io.InputStreamReader;
 import java.io.OutputStream;
 import java.io.PrintStream;
+import java.io.UnsupportedEncodingException;
+import java.nio.charset.Charset;
 import java.util.List;
 
 import net.sourceforge.plantuml.core.Diagram;
@@ -51,89 +55,97 @@ import net.sourceforge.plantuml.security.SFile;
 public class Pipe {
 
 	private final Option option;
-	private final InputStream is;
+	private final BufferedReader br;
 	private final PrintStream ps;
-	private boolean closed = false;
-	private final String charset;
 	private final Stdrpt stdrpt;
 
 	public Pipe(Option option, PrintStream ps, InputStream is, String charset) {
 		this.option = option;
-		this.is = is;
+		try {
+			this.br = new BufferedReader(
+					new InputStreamReader(is, (charset != null) ? charset : Charset.defaultCharset().name()));
+		} catch (UnsupportedEncodingException e) {
+			throw new RuntimeException("Invalid charset provided", e);
+		}
 		this.ps = ps;
-		this.charset = charset;
 		this.stdrpt = option.getStdrpt();
 	}
 
 	public void managePipe(ErrorStatus error) throws IOException {
 		final boolean noStdErr = option.isPipeNoStdErr();
-		int nb = 0;
+
-		do {
+		for(String source = readFirstDiagram(); source != null; source = readSubsequentDiagram()) {
-			final String source = readOneDiagram();
-			if (source == null) {
-				ps.flush();
-				if (nb == 0) {
-					// error.goNoData();
-				}
-				return;
-			}
-			nb++;
 			final Defines defines = option.getDefaultDefines();
 			final SFile newCurrentDir = option.getFileDir() == null ? null : new SFile(option.getFileDir());
 			final SourceStringReader sourceStringReader = new SourceStringReader(defines, source, "UTF-8",
 					option.getConfig(), newCurrentDir);
+
 			if (option.isComputeurl()) {
-				for (BlockUml s : sourceStringReader.getBlocks()) {
+				computeUrlForDiagram(sourceStringReader);
-					ps.println(s.getEncodedUrl());
-				}
 			} else if (option.isSyntax()) {
-				final Diagram system = sourceStringReader.getBlocks().get(0).getDiagram();
+				syntaxCheckDiagram(sourceStringReader, error);
-				if (system instanceof UmlDiagram) {
-					error.goOk();
-					ps.println(((UmlDiagram) system).getUmlDiagramType().name());
-					ps.println(system.getDescription());
-				} else if (system instanceof PSystemError) {
-					error.goWithError();
-					stdrpt.printInfo(ps, (PSystemError) system);
-				} else {
-					error.goOk();
-					ps.println("OTHER");
-					ps.println(system.getDescription());
-				}
 			} else if (option.isPipeMap()) {
-				final String result = sourceStringReader.getCMapData(option.getImageIndex(),
+				createPipeMapForDiagram(sourceStringReader, error);
-						option.getFileFormatOption());
-				// https://forum.plantuml.net/10049/2019-pipemap-diagrams-containing-links-give-zero-exit-code
-				// We don't check errors
-				error.goOk();
-				if (result == null) {
-					ps.println();
-				} else {
-					ps.println(result);
-				}
 			} else {
-				final OutputStream os = noStdErr ? new ByteArrayOutputStream() : ps;
+				generateDiagram(sourceStringReader, error, noStdErr);
-				final DiagramDescription result = sourceStringReader.outputImage(os, option.getImageIndex(),
-						option.getFileFormatOption());
-				printInfo(noStdErr ? ps : System.err, sourceStringReader);
-				if (result != null && "(error)".equalsIgnoreCase(result.getDescription())) {
-					error.goWithError();
-				} else {
-					error.goOk();
-					if (noStdErr) {
-						final ByteArrayOutputStream baos = (ByteArrayOutputStream) os;
-						baos.close();
-						ps.write(baos.toByteArray());
-					}
-				}
-				if (option.getPipeDelimitor() != null) {
-					ps.println(option.getPipeDelimitor());
-				}
 			}
 			ps.flush();
-		} while (closed == false);
+		}
-		if (nb == 0) {
+	}
-			// error.goNoData();
+
+	private void generateDiagram(SourceStringReader sourceStringReader, ErrorStatus error, boolean noStdErr) throws IOException {
+		final OutputStream os = noStdErr ? new ByteArrayOutputStream() : ps;
+		final DiagramDescription result = sourceStringReader.outputImage(os, option.getImageIndex(),
+				option.getFileFormatOption());
+
+		printInfo(noStdErr ? ps : System.err, sourceStringReader);
+		if (result != null && "(error)".equalsIgnoreCase(result.getDescription())) {
+			error.goWithError();
+		} else {
+			error.goOk();
+			if (noStdErr) {
+				final ByteArrayOutputStream baos = (ByteArrayOutputStream) os;
+				baos.close();
+				ps.write(baos.toByteArray());
+			}
+		}
+		if (option.getPipeDelimitor() != null) {
+			ps.println(option.getPipeDelimitor());
+		}
+	}
+
+	private void createPipeMapForDiagram(SourceStringReader sourceStringReader, ErrorStatus error) throws IOException {
+		final String result = sourceStringReader.getCMapData(option.getImageIndex(),
+				option.getFileFormatOption());
+		// https://forum.plantuml.net/10049/2019-pipemap-diagrams-containing-links-give-zero-exit-code
+		// We don't check errors
+		error.goOk();
+		if (result == null) {
+			ps.println();
+		} else {
+			ps.println(result);
+		}
+	}
+
+	private void computeUrlForDiagram(SourceStringReader sourceStringReader) throws IOException {
+		for (BlockUml s : sourceStringReader.getBlocks()) {
+			ps.println(s.getEncodedUrl());
+		}
+	}
+
+	private void syntaxCheckDiagram(SourceStringReader sourceStringReader, ErrorStatus error) {
+		final Diagram system = sourceStringReader.getBlocks().get(0).getDiagram();
+		if (system instanceof UmlDiagram) {
+			error.goOk();
+			ps.println(((UmlDiagram) system).getUmlDiagramType().name());
+			ps.println(system.getDescription());
+		} else if (system instanceof PSystemError) {
+			error.goWithError();
+			stdrpt.printInfo(ps, system);
+		} else {
+			error.goOk();
+			ps.println("OTHER");
+			ps.println(system.getDescription());
 		}
 	}
 
@@ -146,34 +158,55 @@ public class Pipe {
 		}
 	}
 
-	private boolean isFinished(String s) {
+	String readFirstDiagram() throws IOException {
-		return s == null || s.startsWith("@end");
+		return readSingleDiagram(true);
 	}
 
-	String readOneDiagram() throws IOException {
+	String readSubsequentDiagram() throws IOException {
+		return readSingleDiagram(false);
+	}
+
+	String readSingleDiagram(boolean unmarkedAllowed) throws IOException {
+		State state = State.NO_CONTENT;
+		String expectedEnd = null;
 		final StringBuilder sb = new StringBuilder();
-		while (true) {
+
-			final String s = readOneLine();
+		String line;
-			if (s == null) {
+		while(state != State.COMPLETE && (line = br.readLine()) != null) {
-				closed = true;
+
-			} else if (s.startsWith("@@@format ")) {
+			if(line.startsWith("@@@format ")) {
-				manageFormat(s);
+				manageFormat(line);
 			} else {
-				sb.append(s);
+				if (state == State.NO_CONTENT && line.trim().length() > 0) {
-				sb.append(BackSlash.NEWLINE);
+					state = State.START_MARK_NOT_FOUND;
-			}
+				}
-			if (isFinished(s)) {
+
-				break;
+				if (state == State.START_MARK_NOT_FOUND && line.startsWith("@start")) {
+					sb.setLength(0); // discard any previous input
+					state = State.START_MARK_FOUND;
+					expectedEnd = "@end" + line.substring(6).split("^[A-Za-z]")[0];
+				} else if (state == State.START_MARK_FOUND && line.startsWith(expectedEnd)) {
+					state = State.COMPLETE;
+				}
+
+				if (state != State.NO_CONTENT) {
+					sb.append(line).append("\n");
+				}
 			}
 		}
-		String source = sb.toString().trim();
+
-		if (sb.length() == 0) {
+		switch(state) {
-			return null;
+			case NO_CONTENT:
+				return null;
+			case START_MARK_NOT_FOUND:
+				return (unmarkedAllowed) ? "@startuml\n" + sb.toString() + "@enduml\n" : null;
+			case START_MARK_FOUND:
+				return sb.toString() + expectedEnd;
+			case COMPLETE:
+				return sb.toString();
+			default:
+				throw new IllegalStateException("Unexpected value: " + state);
 		}
-		if (source.startsWith("@start") == false) {
-			source = "@startuml\n" + source + "\n@enduml";
-		}
-		return source;
 	}
 
 	void manageFormat(String s) {
@@ -184,27 +217,10 @@ public class Pipe {
 		}
 	}
 
-	String readOneLine() throws IOException {
+	 enum State {
-		final ByteArrayOutputStream baos = new ByteArrayOutputStream();
+		NO_CONTENT,
-		while (true) {
+		START_MARK_NOT_FOUND,
-			final int read = is.read();
+		START_MARK_FOUND,
-			if (read == -1) {
+		COMPLETE
-				if (baos.size() == 0) {
-					return null;
-				}
-				break;
-			}
-			if (read != '\r' && read != '\n') {
-				baos.write(read);
-			}
-			if (read == '\n') {
-				break;
-			}
-		}
-		if (charset == null) {
-			return new String(baos.toByteArray());
-		}
-		return new String(baos.toByteArray(), charset);
-
 	}
 }

test/net/sourceforge/plantuml/PipeTest.java

@@ -3,6 +3,7 @@ package net.sourceforge.plantuml;
 
 import static java.nio.charset.StandardCharsets.UTF_8;
 import static org.assertj.core.api.Assertions.assertThat;
+import static org.junit.jupiter.api.Assertions.assertTrue;
 
 import java.io.ByteArrayInputStream;
 import java.io.ByteArrayOutputStream;
@@ -154,43 +155,109 @@ class PipeTest {
     }
 
     @Test
-    void should_readOneDiagram_return_null_for_empty_input() throws IOException {
+    void should_readFirstDiagram_return_null_for_empty_input() throws IOException {
         pipe = new Pipe(option, null, new ByteArrayInputStream(new byte[0]), UTF_8.name());
 
-        String actual = pipe.readOneDiagram();
+        String actual = pipe.readFirstDiagram();
 
         assertThat(actual).isNull();
     }
 
     @Test
-    void should_readOneDiagram_remove_carriage_returns() throws IOException {
+    void should_readFirstDiagram_decode_a_special_unicode_character_when_provided_charset_is_utf8() throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream("@startuml\na\rb\r\nc\rd\re\n@enduml".getBytes(UTF_8)), UTF_8.name());
+        pipe = new Pipe(option, null, new ByteArrayInputStream("\u2620\n".getBytes(UTF_8)), UTF_8.name());
 
-        String actual = pipe.readOneDiagram();
+        String actual = pipe.readFirstDiagram();
 
-        assertThat(actual).isEqualTo("@startuml\nab\ncde\n@enduml");
+        assertThat(actual).isEqualTo("@startuml\n\u2620\n@enduml\n");
     }
 
-    static List<InputExpected> startAndEndMarks() {
+    // The testing is relevant only if the testing VM is configured in non-UTF-8 defaultCharset()
-        LinkedList<InputExpected> linkedList = new LinkedList<>();
+    @Test
-        linkedList.add(InputExpected.of("\na\rb\r\nc\rd\re\n@enduml", "@startuml\nab\ncde\n@enduml\n@enduml"));
+    void should_readFirstDiagram_uses_current_charset_if_not_provided() throws IOException {
-        linkedList.add(InputExpected.of("\na\rb\r\nc\rd\re\n@endwhatever", "@startuml\nab\ncde\n@endwhatever\n@enduml"));
+        String input = "\u00C1"; // A acute accented. (HTML &Aacute;). Multibyte in UTF-8.
-        linkedList.add(InputExpected.of("\na\rb\r\nc\rd\re\n@enduml\n", "@startuml\nab\ncde\n@enduml\n@enduml"));
+        if(Charset.defaultCharset().newEncoder().canEncode(input)) {
-        linkedList.add(InputExpected.of("@startuml\na\rb\r\nc\rd\re\n@enduml", "@startuml\nab\ncde\n@enduml"));
+            pipe = new Pipe(option, null, new ByteArrayInputStream(input.getBytes(Charset.defaultCharset())), null);
-        linkedList.add(InputExpected.of("@startuml\na\rb\r\nc\rd\re\n@enduml\n", "@startuml\nab\ncde\n@enduml"));
+
-        linkedList.add(InputExpected.of("@startuml\na\rb\r\nc\rd\re\n@enduml\n\n", "@startuml\nab\ncde\n@enduml"));
+            String actual = pipe.readFirstDiagram();
-        linkedList.add(InputExpected.of("@startuml\na\rb\r\nc\rd\re\n@enduml\r\n\r\n", "@startuml\nab\ncde\n@enduml"));
+
-        linkedList.add(InputExpected.of("this-is-garbage\n@startuml\na\rb\r\nc\rd\re\n@enduml\nthis-is-garbage\n", "@startuml\nab\ncde\n@enduml"));
+            assertThat(actual).isEqualTo("@startuml\n\u00C1\n@enduml\n");
-        linkedList.add(InputExpected.of("@startwhatever\na\rb\r\nc\rd\re\n@endwhatever", "@startwhatever\nab\ncde\n@endwhatever"));
+
-        return linkedList;
+        } else {
+            // default charset can't encode &Aacute;. Ignore the test.
+            assertTrue(true);
+        }
     }
 
     @ParameterizedTest
-    @MethodSource("startAndEndMarks")
+    @ValueSource(strings = {
-    void should_readOneDiagram_handle_correctly_start_and_end_marks(InputExpected inputExpected) throws IOException {
+            "ab\nc",    // *nix, macOsX
+            "ab\rc",    // pre-macOsX macs
+            "ab\r\nc",  // Windows
+            // the case \n\r is handled as 2 new lines, thus not added
+
+            "ab\nc\n",
+            "ab\nc\r",
+            "ab\nc\r\n"
+    })
+    void should_readFirstDiagram_decode_correctly_different_line_endings(String input) throws IOException {
+        pipe = new Pipe(option, null, new ByteArrayInputStream(input.getBytes(UTF_8)), UTF_8.name());
+
+        String first = pipe.readFirstDiagram();
+        String second = pipe.readFirstDiagram();
+
+        assertThat(first).isEqualTo("@startuml\nab\nc\n@enduml\n");
+        assertThat(second).isEqualTo(null);                         // no spurious diagram afterwards
+    }
+
+    static List<InputExpected> firstStartAndEndMarks() {
+        List<InputExpected> l = new LinkedList<>();
+        l.add(InputExpected.of("\nab\r\ncde", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("\nab\r\ncde\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("\nab\r\ncde\n@enduml", "@startuml\nab\ncde\n@enduml\n@enduml\n"));
+        l.add(InputExpected.of("\nab\r\ncde\n@endwhatever", "@startuml\nab\ncde\n@endwhatever\n@enduml\n"));
+        l.add(InputExpected.of("\nab\r\ncde\n@enduml\n", "@startuml\nab\ncde\n@enduml\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\n\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\r\n\r\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("this-is-garbage\n@startuml\nab\rcde\n@enduml\nthis-is-garbage\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startwhatever\nab\rcde\n@endwhatever", "@startwhatever\nab\ncde\n@endwhatever\n"));
+        return l;
+    }
+
+    @ParameterizedTest
+    @MethodSource("firstStartAndEndMarks")
+    void should_readFirstDiagram_handle_correctly_start_and_end_marks(InputExpected inputExpected) throws IOException {
         pipe = new Pipe(option, null, new ByteArrayInputStream(inputExpected.getInput().getBytes(UTF_8)), UTF_8.name());
 
-        String actual = pipe.readOneDiagram();
+        String actual = pipe.readFirstDiagram();
+
+        assertThat(actual).isEqualTo(inputExpected.getExpected());
+    }
+
+    static List<InputExpected> subsequentStartAndEndMarks() {
+        List<InputExpected> l = new LinkedList<>();
+        l.add(InputExpected.of("\nab\r\ncde", null));
+        l.add(InputExpected.of("\nab\r\ncde\n", null));
+        l.add(InputExpected.of("\nab\r\ncde\n@enduml", null));
+        l.add(InputExpected.of("\nab\r\ncde\n@endwhatever", null));
+        l.add(InputExpected.of("\nab\r\ncde\n@enduml\n", null));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\n\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startuml\nab\r\ncde\n@enduml\r\n\r\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("this-is-garbage\n@startuml\nab\rcde\n@enduml\nthis-is-garbage\n", "@startuml\nab\ncde\n@enduml\n"));
+        l.add(InputExpected.of("@startwhatever\nab\rcde\n@endwhatever", "@startwhatever\nab\ncde\n@endwhatever\n"));
+        return l;
+    }
+
+    @ParameterizedTest
+    @MethodSource("subsequentStartAndEndMarks")
+    void should_readSubsequentDiagram_handle_correctly_start_and_end_marks(InputExpected inputExpected) throws IOException {
+        pipe = new Pipe(option, null, new ByteArrayInputStream(inputExpected.getInput().getBytes(UTF_8)), UTF_8.name());
+
+        String actual = pipe.readSubsequentDiagram();
 
         assertThat(actual).isEqualTo(inputExpected.getExpected());
     }
@@ -219,81 +286,8 @@ class PipeTest {
         assertThat(option.getFileFormatOption().getFileFormat()).isEqualTo(FileFormat.PNG);
     }
 
-    @Test
-    void should_readOneLine_returns_null_if_no_input() throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream(new byte[0]), UTF_8.toString());
-
-        String result = pipe.readOneLine();
-
-        assertThat(result).isNull();
-    }
-
-    @ParameterizedTest
-    @ValueSource(strings = {"\rab\nc\n", "\ra\rb\r\nc\n", "ab\n\rc"})
-    void should_readOneLine_ignore_carriage_return_and_stop_at_line_feed(String input) throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream(input.getBytes(UTF_8)), UTF_8.name());
-
-        String actual = pipe.readOneLine();
-
-        assertThat(actual).isEqualTo("ab");
-    }
-
-    @Test
-    void should_readOneLine_return_empty_string_if_line_feed_is_the_first_character() throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream("\n".getBytes(UTF_8)), UTF_8.name());
-
-        String actual = pipe.readOneLine();
-
-        assertThat(actual).isEqualTo("");
-    }
-
-    @Test
-    void should_readOneLine_use_the_default_charset_if_not_provided() throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream("ab\n".getBytes(Charset.defaultCharset())), null);
-
-        String actual = pipe.readOneLine();
-
-        assertThat(actual).isEqualTo("ab");
-    }
-
-    @Test
-    void should_readOneLine_use_provided_charset() throws IOException {
-        pipe = new Pipe(option, null, new ByteArrayInputStream("\u2620\n".getBytes(UTF_8)), UTF_8.name());
-
-        String actual = pipe.readOneLine();
-
-        assertThat(actual).isEqualTo("\u2620");
-    }
-
-    static class InputExpected {
-        private final String input;
-        private final String expected;
-
-
-        public InputExpected(String input, String expected) {
-            this.input = input;
-            this.expected = expected;
-        }
-
-        public static InputExpected of(String input, String expected) {
-            return new InputExpected(input, expected);
-        }
-
-        public String getInput() {
-            return input;
-        }
-
-        public String getExpected() {
-            return expected;
-        }
-
-        @Override
-        public String toString() {
-            return "i:'" + input + "', e='" + expected + "'";
-        }
-    }
-
     static class TestCase {
+
         private final String options;
         private final String input;
         private final String expectedOut;
@@ -346,6 +340,34 @@ class PipeTest {
 
     }
 
+    static class InputExpected {
+        private final String input;
+        private final String expected;
+
+
+        public InputExpected(String input, String expected) {
+            this.input = input;
+            this.expected = expected;
+        }
+
+        public static InputExpected of(String input, String expected) {
+            return new InputExpected(input, expected);
+        }
+
+        public String getInput() {
+            return input;
+        }
+
+        public String getExpected() {
+            return expected;
+        }
+
+        @Override
+        public String toString() {
+            return "i:'" + input + "', e='" + expected + "'";
+        }
+    }
+
     enum Verification {
         EXACT,
         REGEX