From db5f055f0c456bd852601bd0de61336c75bd9bd7 Mon Sep 17 00:00:00 2001
From: Robin Stocker <robin@nibor.org>
Date: Tue, 23 Jan 2024 22:43:20 +1100
Subject: [PATCH] Escape special characters at beginning of line
---
.../internal/util/AsciiMatcher.java | 7 +++
.../markdown/CoreMarkdownNodeRenderer.java | 58 ++++++++++++++++++-
.../renderer/markdown/MarkdownWriter.java | 6 +-
.../markdown/MarkdownRendererTest.java | 16 +++++
.../markdown/SpecMarkdownRendererTest.java | 2 +-
5 files changed, 84 insertions(+), 5 deletions(-)
diff --git a/commonmark/src/main/java/org/commonmark/internal/util/AsciiMatcher.java b/commonmark/src/main/java/org/commonmark/internal/util/AsciiMatcher.java
index 82d83ca4..35769f82 100644
--- a/commonmark/src/main/java/org/commonmark/internal/util/AsciiMatcher.java
+++ b/commonmark/src/main/java/org/commonmark/internal/util/AsciiMatcher.java
@@ -29,6 +29,13 @@ public class AsciiMatcher implements CharMatcher {
this.set = set;
}
+ public Builder anyOf(String s) {
+ for (int i = 0; i < s.length(); i++) {
+ c(s.charAt(i));
+ }
+ return this;
+ }
+
public Builder c(char c) {
if (c > 127) {
throw new IllegalArgumentException("Can only match ASCII characters");
diff --git a/commonmark/src/main/java/org/commonmark/renderer/markdown/CoreMarkdownNodeRenderer.java b/commonmark/src/main/java/org/commonmark/renderer/markdown/CoreMarkdownNodeRenderer.java
index 813efb16..fe5725ae 100644
--- a/commonmark/src/main/java/org/commonmark/renderer/markdown/CoreMarkdownNodeRenderer.java
+++ b/commonmark/src/main/java/org/commonmark/renderer/markdown/CoreMarkdownNodeRenderer.java
@@ -10,6 +10,8 @@ import java.util.Arrays;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
/**
* The node renderer that renders all the core nodes (comes last in the order of node renderers).
@@ -20,8 +22,8 @@ import java.util.Set;
*/
public class CoreMarkdownNodeRenderer extends AbstractVisitor implements NodeRenderer {
- private final CharMatcher textEscape =
- AsciiMatcher.builder().c('[').c(']').c('<').c('>').c('`').build();
+ private final AsciiMatcher textEscape =
+ AsciiMatcher.builder().anyOf("[]<>`*&").build();
private final CharMatcher linkDestinationNeedsAngleBrackets =
AsciiMatcher.builder().c(' ').c('(').c(')').c('<').c('>').c('\\').build();
private final CharMatcher linkDestinationEscapeInAngleBrackets =
@@ -29,6 +31,8 @@ public class CoreMarkdownNodeRenderer extends AbstractVisitor implements NodeRen
private final CharMatcher linkTitleEscapeInQuotes =
AsciiMatcher.builder().c('"').build();
+ private final Pattern orderedListMarkerPattern = Pattern.compile("^([0-9]{1,9})([.)])");
+
protected final MarkdownNodeRendererContext context;
private final MarkdownWriter writer;
/**
@@ -319,7 +323,55 @@ public class CoreMarkdownNodeRenderer extends AbstractVisitor implements NodeRen
@Override
public void visit(Text text) {
- writer.writeEscaped(text.getLiteral(), textEscape);
+ String literal = text.getLiteral();
+ if (writer.isAtLineStart() && !literal.isEmpty()) {
+ char c = literal.charAt(0);
+ switch (c) {
+ case '-': {
+ // Would be ambiguous with a bullet list marker, escape
+ writer.write("\\-");
+ literal = literal.substring(1);
+ break;
+ }
+ case '#': {
+ // Would be ambiguous with an ATX heading, escape
+ writer.write("\\#");
+ literal = literal.substring(1);
+ break;
+ }
+ case '=': {
+ // Would be ambiguous with a Setext heading, escape
+ writer.write("\\=");
+ literal = literal.substring(1);
+ break;
+ }
+ case '0':
+ case '1':
+ case '2':
+ case '3':
+ case '4':
+ case '5':
+ case '6':
+ case '7':
+ case '8':
+ case '9': {
+ // Check for ordered list marker
+ Matcher m = orderedListMarkerPattern.matcher(literal);
+ if (m.find()) {
+ writer.write(m.group(1));
+ writer.write("\\" + m.group(2));
+ literal = literal.substring(m.end());
+ }
+ }
+ }
+ }
+
+ if (literal.endsWith("!") && text.getNext() instanceof Link) {
+ writer.writeEscaped(literal.substring(0, literal.length() - 1), textEscape);
+ writer.write("\\!");
+ } else {
+ writer.writeEscaped(literal, textEscape);
+ }
}
@Override
diff --git a/commonmark/src/main/java/org/commonmark/renderer/markdown/MarkdownWriter.java b/commonmark/src/main/java/org/commonmark/renderer/markdown/MarkdownWriter.java
index ba682d46..bc10f020 100644
--- a/commonmark/src/main/java/org/commonmark/renderer/markdown/MarkdownWriter.java
+++ b/commonmark/src/main/java/org/commonmark/renderer/markdown/MarkdownWriter.java
@@ -11,7 +11,7 @@ public class MarkdownWriter {
private int blockSeparator = 0;
private boolean tight;
- private char lastChar;
+ private char lastChar = '\n';
private final LinkedList<String> prefixes = new LinkedList<>();
public MarkdownWriter(Appendable out) {
@@ -22,6 +22,10 @@ public class MarkdownWriter {
return lastChar;
}
+ public boolean isAtLineStart() {
+ return lastChar == '\n' || blockSeparator > 0;
+ }
+
public void write(String s) {
flushBlockSeparator();
append(s);
diff --git a/commonmark/src/test/java/org/commonmark/renderer/markdown/MarkdownRendererTest.java b/commonmark/src/test/java/org/commonmark/renderer/markdown/MarkdownRendererTest.java
index 76a77dbb..59039799 100644
--- a/commonmark/src/test/java/org/commonmark/renderer/markdown/MarkdownRendererTest.java
+++ b/commonmark/src/test/java/org/commonmark/renderer/markdown/MarkdownRendererTest.java
@@ -127,6 +127,22 @@ public class MarkdownRendererTest {
// otherwise result in a different parse result (e.g. a link):
assertRoundTrip("\\[a\\](/uri)\n");
assertRoundTrip("\\`abc\\`\n");
+
+ // Some characters only need to be escaped at the beginning of the line
+ assertRoundTrip("\\- Test\n");
+ assertRoundTrip("\\-\n");
+ assertRoundTrip("Test -\n");
+ assertRoundTrip("Abc\n\n\\- Test\n");
+ assertRoundTrip("\\# Test\n");
+ assertRoundTrip("\\## Test\n");
+ assertRoundTrip("\\#\n");
+ assertRoundTrip("Foo\n\\===\n");
+
+ // This is a bit more tricky as we need to check for a list start
+ assertRoundTrip("1\\. Foo\n");
+ assertRoundTrip("999\\. Foo\n");
+ assertRoundTrip("1\\.\n");
+ assertRoundTrip("1\\) Foo\n");
}
@Test
diff --git a/commonmark/src/test/java/org/commonmark/renderer/markdown/SpecMarkdownRendererTest.java b/commonmark/src/test/java/org/commonmark/renderer/markdown/SpecMarkdownRendererTest.java
index 2a0c46c0..632f4acf 100644
--- a/commonmark/src/test/java/org/commonmark/renderer/markdown/SpecMarkdownRendererTest.java
+++ b/commonmark/src/test/java/org/commonmark/renderer/markdown/SpecMarkdownRendererTest.java
@@ -62,7 +62,7 @@ public class SpecMarkdownRendererTest {
System.out.println();
}
- int expectedPassed = 630;
+ int expectedPassed = 646;
assertTrue("Expected at least " + expectedPassed + " examples to pass but was " + passes.size(), passes.size() >= expectedPassed);
}
--
GitLab