commonmark-java Files

Added

• New option lineSeparator for MarkdownRenderer.Builder to change the default line separator from \n (e.g. to \r\n) (#442)
• YAML front matter extension: Support extracting the raw YAML content so that you can do the YAML parsing using a library (instead of the limited built-in parsing) (#391)

Changed

• Tables extension: Limit max number of parsed table cells by default (1 million cells), see option maxCells in TablesExtension.Builder. Parsing will throw an error if the limit would be exceeded, to protect against malicious input.
• The default for maxOpenBlockParsers for Parser.Builder is now 100; configure the option to remove the limit.
• Limit how deeply inline nodes may nest by default (100), see option maxInlineNesting in Parser.Builder; configure the option to remove the limit. This covers emphasis-like delimiters (e.g. * or the ins extension's +) as well as images, which can otherwise nest arbitrarily deep for a small amount of input. Anything over the limit is treated as plain text instead.

Fixed

• Fix quadratic runtime on various pathological inputs (where inline syntax to start a node is used repeatedly but never finished):
• Inline HTML like "x <!--".repeat(100_000) (#447)
• Autolink starts like "<".repeat(100_000)
• Emphasis like "a**b" + "c* ".repeat(100_000)
• Backticks runs of different lengths
• Fix stack overflow errors when parsing pathological inputs with repeated syntax:
• HTML block attributes
• Autolink email addresses
• Deeply nested emphasis
• Fix stack overflow error when rendering/visiting deeply nested inline nodes, by capping nesting depth during parsing (see maxInlineNesting above). This can happen with emphasis, e.g. "*".repeat(n) + "x" + "*".repeat(n) or "*a ".repeat(n) + " b*".repeat(n), as well as with images, e.g. "![".repeat(n) + "x" + "](u)".repeat(n)
• YAML front matter extension: Fix inefficient string concatenation for literal-block values (|)