Currently, indented code fences are incorrectly represented in the AST. Currently, the actual indents are being put into the CODE_FENCE_CONTENT nodes, which makes language injections much harder, since you need to manually calculate indents for each fence line.
This PR solves this problem by calculating indents during parsing and putting them into separate WHITE_SPACE nodes, so they won't pollute actual content nodes.
Currently, indented code fences are incorrectly represented in the AST. Currently, the actual indents are being put into the
CODE_FENCE_CONTENT
nodes, which makes language injections much harder, since you need to manually calculate indents for each fence line. This PR solves this problem by calculating indents during parsing and putting them into separateWHITE_SPACE
nodes, so they won't pollute actual content nodes.