For a notebook output, Literate will strip references for an input like this: <div

Literate does not strip notebook references when source spans multiple lines about literate.jl HOT 2 CLOSED

kbarros commented on September 27, 2024

Literate does not strip notebook references when source spans multiple lines

from literate.jl.

Comments (2)

fredrikekre commented on September 27, 2024

For regex 2 and 3 it make sense, I think. Do you want to submit a PR? You can probably find some tests for these replacements that you can augment too.

from literate.jl.

kbarros commented on September 27, 2024

After some investigation, I would propose these three modified regexes:

repls = []
push!(repls, r"\[([^]]+?)\]\(@ref\)"s => s"\1")     # [foo](@ref) => foo
push!(repls, r"\[([^]]+?)\]\(@ref .*?\)"s => s"\1") # [foo](@ref bar) => foo
push!(repls, r"\[([^]]+?)\]\(@id .*?\)"s => s"\1")  # [foo](@id bar) => foo

In addition to including the trailing s (which allows the match to cross between lines, https://perldoc.perl.org/perlre#Modifiers) I have also replaced (.*?) with ([^]]+?), which excludes the character ] from the matched ID within brackets.

These have the following behavior:

content = """
    # # [Example](@id example-id)
    # [foo](@ref), [bar](@ref bbaarr)
    x = 1
    """
for repl in repls
    content = replace(content, repl)
end
@assert content == """
    # # Example
    # foo, bar
    x = 1
    """

content = """
    # # [Example](@id example-id)
    #
    # [This is a long example which
    # wraps a line](@ref)
    #
    # [In this case the](@ref reference also
    # wraps a line)
    #    
    # x = 1
    """
for repl in repls
    content = replace(content, repl)
end
@assert content == """
    # # Example
    #
    # This is a long example which
    # wraps a line
    #
    # In this case the
    #    
    # x = 1
    """