Hi, Sounds reasonable but may not work as good. This may change a lot of user formatted code and produce extra diffs in tests. If we could only reformat a part of the AST (one function call in our case) this would be ideal.
Thanks, Serge. On Monday, February 26, 2018 at 2:31:01 PM UTC+2, Louis Pop wrote: > > Hi there > > Given we have the formatter now I would be tempted to avoid editing > strings and instead update the AST of the file, render that to a string, > format it and write it back to the file. Would be more reliable. > > Cheers, > Louis > > On Mon, 26 Feb 2018, 12:11 Serge Smetana, <serge....@gmail.com > <javascript:>> wrote: > >> Hi, >> >> I would like Elixir AST nodes to include begin and end parse character >> offsets. >> This would make it easier to write refactoring tools that modify parts of >> Elixir source files. >> >> Usecase: >> >> We have a library https://github.com/assert-value/assert_value_elixir >> that is >> able to create and update expected values in tests. For example, given the >> following: >> >> assert_value 2 + 2 == 2 + 3 >> >> our code will find the location of `2 + 3` and replace it with `4`: >> >> assert_value 2 + 2 == 4 >> >> Today determining the location of `2 + 3` in the source file is >> difficult. For >> now, we use a custom-made parser which processes code char by char until >> it gets >> value matching AST: >> >> https://github.com/assert-value/assert_value_elixir/blob/master/lib/assert_value/parser.ex >> But if we had the parse offsets it would be much easier. >> >> Proposed interface: >> >> Probably need two sets of offsets, exclusive and inclusive of children. >> For >> each probably best to store beginning offset and length. Will need >> reasonable >> handling for parentheses and other tokens that do not make it into AST. >> >> Existing implementation: >> >> Elixir AST nodes do have useful info on this already. We use the "line:" >> which >> is very helpful. We don't use "column:", it did not seem useful given our >> implementation. We may be missing something obvious here. >> >> Details: >> >> In Elixir 1.6 compiled code AST has only function line number in meta. >> Even >> "columns: true" in Code.string_to_quoted gives only function starting >> column >> without information about arguments. >> >> Consider the following code: >> >> Code.string_to_quoted!("(41.00 == 42.0000)", columns: true) >> #=> {:==, [line: 1, column: 8], [41.0, 42.0]} >> >> From the AST you don't know where to find 41.00 and 42.0000 in a code. >> Column information does not help. AST values of 41.0 and 42.0 don't have >> information about how original values were formatted. >> >> Thanks, >> Serge >> >> -- >> You received this message because you are subscribed to the Google Groups >> "elixir-lang-core" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to elixir-lang-co...@googlegroups.com <javascript:>. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/elixir-lang-core/4dbc37cc-6630-4989-acc0-c4acfd719a5b%40googlegroups.com >> >> <https://groups.google.com/d/msgid/elixir-lang-core/4dbc37cc-6630-4989-acc0-c4acfd719a5b%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> For more options, visit https://groups.google.com/d/optout. >> > -- You received this message because you are subscribed to the Google Groups "elixir-lang-core" group. To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/644251b6-8b27-4836-98a3-a9feddcc29ba%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
