ctags-lang-elm − Random notes about tagging Elm source code with Universal Ctags
ctags ...
−−languages=+Elm ...
ctags ... −−language−force=Elm ...
ctags ... −−map−Elm=+.elm ...
The Elm parser is a PEG parser using PackCC, which is part of the ctags infrastructure. It should correctly process all top level statements, however there is a limitation with functions embedded in let/in blocks. They will mostly be fine, but sometimes a function in a let/in block will be omitted.
Imported modules are tagged, and their role is "imported", not "def". Types, functions, etc which are exposed via imported module have their role as "exposed".
Exposed items are marked as being in the scope of their own module, not the module that's doing the importing.
"input.elm"
module SomeMod exposing (..) import MyMod exposing ( map , Maybe , Result(..) , MyList(Empty) )
"output.tags" with "−−options=NONE −o − −−sort=no −−extras=+r −−fields=+r input.elm"
SomeMod input.elm /ˆmodule SomeMod exposing (..)$/;" m roles:def MyMod input.elm /ˆimport MyMod exposing$/;" m roles:imported map input.elm /ˆ ( map$/;" f module:MyMod roles:exposed Maybe input.elm /ˆ , Maybe$/;" t module:MyMod roles:exposed Result input.elm /ˆ , Result(..)$/;" t module:MyMod roles:exposed MyList input.elm /ˆ , MyList(Empty)$/;" t module:MyMod roles:exposed Empty input.elm /ˆ , MyList(Empty)$/;" c type:MyMod.MyList roles:exposed
Namespaces are tagged and their role is "def".
"input.elm"
module AMod exposing (..) import MyImport as NSpace exposing (impFunc)
"output.tags" with "−−options=NONE −o − −−sort=no −−extras=+r −−fields=+r input.elm"
AMod input.elm /ˆmodule AMod exposing (..)$/;" m roles:def NSpace input.elm /ˆimport MyImport as NSpace exposing (impFunc)$/;" n module:AMod roles:def moduleName:MyImport MyImport input.elm /ˆimport MyImport as NSpace exposing (impFunc)$/;" m roles:imported impFunc input.elm /ˆimport MyImport as NSpace exposing (impFunc)$/;" f module:MyImport roles:exposed
Constructors top level functions will have type names.
"input.elm"
funcA : Int −> Int funcA a = a + 1 type B = B1Cons { x : Float , y : Float } | B2Cons String Integer | B3Cons
"output.tags" with "−−options=NONE −o − −−sort=no −−extras=+r −−fields=+r input.elm"
funcA input.elm /ˆfuncA a = a + 1$/;" f typeref:typename:Int −> Int roles:def B input.elm /ˆtype B$/;" t roles:def B1Cons input.elm /ˆ = B1Cons$/;" c type:B typeref:typename:{ x : Float , y : Float } −> B roles:def B2Cons input.elm /ˆ | B2Cons String Integer$/;" c type:B typeref:typename:String −> Integer −> B roles:def B3Cons input.elm /ˆ | B3Cons$/;" c type:B typeref:typename:B roles:def
Function parameter lists can be extracted into the tags file signature field. They are not really function signatures, but it's the closest concept available in ctags. Use "−−fields=+S".
funcA a1 a2 = a1 + a2
"output.tags" with "−−sort=no −−extras=+r −−fields=+rS"
funcA input.elm /ˆfuncA a1 a2 =$/;" f signature:a1 a2 roles:def
The ctags signature field is used for function parameter lists, even though it's not an idea field. See above.
Elm requires all statements at the same logical level to have the same indentation. If there is additional indentation that line is part of the previous one. Therefore without over−complicating the PEG parser we have the following limitations...
Sometimes functions in let/in blocks will be omitted.
Functions in let/in blocks will be marked as being in the scope of their outer function, regardless of how deeply nested the let/in block is.
Functions in let/in blocks won't have type names.
ctags(1), ctags−client−tools(7)