cannot read .med from Gmsh

[shipengcheng1230]

Hello, I am trying to read a .med (generated using Gmsh tutorial t1.jl in the tutorial and exported to .med format), but I got the following error:

julia> me = aster_read_mesh("t1.med")
ERROR: ArgumentError: invalid base 10 digit 'D' in "0D"
Stacktrace:
 [1] tryparse_internal(::Type{Int64}, ::SubString{String}, ::Int64, ::Int64, ::Int64, ::Bool) at ./parse.jl:128
 [2] #parse#332(::Nothing, ::Function, ::Type{Int64}, ::SubString{String}) at ./parse.jl:228
 [3] parse(::Type{Int64}, ::SubString{String}) at ./parse.jl:228
 [4] get_element_sets(::AsterReader.MEDFile, ::String) at /Users/spc/.julia/packages/AsterReader/cKdBD/src/read_aster_mesh.jl:102
 [5] aster_read_mesh_(::AsterReader.MEDFile, ::Nothing) at /Users/spc/.julia/packages/AsterReader/cKdBD/src/read_aster_mesh.jl:189
 [6] aster_read_mesh at /Users/spc/.julia/packages/AsterReader/cKdBD/src/read_aster_mesh.jl:167 [inlined] (repeats 2 times)
 [7] top-level scope at none:0

Is it a bug in this package or problem with the export of Gmsh? Thank you!

[ahojukka5]

Hard to say based on that error. Can you attach the med file?

[shipengcheng1230]

Here is the file. (Github doesn't allow directly to upload .med)
t1.med.zip

[ahojukka5]

Looks that Gmsh is not making a proper med file in order to determine element sets.

med1 = MEDFile("t1.med")
med2 = MEDFile("quad4.med")

Now, we should find element set names and corresponding id number, which is coded in case of med files created by Salome, to key value, i.e.

julia> med2.data["FAS"]["BLOCK"]["ELEME"]
Dict{String,Any} with 5 entries:
  "FAM_-6_BLOCK"  => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[66, 76, 79, 67, 75, 0, 0, 0, 0, 0  …  0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]))
  "FAM_-9_LEFT"   => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[76, 69, 70, 84, 0, 0, 0, 0, 0, 0  …  0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]))
  "FAM_-8_BOTTOM" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[66, 79, 84, 84, 79, 77, 0, 0, 0, 0  …  0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]))
  "FAM_-10_RIGHT" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[82, 73, 71, 72, 84, 0, 0, 0, 0, 0  …  0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]))
  "FAM_-7_TOP"    => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[84, 79, 80, 0, 0, 0, 0, 0, 0, 0  …  0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]))

This id number, -9, -8, etc. is then used to determine what element set particular element belongs to:

julia> med2.data["ENS_MAA"]["BLOCK"]["-0000000000000000001-0000000000000000001"]["MAI"]
Dict{String,Any} with 2 entries:
  "QU4" => Dict{String,Any}("FAM"=>Int32[-6],"NOD"=>Int32[4, 3, 2, 1],"NUM"=>Int32[5])
  "SE2" => Dict{String,Any}("FAM"=>Int32[-8, -10, -7, -9],"NOD"=>Int32[4, 3, 2, 1, 3, 2, 1, 4],"NUM"=>Int32[1, 2, 3, 4])

For example, element set BLOCK is -6, and from here we can see that element of type QU4 belongs to set id -6. Now, in Gmsh exported med file, we still have id numbers assigned to each element:

julia> med1.data["ENS_MAA"]["t1"]["-0000000000000000001-0000000000000000001"]["MAI"]
Dict{String,Any} with 3 entries:
  "SE2" => Dict{String,Any}("FAM"=>Int32[-5, -5, -5, -5, -5, -5, -5, -5, -5, -5  …  -6, -6, -6, -6, -6, -6, -6, -6, -6, -6],"NOD"=>Int32[1, 5, 6, 7, 8, 9, 10, 11, 12, 13  …  34, 35, 36, 37,…
  "TR3" => Dict{String,Any}("FAM"=>Int32[-9, -9, -9, -9, -9, -9, -9, -9, -9, -9  …  -9, -9, -9, -9, -9, -9, -9, -9, -9, -9],"NOD"=>Int32[214, 105, 106, 122, 135, 60, 170, 170, 183, 54  …  3…
  "PO1" => Dict{String,Any}("FAM"=>Int32[-1, -2],"NOD"=>Int32[1, 2])

So for example, elements of type TR3 belongs to set id -9. But here's the problem. What is the name of the element set with id -9?

julia> med1.data["FAS"]["t1"]["ELEME"]
Dict{String,Any} with 9 entries:
  "F_0D_3" => Dict{String,Any}()
  "F_1D_3" => Dict{String,Any}()
  "F_2D_1" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[77, 121, 32, 115, 117, 114, 102, 97, 99, 101  …  32, 32, 32, 32, 32, 32, 32, 32, 32, 32]]))
  "F_0D_2" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[71, 95, 48, 68, 95, 49, 32, 32, 32, 32  …  32, 32, 32, 32, 32, 32, 32, 32, 32, 32]]))
  "F_0D_4" => Dict{String,Any}()
  "F_1D_4" => Dict{String,Any}()
  "F_1D_1" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[71, 95, 49, 68, 95, 50, 32, 32, 32, 32  …  32, 32, 32, 32, 32, 32, 32, 32, 32, 32]]))
  "F_0D_1" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[71, 95, 48, 68, 95, 49, 32, 32, 32, 32  …  32, 32, 32, 32, 32, 32, 32, 32, 32, 32]]))
  "F_1D_2" => Dict{String,Any}("GRO"=>Dict{String,Any}("NOM"=>Array{Int8,N} where N[[71, 95, 49, 68, 95, 50, 32, 32, 32, 32  …  32, 32, 32, 32, 32, 32, 32, 32, 32, 32]]))

Because med file exported from Gmsh does not contain id value in key, it's unclear for me how to actually map element set names to corresponding id-numbers.

[ahojukka5]

Ok, here's maybe the idea. Looks that the number of element set names matches, so maybe there is some sort of implicit assumption that keys, ordered in alphabetical order or similar, gives id numebrs. So let's test this:

julia> for (i, k) in enumerate(sort(collect(keys(med1.data["FAS"]["t1"]["ELEME"]))))
           v = med1.data["FAS"]["t1"]["ELEME"][k]
           s = ""
           if !isempty(v)
               s = join(to_ascii(v["GRO"]["NOM"][1]))
           end
           println("$k -> -$i $s")
       end
F_0D_1 -> -1 G_0D_1
F_0D_2 -> -2 G_0D_1
F_0D_3 -> -3
F_0D_4 -> -4
F_1D_1 -> -5 G_1D_2
F_1D_2 -> -6 G_1D_2
F_1D_3 -> -7
F_1D_4 -> -8
F_2D_1 -> -9 My surface

julia> med1.data["ENS_MAA"]["t1"]["-0000000000000000001-0000000000000000001"]["MAI"]
Dict{String,Any} with 3 entries:
  "SE2" => Dict{String,Any}("FAM"=>Int32[-5, -5, -5, -5, -5, -5, -5, -5, -5, -5  …  -6, -6, -6, -6, -6, -6, -6, -6, -6, -6],"NOD"=>Int32[1, 5, 6, 7, 8, 9, 10, 11, 12, 13  …  34, 35, 36, 37,…
  "TR3" => Dict{String,Any}("FAM"=>Int32[-9, -9, -9, -9, -9, -9, -9, -9, -9, -9  …  -9, -9, -9, -9, -9, -9, -9, -9, -9, -9],"NOD"=>Int32[214, 105, 106, 122, 135, 60, 170, 170, 183, 54  …  3…
  "PO1" => Dict{String,Any}("FAM"=>Int32[-1, -2],"NOD"=>Int32[1, 2])

This actually makes sense. 0D elements POI1 are in groups -1 and -2, 1D elements SEG2 in groups -5 and -6, and finally, TR3 elements in group -9.

[ahojukka5]

This should be now ok. Please open a new issue if you find that reading Gmesh exported med files is not working as intended.