Mesh overhaul phase1

.gitignore

@@ -11,3 +11,4 @@
 *.g.nem
 *.g.pex
 *.g.*.*
+*.e.*.*

Project.toml

@@ -1,12 +1,13 @@
 name = "FiniteElementContainers"
 uuid = "d08262e4-672f-4e7f-a976-f2cea5767631"
-version = "0.10.1"
 authors = ["Craig M. Hamel <[email protected]> and contributors"]
+version = "0.10.1"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
 Atomix = "a9b6321e-bd34-4604-b9c9-b65b8de01458"
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
+Exodus = "f57ae99e-f805-4780-bdca-96e224be1e5a"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
 Krylov = "ba0b0d4f-ebba-5204-a429-3ac8c609bfb7"
@@ -19,19 +20,22 @@ TimerOutputs = "a759f4b9-e2f1-59dc-863e-4aeb61b1ea8f"
 
 [weakdeps]
 AMDGPU = "21141c5a-9bdb-4563-92ae-f87d6854732e"
+AbaqusReader = "bc6b9049-e460-56d6-94b4-a597b2c0390d"
 BlockArrays = "8e7c35d0-a365-5155-bbbb-fb81a777f24e"
 CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
-Exodus = "f57ae99e-f805-4780-bdca-96e224be1e5a"
 MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
+PartitionedArrays = "5a9dfac6-5c52-46f7-8278-5e2210713be9"
 
 [extensions]
-FiniteElementContainersAMDGPUExt = "AMDGPU"
-FiniteElementContainersBlockArraysExt = "BlockArrays"
-FiniteElementContainersCUDAExt = "CUDA"
-FiniteElementContainersExodusExt = "Exodus"
-FiniteElementContainersMPIExt = ["Exodus", "MPI"]
+AbaqusReaderExt = "AbaqusReader"
+AMDGPUExt = "AMDGPU"
+BlockArraysExt = "BlockArrays"
+CUDAExt = "CUDA"
+MPIExt = "MPI"
+PartitionedArraysExt = "PartitionedArrays"
 
 [compat]
+AbaqusReader = "0.2.7"
 AMDGPU = "2"
 Adapt = "4"
 Aqua = "0.8"
@@ -45,6 +49,7 @@ KernelAbstractions = "0.9"
 Krylov = "0.9"
 LinearAlgebra = "1"
 MPI = "0.20"
+PartitionedArrays = "0.5"
 ReferenceFiniteElements = "0.13"
 SparseArrays = "1"
 StaticArrays = "1"
@@ -58,9 +63,9 @@ AMDGPU = "21141c5a-9bdb-4563-92ae-f87d6854732e"
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
 Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
 CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
-Exodus = "f57ae99e-f805-4780-bdca-96e224be1e5a"
 MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
+PartitionedArrays = "5a9dfac6-5c52-46f7-8278-5e2210713be9"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["AMDGPU", "Aqua", "CUDA", "Exodus", "MPI", "Test"]
+test = ["AMDGPU", "Aqua", "CUDA", "MPI", "PartitionedArrays", "Test"]

ext/FiniteElementContainersAMDGPUExt.jl → ext/AMDGPUExt.jl

@@ -1,4 +1,4 @@
-module FiniteElementContainersAMDGPUExt
+module AMDGPUExt
 
 using Adapt
 using AMDGPU

ext/AbaqusReaderExt.jl

@@ -0,0 +1,306 @@
+module AbaqusReaderExt
+
+using AbaqusReader
+using FiniteElementContainers
+
+function _check_for_command(line)
+    if line[1] == '*' && line[2] != '*'
+        return true
+    else
+        return false
+    end
+end
+
+function _check_for_comment(line)
+    try
+        if line[1:2] == "**"
+            return true
+        end
+    catch BoundsError
+        # do nothing
+    end
+    return false
+end
+
+function _get_command_data(lines, command_line_ids, comment_line_ids, command_id)
+    if _check_for_comment(lines[command_id])
+        return nothing
+    end
+    # substr = split(lines[command_id], '*')[2]
+    substr = lowercase(lines[command_id])
+
+    if contains(substr, "*element") && 
+       !contains(substr, "*element output") &&
+       !contains(substr, "*user element")
+        element_conns = Dict{Int, Vector{Int}}()
+
+        if contains(substr, "elset")
+            elset_name = String(split(split(substr, "elset")[2], '=')[2])
+        else
+            elset_name = nothing
+        end
+
+        if contains(substr, "type")
+            el_type = String(split(split(substr, "type")[2], '=')[2])
+        else
+            @assert false
+        end
+
+        n = command_id + 1
+        while n ∉ command_line_ids
+            if n in comment_line_ids
+                n = n + 1
+                continue
+            end
+            parts = split(lines[n], ',')
+            el_id = parse(Int, parts[1])
+            conn = parse.((Int,), parts[2:end])
+            element_conns[el_id] = conn
+            n = n + 1
+        end
+        return "element", Dict(elset_name => element_conns), el_type
+    elseif contains(substr, "*heading")
+        heading = String[]
+        n = command_id + 1
+        while n ∉ command_line_ids
+            if n in comment_line_ids
+                n = n + 1
+                continue
+            end
+            push!(heading, lines[n])
+            n = n + 1
+        end
+        return "heading", heading
+    elseif contains(substr, "*material")
+        mat_name = String(split(split(substr, ',')[2], '=')[2])
+        props = Dict{String, Float64}()
+        n = command_id + 1
+        if contains(lowercase(lines[n]), "*elastic")
+            n = n + 1
+            props["elastic"] = parse(Float64, lines[n])
+        end
+        return "material", Dict(mat_name => props)
+    elseif contains(substr, "*node") && !contains(substr, "*node output")
+        nodes = Dict{Int, Vector{Float64}}()
+        if contains(substr, "nset")
+            nset_name = split(split(substr, "nset")[2], '=')[2]
+        else
+            nset_name = nothing
+        end
+
+        n = command_id + 1
+        while n ∉ command_line_ids
+            if n in comment_line_ids
+                n = n + 1
+                continue
+            end
+
+            parts = split(lines[n], ',')
+            node_id = parse(Int, parts[1])
+            coords = parse.((Float64,), parts[2:end])
+            nodes[node_id] = coords
+            n = n + 1
+        end
+        return "nodes", Dict(nset_name => nodes)
+    elseif contains(substr, "*nset")
+        nodes = Dict{String, Vector{Int}}()
+        n = command_id + 1
+        if contains(substr, "generate")
+            nset_name = split(split(split(split(substr, "*nset")[2], "generate")[1], ',')[2], '=')[2]
+            parts = parse.((Int,), split(lines[n], ','))
+            nodes[nset_name] = range(start=parts[1], step=parts[3], stop=parts[2])
+        else
+            nset_name = split(split(split(substr, "*nset")[2], ',')[2], '=')[2]
+            temp_nodes = Vector{Int}(undef, 0)
+            while n ∉ command_line_ids
+                if n in comment_line_ids
+                    n = n + 1
+                    continue
+                end
+                for node in split(lines[n], ',')
+                    try
+                        push!(temp_nodes, parse(Int, node))
+                    catch ArgumentError
+                        # do nothing
+                        if node != ""
+                            @warn "Got a potentially bad line reading a nodeset"
+                            @warn "$(lines[n])"
+                        end
+                    end
+                end
+                n = n + 1
+            end
+            nodes[nset_name] = temp_nodes
+        end
+        return "nset", nodes
+    elseif contains(substr, "*parameter")
+        parameters = String[]
+        n = command_id + 1
+        while n ∉ command_line_ids
+            if n in comment_line_ids
+                n = n + 1
+                continue
+            end
+            push!(parameters, lines[n])
+            n = n + 1
+        end
+        return "parameter", parameters
+    else
+        @warn "Implement $substr"
+        substr, nothing
+    end
+end
+
+function FiniteElementContainers.UnstructuredMesh(
+    ::FiniteElementContainers.AbaqusMesh, 
+    file_name::String,
+    create_edges::Bool, create_faces::Bool
+)
+    command_line_ids = Int[]
+    comment_line_ids = Int[]
+    comments = String[]
+
+    sections = Dict{String, Any}()
+    open(file_name, "r") do f
+        lines = readlines(f)
+
+        # do a first pass to find where commands stop
+        for (n, line) in enumerate(lines)
+            if _check_for_command(line)
+                push!(command_line_ids, n)
+            end
+
+            # skip comments
+            if _check_for_comment(line)
+                push!(comment_line_ids, n)
+            end
+        end
+
+        for command_id in command_line_ids
+            key, data = _get_command_data(lines, command_line_ids, comment_line_ids, command_id)
+
+            if key === nothing || data === nothing
+                @warn "Got nothing key/data for line $(lines[command_id])"
+                continue
+            end
+
+            if haskey(sections, key)
+                sections[key] = merge(sections[key], data)
+            else
+                sections[key] = data
+            end
+        end
+
+        for comment_id in comment_line_ids
+            push!(comments, lines[comment_id])
+        end
+    end
+
+    sections
+    # comments
+    # WIP below
+    # AbaqusReader.register_element!("CPE4T", 4, "Quad4")
+    # AbaqusReader.register_element!("C3D8RT", 8, "Hex8")
+    # AbaqusReader.register_element!("C3D8T", 8, "Hex8")
+
+    # # adding a few user element type names
+    # # we've encountered in the wild
+    # # NOTE this is probably dangerous in general
+    # AbaqusReader.register_element!("U1", 8, "Quad4")
+    # # AbaqusReader.register_element!("U3", 8, "Hex8")
+
+    # abq_data = abaqus_read_mesh(file_name)
+
+    # # nodal_coords = abq_data["nodes"]
+    # # nodal_coords = values(abq_data["nodes"])
+    # # node_id_map = keys(abq_data["nodes"])
+    # # perm = sortperm(keys(abq_data["nodes"]) |> collect)
+    # # node_id_map = collect(keys(abq_data["nodes"]))[perm]
+    # # nodal_coords = collect(values(abq_data["nodes"]))[perm]
+
+    # # node stuff
+    # # perm = Int[]
+    # abq_to_fem = Dict{Int, Int}()
+    # for (n, k) in enumerate(keys(abq_data["nodes"]))
+    #     # push!(perm, k)
+    #     abq_to_fem[k] = n
+    # end
+
+    # nodal_coords = Matrix{Float64}(undef, length(abq_data["nodes"][1]), length(abq_data["nodes"]))
+    # for (n, v) in enumerate(values(abq_data["nodes"]))
+    #     nodal_coords[:, n] = v
+    # end
+    # node_id_map = 1:size(nodal_coords, 2) |> collect
+
+    # # element stuff
+    # element_block_names = keys(abq_data["element_sets"])
+    # # NOTE these are mapping el id to type name
+    # # we need to sort through this with the 
+    # element_types = abq_data["element_types"]
+    # # abq_conns = values(abq_data["element_sets"])
+    # blocks = abq_data["element_sets"]
+
+    # # @show length(element_types)
+    # # @show length(blocks)
+
+    # # collect element type by block
+    # for block in values(blocks)
+    #     # el_types = Vector{Symbol}(undef, length(block))
+    #     el_types = Symbol[]
+    #     for (k, v) in element_types
+    #         # el_types[k] = v
+    #         push!(el_types, v)
+    #     end
+    #     display(el_types)
+    #     # @assert length(unique(el_types)) == 1
+    # end
+
+    # # element_conns = L2ElementField[]
+
+    # # for conn in abq_conns
+    # #     # block_el_ids = map(x -> )
+    # #     # n_elem = length(conn)
+    # #     # fem_conns = Matrix{Int}(undef, )
+    # #     # @show el_type
+    # #     # el_type = unique(el_type)
+    # #     # @show el_type
+    # #     # @assert length(el_type) == 1
+    # #     # el_type = el_type[1]
+    # #     # @show el_type
+
+    # #     if el_type == :Hex8
+    # #         nnpe = 8
+    # #     elseif el_type == :Quad4
+    # #         nnpe = 4
+    # #     elseif el_type == :Tri3
+    # #         nnpe = 3
+    # #     elseif el_type == :Tri6
+    # #         nnpe = 6
+    # #     elseif el_type == :Tet4
+    # #         nnpe = 4
+    # #     elseif el_type == :Tet10
+    # #         nnpe = 10
+    # #     else
+    # #         # @assert false "Got unnsupported el_type $el_type"
+    # #     end
+
+    # #     ne = length(conn) ÷ nnpe
+    # #     conn_rs = reshape(conn, nnpe, ne)
+    # #     display(conn_rs )
+    # # end
+    # # # nodesets
+    # # nodeset_nodes = Dict{Symbol, Vector{Int}}()
+    # # for (k, v) in abq_data["node_sets"]
+    # #     nodeset_nodes[Symbol(k)] = v
+    # # end
+
+    # # TODO need to set up sidesets
+
+    # # TODO edge/face connectivities
+    # # abq_data
+    # # nodal_coords
+    # # abq_conns
+    # abq_data
+end
+
+end # module

ext/FiniteElementContainersBlockArraysExt.jl → ext/BlockArraysExt.jl

@@ -1,4 +1,4 @@
-module FiniteElementContainersBlockArraysExt
+module BlockArraysExt
 
 using BlockArrays
 using FiniteElementContainers

ext/FiniteElementContainersCUDAExt.jl → ext/CUDAExt.jl

@@ -1,4 +1,4 @@
-module FiniteElementContainersCUDAExt
+module CUDAExt
 
 using Adapt
 using CUDA