self-hosted

.github/.kodiak.toml

@@ -1,9 +1,9 @@
-# https://kodiakhq.com/docs/recipes
-version = 1
-
-[merge]
-  method                               = "squash"
-  block_on_neutral_required_check_runs = true
-  [merge.message]
-    title = "pull_request_title"
-    body  = "pull_request_body"
+# https://kodiakhq.com/docs/recipes
+version = 1
+
+[merge]
+method = "squash"
+block_on_neutral_required_check_runs = true
+  [merge.message]
+  title = "pull_request_title"
+  body  = "pull_request_body"

.github/ci.jl

@@ -1,27 +1,17 @@
-using JSON
-using Literate
+using Distributed
+using Tables
+using MarkdownTables
+using SHA
 
-ENV["GKSwstype"] = "100"
-
-function main(; rmsvg=true)
-    file = get(ENV, "NB", "test.ipynb")
-    cachedir = get(ENV, "NBCACHE", ".cache")
-    nb = if endswith(file, ".jl")
-        run_literate(file; cachedir)
-    elseif endswith(file, ".ipynb")
-        lit = to_literate(file)
-        run_literate(lit; cachedir)
-    else
-        error("$(file) is not a valid notebook file!")
-    end
-    rmsvg && strip_svg(nb)
-    return nothing
+@everywhere begin
+    ENV["GKSwstype"] = "100"
+    using Literate, JSON
 end
 
 # Strip SVG output from a Jupyter notebook
-function strip_svg(ipynb)
-    oldfilesize = filesize(ipynb)
-    nb = open(JSON.parse, ipynb, "r")
+@everywhere function strip_svg(nbpath)
+    oldfilesize = filesize(nbpath)
+    nb = open(JSON.parse, nbpath, "r")
     for cell in nb["cells"]
         !haskey(cell, "outputs") && continue
         for output in cell["outputs"]
@@ -33,14 +23,34 @@ function strip_svg(ipynb)
             end
         end
     end
-    rm(ipynb; force=true)
-    write(ipynb, JSON.json(nb, 1))
-    @info "Stripped SVG in $(ipynb). The original size is $(Base.format_bytes(oldfilesize)). The new size is $(Base.format_bytes(filesize(ipynb)))."
-    return ipynb
+    rm(nbpath; force=true)
+    write(nbpath, JSON.json(nb, 1))
+    @info "Stripped SVG in $(nbpath). The original size is $(Base.format_bytes(oldfilesize)). The new size is $(Base.format_bytes(filesize(nbpath)))."
+    return nbpath
+end
+
+# Remove cached notebook and sha files if there is no corresponding notebook
+function clean_cache(cachedir)
+    for (root, _, files) in walkdir(cachedir)
+        for file in files
+            fn, ext = splitext(file)
+            if ext == ".sha"
+                target = joinpath(joinpath(splitpath(root)[2:end]), fn)
+                nb = target * ".ipynb"
+                lit = target * ".jl"
+                if !isfile(nb) && !isfile(lit)
+                    cachepath = joinpath(root, fn)
+                    @info "Notebook $(nb) or $(lit) not found. Removing $(cachepath) SHA and notebook."
+                    rm(cachepath * ".sha")
+                    rm(cachepath * ".ipynb"; force=true)
+                end
+            end
+        end
+    end
 end
 
 # Convert a Jupyter notebook into a Literate notebook. Adapted from https://github.com/JuliaInterop/NBInclude.jl.
-function to_literate(nbpath; shell_or_help = r"^\s*[;?]")
+function to_literate(nbpath; shell_or_help=r"^\s*[;?]")
     nb = open(JSON.parse, nbpath, "r")
     jlpath = splitext(nbpath)[1] * ".jl"
     open(jlpath, "w") do io
@@ -62,12 +72,78 @@ function to_literate(nbpath; shell_or_help = r"^\s*[;?]")
     return jlpath
 end
 
-function run_literate(file; cachedir = ".cache")
+# List notebooks without caches in a file tree
+function list_notebooks(basedir, cachedir)
+    list = String[]
+    for (root, _, files) in walkdir(basedir)
+        for file in files
+            name, ext = splitext(file)
+            if ext == ".ipynb" || ext == ".jl"
+                nb = joinpath(root, file)
+                shaval = read(nb, String) |> sha256 |> bytes2hex
+                @info "$(nb) SHA256 = $(shaval)"
+                shafilename = joinpath(cachedir, root, name * ".sha")
+                if isfile(shafilename) && read(shafilename, String) == shaval
+                    @info "$(nb) cache hits and will not be executed."
+                else
+                    @info "$(nb) cache misses. Writing hash to $(shafilename)."
+                    mkpath(dirname(shafilename))
+                    write(shafilename, shaval)
+                    if ext == ".ipynb"
+                        litnb = to_literate(nb)
+                        rm(nb; force=true)
+                        push!(list, litnb)
+                    elseif ext == ".jl"
+                        push!(list, nb)
+                    end
+                end
+            end
+        end
+    end
+    return list
+end
+
+# Run a Literate notebook
+@everywhere function run_literate(file, cachedir; rmsvg=true)
     outpath = joinpath(abspath(pwd()), cachedir, dirname(file))
     mkpath(outpath)
     ipynb = Literate.notebook(file, dirname(file); mdstrings=true, execute=true)
+    rmsvg && strip_svg(ipynb)
     cp(ipynb, joinpath(outpath, basename(ipynb)); force=true)
     return ipynb
 end
 
+function main(;
+    basedir=get(ENV, "DOCDIR", "docs"),
+    cachedir=get(ENV, "NBCACHE", ".cache"),
+    rmsvg=true)
+
+    mkpath(cachedir)
+    clean_cache(cachedir)
+    litnbs = list_notebooks(basedir, cachedir)
+
+    if !isempty(litnbs)
+        # Execute literate notebooks in worker process(es)
+        ts_lit = pmap(litnbs; on_error=identity) do nb
+            @elapsed run_literate(nb, cachedir; rmsvg)
+        end
+        rmprocs(workers()) # Remove worker processes to release some memory
+        failed = false
+        for (nb, t) in zip(litnbs, ts_lit)
+            if t isa ErrorException
+                println("Notebook: ", nb, "failed with error: \n", t.msg)
+                failed = true
+            end
+        end
+
+        if failed
+            error("Please check literate notebook error(s).")
+        else
+            # Print execution result
+            Tables.table([litnbs ts_lit]; header=["Notebook", "Elapsed (s)"]) |> markdown_table(String) |> print
+        end
+    end
+end
+
+# Run code
 main()