Merge pull request #10 from davibarreira/zenodo

Implemented Getting Zenodo citation

Project.toml

@@ -4,14 +4,18 @@ authors = ["Sebastian Micluța-Câmpeanu <[email protected]>"]
 version = "0.1.0"
 
 [deps]
+BibParser = "13533e5b-e1c2-4e57-8cef-cac5e52f6474"
 Bibliography = "f1be7e48-bf82-45af-a471-ae754a193061"
 DataStructures = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
+HTTP = "cd3eb016-35fb-5094-929b-558a96fad6f3"
 InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
 
 [compat]
-DataStructures = "0.18"
 Bibliography = "0.2"
+BibParser = "0.1"
+DataStructures = "0.18"
+HTTP = "0.9"
 julia = "1.6"
 
 [extras]

src/PkgCite.jl

@@ -9,6 +9,8 @@ import Pkg
 using Bibliography: import_bibtex, export_bibtex, Entry
 using DataStructures
 using InteractiveUtils
+using BibParser: parse_entry
+using HTTP
 
 const DEFAULT_CITE = "\\cite"
 

src/citations.jl

@@ -5,32 +5,43 @@ function citation_path(pkg)
     end
 end
 
-function get_citation(pkg)
+
+# Added `badge` flag to avoid breaking current tests
+function get_citation(pkg; badge=false)
     bib_path = citation_path(pkg)
+    if badge == false
+        urlbadge = nothing
+    else
+        urlbadge = get_badge(pkg)
+    end
     if !isnothing(bib_path)
         @debug "Reading CITATION.bib for $(pkg.name)"
         try
             import_bibtex(bib_path)
         catch e
             @warn "There was an error reading the CITATION.bib file for $(pkg.name)" exception=e
         end
+    elseif !isnothing(urlbadge)
+        get_citation_badge(urlbadge)
     end
 end
 
 """
-    collect_citations(only_direct::Bool)
+    collect_citations(only_direct::Bool; badge=false)
 
 Collect the citations from all the dependencies in the current environment.
+Use `badge = true` to get the citations from packages without
+a `Citation.bib` file, but with a DOI badge.
 """
-function collect_citations(only_direct::Bool)
+function collect_citations(only_direct::Bool; badge=false)
     @debug "Generating citation report for the current environment"
     deps = Pkg.dependencies()
     pkg_citations = Dict{String, DataStructures.OrderedDict{String,Entry}}()
     for pkg in values(deps)
         if only_direct && !pkg.is_direct_dep
             continue
         end
-        c = get_citation(pkg)
+        c = get_citation(pkg, badge=badge)
         if !isnothing(c)
             push!(pkg_citations, pkg.name=>c)
         end
@@ -67,10 +78,12 @@ end
 This function will create a .bib file with all the citations collected form
 the CITATION.bib files corresponding to the dependecies of
 the current active environment. Use `filename` to change the name of the
-file. To include just the direct dependencies use `only_direct=true`
+file. To include just the direct dependencies use `only_direct=true`.
+Use `badge = true` to get the citations from packages without
+a `Citation.bib` file, but with a DOI badge.
 """
-function get_citations(;only_direct=false, filename="julia_citations.bib")
-    pkg_citations = collect_citations(only_direct)
+function get_citations(;only_direct=false, filename="julia_citations.bib", badge=false)
+    pkg_citations = collect_citations(only_direct, badge=badge)
 
     if isempty(pkg_citations)
         @warn "No citations found in current environment"
@@ -80,3 +93,50 @@ function get_citations(;only_direct=false, filename="julia_citations.bib")
 
     return nothing
 end
+
+"""
+    get_citation_badge(url::String)
+
+This function will return an `OrderedDict` in the default to BibTeX format
+from `Bibliography.jl`. The `url` argument is the link present in the DOI badge.
+"""
+function get_citation_badge(urlbadge)
+    isnothing(urlbadge) && return urlbadge
+
+    index_doi = findfirst("https://doi.org/", urlbadge)
+    if !isnothing(index_doi)
+        # If badge link is the doi
+        doi =  urlbadge[index_doi[end]+1:end]
+    else
+        # If link is for `lastestdoi` in Zenodo
+        header = HTTP.head(urlbadge, redirect=false).headers
+        doi = last(header[findfirst(i -> isequal("Location", first(i)), header)])[17:end]
+    end
+    url = joinpath("https://data.datacite.org/", doi)
+    resp = HTTP.get(url, ["Accept"=>"application/x-bibtex"]; forwardheaders=true).body |> String
+    bib = parse_entry(resp)
+    return bib
+end
+
+"""
+    get_badge(pkg)
+
+This function extracts the link in the DOI badge
+present in the package's `README.md`.
+"""
+function get_badge(pkg)
+    readme_path = joinpath(pkg.source, "README.md")
+    if isfile(readme_path)
+        readme = open(f->read(f, String), readme_path)
+        getdoi = findfirst("![DOI]", readme)
+        if isnothing(getdoi)
+            urlbadge = nothing
+        else
+            index_init = findfirst('(', readme[getdoi[end]+3:end])[1]+getdoi[end]+3
+            index_final = findfirst(')',readme[index_init:end]) - 2
+            urlbadge = readme[index_init: index_init + index_final]
+        end
+        return urlbadge
+    end
+    return nothing
+end

src/tool_report.jl

@@ -62,7 +62,7 @@ function make_sentence(pkg_citations; cite_commands=Dict{String,String}(), jl=tr
 end
 
 """
-    get_tool_citation(io::IO=stdout; jl = true, texttt = false, copy = true, cite_commands=Dict{String,String}(), filename="julia_citations.bib")
+    get_tool_citation(io::IO=stdout; jl = true, texttt = false, copy = true, cite_commands=Dict{String,String}(), filename="julia_citations.bib"; badge=false)
 
 Print a sentence describing the packages used in the current environment.
 If you only want to consider the direct dependencies, you can set `only_direct=true`.
@@ -71,16 +71,19 @@ The package names have the .jl ending by default. You can ommit it with `jl=fals
 Package names can be wrapped in `texttt` by setting `texttt=true` and you can also customize
 the cite command used for each package by using `cite_commands=Dict("PackageName"=>"custom_cite")`.
 The filename of the .bib file can be passed via the `filename` keyword.
+Use `badge = true` to get the citations from packages without
+a `Citation.bib` file, but with a DOI badge.
 """
 function get_tool_citation(io::IO=stdout;
     jl = true,
     texttt = false,
     copy = true,
     only_direct = false,
     cite_commands=Dict{String,String}(),
-    filename="julia_citations.bib")
+    filename="julia_citations.bib",
+    badge=false)
 
-    pkg_citations = collect_citations(only_direct)
+    pkg_citations = collect_citations(only_direct, badge=badge)
 
     cite_sentence = make_sentence(pkg_citations; cite_commands, jl, texttt)
     println(io, cite_sentence)