Merge pull request #21 from mseri/fix-url

Fix #20

bin/doi2bib.ml

@@ -2,9 +2,46 @@ open Doi2bib
 
 let err s = `Error (false, s)
 
+(* Assumes "    url = { THE_URL }," is always well formed *)
+let url_re =
+  Re.compile
+    Re.(
+      seq
+        [
+          bol;
+          group (seq [ rep blank; str "url = {" ]);
+          group (rep (compl [ char '}' ]));
+          str "},";
+        ])
+
+let escape_table =
+  let escapes =
+    List.map
+      (fun s -> Re.compile @@ Re.str s)
+      [ "%2F"; "%28"; "%29"; "%3C"; "%3E"; "%3A"; "%3B" ]
+  in
+  let chars = [ "/"; "("; ")"; "<"; ">"; ":"; ";" ] in
+  List.combine escapes chars
+
+let unescape s =
+  List.fold_left
+    (fun s (re_code, chr) -> Re.replace_string ~all:true re_code ~by:chr s)
+    s escape_table
+
 let process_id outfile id =
   let open Lwt.Syntax in
   let* bibtex = Http.get_bib_entry @@ Parser.parse_id id in
+  let f grp =
+    let prefix = Re.Group.get grp 1 in
+    let url = Re.Group.get grp 2 in
+    String.concat "" [ prefix; unescape url; "}," ]
+  in
+  (* The doi produced by crossref contains %-escaped urls,
+     but %s are comments in bibtex/latex. This is perhaps a
+     brute way of dealing with it but it has been working
+     fine for me on a huge number of examples.
+     See https://github.com/mseri/doi2bib/issues/20 for context*)
+  let bibtex = Re.replace ~all:true url_re ~f bibtex in
   match outfile with
   | "stdout" -> Lwt_io.print bibtex
   | outfile ->

bin/dune

@@ -1,5 +1,5 @@
 (executable
  (name doi2bib)
  (public_name doi2bib)
- (libraries cmdliner doi2bib unix)
+ (libraries cmdliner doi2bib re unix)
  (preprocess future_syntax))

tests/arxiv.t/run.t

@@ -2,7 +2,7 @@ Arxiv entry (with DOI entry) executed with prefix
   $ doi2bib arXiv:1609.01724
   @article{Prandi_2018,
   	doi = {10.4171/jst/226},
-  	url = {https://doi.org/10.4171%2Fjst%2F226},
+  	url = {https://doi.org/10.4171/jst/226},
   	year = 2018,
   	month = {jul},
   	publisher = {European Mathematical Society - {EMS} - Publishing House {GmbH}},
@@ -17,7 +17,7 @@ Arxiv Entry (with DOI entry) executed without prefix
   $ doi2bib 1902.00436
   @article{Vermeeren_2019,
   	doi = {10.1088/1751-8121/ab4767},
-  	url = {https://doi.org/10.1088%2F1751-8121%2Fab4767},
+  	url = {https://doi.org/10.1088/1751-8121/ab4767},
   	year = 2019,
   	month = {oct},
   	publisher = {{IOP} Publishing},

tests/doi.t/run.t

@@ -2,7 +2,7 @@ DOI entry executed without prefix
   $ doi2bib 10.1007/s10569-019-9946-9
   @article{Bravetti_2020,
   	doi = {10.1007/s10569-019-9946-9},
-  	url = {https://doi.org/10.1007%2Fs10569-019-9946-9},
+  	url = {https://doi.org/10.1007/s10569-019-9946-9},
   	year = 2020,
   	month = {jan},
   	publisher = {Springer Science and Business Media {LLC}},
@@ -16,7 +16,7 @@ DOI entry executed with prefix
   $ doi2bib doi:10.4171/JST/226
   @article{Prandi_2018,
   	doi = {10.4171/jst/226},
-  	url = {https://doi.org/10.4171%2Fjst%2F226},
+  	url = {https://doi.org/10.4171/jst/226},
   	year = 2018,
   	month = {jul},
   	publisher = {European Mathematical Society - {EMS} - Publishing House {GmbH}},
@@ -27,3 +27,17 @@ DOI entry executed with prefix
   	title = {Quantum confinement on non-complete Riemannian manifolds},
   	journal = {Journal of Spectral Theory}
   }
+DOI entry containing parentheses
+  $ doi2bib "doi:10.1016/0393-0440(89)90029-6"
+  @article{Albert_1989,
+  	doi = {10.1016/0393-0440(89)90029-6},
+  	url = {https://doi.org/10.1016/0393-0440(89)90029-6},
+  	year = 1989,
+  	publisher = {Elsevier {BV}},
+  	volume = {6},
+  	number = {4},
+  	pages = {627--649},
+  	author = {Claude Albert},
+  	title = {Le th{\'{e}}or{\`{e}}me de r{\'{e}}duction de Marsden-Weinstein en g{\'{e}}om{\'{e}}trie cosymplectique et de contact},
+  	journal = {Journal of Geometry and Physics}
+  }

tests/pubmed.t/run.t

@@ -2,7 +2,7 @@ Pubmed entry with 'PMC' prefix.
   $ doi2bib PMC2883744
   @article{Comas_2010,
   	doi = {10.1038/ng.590},
-  	url = {https://doi.org/10.1038%2Fng.590},
+  	url = {https://doi.org/10.1038/ng.590},
   	year = 2010,
   	month = {may},
   	publisher = {Springer Science and Business Media {LLC}},