Merge pull request #24 from citation-file-format/jspaaks-codemeta-export

Jspaaks codemeta export

.gitignore

@@ -99,3 +99,7 @@ ENV/
 
 # mypy
 .mypy_cache/
+
+# IDE
+.idea
+

cff_converter_python/citation.py

@@ -20,7 +20,7 @@ def _get_baseurl(self):
         if self.url[0:18] == "https://github.com":
             self.baseurl = "https://raw.githubusercontent.com"
         else:
-            raise Exception("Only GitHub is supported at the moment.")
+            raise Exception("Only 'https://github.com' URLs are supported at the moment.")
 
     def _retrieve_file(self):
 
@@ -29,22 +29,30 @@ def _retrieve_file(self):
                             "(?P<user>[^/\n]*)/" +
                             "(?P<repo>[^/\n]*)" +
                             "(/(?P<branch>[^/\n]*))?", re.IGNORECASE)
-        matched = re.match(regexp, self.url).groupdict()
+
+        matched = re.match(regexp, self.url)
+        if matched is None:
+            raise Exception("Error extracting (user|organization) and/or repository " +\
+                            "information from the provided URL ({0}).".format(self.url))
+        else:
+            url_parts = matched.groupdict()
 
         self.file_url = "/".join([self.baseurl,
-                                  matched["user"],
-                                  matched["repo"],
-                                  matched["branch"] if matched["branch"] is not None else "master",
+                                  url_parts["user"],
+                                  url_parts["repo"],
+                                  url_parts["branch"] if url_parts["branch"] is not None else "master",
                                   "CITATION.cff"])
 
         r = requests.get(self.file_url)
         if r.ok:
             self.file_contents = r.text
         else:
-            raise Warning("status not '200 OK'")
+            raise Exception("Error requesting file: {0}".format(self.file_url))
 
     def _parse_yaml(self):
         self.as_yaml = yaml.safe_load(self.file_contents)
+        if not isinstance(self.as_yaml, dict):
+            raise Exception("Provided CITATION.cff does not seem valid YAML.")
 
     def as_bibtex(self):
 
@@ -61,26 +69,30 @@ def get_author_string():
                 if "name-suffix" in author:
                     authors.append(author["name-suffix"])
                 arr.append(" " * 12 + " ".join(authors))
-            return " and\n".join(arr) + "\n"
+            return " and\n".join(arr)
 
-        width = 8
         s = ""
         s += "@misc{"
-        s += "YourReferenceHere,\n"
-        s += "author".ljust(width, " ") + " = {\n"
-        s += get_author_string()
-        s += " " * 11 + "},\n"
-        s += "title".ljust(width, " ") + " = {"
-        s += self.as_yaml["title"] + "},\n"
-        s += "month".ljust(width, " ") + " = {"
-        s += str(self.as_yaml["date-released"].month) + "},\n"
-        s += "year".ljust(width, " ") + " = {"
-        s += str(self.as_yaml["date-released"].year) + "},\n"
-        s += "doi".ljust(width, " ") + " = {"
-        s += self.as_yaml["doi"] + "},\n"
-        s += "url".ljust(width, " ") + " = {"
-        s += self.as_yaml["repository-code"] + "}\n"
-        s += "}\n"
+        s += "YourReferenceHere"
+        if "authors" in self.as_yaml:
+            s += ",\nauthor = {\n"
+            s += get_author_string()
+            s += "\n         }"
+        if "title" in self.as_yaml:
+            s += ",\ntitle  = {"
+            s += self.as_yaml["title"] + "}"
+        if "date-released" in self.as_yaml:
+            s += ",\nmonth  = {"
+            s += str(self.as_yaml["date-released"].month) + "}"
+            s += ",\nyear   = {"
+            s += str(self.as_yaml["date-released"].year) + "}"
+        if "doi" in self.as_yaml:
+            s += ",\ndoi    = {"
+            s += self.as_yaml["doi"] + "}"
+        if "repository-code" in self.as_yaml:
+            s += ",\nurl    = {"
+            s += self.as_yaml["repository-code"] + "}"
+        s += "\n}\n"
 
         return s
 
@@ -94,7 +106,7 @@ def construct_author_string():
                 if "family-names" in author:
                     name += author["family-names"]
                 if "name-suffix" in author:
-                    name += author["name-suffix"]
+                    name += " " + author["name-suffix"]
                 if "given-names" in author:
                     name += ", " + author["given-names"]
                 name += "\n"
@@ -115,16 +127,43 @@ def construct_date_string():
 
         s = ""
         s += "TY  - COMP\n"
-        s += construct_author_string()
-        s += "DO  - " + self.as_yaml["doi"] + "\n"
-        s += construct_keywords_string()
+
+        if "authors" in self.as_yaml:
+            s += construct_author_string()
+        else:
+            s += "AU  -\n"
+
+        if "doi" in self.as_yaml:
+            s += "DO  - " + self.as_yaml["doi"] + "\n"
+        else:
+            s += "DO  -\n"
+
+        if "keywords" in self.as_yaml:
+            s += construct_keywords_string()
+        else:
+            s += "KW  -\n"
+
         s += "M3  - software\n"
         s += "PB  - GitHub Inc.\n"
         s += "PP  - San Francisco, USA\n"
-        s +=  construct_date_string()
-        s += "T1  - " + self.as_yaml["title"] + "\n"
-        s += "UR  - " + self.as_yaml["repository-code"] + "\n"
+
+        if "date-released" in self.as_yaml:
+            s += construct_date_string()
+        else:
+            s += "PY  -\n"
+
+        if "title" in self.as_yaml:
+            s += "T1  - " + self.as_yaml["title"] + "\n"
+        else:
+            s += "T1  -\n"
+
+        if "repository-code" in self.as_yaml:
+            s += "UR  - " + self.as_yaml["repository-code"] + "\n"
+        else:
+            s += "UR  -\n"
+
         s += "ER  -\n"
+
         return s
 
     def as_enw(self):
@@ -138,24 +177,34 @@ def construct_author_string():
                 if "family-names" in author:
                     name += author["family-names"]
                 if "name-suffix" in author:
-                    name += author["name-suffix"]
+                    name += " " + author["name-suffix"]
                 if "given-names" in author:
                     name += ", " + author["given-names"]
                 names.append(name)
             return " & ".join(names)
 
         def construct_keywords_string():
-            if "keywords" in self.as_yaml:
-                return ", ".join(["\"" + keyword + "\"" for keyword in self.as_yaml["keywords"]])
-            else:
-                return ""
+            return ", ".join(["\"" + keyword + "\"" for keyword in self.as_yaml["keywords"]])
 
         s = ""
         s += "%0\n"
         s += "%0 Generic\n"
-        s += "%A " + construct_author_string() + "\n"
-        s += "%D " + str(self.as_yaml["date-released"].year) + "\n"
-        s += "%T " + self.as_yaml["title"] + "\n"
+
+        if "authors" in self.as_yaml:
+            s += "%A " + construct_author_string() + "\n"
+        else:
+            s += "%A\n"
+
+        if "date-released" in self.as_yaml:
+            s += "%D " + str(self.as_yaml["date-released"].year) + "\n"
+        else:
+            s += "%D\n"
+
+        if "title" in self.as_yaml:
+            s += "%T " + self.as_yaml["title"] + "\n"
+        else:
+            s += "%T\n"
+
         s += "%E\n"
         s += "%B\n"
         s += "%C\n"
@@ -168,7 +217,11 @@ def construct_keywords_string():
         s += "%Y\n"
         s += "%S\n"
         s += "%7\n"
-        s += "%8 " + str(self.as_yaml["date-released"].month) + "\n"
+        if "date-released" in self.as_yaml:
+            s += "%8 " + str(self.as_yaml["date-released"].month) + "\n"
+        else:
+            s += "%8\n"
+
         s += "%9\n"
         s += "%?\n"
         s += "%!\n"
@@ -187,9 +240,75 @@ def construct_keywords_string():
         s += "%#\n"
         s += "%$\n"
         s += "%F YourReferenceHere\n"
-        s += "%K " + construct_keywords_string() + "\n"
+        if "keywords" in self.as_yaml:
+            s += "%K " + construct_keywords_string() + "\n"
+        else:
+            s += "%K\n"
+
         s += "%X\n"
         s += "%Z\n"
-        s += "%U " + self.as_yaml["repository-code"] + "\n"
+        if "repository-code" in self.as_yaml:
+            s += "%U " + self.as_yaml["repository-code"] + "\n"
+        else:
+            s += "%U\n"
 
         return s
+
+    def as_codemeta(self):
+
+        def resolve_spdx_license(spdx_license_code):
+            licenses_url = "https://raw.githubusercontent.com/spdx/license-list-data" + \
+                           "/b541ee8a345aa93b70a08765c7bf5e423bb4d558/json/licenses.json"
+            r = requests.get(licenses_url)
+            if r.ok:
+                data = r.json()
+                return [license["seeAlso"][0] for license in data["licenses"] if license["licenseId"] == spdx_license_code][0]
+            else:
+                raise Warning("status not '200 OK'")
+
+        def convert(author):
+
+            family_names = list()
+            for name_part in ["name-particle", "family-names", "name-suffix"]:
+                if name_part in author.keys() and author[name_part] is not "":
+                    family_names.append(author[name_part])
+
+            author_str = ''
+            author_str += '{\n'
+            author_str += '        "@type": "Person"'
+            if "given-names" in author:
+                author_str += ',\n        "givenName": "{0}"'.format(author["given-names"])
+            author_str += ',\n        "familyName": "{0}"'.format(" ".join(family_names))
+            if "affiliation" in author:
+                author_str += ',\n'
+                author_str += '        "affiliation": {\n'
+                author_str += '            "@type": "Organization",\n'
+                author_str += '            "legalName": "{0}"\n'.format(author["affiliation"])
+                author_str += '        }'
+            author_str += '\n    }'
+            return author_str
+
+        s = ''
+        s += '{\n'
+        s += '    "@context": "http://schema.org",\n'
+        s += '    "@type": "SoftwareSourceCode"'
+        if "repository-code" in self.as_yaml:
+            s += ',\n    "codeRepository": "{0}"'.format(self.as_yaml["repository-code"])
+        if "date-released" in self.as_yaml:
+            s += ',\n    "datePublished": "{0}"'.format(self.as_yaml["date-released"])
+        if "authors" in self.as_yaml:
+            s += ',\n    "author": [{0}]'.format(", ".join([convert(author) for author in self.as_yaml["authors"]]))
+        if "keywords" in self.as_yaml:
+            s += ',\n    "keywords": [{0}]'.format(", ".join(['"{0}"'.format(kw) for kw in self.as_yaml["keywords"]]))
+        if "license" in self.as_yaml:
+            s += ',\n    "license": "{0}"'.format(resolve_spdx_license(self.as_yaml["license"]))
+        if "version" in self.as_yaml:
+            s += ',\n    "version": "{0}"'.format(self.as_yaml["version"])
+        if "doi" in self.as_yaml:
+            s += ',\n    "identifier": "https://doi.org/{0}"'.format(self.as_yaml["doi"])
+        if "title" in self.as_yaml:
+            s += ',\n    "name": "{0}"'.format(self.as_yaml["title"])
+        s += '\n}\n'
+
+        return s
+

fixtures/bibtex-1

@@ -0,0 +1,11 @@
+@misc{YourReferenceHere,
+author = {
+            Jurriaan H. Spaaks and
+            Tom Klaver
+         },
+title  = {cff-converter-python},
+month  = {1},
+year   = {2018},
+doi    = {10.5281/zenodo.1162057},
+url    = {https://github.com/citation-file-format/cff-converter-python}
+}

fixtures/bibtex-2

@@ -0,0 +1,6 @@
+@misc{YourReferenceHere,
+author = {
+            Gonzalo Fernández de Córdoba Jr.
+         },
+title  = {example title}
+}

fixtures/bibtex-3

@@ -0,0 +1,10 @@
+@misc{YourReferenceHere,
+author = {
+            Dafne van Kuppevelt and
+            Christiaan Meijer and
+            Vincent van Hees and
+            Mateusz Kuzak
+         },
+title  = {mcfly},
+url    = {https://github.com/NLeSC/mcfly}
+}

fixtures/cff-converter-python-citationcff-1 → fixtures/citationcff-1

@@ -1,22 +1,22 @@
 # YAML 1.2
----
-authors:
-  -
+--- 
+authors: 
+  - 
     affiliation: "Netherlands eScience Center"
     family-names: Spaaks
-    given-names: Jurriaan H.
-  -
+    given-names: "Jurriaan H."
+  - 
     affiliation: "Netherlands eScience Center"
     family-names: Klaver
     given-names: Tom
 cff-version: "1.0.3"
 date-released: 2018-01-16
 doi: 10.5281/zenodo.1162057
-keywords:
-  - "citation"
-  - "bibliography"
-  - "cff"
-  - "CITATION.cff"
+keywords: 
+  - citation
+  - bibliography
+  - cff
+  - CITATION.cff
 license: Apache-2.0
 message: "If you use this software, please cite it using these metadata."
 repository-code: "https://github.com/citation-file-format/cff-converter-python"

fixtures/citationcff-2

@@ -0,0 +1,11 @@
+# YAML 1.2
+--- 
+authors: 
+  - 
+    family-names: "Fernández de Córdoba"
+    given-names: Gonzalo
+    name-suffix: Jr.
+cff-version: "1.0.3"
+message: "Example data from the CITATION.cff spec 1.0.3"
+title: "example title"
+version: "1.0.0"