Updating to fix bug that we do not allow repos to end in .git (#42)

* updating to fix bug that we do not allow repos to end in .git
* making sure that zenodo update also updates orcid identifiers

Signed-off-by: vsoch <vsochat@stanford.edu>

.tributors

@@ -2,15 +2,19 @@
     "yarikoptic": {
         "name": "Yaroslav Halchenko",
         "blog": "www.onerussian.com",
+        "email": "debian@onerussian.com",
+        "bio": "Cheers!",
         "orcid": "0000-0003-3456-2493",
         "affiliation": "Dartmouth College"
     },
     "vsoch": {
         "name": "Vanessasaurus",
-        "blog": "https://vsoch.github.io"
+        "blog": "https://vsoch.github.io",
+        "bio": "I'm the Vanessasaurus!"
     },
     "pgrimaud": {
         "name": "Pierre Grimaud",
-        "blog": "https://github.com/pgrimaud"
+        "blog": "https://github.com/pgrimaud",
+        "bio": "Technical Manager @bigyouth / IT Teacher / Retired gladiator"
     }
 }

CHANGELOG.md

@@ -14,6 +14,7 @@ and **Merged pull requests**. Critical items to know are:
 The versions coincide with releases on pip.
 
 ## [0.0.x](https://github.com/con/tributors/tree/master) (0.0.x)
+ - should allow for repository names with .git extension (0.0.15)
  - adding support to do updates `--from` (between) resources (0.0.14)
  - adding github to update-lookup, not doing by default (0.0.13)
  - support for updating .tributors from a lookup (0.0.12)

tributors/main/github.py

@@ -251,4 +251,4 @@ def get_github_repository(repo):
     if not match:
         sys.exit("Malformed repository address %s" % repo)
     owner, repo = match.groups()
-    return "%s/%s" % (owner, repo)
+    return "%s/%s" % (owner, re.sub(".git$", "", repo, 1))

tributors/main/parsers/zenodo.py

@@ -15,6 +15,7 @@
 
 from tributors.utils.file import write_json
 from .base import ParserBase
+from tributors.main.orcid import get_orcid
 
 bot = logging.getLogger("    zenodo")
 
@@ -108,9 +109,26 @@ def update_from_orcids(self, orcids):
                 self.data["creators"].append(entry)
         return self.data["creators"]
 
+    def update_orcids(self):
+        """Zenodo is a special case that has emails and real usernames, so we
+           can parse through the existing file and look for orcid identifiers
+        """
+        creators = []
+        for user in self.data.get("creators", []):
+            name = user.get("name")
+            email = user.get("email")
+            if email or name and self.orcid_token is not None:
+                orcid = get_orcid(email=email, token=self.orcid_token, name=name)
+                if orcid:
+                    user["orcid"] = orcid
+            creators.append(user)
+        self.data["creators"] = creators
+
     def update_from_emails(self, emails):
-        """Given a list of emails, update the contributor file from it
+        """Given a list of emails, update the contributor file from it. We also
+           look for new orcid ids for emails that don't have them.
         """
+        # First loop through emails in the cache
         lookup = {x["email"]: x for _, x in self.cache.items() if "email" in x}
         for email in emails:
             if email in self.email_lookup:
@@ -124,7 +142,9 @@ def update_from_emails(self, emails):
         return self.data["creators"]
 
     def update_from_logins(self, logins):
-        """Given a list of logins, update the contributor file from it
+        """Given a list of logins, update the zenodo.json from it. We only
+           do this on init when we haven't added /updated logins with
+           people's actual names.
         """
         # GitHub contributors are the source of truth
         for login in logins:
@@ -169,8 +189,9 @@ def update(self, thresh=1, from_resources=None, save=True):
 
         self.update_cache()
 
-        # Update zenodo file from GitHub logins (default) or other
-        self.update_from_logins(from_resources.get("login", []))
+        # Here we can only reasonable update from orcids (not logins)
+        self.update_orcids()
+        self.update_from_emails(from_resources.get("email", []))
         self.update_from_orcids(from_resources.get("orcid", []))
         self.update_from_names(from_resources.get("names", []))
         if save:

tributors/version.py

@@ -8,7 +8,7 @@
 
 """
 
-__version__ = "0.0.14"
+__version__ = "0.0.15"
 AUTHOR = "Vanessa Sochat"
 AUTHOR_EMAIL = "vsochat@stanford.edu"
 NAME = "tributors"