handle versions in Figshare properly

Author: nuest

repo2docker/contentproviders/figshare.py

@@ -20,6 +20,7 @@ class Figshare(DoiProvider):
 
     Examples:
       - https://doi.org/10.6084/m9.figshare.9782777
+      - https://doi.org/10.6084/m9.figshare.9782777.v2
       - https://figshare.com/articles/binder-examples_requirements/9784088 (only one zipfile, no DOI)
     """
 
@@ -38,7 +39,7 @@ def __init__(self):
             }
         ]
 
-    url_regex = re.compile(r"(.*)/articles/([^/]+)/(\d+)(/\d)?")
+    url_regex = re.compile(r"(.*)/articles/([^/]+)/(\d+)(/)?(\d)?")
 
     def detect(self, doi, ref=None, extra_args=None):
         """Trigger this provider for things that resolve to a Figshare article"""
@@ -53,7 +54,14 @@ def detect(self, doi, ref=None, extra_args=None):
                 match = self.url_regex.match(url)
                 if match:
                     self.article_id = match.groups()[2]
-                    return {"article": self.article_id, "host": host}
+                    self.article_version = match.groups()[4]
+                    if not self.article_version:
+                        self.article_version = "1"
+                    return {
+                        "article": self.article_id,
+                        "host": host,
+                        "version": self.article_version,
+                    }
                 else:
                     return None
 
@@ -62,9 +70,13 @@ def fetch(self, spec, output_dir, yield_output=False):
         article_id = spec["article"]
         host = spec["host"]
 
-        yield "Fetching Figshare article {}.\n".format(article_id)
+        yield "Fetching Figshare article {} in version {}.\n".format(
+            self.article_id, self.article_version
+        )
         req = Request(
-            "{}{}".format(host["api"], article_id),
+            "{}{}/versions/{}".format(
+                host["api"], self.article_id, self.article_version
+            ),
             headers={"accept": "application/json"},
         )
         resp = self.urlopen(req)
@@ -84,3 +96,8 @@ def fetch(self, spec, output_dir, yield_output=False):
     def content_id(self):
         """The Figshare article ID"""
         return self.article_id
+
+    @property
+    def content_version(self):
+        """The Figshare article ID"""
+        return self.article_version

tests/unit/contentproviders/test_figshare.py

@@ -1,5 +1,6 @@
 import json
 import os
+import re
 import pytest
 
 from contextlib import contextmanager
@@ -10,6 +11,7 @@
 from zipfile import ZipFile
 
 from repo2docker.contentproviders import Figshare
+from repo2docker.__main__ import make_r2d
 
 
 def test_content_id():
@@ -21,31 +23,49 @@ def test_content_id():
 
 
 test_fig = Figshare()
+test_fig.article_id = "123456"
+test_fig.article_version = "42"
+
 test_dois_links = [
-    ("10.6084/m9.figshare.9782777", {"host": test_fig.hosts[0], "article": "9782777"}),
+    (
+        "10.6084/m9.figshare.9782777",
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "1"},
+    ),
     (
         "10.6084/m9.figshare.9782777.v1",
-        {"host": test_fig.hosts[0], "article": "9782777"},
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "1"},
+    ),
+    (
+        "10.6084/m9.figshare.9782777.v2",
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "2"},
     ),
     (
-        "https://doi.org/10.6084/m9.figshare.9782777",
-        {"host": test_fig.hosts[0], "article": "9782777"},
+        "https://doi.org/10.6084/m9.figshare.9782777.v1",
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "1"},
+    ),
+    (
+        "https://doi.org/10.6084/m9.figshare.9782777.v3",
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "3"},
     ),
     (
         "https://figshare.com/articles/title/97827771234",
-        {"host": test_fig.hosts[0], "article": "97827771234"},
+        {"host": test_fig.hosts[0], "article": "97827771234", "version": "1"},
     ),
     (
         "https://figshare.com/articles/title/9782777/1",
-        {"host": test_fig.hosts[0], "article": "9782777"},
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "1"},
+    ),
+    (
+        "https://figshare.com/articles/title/9782777/2",
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "2"},
     ),
     (
         "https://figshare.com/articles/title/9782777/",
-        {"host": test_fig.hosts[0], "article": "9782777"},
+        {"host": test_fig.hosts[0], "article": "9782777", "version": "1"},
     ),
 ]
 
-test_spec = {"host": test_fig.hosts[0], "article": "1234"}
+test_spec = {"host": test_fig.hosts[0], "article": "1234", "version": "42"}
 
 
 @pytest.mark.parametrize("test_input,expected", test_dois_links)
@@ -96,9 +116,8 @@ def mock_urlopen(self, req):
 
         with patch.object(Figshare, "urlopen", new=mock_urlopen):
             with TemporaryDirectory() as d:
-                fig = Figshare()
                 output = []
-                for l in fig.fetch(test_spec, d):
+                for l in test_fig.fetch(test_spec, d):
                     output.append(l)
 
                 unpacked_files = set(os.listdir(d))
@@ -137,10 +156,8 @@ def mock_urlopen(self, req):
 
             with patch.object(Figshare, "urlopen", new=mock_urlopen):
                 with TemporaryDirectory() as d:
-                    fig = Figshare()
-
                     output = []
-                    for l in fig.fetch(test_spec, d):
+                    for l in test_fig.fetch(test_spec, d):
                         output.append(l)
 
                     unpacked_files = set(os.listdir(d))