crossref importer checks for existing DOIs

2 files changed, 19 insertions, 4 deletions

diff --git a/python/fatcat_tools/importers/crossref.py b/python/fatcat_tools/importers/crossref.py
index 475afdb0..385a8235 100644
--- a/python/fatcat_tools/importers/crossref.py
+++ b/python/fatcat_tools/importers/crossref.py
@@ -40,7 +40,7 @@ class CrossrefImporter(FatcatImporter):
     See https://github.com/CrossRef/rest-api-doc for JSON schema notes
     """
 
-    def __init__(self, host_url, issn_map_file, extid_map_file=None, create_containers=True):
+    def __init__(self, host_url, issn_map_file, extid_map_file=None, create_containers=True, check_existing=True):
         super().__init__(host_url, issn_map_file)
         self.extid_map_db = None
         if extid_map_file:
@@ -50,6 +50,7 @@ class CrossrefImporter(FatcatImporter):
         else:
             print("Not using external ID map")
         self.create_containers = create_containers
+        self.check_existing = check_existing
 
     def lookup_ext_ids(self, doi):
         if self.extid_map_db is None:
@@ -85,6 +86,20 @@ class CrossrefImporter(FatcatImporter):
                 'book-track', 'proceedings-series'):
             return None
 
+        # lookup existing DOI
+        existing_release = None
+        if self.check_existing:
+            try:
+                existing_release = self.api.lookup_release(doi=obj['DOI'].lower())
+            except fatcat_client.rest.ApiException as err:
+                if err.status != 404:
+                    raise err
+
+        # eventually we'll want to support "updates", but for now just skip if
+        # entity already exists
+        if existing_release:
+            return None
+
         # contribs
         def do_contribs(obj_list, ctype):
             contribs = []
diff --git a/python/fatcat_tools/importers/matched.py b/python/fatcat_tools/importers/matched.py
index 732fccbe..6f83dd23 100644
--- a/python/fatcat_tools/importers/matched.py
+++ b/python/fatcat_tools/importers/matched.py
@@ -37,12 +37,12 @@ class MatchedImporter(FatcatImporter):
     - core_id, wikidata_id, pmcid, pmid: not as lists
     """
 
-    def __init__(self, host_url, skip_file_update=False, default_mime=None,
+    def __init__(self, host_url, skip_file_updates=False, default_mime=None,
             default_link_rel="web"):
         super().__init__(host_url)
         self.default_mime = default_mime
         self.default_link_rel = default_link_rel
-        self.skip_file_update = skip_file_update
+        self.skip_file_updates = skip_file_updates
 
     def make_url(self, raw):
         rel = self.default_link_rel
@@ -61,7 +61,7 @@ class MatchedImporter(FatcatImporter):
 
         # lookup sha1, or create new entity
         fe = None
-        if not self.skip_file_update:
+        if not self.skip_file_updates:
             try:
                 fe = self.api.lookup_file(sha1=sha1)
             except fatcat_client.rest.ApiException as err: