refactor DOI domain lookup into python code; expand table

3 files changed, 101 insertions, 29 deletions

diff --git a/fatcat_scholar/biblio_hacks.py b/fatcat_scholar/biblio_hacks.py
new file mode 100644
index 0000000..935d1ff
--- /dev/null
+++ b/fatcat_scholar/biblio_hacks.py
@@ -0,0 +1,82 @@
+from typing import Optional
+
+DOI_PREFIX_MAP = {
+    # simple entries (mostly OA and platforms)
+    "10.2307": {"domain": "jstor.org"},
+    "10.11501": {"domain": "ndl.go.jp"},
+    "10.6084": {"domain": "figshare.com"},
+    "10.5281": {"domain": "zenodo.org"},
+    "10.1590": {"domain": "scielo.br"},
+    "10.1371": {"domain": "plos.org"},
+    "10.1155": {"domain": "hindawi.com"},
+    "10.7554": {"domain": "elifesciences.com"},
+    "10.1145": {"domain": "acm.org"},
+    # more complex publisher mappings (verify journal/publisher)
+    "10.1016": {"domain": "elsevier.com", "publisher": "elsevier"},
+    "10.1007": {"domain": "springer.com", "publisher": "springer"},
+    "10.1186": {"domain": "springer.com", "publisher": "springer"},
+    "10.1002": {"domain": "wiley.com", "publisher": "wiley"},
+    "10.1109": {"domain": "ieee.com", "publisher": "ieee"},
+    "10.1080": {"domain": "tandfonline.com", "publisher": "informa"},
+    "10.1093": {"domain": "oup.com", "publisher": "oxford"},
+    "10.1111": {"domain": "sagepub.com", "publisher": "sage"},
+    "10.1042": {"domain": "sagepub.com", "publisher": "sage"},
+    "10.1177": {"domain": "sagepub.com", "publisher": "sage"},
+    "10.1021": {"domain": "acs.org", "publisher": "acs"},
+    "10.1017": {"domain": "cambridge.org", "publisher": "cambridge"},
+    # "10.1097": {"domain": "lww.org",    "publisher": "wolters"},
+    "10.1515": {"domain": "degruyter.com", "publisher": "gruyter"},
+    "10.1038": {"domain": "nature.com", "container_name": "nature"},
+    "10.1163": {"domain": "brill.com", "publisher": "brill"},
+    "10.3390": {"domain": "mdpi.com", "publisher": "mdpi"},
+    "10.1128": {"domain": "asm.org", "publisher": "microbiology"},
+    "10.1103": {"domain": "aps.org", "publisher": "physical"},
+    "10.3389": {"domain": "frontiersin.org", "publisher": "frontiers"},
+    "10.1136": {"domain": "bmj.org", "publisher": "bmj"},
+    "10.1088": {"domain": "iop.org", "publisher": "iop"},
+    "10.1086": {"domain": "iop.org", "publisher": "iop"},
+    "10.1142": {"domain": "worldscientific.com", "publisher": "world"},
+    "10.1126": {"domain": "sciencemag.org", "container_name": "science"},
+    "10.1162": {"domain": "mitpressjournals.org", "publisher": "mit"},
+    "10.1045": {"domain": "dlib.org", "container_name": "d-lib"},
+    "10.17723": {"domain": "archivists.org", "publisher": "archiv"},
+    "10.2139": {"domain": "ssrn.com", "container_name": "social science"},
+}
+
+
+def doi_link_domain(
+    doi_prefix: str, container_name: Optional[str], publisher: Optional[str]
+) -> Optional[str]:
+    """
+    Takes a DOI prefix and a publisher name, and tries to guess which domain
+    name the DOI will resolve to. This is used for display only.
+
+    helpful: https://gist.github.com/TomDemeranville/8699224
+
+    TODO: JSTOR, biorxiv, medrxiv, zenodo, figshare, dryad, etc
+    """
+
+    # manual cases first
+    if doi_prefix == "10.1101" and container_name:
+        if "biorxiv" in container_name.lower():
+            return "biorxiv.org"
+        elif "medrxiv" in container_name.lower():
+            return "medrxiv.org"
+        else:
+            return None
+    elif doi_prefix == "10.1101" and container_name:
+        if "lancet" in container_name.lower():
+            return "thelancet.com"
+
+    # then the map
+    meta = DOI_PREFIX_MAP.get(doi_prefix)
+    if not meta:
+        return None
+
+    if meta.get("publisher"):
+        if not publisher or meta["publisher"] not in publisher.lower():
+            return None
+    if meta.get("container_name"):
+        if not container_name or meta["container_name"] not in container_name.lower():
+            return None
+    return meta.get("domain")
diff --git a/fatcat_scholar/schema.py b/fatcat_scholar/schema.py
index d3a91a7..480b8fa 100644
--- a/fatcat_scholar/schema.py
+++ b/fatcat_scholar/schema.py
@@ -19,6 +19,7 @@ from pydantic import BaseModel
 
 from fatcat_openapi_client import ReleaseEntity, ReleaseContrib
 from fatcat_scholar.api_entities import entity_to_dict
+from fatcat_scholar.biblio_hacks import doi_link_domain
 
 
 class DocType(str, Enum):
@@ -104,6 +105,19 @@ class ScholarBiblio(BaseModel):
     contrib_names: List[str]
     affiliations: List[str]
 
+    def doi_link_domain(self, default: str = "doi.org") -> str:
+        if not self.doi_prefix:
+            return default
+        domain = doi_link_domain(
+            self.doi_prefix,
+            container_name=self.container_name,
+            publisher=self.publisher,
+        )
+        if domain:
+            return domain
+        else:
+            return default
+
     def citation_str(self, style: str) -> Optional[str]:
         """
         Tries to format this biblio metadata as a citation string. If it fails,
diff --git a/fatcat_scholar/templates/search_macros.html b/fatcat_scholar/templates/search_macros.html
index f580cee..25ba7f5 100644
--- a/fatcat_scholar/templates/search_macros.html
+++ b/fatcat_scholar/templates/search_macros.html
@@ -99,40 +99,16 @@
   </a>
 {% endmacro %}
 
-{% macro doi_access_button(biblio, is_oa=False) %}
-  {% if biblio.doi %}
-    {% set publisher = "" or (biblio.publisher and biblio.publisher.lower()) %}
-    {% set container_name = "" or (biblio.container_name and biblio.container_name.lower()) %}
-    <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/{{ biblio.doi }}">
+{% macro doi_access_button(paper, is_oa=False) %}
+  {% if paper.biblio and paper.biblio.doi %}
+    <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/{{ paper.biblio.doi }}">
       <button class="ui left aligned compact blue labeled icon button serp-button">
         {% if is_oa %}
           <i class="unlock alternate icon" style="background-color: #fb971f;"></i>
         {% else %}
           <i class="external alternate icon"></i>
         {% endif %}
-        {# TODO: detect prefix? JSTOR, biorxiv, medrxiv, zenodo, figshare, dryad, etc #}
-        {# helpful: https://gist.github.com/TomDemeranville/8699224 #}
-        {% if biblio.doi_prefix == "10.6084" %}
-          figshare.com
-        {% elif biblio.doi_prefix == "10.5281" %}
-          zenodo.org
-        {% elif biblio.doi_prefix == "10.1371" %}
-          plos.org
-        {% elif biblio.doi_prefix == "10.1101" and "biorxiv" in container_name %}
-          biorxiv.org
-        {% elif biblio.doi_prefix == "10.1101" and "medrxiv" in container_name %}
-          medrxiv.org
-        {% elif biblio.doi_prefix == "10.1016" and "elsevier" in publisher %}
-          elsevier.com
-        {% elif biblio.doi_prefix in ["10.1186", "10.1007"] and "springer" in publisher %}
-          springer.com
-        {% elif biblio.doi_prefix in ["10.1042", "10.1111", "10.1177"] and "sage" in publisher %}
-          sagepub.com
-        {% elif biblio.doi_prefix in ["10.1080"] and "taylor" in publisher %}
-          tandfonline.com
-        {% else %}
-          Publisher / doi.org
-        {% endif %}
+        {{ paper._obj.biblio.doi_link_domain("Publisher / doi.org") }}
       </button>
     </a>
   {% endif %}
@@ -453,7 +429,7 @@
   {% endif %}
 
   {# publisher / repository #}
-  {{ doi_access_button(paper.biblio, is_oa=("oa" in paper.tags)) }}
+  {{ doi_access_button(paper, is_oa=("oa" in paper.tags)) }}
 
   {# trusted platform fulltext links #}
   {{ platform_access_button(paper.biblio) }}