From 0591b388a55771ed0581102989e23ba22a720a27 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Fri, 22 Jan 2021 19:23:51 -0800
Subject: add container_sherpa_color field, and populate it

---
 fatcat_scholar/schema.py         | 36 ++++++++++++++++++------------------
 fatcat_scholar/transform.py      |  1 +
 schema/scholar_fulltext.v01.json |  1 +
 3 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/fatcat_scholar/schema.py b/fatcat_scholar/schema.py
index 480b8fa..434f735 100644
--- a/fatcat_scholar/schema.py
+++ b/fatcat_scholar/schema.py
@@ -99,6 +99,7 @@ class ScholarBiblio(BaseModel):
     container_ident: Optional[str]
     container_issnl: Optional[str]
     container_wikidata_qid: Optional[str]
+    container_sherpa_color: Optional[str]
     issns: List[str]
     container_type: Optional[str]
     contrib_count: Optional[int]
@@ -469,38 +470,36 @@ def es_abstracts_from_release(release: ReleaseEntity) -> List[ScholarAbstract]:
 
 def es_biblio_from_release(release: ReleaseEntity) -> ScholarBiblio:
 
+    container_name = release.extra and release.extra.get("container_name")
+
     if release.container:
-        publisher = release.container.publisher or release.publisher
-        publisher_type = release.container.extra and release.container.extra.get(
-            "publisher_type", None
-        )
-        if not publisher_type or not isinstance(publisher_type, str):
-            publisher_type = None
-        container_name = release.container.name
-        container_original_name = (
-            release.container.extra and release.container.extra.get("original_name")
-        )
-        if not container_original_name or not isinstance(container_original_name, str):
-            container_original_name = None
+        publisher = release.publisher or release.container.publisher
+        container_name = container_name or release.container.name
         container_ident = release.container.ident
         container_type = release.container.container_type
         container_issnl = release.container.issnl
         issns = []
         if container_issnl:
             issns.append(container_issnl)
-        if release.container.extra and release.container.extra.get("issne"):
-            issns.append(release.container.extra["issne"])
-        if release.container.extra and release.container.extra.get("issnp"):
-            issns.append(release.container.extra["issnp"])
+        publisher_type = None
+        container_original_name = None
+        if release.container.extra:
+            publisher_type = release.container.extra.get("publisher_type")
+            container_original_name = release.container.extra.get("original_name")
+            container_sherpa_color = release.container.extra.get("sherpa_romeo", {}).get("color")
+            if release.container.extra.get("issne"):
+                issns.append(release.container.extra["issne"])
+            if release.container.extra.get("issnp"):
+                issns.append(release.container.extra["issnp"])
         issns = list(set(issns))
     else:
-        publisher = release.extra and release.extra.get("publisher")
         publisher_type = None
-        container_name = release.extra and release.extra.get("container_name")
+        publisher = release.publisher
         container_original_name = None
         container_ident = None
         container_type = None
         container_issnl = None
+        container_sherpa_color = None
         issns = []
 
     first_page: Optional[str] = None
@@ -547,6 +546,7 @@ def es_biblio_from_release(release: ReleaseEntity) -> ScholarBiblio:
         container_ident=container_ident,
         container_type=container_type,
         container_issnl=container_issnl,
+        container_sherpa_color=container_sherpa_color,
         issns=issns,
         # TODO; these filters sort of meh. refactor to be above?
         contrib_names=list(
diff --git a/fatcat_scholar/transform.py b/fatcat_scholar/transform.py
index 6c2df90..ba79e50 100644
--- a/fatcat_scholar/transform.py
+++ b/fatcat_scholar/transform.py
@@ -168,6 +168,7 @@ def es_biblio_from_sim(sim: Dict[str, Any]) -> ScholarBiblio:
         container_ident=None,  # TODO
         container_type=None,  # TODO
         container_issnl=None,  # TODO
+        # container_sherpa_color
         issns=issns,
         # no contrib/affiliation info
         contrib_names=[],
diff --git a/schema/scholar_fulltext.v01.json b/schema/scholar_fulltext.v01.json
index 91afdc1..9e99026 100644
--- a/schema/scholar_fulltext.v01.json
+++ b/schema/scholar_fulltext.v01.json
@@ -118,6 +118,7 @@
             "container_ident":      { "type": "keyword", "normalizer": "default" },
             "container_issnl":      { "type": "keyword", "normalizer": "default" },
             "container_wikidata_qid": { "type": "keyword", "normalizer": "default" },
+            "container_sherpa_color": { "type": "keyword", "normalizer": "default" },
             "issns":                { "type": "keyword", "normalizer": "default" },
             "container_type":       { "type": "keyword", "normalizer": "default" },
             "contrib_count":        { "type": "short" },
-- 
cgit v1.2.3