Native language parsing for author names (#125)

* Native language parsing for author names * lint fix --------- Co-authored-by: Mugdha Polimera <[email protected]>
adsabs · Aug 21, 2024 · 8e1796d · 8e1796d
1 parent 5f376ac
commit 8e1796d
Show file tree

Hide file tree

Showing 6 changed files with 9,320 additions and 1 deletion.
diff --git a/adsingestp/parsers/base.py b/adsingestp/parsers/base.py
@@ -299,7 +299,7 @@ def format(self, input_dict, format):
                     "prefix": i.get("prefix", ""),
                     "suffix": i.get("suffix", ""),
                     "pubraw": i.get("nameraw", ""),
-                    # "native_lang": "XXX",
+                    "native_lang": i.get("native_lang", ""),
                     "collab": i.get("collab", ""),
                 },
                 "affiliation": [

diff --git a/adsingestp/parsers/jats.py b/adsingestp/parsers/jats.py
@@ -339,6 +339,7 @@ def parse(self, article_metadata):
                     surname = contrib.find("string-name").find("surname").get_text()
                 else:
                     surname = ""
+
                 if contrib.find("name") and contrib.find("name").find("given-names"):
                     given = contrib.find("name").find("given-names").get_text()
                 elif contrib.find("string-name") and contrib.find("string-name").find(
@@ -348,6 +349,12 @@ def parse(self, article_metadata):
                 else:
                     given = ""
 
+                # get native language author name
+                if contrib.find("name-alternatives"):
+                    native_lang = contrib.find("name-alternatives").get_text().strip()
+                else:
+                    native_lang = ""
+
                 # NOTE: institution-id is actually useful, but at
                 # at the moment, strip it
                 # contrib = self._decompose(soup=contrib, tag="institution-id")
@@ -425,6 +432,7 @@ def parse(self, article_metadata):
                 auth["corresp"] = l_correspondent
                 auth["surname"] = surname
                 auth["given"] = given
+                auth["native_lang"] = native_lang
                 auth["aff"] = aff_text
                 auth["affid"] = aff_extids
                 auth["xaff"] = xref_aff