Ingest jep/taln/recital 2019 and 2020 (acl-org#957)

* Added jepn/taln/recital 2019 * Added jepn/taln/recital 2020 * added language field; lowercase volume names
namrathaurs · Aug 30, 2020 · 76854fe · 76854fe
1 parent 732ccee
commit 76854fe
Show file tree

Hide file tree

Showing 4 changed files with 2,588 additions and 6 deletions.
diff --git a/bin/create_hugo_pages.py b/bin/create_hugo_pages.py
@@ -113,10 +113,7 @@ def create_volumes(srcdir, clean=False):
                 {
                     "anthology_id": anthology_id,
                     "title": entry["title"],
-                    "aliases": [
-                        "/volumes/{}/".format(slugify(entry["title"])),
-                        paper_dir,
-                    ],
+                    "aliases": [paper_dir,],
                 },
                 default_flow_style=False,
                 stream=f,

diff --git a/bin/ingest.py b/bin/ingest.py
@@ -94,6 +94,7 @@ def bib2xml(bibfilename, anthology_id):
         'abstract',
         'url',
         'doi',
+        'language',
     ]
 
     collection_id, volume_name, paper_no = deconstruct_anthology_id(anthology_id)
@@ -175,7 +176,7 @@ def main(args):
         meta["collection_id"] = collection_id = (
             meta["year"] + "." + meta["abbrev"].lower()
         )
-        volume_name = meta["volume"]
+        volume_name = meta["volume"].lower()
         volume_full_id = f"{collection_id}-{volume_name}"
 
         if volume_full_id in volumes:
@@ -197,7 +198,7 @@ def main(args):
         root_path = os.path.join(meta["path"], "cdrom")
         collection_id = meta["collection_id"]
         venue_name = meta["abbrev"].lower()
-        volume_name = meta["volume"]
+        volume_name = meta["volume"].lower()
         year = meta["year"]
 
         pdfs_dest_dir = os.path.join(args.pdfs_dir, venue_name)