Bugfix: Fix export of database to csv so the name field is populated

strangetom · Jan 31, 2025 · 2bc7bc1 · 2bc7bc1
1 parent ff4dcbd
commit 2bc7bc1
Show file tree

Hide file tree

Showing 7 changed files with 81,319 additions and 81,315 deletions.
diff --git a/train/data/allrecipes/allrecipes-ingredients-snapshot-2017.csv b/train/data/allrecipes/allrecipes-ingredients-snapshot-2017.csv
diff --git a/train/data/bbc/bbc-ingredients-snapshot-2017.csv b/train/data/bbc/bbc-ingredients-snapshot-2017.csv
diff --git a/train/data/cookstr/cookstr-ingredients-snapshot-2017.csv b/train/data/cookstr/cookstr-ingredients-snapshot-2017.csv
diff --git a/train/data/db_to_csv.py b/train/data/db_to_csv.py
@@ -81,9 +81,14 @@ def create_csv_value(db_row: dict[str, str], label: str) -> str:
     str
         Constructed value to insert into csv
     """
-    tokens = [
-        tok for tok, lab in zip(db_row["tokens"], db_row["labels"]) if lab == label
-    ]
+    if label == "NAME":
+        tokens = [
+            tok for tok, lab in zip(db_row["tokens"], db_row["labels"]) if label in lab
+        ]
+    else:
+        tokens = [
+            tok for tok, lab in zip(db_row["tokens"], db_row["labels"]) if lab == label
+        ]
     return " ".join(tokens)
 
 

diff --git a/train/data/nytimes/nyt-ingredients-snapshot-2015.csv b/train/data/nytimes/nyt-ingredients-snapshot-2015.csv
diff --git a/train/data/tastecooking/tastecooking-ingredients-snapshot-2024.csv b/train/data/tastecooking/tastecooking-ingredients-snapshot-2024.csv
diff --git a/train/data/training.sqlite3 b/train/data/training.sqlite3