add extraction to outs

biobricks-ai · Oct 26, 2024 · 8f7c75b · 8f7c75b
1 parent c790bb8
commit 8f7c75b
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 2 deletions.
diff --git a/dvc.yaml b/dvc.yaml
@@ -23,4 +23,11 @@ stages:
     outs:
       - brick/riskder.pdf:
           persist: true
-      - brick/riskder.parquet
+      - brick/riskder.parquet
+
+  extract_data:
+    cmd: python stages/03_data_extractor.py
+    deps:
+      - brick/riskder.pdf
+    outs:
+      - brick/extraction.parquet
diff --git a/notebook.py b/notebook.py
@@ -0,0 +1,3 @@
+import pandas as pd
+
+results = pd.read_parquet('brick/riskder.parquet')
diff --git a/stages/03_data_extractor.py b/stages/03_data_extractor.py
@@ -128,5 +128,5 @@ def extract_testing_results(pdf_path):
         continue
 
 
-aggdf.to_csv('extraction.csv')
+aggdf.to_csv('brick/extraction.csv')
 # endregion
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,3 @@
		import pandas as pd

		results = pd.read_parquet('brick/riskder.parquet')