From abe3de09045c6dd96febe738bc965175036f460d Mon Sep 17 00:00:00 2001
From: Khoroshevskyi <sasha99250@gmail.com>
Date: Tue, 3 Dec 2024 15:57:03 -0500
Subject: [PATCH] Updated bedms to new version + stability improvements

---
 pephub/_version.py                |  2 +-
 pephub/const.py                   |  4 +++
 pephub/routers/api/v1/project.py  | 43 +++++++++++++++++++++----------
 requirements/requirements-all.txt |  2 +-
 4 files changed, 36 insertions(+), 15 deletions(-)

diff --git a/pephub/_version.py b/pephub/_version.py
index f075dd36..745162e7 100644
--- a/pephub/_version.py
+++ b/pephub/_version.py
@@ -1 +1 @@
-__version__ = "0.14.1"
+__version__ = "0.14.2"
diff --git a/pephub/const.py b/pephub/const.py
index 18c1ad76..89af9111 100644
--- a/pephub/const.py
+++ b/pephub/const.py
@@ -142,3 +142,7 @@
 ARCHIVE_URL_PATH = "https://cloud2.databio.org/pephub/"
 
 MAX_PROCESSED_PROJECT_SIZE = 5000
+
+MAX_STANDARDIZED_PROJECT_SIZE = 100
+
+BEDMS_REPO_URL = "databio/attribute-standardizer-model6"
diff --git a/pephub/routers/api/v1/project.py b/pephub/routers/api/v1/project.py
index 5e7a44d7..5d873229 100644
--- a/pephub/routers/api/v1/project.py
+++ b/pephub/routers/api/v1/project.py
@@ -55,7 +55,11 @@
     ConfigResponseModel,
     StandardizerResponse,
 )
-from ....const import MAX_PROCESSED_PROJECT_SIZE
+from ....const import (
+    MAX_PROCESSED_PROJECT_SIZE,
+    BEDMS_REPO_URL,
+    MAX_STANDARDIZED_PROJECT_SIZE,
+)
 from .helpers import verify_updated_project
 
 from bedms import AttrStandardizer
@@ -1182,33 +1186,46 @@ def delete_full_history(
     response_model=StandardizerResponse,
 )
 async def get_standardized_cols(
-    pep: peppy.Project = Depends(get_project),
+    pep: dict = Depends(get_project),
     schema: str = "",
 ):
     """
-    Standardize PEP metadata column headers using BEDmess.
+    Standardize PEP metadata column headers using BEDms.
 
-    :param namespace: pep: PEP string to be standardized
+    :param pep: PEP string to be standardized
     :param schema: Schema for AttrStandardizer
 
     :return dict: Standardized results
     """
 
-    if schema == "":
+    if schema == "" or schema not in ["ENCODE", "BEDBASE", "FAIRTRACKS"]:
         raise HTTPException(
-            code=500,
-            detail="Schema is required! Available schemas are ENCODE and Fairtracks",
+            status_code=404,
+            detail="Schema not available! Available schemas are ENCODE, BEDBASE and FAIRTRACKS.",
+        )
+
+    if len(pep["_sample_dict"]) > MAX_STANDARDIZED_PROJECT_SIZE:
+        # raise HTTPException(
+        #     status_code=400,
+        #     detail=f"Project is too large. Cannot standardize. "
+        #            f"Limit is {MAX_STANDARDIZED_PROJECT_SIZE} samples.",
+        # )
+        prj = peppy.Project.from_dict(
+            {
+                "_config": pep["_config"],
+                "_sample_dict": pep["_sample_dict"][:50],
+            }
         )
-        return {}
-
-    prj = peppy.Project.from_dict(pep)
-    model = AttrStandardizer(schema)
+    else:
+        prj = peppy.Project.from_dict(pep)
+    model = AttrStandardizer(repo_id=BEDMS_REPO_URL, model_name=schema.lower())
 
     try:
         results = model.standardize(pep=prj)
-    except Exception:
+    except Exception as e:
+        _LOGGER.error(f"Error standardizing PEP. {e}")
         raise HTTPException(
-            code=400,
+            status_code=400,
             detail=f"Error standardizing PEP.",
         )
 
diff --git a/requirements/requirements-all.txt b/requirements/requirements-all.txt
index d7903965..25631478 100644
--- a/requirements/requirements-all.txt
+++ b/requirements/requirements-all.txt
@@ -16,4 +16,4 @@ fastembed
 numpy<2.0.0
 slowapi
 cachetools>=4.2.4
-bedms>=0.1.0
\ No newline at end of file
+bedms>=0.2.0
\ No newline at end of file