added license table to the database

bbconf/db_utils.py

@@ -18,6 +18,9 @@
 
 POSTGRES_DIALECT = "postgresql+psycopg"
 
+# set if the db schema was created
+first_db_init: bool = False
+
 
 class SchemaError(Exception):
     def __init__(self):
@@ -49,8 +52,10 @@ def receive_after_create(target, connection, tables, **kw):
     """
     listen for the 'after_create' event
     """
+    global first_db_init
     if tables:
         _LOGGER.info("A table was created")
+        first_db_init = True
     else:
         _LOGGER.info("A table was not created")
 
@@ -103,6 +108,10 @@ class Bed(Base):
     tokenized: Mapped["TokenizedBed"] = relationship(
         back_populates="bed", cascade="all, delete-orphan"
     )
+    license: Mapped["str"] = mapped_column(
+        ForeignKey("licenses.id", ondelete="CASCADE"), nullable=True, index=True
+    )
+    license_map: Mapped["License"] = relationship("License", back_populates="bed")
 
 
 class BedStats(Base):
@@ -260,6 +269,19 @@ class TokenizedBed(Base):
     )
 
 
+class License(Base):
+    __tablename__ = "licenses"
+
+    id: Mapped[str] = mapped_column(primary_key=True, index=True)
+    shorthand: Mapped[str] = mapped_column(nullable=True, comment="License shorthand")
+    label: Mapped[str] = mapped_column(nullable=False, comment="License label")
+    description: Mapped[str] = mapped_column(
+        nullable=False, comment="License description"
+    )
+
+    bed: Mapped[List["Bed"]] = relationship("Bed", back_populates="license_map")
+
+
 @listens_for(Universes, "after_insert")
 @listens_for(Universes, "after_update")
 def add_bed_universe(mapper, connection, target):
@@ -331,6 +353,11 @@ def create_schema(self, engine=None):
         if not engine:
             engine = self._engine
         Base.metadata.create_all(engine)
+
+        global first_db_init
+        if first_db_init:
+            self._upload_licenses()
+
         return None
 
     def delete_schema(self, engine=None) -> None:
@@ -398,3 +425,17 @@ def create_schema_graph(self, output_file: str = "schema.svg"):
         graph = create_schema_graph(engine=self.engine, metadata=Base.metadata)
         graph.write(output_file, format="svg", prog="dot")
         return None
+
+    def _upload_licenses(self):
+        """
+        Upload licenses to the database.
+        """
+
+        import pandas as pd
+
+        csv_file_path = "https://raw.githubusercontent.com/EBISPOT/DUO/master/duo.csv"
+        df = pd.read_csv(csv_file_path)
+
+        with Session(self.engine) as session:
+            df.to_sql("licenses", self.engine, if_exists="append", index=False)
+            session.commit()

requirements/requirements-all.txt

@@ -13,3 +13,4 @@ pyyaml >= 6.0.1 # for s3fs because of the errors
 s3fs >= 2024.3.1
 # quick fix for gesnim
 scipy <= 1.11.0
+pandas