Merge pull request #89 from databio/dev

Release 0.11.4

Author: khoroshevskyi

bbconf/_version.py

@@ -1 +1 @@
-__version__ = "0.11.3"
+__version__ = "0.11.4"

bbconf/bbagent.py

@@ -229,10 +229,14 @@ def get_list_genomes(self) -> List[str]:
 
         :return: list of genomes
         """
-        statement = select(distinct(Bed.genome_alias))
+        statement = (
+            select(Bed.genome_alias)
+            .group_by(Bed.genome_alias)
+            .order_by(func.count(Bed.genome_alias).desc())
+        )
         with Session(self.config.db_engine.engine) as session:
             genomes = session.execute(statement).all()
-        return [result[0] for result in genomes]
+        return [result[0] for result in genomes if result[0]]
 
     @cached_property
     def list_of_licenses(self) -> List[str]:

bbconf/modules/bedfiles.py

@@ -1183,7 +1183,7 @@ def text_to_bed_search(
             if result_meta:
                 results_list.append(QdrantSearchResult(**result, metadata=result_meta))
         return BedListSearchResult(
-            count=self.bb_agent.get_stats().bedfiles_number,
+            count=self.bb_agent.get_detailed_stats().file_genome.get("hg38", 0),
             limit=limit,
             offset=offset,
             results=results_list,
@@ -1218,34 +1218,49 @@ def bed_to_bed_search(
         )
 
     def sql_search(
-        self, query: str, limit: int = 10, offset: int = 0
+        self,
+        query: str,
+        genome: str = None,
+        limit: int = 10,
+        offset: int = 0,
     ) -> BedListSearchResult:
         """
         Search for bed files by using sql exact search.
         This search will search files by id, name, and description
 
         :param query: text query
+        :param genome: genome alias to filter results
         :param limit: number of results to return
         :param offset: offset to start from
 
         :return: list of bed file metadata
         """
         _LOGGER.debug(f"Looking for: {query}")
 
+        statement = select(Bed).join(BedMetadata)
+
         sql_search_str = f"%{query}%"
+        or_statement = or_(
+            Bed.id.ilike(sql_search_str),
+            Bed.name.ilike(sql_search_str),
+            Bed.description.ilike(sql_search_str),
+            BedMetadata.cell_line.ilike(sql_search_str),
+            BedMetadata.tissue.ilike(sql_search_str),
+            BedMetadata.cell_type.ilike(sql_search_str),
+        )
+
+        if genome_alias := genome:
+            _LOGGER.debug(f"Filtering by genome: {genome_alias}")
+
+            condition_statement = and_(Bed.genome_alias == genome_alias, or_statement)
+
+        else:
+            condition_statement = or_statement
+
+        statement = statement.where(condition_statement)
+
         with Session(self._sa_engine) as session:
-            statement = (
-                select(Bed)
-                .where(
-                    or_(
-                        Bed.id.ilike(sql_search_str),
-                        Bed.name.ilike(sql_search_str),
-                        Bed.description.ilike(sql_search_str),
-                    )
-                )
-                .limit(limit)
-                .offset(offset)
-            )
+            statement = statement.limit(limit).offset(offset)
             bed_objects = session.scalars(statement)
             results = [
                 BedMetadataBasic(
@@ -1266,32 +1281,27 @@ def sql_search(
             ]
 
         return BedListSearchResult(
-            count=self._sql_search_count(query),
+            count=self._sql_search_count(condition_statement),
             limit=limit,
             offset=offset,
             results=result_list,
         )
 
-    def _sql_search_count(self, query: str) -> int:
+    def _sql_search_count(self, condition_statement) -> int:
         """
         Get number of total found files in the database.
 
-        :param query: text query
+        :param condition_statement: sql alchemy condition statement to filter results
 
         :return: number of found files
         """
-        sql_search_str = f"%{query}%"
+
         with Session(self._sa_engine) as session:
             statement = (
                 select(func.count())
                 .select_from(Bed)
-                .where(
-                    or_(
-                        Bed.id.ilike(sql_search_str),
-                        Bed.name.ilike(sql_search_str),
-                        Bed.description.ilike(sql_search_str),
-                    )
-                )
+                .join(BedMetadata)
+                .where(condition_statement)
             )
             count = session.execute(statement).one()
         return count[0]

bbconf/modules/bedsets.py

@@ -482,6 +482,7 @@ def get_ids_list(
         statement = select(BedSets.id)
         count_statement = select(func.count(BedSets.id))
         if query:
+            query = query.strip()
             sql_search_str = f"%{query}%"
             statement = statement.where(
                 or_(

docs/changelog.md

@@ -2,7 +2,12 @@
 
 This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html) and [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) format.
 
-### [0.11.3] - 2025-06-27
+### [0.11.4] - 2025-06-01
+### Fixed:
+- SQL search
+
+
+### [0.11.3] - 2025-05-27
 ### Fixed:
 - Usage tracker
 - Order of comprehensive stats

manual_testing.py

@@ -197,7 +197,7 @@ def sql_search():
     from bbconf import BedBaseAgent
 
     agent = BedBaseAgent(config="/home/bnt4me/virginia/repos/bedhost/config.yaml")
-    results = agent.bed.sql_search("K562", limit=100)
+    results = agent.bed.sql_search("", limit=100, genome="mm39")
     results
 
 
@@ -226,6 +226,15 @@ def get_unprocessed_files():
     print(results)
 
 
+def get_genomes():
+
+    from bbconf import BedBaseAgent
+
+    agent = BedBaseAgent(config="/home/bnt4me/virginia/repos/bedhost/config.yaml")
+    results = agent.get_list_genomes()
+    print(results)
+
+
 if __name__ == "__main__":
     # zarr_s3()
     # add_s3()
@@ -234,7 +243,8 @@ def get_unprocessed_files():
     # get_pep()
     # get_id_plots_missing()
     # neighbour_beds()
-    # sql_search()
+    sql_search()
     # config_t()
     # compreh_stats()
-    get_unprocessed_files()
+    # get_unprocessed_files()
+    # get_genomes()