sws-spark-dissemination-helper 0.0.152__tar.gz → 0.0.154__tar.gz

{sws_spark_dissemination_helper-0.0.152 → sws_spark_dissemination_helper-0.0.154}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sws-spark-dissemination-helper
-Version: 0.0.152
+Version: 0.0.154
 Summary: A Python helper package providing streamlined Spark functions for efficient data dissemination processes
 Project-URL: Repository, https://github.com/un-fao/fao-sws-it-python-spark-dissemination-helper
 Author-email: Daniele Mansillo <danielemansillo@gmail.com>

{sws_spark_dissemination_helper-0.0.152 → sws_spark_dissemination_helper-0.0.154}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sws-spark-dissemination-helper"
-version = "0.0.152"
+version = "0.0.154"
 dependencies = [
     "annotated-types==0.7.0",
     "boto3>=1.36.18",

{sws_spark_dissemination_helper-0.0.152 → sws_spark_dissemination_helper-0.0.154}/src/sws_spark_dissemination_helper/SWSBronzeIcebergSparkHelper.py

@@ -1,7 +1,7 @@
 import logging
 import time
 from copy import copy
-from typing import Dict, List, Tuple
+from typing import Dict, List, Tuple, Union
 
 import pyspark.sql.functions as F
 from pyspark.sql import DataFrame, SparkSession
@@ -26,6 +26,9 @@ class SWSBronzeIcebergSparkHelper:
         domain_code: str,
         dataset_details: dict = None,
         dataset_tables: DatasetTables = None,
+        keep_history: bool = False,
+        write_csv: bool = True,
+        source_tag: Union[str, None] = None,
     ) -> None:
         self.spark: SparkSession = spark
         self.dataset_details: dict = dataset_details
@@ -36,6 +39,9 @@ class SWSBronzeIcebergSparkHelper:
         self.dataset_tables: DatasetTables = dataset_tables
         self.iceberg_tables: IcebergTables = iceberg_tables
         self.domain_code = domain_code
+        self.keep_history: bool = keep_history
+        self.write_csv: bool = write_csv
+        self.source_tag: Union[str, None] = source_tag
 
         if dataset_details is not None:
             (
@@ -83,6 +89,7 @@ class SWSBronzeIcebergSparkHelper:
                 self.df_obs_coord,
                 self.df_metadata,
                 self.df_meta_elem,
+                self.df_tag_observation,
             ) = self.raw_data
 
             (
@@ -92,10 +99,11 @@ class SWSBronzeIcebergSparkHelper:
                 self.df_meta_elem_type,
                 self.df_language,
                 self.df_unit_of_measure,
+                self.df_dataset,
                 self.dfs_dimension,
             ) = self.raw_reference_data
 
-            self.df_user = self.raw_operational_data
+            (self.df_user, self.df_tag) = self.raw_operational_data
 
     def _get_dim_time_flag_columns(self) -> Tuple[List[str], List[str], str, List[str]]:
         """Extract the dimension columns with time, without time, the time column and the flag columns names."""
@@ -284,6 +292,148 @@ class SWSBronzeIcebergSparkHelper:
 
         return df_obs_denorm
 
+    def _gen_denormalized_observation_sql(self) -> DataFrame:
+        # ----------------
+        # Prepare dataframes for the joins
+        # ----------------
+
+        select_statement = """
+            o.id,
+            o.value,
+            u.email,
+            o.created_on,
+            o.replaced_on,
+            o.version"""
+
+        from_statement = f"""
+        FROM {self.dataset_tables.OBSERVATION.iceberg_id} o
+            JOIN {self.dataset_tables.USER.iceberg_id} u ON u.id = o.created_by
+            LEFT JOIN {self.dataset_tables.OBSERVATION_COORDINATE.iceberg_id} AS oc ON oc.id = o.observation_coordinates"""
+
+        hint_statement = ""
+
+        id_to_flag_col_mapping = {v: k for k, v in self.flag_col_to_id_mapping.items()}
+        for flag_col in self.flag_columns:
+            select_statement += f",\no.{id_to_flag_col_mapping[flag_col]} AS {flag_col}"
+
+        id_to_dim_col_mapping = {v: k for k, v in self.dim_col_to_id_mapping.items()}
+        for i, (dim_col, cl) in enumerate(
+            zip(self.dim_columns_w_time, self.dataset_tables.CODELISTS)
+        ):
+            select_statement += f",\nd{i}.code AS {dim_col}"
+            from_statement += f"\nLEFT JOIN {cl.iceberg_id} d{i} ON d{i}.id = oc.{id_to_dim_col_mapping[dim_col]}"
+            hint_statement = (
+                hint_statement + f", BROADCAST({cl.iceberg_id})"
+                if hint_statement
+                else f"BROADCAST({cl.iceberg_id})"
+            )
+
+        hint_statement = "/*+ " + hint_statement + " */"
+
+        final_query = "SELECT " + hint_statement + select_statement + from_statement
+        if not self.keep_history:
+            final_query += "\nWHERE o.replaced_on IS NULL"
+
+        logging.info("Final query for merging observation and observation_coordinares")
+        logging.info(final_query)
+
+        df_obs_denorm = self.spark.sql(final_query)
+
+        df_element_uom = self._prepare_element_uom()
+
+        dfs_dimension_w_validity = self._convert_dim_start_end_date_to_data()
+
+        # Join all the dimension codelists
+        for dimension_column, df_dimension in zip(
+            self.dim_columns_w_time, dfs_dimension_w_validity
+        ):
+            df_obs_denorm = (
+                df_obs_denorm.alias("o")
+                .join(
+                    F.broadcast(df_dimension.withColumnRenamed("id", "join_id")).alias(
+                        "d"
+                    ),
+                    col(f"{dimension_column}") == col("d.join_id"),
+                )
+                .drop(f"{dimension_column}", "join_id")
+                .withColumnRenamed("code", dimension_column)
+            )
+        
+        df_obs_denorm = (
+            df_obs_denorm.alias("d")
+            .join(
+                F.broadcast(df_element_uom).alias("e"),
+                col(f"d.{self.element_column}") == col("e.element_code"),
+                "left",
+            )
+            .drop("element_code")
+        )
+
+        return df_obs_denorm
+
+    def _gen_denormalized_observation_sql_from_tag(self) -> DataFrame:
+        # ----------------
+        # Prepare dataframes for the joins
+        # ----------------
+
+        select_statement = """
+            o.id,
+            o.value,
+            u.email,
+            o.created_on,
+            o.replaced_on,
+            o.version"""
+
+        from_statement = f"""
+        FROM {self.dataset_tables.OBSERVATION.iceberg_id} o
+            INNER JOIN {self.dataset_tables.TAG_OBSERVATION.iceberg_id} to ON o.id = to.observation
+            INNER JOIN {self.dataset_tables.TAG.iceberg_id} t ON to.tag = t.id
+            INNER JOIN {self.dataset_tables.DATASET.iceberg_id} d ON t.dataset = d.id
+            LEFT JOIN {self.dataset_tables.USER.iceberg_id} u ON u.id = o.created_by
+            LEFT JOIN {self.dataset_tables.OBSERVATION_COORDINATE.iceberg_id} AS oc ON oc.id = o.observation_coordinates"""
+
+        hint_statement = ""
+
+        id_to_flag_col_mapping = {v: k for k, v in self.flag_col_to_id_mapping.items()}
+        for flag_col in self.flag_columns:
+            select_statement += f",\no.{id_to_flag_col_mapping[flag_col]} AS {flag_col}"
+
+        id_to_dim_col_mapping = {v: k for k, v in self.dim_col_to_id_mapping.items()}
+        for i, (dim_col, cl) in enumerate(
+            zip(self.dim_columns_w_time, self.dataset_tables.CODELISTS)
+        ):
+            select_statement += f",\nd{i}.code AS {dim_col}"
+            from_statement += f"\nLEFT JOIN {cl.iceberg_id} d{i} ON d{i}.id = oc.{id_to_dim_col_mapping[dim_col]}"
+            hint_statement = (
+                hint_statement + f", BROADCAST({cl.iceberg_id})"
+                if hint_statement
+                else f"BROADCAST({cl.iceberg_id})"
+            )
+
+        hint_statement = "/*+ " + hint_statement + " */"
+
+        # TODO Add tag name as a parameter
+        where_statement = (
+            f"\nWHERE t.name = '{self.source_tag}' AND d.xml_name = '{self.dataset_id}'"
+        )
+
+        final_query = (
+            "SELECT "
+            + hint_statement
+            + select_statement
+            + from_statement
+            + where_statement
+        )
+        if not self.keep_history:
+            final_query += "\n AND o.replaced_on IS NULL"
+
+        logging.info("Final query for merging observation and observation_coordinares")
+        logging.info(final_query)
+
+        df_obs_denorm = self.spark.sql(final_query)
+
+        return df_obs_denorm
+
     def _gen_denormalized_metadata(self) -> DataFrame:
         """Original query upon which the below computation is based
 
@@ -347,6 +497,32 @@ class SWSBronzeIcebergSparkHelper:
 
         return df_meta_denorm
 
+    def _gen_denormalized_metadata_sql(self) -> DataFrame:
+        # ----------------
+        # Generate denormalized observation table
+        # ----------------
+
+        logging.info("meta_denorm start")
+
+        df_meta_denorm = self.spark.sql(
+            f"""
+        select m.observation as observation_id,
+            mt.code as type,
+            met.code as element_type,
+            l.country_code as language,
+            me.value
+        from {self.dataset_tables.METADATA_ELEMENT.iceberg_id} me
+            left join {self.dataset_tables.METADATA.iceberg_id} m on m.id = me.metadata
+            left join {self.dataset_tables.METADATA_ELEMENT_TYPE.iceberg_id} met on met.id = me.metadata_element_type
+            left join {self.dataset_tables.METADATA_TYPE.iceberg_id} mt on mt.id = m.metadata_type
+            left join {self.dataset_tables.LANGUAGE.iceberg_id} l on l.id = m.language
+        """
+        )
+
+        logging.info("meta_denorm write")
+
+        return df_meta_denorm
+
     def _gen_grouped_metadata(self) -> DataFrame:
         return (
             self._gen_denormalized_metadata()
@@ -367,6 +543,26 @@ class SWSBronzeIcebergSparkHelper:
             .agg(F.collect_list("metadata").alias("metadata"))
         )
 
+    def _gen_grouped_metadata_sql(self) -> DataFrame:
+        return (
+            self._gen_denormalized_metadata_sql()
+            .select(
+                col("observation_id"),
+                F.create_map(
+                    lit("type"),
+                    col("type"),
+                    lit("element_type"),
+                    col("element_type"),
+                    lit("language"),
+                    col("language"),
+                    lit("value"),
+                    col("value"),
+                ).alias("metadata"),
+            )
+            .groupby("observation_id")
+            .agg(F.collect_list("metadata").alias("metadata"))
+        )
+
     def _gen_bronze_data(self) -> DataFrame:
         return (
             self._gen_denormalized_observation()
@@ -379,9 +575,37 @@ class SWSBronzeIcebergSparkHelper:
             .drop("m.observation_id")
         )
 
+    def _gen_bronze_data_sql(self) -> DataFrame:
+        return (
+            self._gen_denormalized_observation_sql()
+            .alias("o")
+            .join(
+                self._gen_grouped_metadata_sql().alias("m"),
+                col("o.id") == col("m.observation_id"),
+                "left",
+            )
+            .drop("m.observation_id")
+        )
+
+    def _gen_bronze_data_sql_from_tag(self) -> DataFrame:
+        return (
+            self._gen_denormalized_observation_sql_from_tag()
+            .alias("o")
+            .join(
+                self._gen_grouped_metadata_sql().alias("m"),
+                col("o.id") == col("m.observation_id"),
+                "left",
+            )
+            .drop("m.observation_id")
+        )
+
     # TODO decouple data generation and data writing
-    def write_bronze_data_to_iceberg_and_csv(self) -> DataFrame:
-        self.df_bronze = self._gen_bronze_data()
+    def write_bronze_data_to_iceberg_and_csv(self, sql=True) -> DataFrame:
+
+        if sql:
+            self.df_bronze = self._gen_bronze_data_sql()
+        else:
+            self.df_bronze = self._gen_bronze_data()
 
         self.df_bronze.writeTo(self.iceberg_tables.BRONZE.iceberg_id).createOrReplace()
 
@@ -449,12 +673,15 @@ class SWSBronzeIcebergSparkHelper:
         logging.info("Bronze Dissemination tags successfully written")
 
     def write_bronze_disseminated_tag_data_to_iceberg_and_csv(
-        self, dimensions: Dict[str, List[str]]
+        self, dimensions: Dict[str, List[str]] = {}, from_tag=False
     ) -> DataFrame:
 
-        self.disseminated_tag_df = self.df_bronze
+        if from_tag:
+            self.disseminated_tag_df = self._gen_bronze_data_sql_from_tag()
+        else:
+            self.disseminated_tag_df = self.df_bronze
 
-        if isinstance(dimensions, dict):
+        if not from_tag and len(dimensions) != 0:
             for dimension_name, codes in dimensions.items():
                 logging.info(f"dimension_name: {dimension_name}")
                 logging.info(f"codes: {codes}")