Reduce DB size

db/export-from-original-database.sql

@@ -6,19 +6,6 @@
 
 SELECT
   OBJECTID AS id,
-  SOURCE_SYSTEM_TYPE AS source_system_type,
-  SOURCE_SYSTEM AS source_system,
-  SOURCE_REPORTING_UNIT AS source_reporting_unit,
-  SOURCE_REPORTING_UNIT_NAME AS source_reporting_unit_name,
-  NWCG_REPORTING_AGENCY AS nwcg_reporting_agency,
-  NWCG_REPORTING_UNIT_ID AS nwcg_reporting_unit_id,
-  NWCG_REPORTING_UNIT_NAME AS nwcg_reporting_unit_name,
-  LOCAL_FIRE_REPORT_ID AS local_fire_report_id,
-  LOCAL_INCIDENT_ID AS local_incident_id,
-  CAST(OWNER_CODE AS INTEGER) AS owner_code,
-  OWNER_DESCR AS owner_descr,
-  FIRE_CODE AS fire_code,
-  FIRE_NAME AS fire_name,
   FIRE_YEAR AS fire_year,
   CAST(CASE WHEN DISCOVERY_TIME IS NULL THEN STRFTIME('%s', DATETIME(DISCOVERY_DATE)) ELSE STRFTIME('%s', DATETIME(DISCOVERY_DATE, CAST(CAST(DISCOVERY_TIME AS integer)/100 AS STRING) || ' hours', CAST(CAST(DISCOVERY_TIME AS integer)%100 AS STRING) || ' minutes')) END AS INTEGER) AS discovery_datetime,
   CAST(CASE WHEN CONT_TIME IS NULL THEN STRFTIME('%s', DATETIME(CONT_DATE)) ELSE STRFTIME('%s', DATETIME(CONT_DATE, CAST(CAST(CONT_TIME AS integer)/100 AS STRING) || ' hours', CAST(CAST(CONT_TIME AS integer) %100 AS STRING) || ' minutes')) END AS INTEGER) AS contained_datetime,

db/fires-schema.sql

@@ -1,18 +1,5 @@
 CREATE TABLE fires(
   id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
-  source_system_type TEXT(255),
-  source_system TEXT(30),
-  source_reporting_unit TEXT(30),
-  source_reporting_unit_name TEXT(255),
-  nwcg_reporting_agency TEXT(255),
-  nwcg_reporting_unit_id TEXT(255),
-  nwcg_reporting_unit_name TEXT(255),
-  local_fire_report_id TEXT(255),
-  local_incident_id TEXT(255),
-  owner_code INTEGER,
-  owner_descr TEXT(100),
-  fire_code TEXT(10),
-  fire_name TEXT(255),
   fire_year INTEGER,
   discovery_datetime INTEGER,
   contained_datetime INTEGER,

db/import-to-new-database.sql

@@ -6,36 +6,23 @@
 --check for "<null>" to replace with NULL
 
 SELECT
-  SUM(CASE WHEN id = '<null>' THEN 1 ELSE 0 END) AS id_null_count,
-  SUM(CASE WHEN source_system_type = '<null>' THEN 1 ELSE 0 END) AS source_system_type_null_count,
-  SUM(CASE WHEN source_system = '<null>' THEN 1 ELSE 0 END) AS source_system_null_count,
-  SUM(CASE WHEN source_reporting_unit = '<null>' THEN 1 ELSE 0 END) AS source_reporting_unit_null_count,
-  SUM(CASE WHEN source_reporting_unit_name = '<null>' THEN 1 ELSE 0 END) AS source_reporting_unit_name_null_count,
-  SUM(CASE WHEN nwcg_reporting_agency = '<null>' THEN 1 ELSE 0 END) AS nwcg_reporting_agency_null_count,
-  SUM(CASE WHEN nwcg_reporting_unit_id = '<null>' THEN 1 ELSE 0 END) AS nwcg_reporting_unit_id_null_count,
-  SUM(CASE WHEN nwcg_reporting_unit_name = '<null>' THEN 1 ELSE 0 END) AS nwcg_reporting_unit_name_null_count,
-  SUM(CASE WHEN local_fire_report_id = '<null>' THEN 1 ELSE 0 END) AS local_fire_report_id_null_count,
-  SUM(CASE WHEN local_incident_id = '<null>' THEN 1 ELSE 0 END) AS local_incident_id_null_count,
-  SUM(CASE WHEN owner_code = '<null>' THEN 1 ELSE 0 END) AS owner_code_null_count,
-  SUM(CASE WHEN owner_descr = '<null>' THEN 1 ELSE 0 END) AS owner_descr_null_count,
-  SUM(CASE WHEN fire_code = '<null>' THEN 1 ELSE 0 END) AS fire_code_null_count,
-  SUM(CASE WHEN fire_name = '<null>' THEN 1 ELSE 0 END) AS fire_name_null_count,
-  SUM(CASE WHEN fire_year = '<null>' THEN 1 ELSE 0 END) AS fire_year_null_count,
-  SUM(CASE WHEN discovery_datetime = '<null>' THEN 1 ELSE 0 END) AS discovery_datetime_null_count,
-  SUM(CASE WHEN contained_datetime = '<null>' THEN 1 ELSE 0 END) AS contained_datetime_null_count,
-  SUM(CASE WHEN stat_cause_code = '<null>' THEN 1 ELSE 0 END) AS stat_cause_code_null_count,
-  SUM(CASE WHEN stat_cause_descr = '<null>' THEN 1 ELSE 0 END) AS stat_cause_descr_null_count,
-  SUM(CASE WHEN fire_size = '<null>' THEN 1 ELSE 0 END) AS fire_size_null_count,
-  SUM(CASE WHEN fire_size_class = '<null>' THEN 1 ELSE 0 END) AS fire_size_class_null_count,
-  SUM(CASE WHEN longitude = '<null>' THEN 1 ELSE 0 END) AS longitude_null_count,
-  SUM(CASE WHEN latitude = '<null>' THEN 1 ELSE 0 END) AS latitude_null_count,
-  SUM(CASE WHEN state_name = '<null>' THEN 1 ELSE 0 END) AS state_name_null_count,
-  SUM(CASE WHEN state_alpha_code = '<null>' THEN 1 ELSE 0 END) AS state_alpha_code_null_count,
-  SUM(CASE WHEN state_fips_code = '<null>' THEN 1 ELSE 0 END) AS state_fips_code_null_count,
-  SUM(CASE WHEN county_name = '<null>' THEN 1 ELSE 0 END) AS county_name_null_count,
-  SUM(CASE WHEN county_fips_name = '<null>' THEN 1 ELSE 0 END) AS county_fips_name_null_count,
-  SUM(CASE WHEN county_fips_code = '<null>' THEN 1 ELSE 0 END) AS county_fips_code_null_count,
-  SUM(CASE WHEN combined_fips_code = '<null>' THEN 1 ELSE 0 END) AS combined_fips_code_null_count
+  SUM(CASE WHEN id is NULL THEN 1 ELSE 0 END) AS id_null_count,
+  SUM(CASE WHEN fire_year is NULL THEN 1 ELSE 0 END) AS fire_year_null_count,
+  SUM(CASE WHEN discovery_datetime is NULL THEN 1 ELSE 0 END) AS discovery_datetime_null_count,
+  SUM(CASE WHEN contained_datetime is NULL THEN 1 ELSE 0 END) AS contained_datetime_null_count,
+  SUM(CASE WHEN stat_cause_code is NULL THEN 1 ELSE 0 END) AS stat_cause_code_null_count,
+  SUM(CASE WHEN stat_cause_descr is NULL THEN 1 ELSE 0 END) AS stat_cause_descr_null_count,
+  SUM(CASE WHEN fire_size is NULL THEN 1 ELSE 0 END) AS fire_size_null_count,
+  SUM(CASE WHEN fire_size_class is NULL THEN 1 ELSE 0 END) AS fire_size_class_null_count,
+  SUM(CASE WHEN longitude is NULL THEN 1 ELSE 0 END) AS longitude_null_count,
+  SUM(CASE WHEN latitude is NULL THEN 1 ELSE 0 END) AS latitude_null_count,
+  SUM(CASE WHEN state_name is NULL THEN 1 ELSE 0 END) AS state_name_null_count,
+  SUM(CASE WHEN state_alpha_code is NULL THEN 1 ELSE 0 END) AS state_alpha_code_null_count,
+  SUM(CASE WHEN state_fips_code is NULL THEN 1 ELSE 0 END) AS state_fips_code_null_count,
+  SUM(CASE WHEN county_name is NULL THEN 1 ELSE 0 END) AS county_name_null_count,
+  SUM(CASE WHEN county_fips_name is NULL THEN 1 ELSE 0 END) AS county_fips_name_null_count,
+  SUM(CASE WHEN county_fips_code is NULL THEN 1 ELSE 0 END) AS county_fips_code_null_count,
+  SUM(CASE WHEN combined_fips_code is NULL THEN 1 ELSE 0 END) AS combined_fips_code_null_count
 FROM fires;
 
 -- then, for each column with value "<null>", run (replacing **colname** with the actual column name):