Mark escaped strings as so

diff --git a/os_postgresql_import.plpgsql b/os_postgresql_import.plpgsql
index 8b7dd19e5b6995ea54e692cedfd62ab65041b4d1..cb15a30f006495635381613971f24a3dfb57ddc0 100644 (file)
--- a/os_postgresql_import.plpgsql
+++ b/os_postgresql_import.plpgsql
@@ -101,8 +101,8 @@ BEGIN
        INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('S92000003', 'CNY', 'Scotland');
        INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008322', 'DIW', 'Bryher Ward');
        INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008323', 'DIW', 'St. Agnes Ward');
-       INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008324', 'DIW', 'St. Martin\'s Ward');
-       INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008325', 'DIW', 'St. Mary\'s Ward');
+       INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008324', 'DIW', E'St. Martin\'s Ward');
+       INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008325', 'DIW', E'St. Mary\'s Ward');
        INSERT INTO public.pc_opendata_areas (code, type, description) VALUES ('E05008326', 'DIW', 'Tresco Ward');
        
        -- Finally our postcode data
@@ -212,8 +212,8 @@ BEGIN
        INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('S92000003', 'CNY', 'Scotland');
        INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008322', 'DIW', 'Bryher Ward');
        INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008323', 'DIW', 'St. Agnes Ward');
-       INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008324', 'DIW', 'St. Martin\'s Ward');
-       INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008325', 'DIW', 'St. Mary\'s Ward');
+       INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008324', 'DIW', E'St. Martin\'s Ward');
+       INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008325', 'DIW', E'St. Mary\'s Ward');
        INSERT INTO tmp_pc_opendata_areas (code, type, description) VALUES ('E05008326', 'DIW', 'Tresco Ward');
        
        -- Finally our postcode data
@@ -243,4 +243,4 @@ BEGIN
 END;
 
 $BODY$
-LANGUAGE 'plpgsql' VOLATILE;
\ No newline at end of file
+LANGUAGE 'plpgsql' VOLATILE;

diff --git a/paf_postgresql_import.plpgsql b/paf_postgresql_import.plpgsql
index 7955083a5223325e6ac7c1baf3481f9b00f21451..13f47cbbfebf752f7ed6964b9dd82446fc683476 100644 (file)
--- a/paf_postgresql_import.plpgsql
+++ b/paf_postgresql_import.plpgsql
@@ -334,7 +334,7 @@ BEGIN
                COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_reference_east IS 'BT postcodes reflect the Irish Grid public (different origin and scale); subtract 17000 to approximate onto OS National Grid. For latlong do not approximate but do proper conversion';
                COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_reference_north IS 'BT postcodes reflect the Irish Grid public (different origin and scale); add 13000 to approximate onto OS National Grid. For latlong do not approximate but do proper conversion';
                COMMENT ON COLUMN public.pc_paf_postzon_100m.user_type IS '0 Small User 1 Large User';
-               COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_status IS '0  Status not supplied by OS \n
+               COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_status IS E'0  Status not supplied by OS \n
                        1  Within the building of the matched address closest to the Postcode mean. \n
                        2  Co-ordinates allocated by GROS during Postcode boundary creation to the building nearest the centre of the populated part of the Postcode (Scotland only) \n
                        3  Approximate to within 50m of true position \n
@@ -1032,7 +1032,7 @@ BEGIN
        -- If I did have a postzon changes file, it'd most likely be best to import the data directly and use postgis to do our conversion
        -- as the data is selected out of the data_stage table (rather than using the convert_paf.pl script). We could get those fields with
        -- something like:
-       -- CASE WHEN substring(northing,1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(northing, 'POUTZY', '221100') ELSE northing END || '0' AS f_notrhing,
+       -- CASE WHEN substring(northing,1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(northing, 'POUTZY', '221100') ELSE northing END || '0' AS f_notrhing,
        -- ST_x(ST_transform(ST_GeomFromText('POINT('||easting||' '||f_northing||')',CASE WHEN postcode LIKE 'BT%' THEN 29903 ELSE 27700 END),4326)) AS longitude,
        -- ST_y(ST_transform(ST_GeomFromText('POINT('||easting||' '||f_northing||')',CASE WHEN postcode LIKE 'BT%' THEN 29903 ELSE 27700 END),4326)) AS latitude;
        

diff --git a/paf_postgresql_import_postgis.plpgsql b/paf_postgresql_import_postgis.plpgsql
index 36c764c4e9e4a1de2db9cdff050303b1741a4176..ffc4db88bcca31d3ad7ebca3cb5a234102c61724 100644 (file)
--- a/paf_postgresql_import_postgis.plpgsql
+++ b/paf_postgresql_import_postgis.plpgsql
@@ -309,7 +309,7 @@ BEGIN
                nullif(trim(substring(data,5,3)),'') AS inward_code,
                to_date(substring(data,8,6), 'YYYYMM') AS introduction_date,
                (nullif(trim(substring(data,14,5)),'') || '0')::integer AS grid_reference_east,         
-               (CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')::integer AS grid_reference_north,
+               (CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')::integer AS grid_reference_north,
                nullif(trim(substring(data,24,9)),'') AS country_code,
                nullif(trim(substring(data,33,9)),'') AS area_code_county,
                nullif(trim(substring(data,42,9)),'') AS area_code_district,
@@ -318,8 +318,8 @@ BEGIN
                nullif(trim(substring(data,69,9)),'') AS nhs_code,
                nullif(trim(substring(data,78,1)),'')::smallint AS user_type,
                nullif(trim(substring(data,79,1)),'')::smallint AS grid_status,
-               ST_y(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS latitude,
-               ST_x(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS longitude
+               ST_y(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS latitude,
+               ST_x(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS longitude
        FROM data_stage;
 
        GET DIAGNOSTICS v_processed = ROW_COUNT;
@@ -334,7 +334,7 @@ BEGIN
                COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_reference_east IS 'BT postcodes reflect the Irish Grid public (different origin and scale); subtract 17000 to approximate onto OS National Grid. For latlong do not approximate but do proper conversion';
                COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_reference_north IS 'BT postcodes reflect the Irish Grid public (different origin and scale); add 13000 to approximate onto OS National Grid. For latlong do not approximate but do proper conversion';
                COMMENT ON COLUMN public.pc_paf_postzon_100m.user_type IS '0 Small User 1 Large User';
-               COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_status IS '0  Status not supplied by OS \n
+               COMMENT ON COLUMN public.pc_paf_postzon_100m.grid_status IS E'0  Status not supplied by OS \n
                        1  Within the building of the matched address closest to the Postcode mean. \n
                        2  Co-ordinates allocated by GROS during Postcode boundary creation to the building nearest the centre of the populated part of the Postcode (Scotland only) \n
                        3  Approximate to within 50m of true position \n
@@ -1076,7 +1076,7 @@ BEGIN
                nullif(trim(substring(data,5,3)),'') AS inward_code,
                to_date(substring(data,8,6), 'YYYYMM') AS introduction_date,
                (nullif(trim(substring(data,14,5)),'') || '0')::integer AS grid_reference_east,         
-               (CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')::integer AS grid_reference_north,
+               (CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')::integer AS grid_reference_north,
                nullif(trim(substring(data,24,9)),'') AS country_code,
                nullif(trim(substring(data,33,9)),'') AS area_code_county,
                nullif(trim(substring(data,42,9)),'') AS area_code_district,
@@ -1085,8 +1085,8 @@ BEGIN
                nullif(trim(substring(data,69,9)),'') AS nhs_code,
                nullif(trim(substring(data,78,1)),'')::smallint AS user_type,
                nullif(trim(substring(data,79,1)),'')::smallint AS grid_status,
-               ST_y(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS latitude,
-               ST_x(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ '[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS longitude
+               ST_y(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS latitude,
+               ST_x(ST_transform(ST_GeomFromText('POINT('||(nullif(trim(substring(data,14,5)),'') || '0')||' '||(CASE WHEN substring(nullif(trim(substring(data,19,5)),''),1,1) ~ E'[\x4f-\x5a]' THEN '1' || translate(nullif(trim(substring(data,19,5)),''), 'POUTZY', '221100') ELSE nullif(trim(substring(data,19,5)),'') END || '0')||')',CASE WHEN nullif(trim(substring(data,1,4)),'') LIKE 'BT%' THEN 29903 ELSE 27700 END),4326))::numeric(8,6) AS longitude
        FROM data_stage;
        
        RAISE NOTICE '%: Done staging postzon with lat long data, updating', clock_timestamp();