merge with wk

coa.md

@@ -1,3 +1,5 @@
+Balance Sheet
+-----------------------
 * Cash
     - On-hand
     - Hunt Checking
@@ -20,5 +22,10 @@
     - Discover
     - Kohls
     - Lowes
+    - BestBuy
+    - Target
+    - TheHomeDepot
 * Mortgage
     - Principle
+
+

do_map.pgsql

@@ -0,0 +1,85 @@
+
+WITH
+
+--------------------apply regex operations to transactions-----------------------------------------------------------------------------------
+
+rx AS (
+SELECT 
+    m.srce,
+    m.target,
+    t.id,
+    jsonb_build_object(
+        e.v ->> 'key',
+        (t.rec #> ((e.v ->> 'key')::text[]))
+    ) AS rkey,
+    jsonb_build_object(
+        e.v->>'field',
+        CASE WHEN array_upper(mt.mt,1)=1 
+            THEN to_json(mt.mt[1]) 
+            ELSE array_to_json(mt.mt) 
+        END
+    ) retval,
+    m.seq,
+    e.v->>'retain' retain
+FROM 
+    tps.map_rm m
+    LEFT JOIN LATERAL jsonb_array_elements(m.regex->'where') w(v) ON TRUE
+    JOIN tps.trans t ON 
+        t.srce = m.srce AND
+        t.rec @> w.v
+    LEFT JOIN LATERAL jsonb_array_elements(m.regex->'defn') WITH ORDINALITY e(v, rn) ON true
+    LEFT JOIN LATERAL regexp_matches(t.rec #>> ((e.v ->> 'key')::text[]), e.v ->> 'regex'::text) WITH ORDINALITY mt(mt, rn) ON true
+WHERE
+    t.srce = 'PNCC'     
+ORDER BY 
+    m.srce, 
+    m.seq,
+    m.target, 
+    t.id, 
+    e.rn
+),
+
+----------aggregate regex back to the target level (may be several targets per row)---------------------------------------------------------------
+
+
+agg_rx AS (
+    SELECT 
+        rx.srce,
+        rx.target,
+        rx.id, 
+        tps.jsonb_concat_obj(rx.rkey) rkey,
+        tps.jsonb_concat_obj(rx.retval) AS retval,
+        tps.jsonb_concat_obj(CASE rx.retain WHEN 'y' THEN rx.retval ELSE '{}'::jsonb END) retain,
+        rx.seq        
+    FROM 
+        --unwrap json instruction and apply regex using a count per original line for re-aggregation
+        --need to look at integrating regex option like 'g' that would then need aggegated back as an array, or adding the ordinality number to the title
+        rx
+    GROUP BY 
+        rx.srce, 
+        rx.target, 
+        rx.id,
+        rx.seq
+)
+
+
+-------------aggregate all targets back to row level (id)------------------------------------------------------------------------------------------------
+
+    SELECT 
+        u.srce,
+        u.id,
+        string_agg(u.target,',') target,
+        jsonb_pretty(tps.jsonb_concat_obj(coalesce(v.map,'{}'::jsonb) ORDER BY seq )) map,
+        jsonb_pretty(tps.jsonb_concat_obj(u.retval||coalesce(v.map,'{}'::jsonb) ORDER BY seq)) comb,
+        jsonb_pretty(tps.jsonb_concat_obj(u.retain||coalesce(v.map,'{}'::jsonb) ORDER BY seq)) retain
+    FROM 	
+        --re-aggregate return values and explude any records where one or more regex failed with a null result
+        agg_rx u
+        LEFT OUTER JOIN tps.map_rv v ON
+            v.target = u.target AND
+            v.srce = u.srce AND
+            v.retval <@ u.retval
+    GROUP BY
+        u.srce,
+        u.id
+    LIMIT 1000

map_rm.pgsql

@@ -1,40 +1,166 @@
-SELECT
+INSERT INTO
-jsonb_pretty(
+tps.map_rm
-$$
+SELECT *
-{
+FROM
+(VALUES 
+    ('PNCC', 'ACH Debits', 
+    $j$
+    {
         "defn": [
             {
                 "key": "{Description}",
                 "field": "ini",
                 "regex": "([\\w].*?)(?=$| -|\\s[0-9].*?|\\s[\\w/]+?:)"
+                ,"retain":"y"
             },
             {
                 "key": "{Description}",
                 "field": "compn",
                 "regex": "Comp Name:(.+?)(?=$| Comp|\\w+?:)"
+                ,"retain":"y"
             },
             {
                 "key": "{Description}",
                 "field": "adp_comp",
                 "regex": "Cust ID:.*?(B3X|UDV|U7E|U7C|U7H|U7J).*?(?=$|\\w+?:)"
+                ,"retain":"y"
             },
             {
                 "key": "{Description}",
                 "field": "desc",
                 "regex": "Desc:(.+?) Comp"
+                ,"retain":"y"
             },
             {
                 "key": "{Description}",
                 "field": "discr",
                 "regex": "Discr:(.+?)(?=$| SEC:|\\w+?:)"
+                ,"retain":"y"
             }
         ],
-    "type": "extract",
         "where": [
             {
                 "Transaction": "ACH Debits"
             }
         ]
-}
+    }
-$$::jsonb
+    $j$::jsonb
-)
+    , 2)
+    ,('PNCC', 'Trans Type', 
+    $j$
+    {
+        "defn": [
+            {
+                "key": "{AccountName}",
+                "field": "acctn",
+                "regex": "(.*)"
+                ,"retain":"n"
+            },
+            {
+                "key": "{Transaction}",
+                "field": "trans",
+                "regex": "(.*)"
+                ,"retain":"n"
+            },
+            {
+                "key": "{Description}",
+                "field": "ini",
+                "regex": "([\\w].*?)(?=$| -|\\s[0-9].*?|\\s[\\w/]+?:)"
+                ,"retain":"y"
+            }
+        ],
+        "where": [
+            {
+            }
+        ]
+    }
+    $j$::jsonb
+    , 1)
+    ,('PNCC', 'Wires Out', 
+    $j$
+    {
+        "defn": [
+            {
+                "key": "{Description}",
+                "field": "ini",
+                "regex": "([\\w].*?)(?=$| -|\\s[0-9].*?|\\s[\\w/]+?:)"
+                ,"retain":"y"
+            },
+            {
+                "key": "{Description}",
+                "field": "bene",
+                "regex": "BENEFICIARY:(.+?) AC/"
+                ,"retain":"y"
+            },
+            {
+                "key": "{Description}",
+                "field": "accts",
+                "regex": "AC/(\\w*) .*AC/(\\w*) "
+                ,"retain":"y"
+            }
+        ],
+        "where": [
+            {
+                "Transaction": "Intl Money Transfer Debits"
+            },
+            {
+                "Transaction": "Money Transfer DB - Wire"
+            }
+        ]
+    }
+    $j$::jsonb
+    , 2)
+    ,('PNCC', 'Currency', 
+    $j$
+    {
+        "defn": [
+            {
+                "key": "{Description}",
+                "field": "ini",
+                "regex": "([\\w].*?)(?=$| -|\\s[0-9].*?|\\s[\\w/]+?:)"
+                ,"retain":"y"
+            },
+            {
+                "key": "{Description}",
+                "field": "curr1",
+                "regex": ".*(DEBIT|CREDIT).*(USD|CAD).*(?=DEBIT|CREDIT).*(?=USD|CAD).*"
+                ,"retain":"y"
+            },
+            {
+                "key": "{Description}",
+                "field": "curr2",
+                "regex": ".*(?=DEBIT|CREDIT).*(?=USD|CAD).*(DEBIT|CREDIT).*(USD|CAD).*"
+                ,"retain":"y"
+            }
+        ],
+        "where": [
+            {
+                "Transaction": "Miscellaneous Credits"
+            },
+            {
+                "Transaction": "Miscellaneous Debits"
+            }
+        ]
+    }
+    $j$::jsonb
+    , 2)
+    ,('PNCC', 'Check Number', 
+    $j$
+    {
+        "defn": [
+            {
+                "key": "{Description}",
+                "field": "checkn",
+                "regex": "[^0-9]*([0-9]*)\\s|$"
+                ,"retain":"y"
+            }
+        ],
+        "where": [
+            {
+                "Transaction": "Checks Paid"
+            }
+        ]
+    }
+    $j$::jsonb
+    , 2)
+) x

ubm_data.sql

@@ -2,8 +2,8 @@
 -- PostgreSQL database dump
 --
 
--- Dumped from database version 10beta4
+-- Dumped from database version 10rc1
--- Dumped by pg_dump version 10beta4
+-- Dumped by pg_dump version 10rc1
 
 SET statement_timeout = 0;
 SET lock_timeout = 0;
@@ -20,7 +20,6 @@ SET search_path = evt, pg_catalog;
 -- Data for Name: log; Type: TABLE DATA; Schema: evt; Owner: -
 --
 
-INSERT INTO log (id, rec) VALUES (1, '{"date": "2017-08-20", "item": [{"item": "Green Chili", "amount": 1.49, "account": "food"}, {"item": "Black Beans", "amount": 1.6, "account": "food"}, {"item": "Distilled Water", "amount": 7.12, "account": "food"}, {"item": "Fruit Preservative", "amount": 3.99, "account": "food"}, {"item": "Watch Battery", "amount": 3.79, "account": "stuff"}, {"item": "Sales Tax", "amount": "0.26", "account": "taxes"}, {"item": "Green Chili", "amount": -1.49, "account": "dcard"}, {"item": "Black Beans", "amount": -1.6, "account": "dcard"}, {"item": "Distilled Water", "amount": -7.12, "account": "dcard"}, {"item": "Fruit Preservative", "amount": -3.99, "account": "dcard"}, {"item": "Watch Battery", "amount": -3.79, "account": "dcard"}, {"item": "Sales Tax", "amount": -0.26, "account": "dcard"}], "vendor": "Drug Mart", "instrument": "Discover Card"}');
 
 
 SET search_path = tps, pg_catalog;
@@ -35,7 +34,7 @@ INSERT INTO srce (srce, defn) VALUES ('HUNT', '{"name": "HUNT", "type": "csv", "
 
 
 --
--- Data for Name: trans; Type: TABLE DATA; Schema: tps; Owner: -
+-- Data for Name: map_rm; Type: TABLE DATA; Schema: tps; Owner: -
 --
 
 INSERT INTO trans (id, srce, rec, map) VALUES (2094, 'HUNT', '{"Date": "2017-08-07", "Memo": "SUBSTITUTE CHECK", "Amount": "-610.01", "Payee Name": "", "Category Name": "", "Reference Number": 1826}', NULL);
@@ -2683,7 +2682,7 @@ SET search_path = evt, pg_catalog;
 -- Name: log_id_seq; Type: SEQUENCE SET; Schema: evt; Owner: -
 --
 
-SELECT pg_catalog.setval('log_id_seq', 1, true);
+SELECT pg_catalog.setval('log_id_seq', 1, false);
 
 
 SET search_path = tps, pg_catalog;

ubm_schema.sql

@@ -2,8 +2,8 @@
 -- PostgreSQL database dump
 --
 
--- Dumped from database version 10beta4
+-- Dumped from database version 10rc1
--- Dumped by pg_dump version 10beta4
+-- Dumped by pg_dump version 10rc1
 
 SET statement_timeout = 0;
 SET lock_timeout = 0;
@@ -56,6 +56,20 @@ CREATE EXTENSION IF NOT EXISTS plpgsql WITH SCHEMA pg_catalog;
 COMMENT ON EXTENSION plpgsql IS 'PL/pgSQL procedural language';
 
 
+--
+-- Name: plprofiler; Type: EXTENSION; Schema: -; Owner: -
+--
+
+CREATE EXTENSION IF NOT EXISTS plprofiler WITH SCHEMA public;
+
+
+--
+-- Name: EXTENSION plprofiler; Type: COMMENT; Schema: -; Owner: -
+--
+
+COMMENT ON EXTENSION plprofiler IS 'server-side support for profiling PL/pgSQL functions';
+
+
 SET search_path = tps, pg_catalog;
 
 --
@@ -95,6 +109,32 @@ CREATE TYPE srce_defn_schema AS (
 );
 
 
+--
+-- Name: jsonb_concat(jsonb, jsonb); Type: FUNCTION; Schema: tps; Owner: -
+--
+
+CREATE FUNCTION jsonb_concat(state jsonb, concat jsonb) RETURNS jsonb
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+	--RAISE notice 'state is %', state;
+	--RAISE notice 'concat is %', concat;
+	RETURN state || concat;
+END;
+$$;
+
+
+--
+-- Name: jsonb_concat_obj(jsonb); Type: AGGREGATE; Schema: tps; Owner: -
+--
+
+CREATE AGGREGATE jsonb_concat_obj(jsonb) (
+    SFUNC = jsonb_concat,
+    STYPE = jsonb,
+    INITCOND = '{}'
+);
+
+
 SET search_path = evt, pg_catalog;
 
 SET default_tablespace = '';
@@ -127,6 +167,30 @@ ALTER TABLE log ALTER COLUMN id ADD GENERATED BY DEFAULT AS IDENTITY (
 
 SET search_path = tps, pg_catalog;
 
+--
+-- Name: map_rm; Type: TABLE; Schema: tps; Owner: -
+--
+
+CREATE TABLE map_rm (
+    srce text NOT NULL,
+    target text NOT NULL,
+    regex jsonb,
+    seq integer NOT NULL
+);
+
+
+--
+-- Name: map_rv; Type: TABLE; Schema: tps; Owner: -
+--
+
+CREATE TABLE map_rv (
+    srce text NOT NULL,
+    target text NOT NULL,
+    retval jsonb NOT NULL,
+    map jsonb
+);
+
+
 --
 -- Name: srce; Type: TABLE; Schema: tps; Owner: -
 --
@@ -175,6 +239,22 @@ ALTER TABLE ONLY log
 
 SET search_path = tps, pg_catalog;
 
+--
+-- Name: map_rm map_rm_pk; Type: CONSTRAINT; Schema: tps; Owner: -
+--
+
+ALTER TABLE ONLY map_rm
+    ADD CONSTRAINT map_rm_pk PRIMARY KEY (srce, target);
+
+
+--
+-- Name: map_rv map_rv_pk; Type: CONSTRAINT; Schema: tps; Owner: -
+--
+
+ALTER TABLE ONLY map_rv
+    ADD CONSTRAINT map_rv_pk PRIMARY KEY (srce, target, retval);
+
+
 --
 -- Name: srce srce_pkey; Type: CONSTRAINT; Schema: tps; Owner: -
 --
@@ -191,6 +271,29 @@ ALTER TABLE ONLY trans
     ADD CONSTRAINT trans_pkey PRIMARY KEY (id);
 
 
+--
+-- Name: trans_rec; Type: INDEX; Schema: tps; Owner: -
+--
+
+CREATE INDEX trans_rec ON trans USING gin (rec);
+
+
+--
+-- Name: map_rm map_rm_fk_srce; Type: FK CONSTRAINT; Schema: tps; Owner: -
+--
+
+ALTER TABLE ONLY map_rm
+    ADD CONSTRAINT map_rm_fk_srce FOREIGN KEY (srce) REFERENCES srce(srce);
+
+
+--
+-- Name: map_rv map_rv_fk_rm; Type: FK CONSTRAINT; Schema: tps; Owner: -
+--
+
+ALTER TABLE ONLY map_rv
+    ADD CONSTRAINT map_rv_fk_rm FOREIGN KEY (srce, target) REFERENCES map_rm(srce, target);
+
+
 --
 -- Name: trans trans_srce_fkey; Type: FK CONSTRAINT; Schema: tps; Owner: -
 --