drop old files and start working on json storage scripts

interface/import/srce_import_do.sql

@@ -0,0 +1,184 @@
+DO 
+$f$
+DECLARE
+    _t text;
+    _c text;
+    _log_info jsonb;
+    _log_id text;
+    _cnt numeric;
+    _message jsonb;
+    _recs jsonb;
+    _srce text;
+    _defn jsonb;
+    _MESSAGE_TEXT text;
+    _PG_EXCEPTION_DETAIL text;
+    _PG_EXCEPTION_HINT text;
+
+BEGIN
+
+    _srce := 'DMAPI';
+    _recs:= $${"id":1,"doc":{"rows":[{"elements":[{"status":"OK","distance":{"text":"225 mi","value":361940},"duration":{"text":"3 hours 50 mins","value":13812}}]}],"status":"OK","origin_addresses":["Washington, DC, USA"],"destination_addresses":["New York, NY, USA"]}}$$::jsonb;
+
+----------------------------------------------------test if source exists----------------------------------------------------------------------------------
+
+    SELECT
+        defn
+    INTO
+        _defn
+    FROM
+        tps.srce    
+    WHERE
+        srce = _srce;
+
+    IF _defn IS NULL THEN
+        _message:= 
+        format(
+            $$
+                {
+                    "status":"fail",
+                    "message":"source %L does not exists"
+                }
+            $$,
+            _srce
+        )::jsonb;
+        RAISE NOTICE '%s', _message;
+    END IF;
+
+    -------------unwrap the json record and apply the path(s) of the constraint to build a constraint key per record-----------------------------------------------------------------------------------
+
+    WITH
+    pending_list AS (
+        SELECT
+            _srce srce
+            ,j.rec
+            ,j.id
+            --aggregate back to the record since multiple paths may be listed in the constraint
+            --it is unclear why the "->>0" is required to correctly extract the text array from the jsonb
+            ,tps.jsonb_concat_obj(
+                jsonb_build_object(
+                    --the new json key is the path itself
+                    cons.path->>0
+                    ,j.rec#>((cons.path->>0)::text[])
+                ) 
+            ) json_key
+        FROM
+            jsonb_array_elements(_recs) WITH ORDINALITY j(rec,id)
+            JOIN LATERAL jsonb_array_elements(_defn->'constraint') WITH ORDINALITY cons(path, seq)  ON TRUE
+        GROUP BY
+            j.rec
+            ,j.id
+    )
+
+    -----------create a unique list of keys from staged rows------------------------------------------------------------------------------------------
+
+    , pending_keys AS (
+        SELECT DISTINCT
+            json_key
+        FROM 
+            pending_list
+    )
+
+    -----------list of keys already loaded to tps-----------------------------------------------------------------------------------------------------
+
+    , matched_keys AS (
+        SELECT DISTINCT
+            k.json_key
+        FROM
+            pending_keys k
+            INNER JOIN tps.trans t ON
+                t.ic = k.json_key
+    )
+
+    -----------return unique keys that are not already in tps.trans-----------------------------------------------------------------------------------
+
+    , unmatched_keys AS (
+    SELECT
+        json_key
+    FROM
+        pending_keys
+
+    EXCEPT
+
+    SELECT
+        json_key
+    FROM
+        matched_keys
+    )
+
+    --------build log record-------------------+------------------------------------------------------------------------------------------------
+
+    , logged AS (
+    INSERT INTO
+        tps.trans_log (info)
+    SELECT
+        JSONB_BUILD_OBJECT('time_stamp',CURRENT_TIMESTAMP)
+        ||JSONB_BUILD_OBJECT('srce',_srce)
+        --||JSONB_BUILD_OBJECT('path',_path)
+        ||JSONB_BUILD_OBJECT('not_inserted',
+            (
+                SELECT 
+                    jsonb_agg(json_key)
+                FROM
+                    matched_keys
+            )
+        )
+        ||JSONB_BUILD_OBJECT('inserted',
+            (
+                SELECT 
+                    jsonb_agg(json_key)
+                FROM
+                    unmatched_keys
+            )
+        )
+    RETURNING *
+    )
+
+    -----------insert pending rows that have key with no trans match-----------------------------------------------------------------------------------
+    --need to look into mapping the transactions prior to loading
+
+    , inserted AS (
+        INSERT INTO
+            tps.trans (srce, rec, ic, logid)
+        SELECT
+            pl.srce
+            ,pl.rec
+            ,pl.json_key
+            ,logged.id
+        FROM 
+            pending_list pl
+            INNER JOIN unmatched_keys u ON
+                u.json_key = pl.json_key
+            CROSS JOIN logged
+        ORDER BY
+            pl.id ASC
+        ----this conflict is only if an exact duplicate rec json happens, which will be rejected
+        ----therefore, records may not be inserted due to ay matches with certain json fields, or if the entire json is a duplicate, reason is not specified
+        RETURNING *
+    )
+
+    SELECT
+        id
+        ,info
+    INTO
+        _log_id
+        ,_log_info
+    FROM
+        logged;
+
+    --RAISE NOTICE 'import logged under id# %, info: %', _log_id, _log_info;
+
+    _message:= 
+    (
+        $$
+            {
+            "status":"complete"
+            }
+        $$::jsonb
+    )||jsonb_build_object('details',_log_info);
+
+    RAISE NOTICE '%s', _message;
+
+END;
+$f$
+LANGUAGE plpgsql
+

reports/col_balance.sql

@@ -1,22 +0,0 @@
-
-SELECT 
-    id
-    ,rec->>'id'
-    ,r.*
-    ,CASE "Schedule#"
-        WHEN '02IN Raw Material' THEN 13097563.42
-        WHEN '03IN Finished Goods' THEN 35790696.52
-        ELSE 0
-    END + SUM("Sales"+"Credits & Adjustments"-"Gross Collections") OVER (PARTITION BY "Schedule#" ORDER BY "Schedule#" ASC, "PostDate" ASC, rec->>'id' ASC) running_bal
-    ,(LEAST("CollateralBalance" - "Ineligible Amount","MaxEligible")*("AdvanceRate"/100))::NUMERIC(20,2) qualified_collateral
-    ,(("CollateralBalance" - "Ineligible Amount")*("AdvanceRate"/100))::NUMERIC(20,2) qualified_collateral_nl
-FROM 
-    tps.trans
-    LEFT JOIN LATERAL jsonb_populate_record(null::tps.pncl, rec) r ON TRUE
-WHERE 
-    srce = 'PNCL'
-    --AND rec @> '{"Schedule#":"03IN Finished Goods"}' 
-ORDER BY 
-    "Schedule#" asc
-    ,r."PostDate" asc
-    ,rec->>'id' asc

reports/dcard_bal.sql

@@ -1,17 +0,0 @@
-\timing
-SELECT 
-    r."Trans. Date",
-    r."Post Date",
-    r."Description",
-    r."Amount",
-    r."Category",
-    rec->'id' id,
-    SUM(r."Amount") OVER (PARTITION BY srce ORDER BY r."Post Date" asc , rec->>'id' asc, r."Description") + 1061.1 + 22.40 balance
-FROM 
-    tps.trans
-    LEFT JOIN LATERAL jsonb_populate_record(null::tps.dcard, rec) r ON TRUE
-WHERE 
-    srce = 'DCARD'
-ORDER BY 
-    r."Post Date" asc
-    ,rEC->>'id' asc

reports/loan_bal.sql

@@ -1,14 +0,0 @@
-\timing
-SELECT 
-    r.*,
-    SUM(r."Advances"+r."Adjustments"-r."Payments") OVER (PARTITION BY "Loan#" ORDER BY r."Post Date" asc ,rec->>'id' asc, r."Reference #" asc)
-FROM 
-    tps.trans
-    LEFT JOIN LATERAL jsonb_populate_record(null::tps.pnco, rec) r ON TRUE
-WHERE 
-    rec @> '{"Loan#":"606780281"}' 
-ORDER BY 
-    r."Loan#"
-    ,r."Post Date" ASC
-    ,rec->>'id' ASC
-    ,r."Reference #" ASC

reports/srce_unq.sql

@@ -1,19 +0,0 @@
-WITH
-ext AS (
-SELECT 
-    srce
-    ,defn->'unique_constraint'->>'fields'
-    ,ARRAY(SELECT ae.e::text[] FROM jsonb_array_elements_text(defn->'unique_constraint'->'fields') ae(e)) txa
-FROM
-    tps.srce
-)
-
-
-SELECT
-    t.srce
-    ,jsonb_pretty(t.rec)
-    ,jsonb_pretty(public.jsonb_extract(rec,txa))
-FROM
-    tps.trans t
-    INNER JOIN ext ON
-        t.srce = ext.srce

reports/transaction_range.sql

@@ -1,18 +0,0 @@
-\timing
-SELECT
-    t.srce
-    ,(ae.e::text[])[1] unq_constr
-    ,MIN(rec #>> ae.e::text[]) min_text
-    ,COUNT(*) cnt
-    ,MAX(rec #>> ae.e::text[]) max_text
-FROM
-    tps.trans t
-    INNER JOIN tps.srce s ON
-        s.srce = t.srce
-    LEFT JOIN LATERAL JSONB_ARRAY_ELEMENTS_TEXT(defn->'unique_constraint'->'fields') WITH ORDINALITY ae(e, rn) ON TRUE
-GROUP BY
-    t.srce
-    ,(ae.e::text[])[1]
-ORDER BY
-    t.srce
-    ,(ae.e::text[])[1]

sample_google_api/data.json

@@ -0,0 +1,29 @@
+{
+    "id": 1,
+    "doc": {
+        "rows": [
+            {
+                "elements": [
+                    {
+                        "status": "OK",
+                        "distance": {
+                            "text": "225 mi",
+                            "value": 361940
+                        },
+                        "duration": {
+                            "text": "3 hours 50 mins",
+                            "value": 13812
+                        }
+                    }
+                ]
+            }
+        ],
+        "status": "OK",
+        "origin_addresses": [
+            "Washington, DC, USA"
+        ],
+        "destination_addresses": [
+            "New York, NY, USA"
+        ]
+    }
+}

sample_google_api/def.json

@@ -0,0 +1,31 @@
+{
+    "name": "DMAPI",
+    "type": "csv",
+    "schemas": {
+        "default": [
+            {
+                "path": "{doc,origin_addresses,0}",
+                "type": "text",
+                "column_name": "origin_address"
+            },
+            {
+                "path": "{doc,destination_addresses,0}",
+                "type": "text",
+                "column_name": "destination_address"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,distance,value}",
+                "type": "numeric",
+                "column_name": "distince"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,duration,value}",
+                "type": "numeric",
+                "column_name": "duration"
+            }
+        ],
+        "constraint": [
+            "{doc}"
+        ]
+    }
+}

sample_google_api/test/srce.sql

@@ -0,0 +1,38 @@
+SELECT 
+    jsonb_pretty(r.x) 
+FROM
+    tps.srce_set(
+    $$
+{
+    "name": "DMAPI",
+    "type": "csv",
+    "schemas": {
+        "default": [
+            {
+                "path": "{doc,origin_addresses,0}",
+                "type": "text",
+                "column_name": "origin_address"
+            },
+            {
+                "path": "{doc,destination_addresses,0}",
+                "type": "text",
+                "column_name": "destination_address"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,distance,value}",
+                "type": "numeric",
+                "column_name": "distince"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,duration,value}",
+                "type": "numeric",
+                "column_name": "duration"
+            }
+        ],
+        "constraint": [
+            "{doc}"
+        ]
+    }
+}
+    $$
+) r(x);

sample_google_api/test_scipt.sql

@@ -5,24 +5,38 @@ SELECT
     jsonb_pretty(r.x) 
 FROM
     tps.srce_set(
-    'DMAPI',
     $$
-    {
+{
     "name": "DMAPI",
     "type": "csv",
-        "schema": [
+    "schemas": {
+        "default": [
             {
-                "key": "doc",
+                "path": "{doc,origin_addresses,0}",
-                "type": "jsonb"
+                "type": "text",
+                "column_name": "origin_address"
+            },
+            {
+                "path": "{doc,destination_addresses,0}",
+                "type": "text",
+                "column_name": "destination_address"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,distance,value}",
+                "type": "numeric",
+                "column_name": "distince"
+            },
+            {
+                "path": "{doc,rows,0,elements,0,duration,value}",
+                "type": "numeric",
+                "column_name": "duration"
             }
         ],
-        "unique_constraint": {
+        "constraint": [
-            "type": "key",
-            "fields": [
             "{doc}"
         ]
     }
-    }
+}
     $$
 ) r(x);
 --------------------------build a csv file---------------------

templates/insert_constraint.json

@@ -1,10 +1,6 @@
 {
-    "unique_constraint": {
+    "constraint": [
-        "{doc,origin_addresses}": [
+        "{doc,origin_addresses}",
-            "Washington, DC, USA"
+        "{doc,destination_addresses}"
-        ],
-        "{doc,destination_addresses}": [
-            "New York, NY, USA"
     ]
-    }
 }

templates/map_rm_template.json

@@ -1,6 +1,27 @@
-{
+[
-	"strip commas":
     {
+        "srce": "dcard",
+        "sequence": 1,
+        "defn": [
+            {
+                "key": "{Description}",
+                "map": "y",
+                "flag": "",
+                "field": "f20",
+                "regex": ".{1,20}",
+                "retain": "y"
+            }
+        ],
+        "name": "First 20",
+        "where": [
+            {}
+        ],
+        "function": "extract",
+        "description": "pull first 20 characters from description for mapping"
+    },
+    {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Strip Amount Commas",
         "description": "the Amount field come from PNC with commas embeded so it cannot be cast to numeric",
         "defn": [
@@ -19,8 +40,9 @@
             {}
         ]
     },
-	"Parse ACH Credits":
     {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Parse ACH Credits",
         "description": "parse select components of the description for ACH Credits Receieved",
         "defn": [
@@ -112,8 +134,9 @@
             }
         ]
     },
-	"Parse ACH Debits":
     {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Parse ACH Debits",
         "description": "parse select components of the description for ACH Credits Receieved",
         "defn": [
@@ -205,8 +228,9 @@
             }
         ]
     },
-	"Parse Wires":
     {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Parse Wires",
         "description": "pull out whatever follows OBI in the description until atleast 3 capital letters followed by a colon are encountered",
         "defn": [
@@ -337,8 +361,9 @@
             }
         ]
     },
-	"Trans Type":
     {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Trans Type",
         "description": "extract intial description in conjunction with account name and transaction type for mapping",
         "defn": [
@@ -369,8 +394,9 @@
         ],
         "function": "extract"
     },
-	"Currency":
     {
+        "srce": "pncc",
+        "sequence": 1,
         "name": "Currency",
         "description": "pull out currency indicators from description of misc items and map",
         "defn": [
@@ -406,26 +432,9 @@
         ],
         "function": "extract"
     },
-	"check number":
-	{
-		"defn": [
-			{
-				"key": "{Description}",
-				"field": "checkn",
-				"regex": "[^0-9]*([0-9]*)\\s|$",
-				"retain": "y",
-				"map": "n"
-			}
-		],
-		"where": [
-			{
-				"Transaction": "Checks Paid"
-			}
-		],
-		"function": "extract"
-	},
-	"ADP Codes":
     {
+        "srce": "adprp",
+        "sequence": 1,
         "name": "ADP Codes",
         "description": "link to adp code definitions",
         "defn": [
@@ -459,4 +468,4 @@
             {}
         ]
     }
-}
+]

templates/srce_template.json

@@ -1,121 +1,76 @@
 {
-    "name": "WMPD",
+    "name": "dcard",
-    "descr": "Williams Paid File",
+    "source": "client_file",
-    "type":"csv",
+    "loading_function": "csv",
-    "schema": [
+    "constraint": [
+        "{Trans. Date}",
+        "{Post Date}",
+        "{Description}"
+    ],
+    "schemas": {
+        "default": [
             {
-            "key": "Carrier",
+                "path": "{Trans. Date}",
-            "type": "text"
+                "type": "date",
+                "column_name": "Trans. Date"
             },
             {
-            "key": "SCAC",
+                "path": "{Post Date}",
-            "type": "text"
+                "type": "date",
+                "column_name": "Post Date"
             },
             {
-            "key": "Mode",
+                "path": "{Description}",
-            "type": "text"
+                "type": "text",
+                "column_name": "Description"
             },
             {
-            "key": "Pro #",
+                "path": "{Amount}",
-            "type": "text"
+                "type": "numeric",
+                "column_name": "Amount"
             },
             {
-            "key": "B/L",
+                "path": "{Category}",
-            "type": "text"
+                "type": "text",
-        },
+                "column_name": "Category"
-        {
-            "key": "Pd Amt",
-            "type": "numeric"
-        },
-        {
-            "key": "Loc#",
-            "type": "text"
-        },
-        {
-            "key": "Pcs",
-            "type": "numeric"
-        },
-        {
-            "key": "Wgt",
-            "type": "numeric"
-        },
-        {
-            "key": "Chk#",
-            "type": "numeric"
-        },
-        {
-            "key": "Pay Dt",
-            "type": "date"
-        },
-        {
-            "key": "Acct #",
-            "type": "text"
-        },
-        {
-            "key": "I/O",
-            "type": "text"
-        },
-        {
-            "key": "Sh Nm",
-            "type": "text"
-        },
-        {
-            "key": "Sh City",
-            "type": "text"
-        },
-        {
-            "key": "Sh St",
-            "type": "text"
-        },
-        {
-            "key": "Sh Zip",
-            "type": "text"
-        },
-        {
-            "key": "Cons Nm",
-            "type": "text"
-        },
-        {
-            "key": "D City ",
-            "type": "text"
-        },
-        {
-            "key": "D St",
-            "type": "text"
-        },
-        {
-            "key": "D Zip",
-            "type": "text"
-        },
-        {
-            "key": "Sh Dt",
-            "type": "date"
-        },
-        {
-            "key": "Inv Dt",
-            "type": "date"
-        },
-        {
-            "key": "Customs Entry#",
-            "type": "text"
-        },
-        {
-            "key": "Miles",
-            "type": "numeric"
-        },
-        {
-            "key": "Frt Class",
-            "type": "text"
-        },
-        {
-            "key": "Master B/L",
-            "type": "text"
             }
         ],
-    "unique_constraint": {
+        "mapped": [
-        "fields":[
+            {
-            "{Pay Dt}",
+                "path": "{Trans. Date}",
-            "{Carrier}" 
+                "type": "date",
+                "column_name": "Trans. Date"
+            },
+            {
+                "path": "{Post Date}",
+                "type": "date",
+                "column_name": "Post Date"
+            },
+            {
+                "path": "{Description}",
+                "type": "text",
+                "column_name": "Description"
+            },
+            {
+                "path": "{Amount}",
+                "type": "numeric",
+                "column_name": "Amount"
+            },
+            {
+                "path": "{Category}",
+                "type": "text",
+                "column_name": "Category"
+            },
+            {
+                "path": "{party}",
+                "type": "text",
+                "column_name": "Party"
+            },
+            {
+                "path": "{reason}",
+                "type": "text",
+                "column_name": "Reason"
+            }
         ]
     }
 }