refactor: total change, no longer ripping gigabytes but going directly to datasets, must find way to automate download.

2024-08-31 15:32:57 -04:00
parent b5e0af3ca5
commit f31d901201
3 changed files with 89 additions and 180 deletions
--- a/db_schema.sql
+++ b/db_schema.sql
@@ -1,55 +1,75 @@
-CREATE TABLE IF NOT EXISTS entities (
-    cik INT PRIMARY KEY,               -- CIK is now the primary key, ensuring uniqueness
-    name VARCHAR(255) NOT NULL  -- Name of the company
+CREATE TABLE num (
+    adsh      VARCHAR(255),
+    tag       VARCHAR(255),
+    version   VARCHAR(255),
+    coreg     VARCHAR(255),
+    ddate     BIGINT,
+    qtrs      BIGINT,
+    uom       VARCHAR(50),
+    value     DOUBLE PRECISION,
+    footnote  TEXT      
 );

-CREATE TABLE IF NOT EXISTS facts (
-    id VARCHAR(255) PRIMARY KEY, -- Unique identifier for the fact
-    taxonomy VARCHAR(255),       -- Taxonomy of the fact
-    label VARCHAR(255),       -- Label of the fact
-    description TEXT, -- Description of the fact
-    unit VARCHAR(255)           -- Unit of the fact
+CREATE TABLE pre (
+    adsh      VARCHAR(255),
+    report    BIGINT,
+    line      BIGINT,
+    stmt      VARCHAR(255),
+    inpth     BIGINT,
+    rfile     VARCHAR(255),
+    tag       VARCHAR(255),
+    version   VARCHAR(255),
+    plabel    VARCHAR(255),
+    negating  BIGINT
 );

-CREATE TABLE IF NOT EXISTS data (
-    cik INT,                            -- CIK of the company
-    fact_id VARCHAR(255),
-    end DATE,                 
-    start DATE,                        -- Start date of the fact
-    val INT,
-    accn VARCHAR(255),
-    fy INT,
-    fp VARCHAR(255),
-    form VARCHAR(255),
-    filed DATE,
-    frame VARCHAR(255),
-    PRIMARY KEY (cik, fact_id, end),
-    FOREIGN KEY (cik) REFERENCES entities(cik),
-    FOREIGN KEY (fact_id) REFERENCES facts(id)
+CREATE TABLE sub (
+    adsh        VARCHAR(255),
+    cik         BIGINT,
+    name        VARCHAR(255),
+    sic         DOUBLE PRECISION,
+    countryba   VARCHAR(100),
+    stprba      VARCHAR(100),
+    cityba      VARCHAR(255),
+    zipba       VARCHAR(50),
+    bas1        VARCHAR(255),
+    bas2        VARCHAR(255),
+    baph        VARCHAR(255),
+    countryma   VARCHAR(100),
+    stprma      VARCHAR(100),
+    cityma      VARCHAR(255),
+    zipma       VARCHAR(50),
+    mas1        VARCHAR(255),
+    mas2        VARCHAR(255),
+    countryinc  VARCHAR(100),
+    stprinc     VARCHAR(100),
+    ein         BIGINT,
+    former      VARCHAR(255),
+    changed     DOUBLE PRECISION,
+    afs         VARCHAR(255),
+    wksi        BIGINT,
+    fye         DOUBLE PRECISION,
+    form        VARCHAR(50),
+    period      DOUBLE PRECISION,
+    fy          DOUBLE PRECISION,
+    fp          VARCHAR(50),
+    filed       BIGINT,
+    accepted    VARCHAR(255),
+    prevrpt     BIGINT,
+    detail      BIGINT,
+    instance    VARCHAR(255),
+    nciks       BIGINT,
+    aciks       VARCHAR(255)
 );

-
-- @block
-CREATE TABLE IF NOT EXISTS data (
-    entity_cik INT,
-    entity_name VARCHAR(255),
-    fact_id VARCHAR(255),
-    fact_taxonomy VARCHAR(255),
-    fact_label VARCHAR(255),
-    fact_description TEXT,
-    fact_unit VARCHAR(255)
-    end DATE,
-    val FLOAT,
-    accn VARCHAR(50),
-    fy INT,
-    fp VARCHAR(255),
-    form VARCHAR(255),
-    filed DATE,
-    frame VARCHAR(255),
-    start DATE,
-    PRIMARY KEY (entity_cik, fact_id, end, form),
-    
-)
-
-- @block
-CREATE TABLE IF NOT EXISTS data ();
+CREATE TABLE tag (
+    tag       VARCHAR(255),
+    version   VARCHAR(255),
+    custom    BIGINT,
+    abstract  BIGINT,
+    datatype  VARCHAR(255),
+    iord      VARCHAR(50),
+    crdr      VARCHAR(50),
+    tlabel    VARCHAR(255),
+    doc       TEXT
+);