Get HTML - parsed_html_t.sql

-- ------------------------------------------------------------------------- -- -- -- -- Title: parsed_html_t.sql -- -- Author: Steve Roach -- -- Date: 24-JUL-2006 -- -- Version: 01.00 -- -- -- -- Description: Create table PARSED_HTML. -- -- -- -- History: -- -- -- -- Date Ver Author Description -- -- --------- ----- ------ -------------------------------------------------- -- -- 24-JUL-06 01.00 SRR First release -- -- -- -- ------------------------------------------------------------------------- -- DROP TABLE parsed_html CASCADE CONSTRAINTS / CREATE TABLE parsed_html ( parse_run_no NUMBER NOT NULL ,raw_run_no NUMBER NOT NULL ,component_seq NUMBER NOT NULL ,indent NUMBER NOT NULL ,html_component VARCHAR2(4000) NOT NULL ,ins_tsp DATE NOT NULL ) / COMMENT ON TABLE parsed_html IS 'Holds parsed external-site HTML. This is reasonably well structured and contains indentation information should a report be run against it. Each row contains a tag or data.' / COMMENT ON COLUMN parsed_html.parse_run_no IS 'Part of primary key. The Run Number that created this row.' / COMMENT ON COLUMN parsed_html.raw_run_no IS 'Part of primary key. The Run Number of the raw HTML that this row is sourced from.' / COMMENT ON COLUMN parsed_html.component_seq IS 'Part of primary key. Maintains the order of the component in the HTML page.' / COMMENT ON COLUMN parsed_html.indent IS 'Indentation information for report layouts. The lowest indent is zero and components within outer components are given numbers comesurate with their place within the web page. For example, the BODY tag is at level 0, a TABLE tag could be level 1, TR at level 2, and TD at level 3.' / COMMENT ON COLUMN parsed_html.html_component IS 'The HTML tag or data from the web page. Long strings of data could be split over several rows and no attempt at parapraph layout is made. i.e. words could be split over rows.' / COMMENT ON COLUMN parsed_html.ins_tsp IS 'Insert timestamp.' /