DON'T USE THIS -- We've used the lessons learned from this scraper to write a
new one; it's much, much better, and it uses the Scrapy framework. 
[https://github.com/tigeorgia/GeorgiaCorporationScraper](https://github.com/tigeorgia/GeorgiaCorporationScraper)

A scraper for the Georgian company registry: https://enreg.reestri.gov.ge

The util folder contains handy scripts for dealing with the database that
this scraper creates.

DB descriptions:

People(pid, name, address, p/n, date_of_birth)

company(cid, id_code, p_code, state_reg_code, comp_name, legal_form, type, state_reg_date, status, scrap_date)

App_status(aid, cid, date, status, text)
filed(pid, aid, cid, date)

scans(cid, sid, date, link_to_scan, text)
extracts(cid, eid, reg_number, application_num, prep_date, address, email, reg_authority, tax_inpsection, seizure, lien, property, debtor)
pages(cid, page_id, property_num, B_number, entity_name, legal_form, reorg_type, number_of, replacement_info, attached_docs, backed_docs, notes)
page_to_person(cid, page_id, pid, role)
person_to_extract(cid, eid, pid, role)
comp_to_person(rid, pid, cid, since, until, relation)

Users(email, first_name, last_name, password, account_type)
watches(email, cid)


--------------QUERIES-------------------------

 
CREATE TABLE people
(pid INTEGER NOT NULL,
name TEXT(40),
address TEXT(64),
personal_number TEXT(16),
birth_date TEXT(25),
PRIMARY KEY (pid))


CREATE TABLE company
(cid INTEGER NOT NULL,
id_code TEXT(16),
p_code VARCHAR(16),
state_reg_code TEXT(25),
comp_name TEXT(64),
legal_form TEXT(25),
type TEXT(25),
state_reg_date TEXT(25),
status VARCHAR(25),
scrap_date TEXT(25),
PRIMARY KEY (cid))


CREATE TABLE app_status
(aid INTEGER NOT NULL,
cid INTEGER NOT NULL,
date TEXT(25),
status TEXT(25),
text TEXT(2048),
PRIMARY KEY (aid, cid),
FOREIGN KEY(cid) REFERENCES company(cid))


CREATE TABLE filed
(pid INTEGER NOT NULL,
aid INTEGER NOT NULL,
cid INTEGER NOT NULL,
date TEXT(25),
PRIMARY KEY (pid, aid, cid)
FOREIGN KEY(cid) REFERENCES company(cid)
FOREIGN KEY(aid) REFERENCES app_status(aid)
FOREIGN KEY(pid) REFERENCES people(pid))

CREATE TABLE scans
(cid INTEGER NOT NULL,
sid INTEGER NOT NULL,
date TEXT(25),
links_to_scan VARCHAR(128),
text TEXT(2048),
PRIMARY KEY (cid, sid)
FOREIGN KEY(cid) REFERENCES company(cid))

CREATE TABLE extracts
(cid INTEGER NOT NULL,
eid INTEGER NOT NULL,
reg_number INTEGER,
application_num VARCHAR(10),
prep_date TEXT(25),
address TEXT(64),
email VARCHAR(25),
reg_authority TEXT(64),
tax_inspection TEXT(64),
seizure TEXT(64),
lien TEXT(64),
property TEXT(64),
debtor TEXT(64),
PRIMARY KEY (cid, eid)
FOREIGN KEY(cid) REFERENCES company(cid))


CREATE TABLE pages
(cid INTEGER NOT NULL,
page_id INTEGER NOT NULL,
property_num TEXT(16),
B_number VARCHAR(10),
entity_name TEXT(64),
legal_form TEXT(25),
reor_type TEXT(25),
number_of TEXT(25),
replacement_info TEXT(256),
attached_docs TEXT(1024),
backed_docs TEXT(1024),
notes TEXT(1024),
PRIMARY KEY (cid, page_id)
FOREIGN KEY(cid) REFERENCES company(cid))


CREATE TABLE page_to_person
(cid INTEGER NOT NULL,
page_id INTEGER NOT NULL,
pid INTEGER NOT NULL,
role TEXT(64),
PRIMARY KEY (cid, page_id, pid)
FOREIGN KEY(cid) REFERENCES company(cid)
FOREIGN KEY(page_id) REFERENCES pages(page_id)
FOREIGN KEY(pid) REFERENCES people(pid))

CREATE TABLE person_to_extract
(cid INTEGER NOT NULL,
eid INTEGER NOT NULL,
pid INTEGER NOT NULL,
role TEXT(64),
PRIMARY KEY (cid, eid, pid)
FOREIGN KEY(cid) REFERENCES company(cid)
FOREIGN KEY(eid) REFERENCES extracts(eid)
FOREIGN KEY(pid) REFERENCES people(pid))

CREATE TABLE comp_to_person
(rid INTEGER NOT NULL,
pid INTEGER NOT NULL,
cid INTEGER NOT NULL,
since TEXT(25),
until TEXT(25),
relation TEXT(64),
PRIMARY KEY (rid, pid, cid)
FOREIGN KEY(cid) REFERENCES company(cid)
FOREIGN KEY(pid) REFERENCES people(pid))


CREATE TABLE users
(email VARCHAR(32) NOT NULL,
first_name TEXT(25),
last_name TEXT(25),
password VARCHAR(12),
account_type VARCHAR(8),
PRIMARY KEY (email))


CREATE TABLE watches
(email VARCHAR(25) NOT NULL,
cid INTEGER NOT NULL,
PRIMARY KEY (email, cid)
FOREIGN KEY(email) REFERENCES users(email)
FOREIGN KEY(cid) REFERENCES company(cid))