xref: /btstack/tool/bluetooth_company_id.py (revision e6c9673776c7f85e24c76da7f4b1b83031102d2b) 
1#!/usr/bin/env python
2#
3# Scrape GATT UUIDs from Bluetooth SIG page
4# https://www.bluetooth.com/specifications/assigned-numbers/company-identifiers
5#
6# Copyright 2017 BlueKitchen GmbH
7#
8
9from lxml import html
10import datetime
11import requests
12import sys
13import codecs
14import os
15import re
16
17program_info = '''
18BTstack Company ID Scraper for BTstack
19Copyright 2017, BlueKitchen GmbH
20'''
21
22header = '''
23/**
24 * bluetooth_company_id.h generated from Bluetooth SIG website for BTstack
25 */
26
27#ifndef __BLUETOOTH_COMPANY_ID_H
28#define __BLUETOOTH_COMPANY_ID_H
29'''
30
31page_info = '''
32/**
33 * Assigned numbers from {page}
34 */
35'''
36
37trailer = '''
38#endif
39'''
40
41tags = []
42
43def create_name(company):
44    # remove parts in braces
45    p = re.compile('\(.*\)')
46    tag = p.sub('',company).rstrip().upper()
47    tag = tag.replace('&',' AND ')
48    tag = tag.replace(''','')
49    tag = tag.replace('"',' ')
50    tag = tag.replace('+',' AND ')
51    tag = tag.replace(' - ', ' ')
52    tag = tag.replace('/', ' ')
53    tag = tag.replace(';',' ')
54    tag = tag.replace(',','')
55    tag = tag.replace('.', '')
56    tag = tag.replace('-','_')
57    tag = tag.replace('  ',' ')
58    tag = tag.replace('  ',' ')
59    tag = tag.replace('  ',' ')
60    tag = tag.replace(' ','_')
61    return "BLUETOOTH_COMPANY_ID_" + tag
62
63def scrape_page(fout, url):
64    print("Parsing %s" % url)
65    fout.write(page_info.format(page=url))
66
67    # get from web
68    r = requests.get(url)
69    content = r.text
70
71    # test: fetch from local file 'service-discovery.html'
72    # f = codecs.open("company-identifiers.html", "r", "utf-8")
73    # content = f.read();
74
75    tree = html.fromstring(content)
76    # get all java script
77    rows = tree.xpath('//script')
78    for row in rows:
79        script = row.text_content()
80        if not 'DataTable' in script:
81            continue
82        start_tag = 'data:  ['
83        end_tag   = '["65535","0xFFFF",'
84        start = script.find(start_tag)
85        end   = script.find(end_tag)
86        company_list = script[start + len(start_tag):end]
87        for entry in company_list.split('],'):
88            if len(entry) < 5:
89                break
90            entry = entry[1:]
91            fields = entry.split('","')
92            id_hex = fields[1]
93            company = create_name(fields[2][:-1])
94            if company in tags:
95                company = company + "2"
96            else:
97                tags.append(company)
98            if len(company) < 2:
99                continue
100            fout.write("#define %-80s %s\n" %  (company, id_hex))
101
102btstack_root = os.path.abspath(os.path.dirname(sys.argv[0]) + '/..')
103gen_path = btstack_root + '/src/bluetooth_company_id.h'
104
105print(program_info)
106
107with open(gen_path, 'wt') as fout:
108    fout.write(header.format(datetime=str(datetime.datetime.now())))
109    scrape_page(fout, 'https://www.bluetooth.com/specifications/assigned-numbers/company-identifiers')
110    fout.write(trailer)
111
112print('Scraping successful!\n')