-
-
Save Drunkar/0b1ab1f8df6bbe28cbef4c79420ec753 to your computer and use it in GitHub Desktop.
KiCad csv BOM generter plugin
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/env python | |
# (c) 2015 Productize <[email protected]> | |
# edited by Drunkar <[email protected]> | |
import os | |
import sys | |
import copy | |
import collections | |
import codecs | |
from bs4 import BeautifulSoup | |
xml_file = sys.argv[1] | |
with codecs.open(sys.argv[1], encoding='utf-8') as f: | |
soup = BeautifulSoup(f, "lxml") | |
date = soup.design.date.contents[0] | |
comp = {} | |
comp_nc = {} | |
standard_fields = ['Reference', 'Identifier', 'Package', | |
'Value', 'Tolerance', 'Voltage', 'Current', 'Power', 'PN'] | |
extra_fields = [] | |
i = 1 | |
for c in soup.components.find_all("comp"): | |
new_comp = collections.defaultdict(str) | |
new_comp['Reference'] = [c['ref']] | |
new_comp['Value'] = c.value.contents[0] | |
new_comp['Identifier'] = '' | |
new_comp['Package'] = '' | |
new_comp['Tolerance'] = '' | |
new_comp['Voltage'] = '' | |
new_comp['Current'] = '' | |
new_comp['Power'] = '' | |
new_comp['PN'] = '' | |
for f in c.find_all("field"): | |
new_comp[f['name']] = f.contents[0] | |
if f['name'] not in standard_fields: | |
if f['name'] not in extra_fields: | |
extra_fields.append(f['name']) | |
found = None | |
new_comp_no_ref = {} | |
for k in new_comp.keys(): | |
if k != 'Reference': | |
new_comp_no_ref[k] = new_comp[k] | |
for c2 in comp.values(): | |
c2_no_ref = {} | |
for k in c2.keys(): | |
if k != 'Reference': | |
c2_no_ref[k] = c2[k] | |
if str(c2_no_ref) == str(new_comp_no_ref): | |
found = c2 | |
c2['Reference'].append(c['ref']) | |
break | |
if found == None: | |
comp[i] = new_comp | |
i += 1 | |
csv_file = os.path.splitext(xml_file)[0] + "_bom.csv" | |
with codecs.open(csv_file, "w", encoding='utf-8') as f: | |
f.write( | |
"Reference,Quantity,Identifier,Package,Value,Tolerance,Voltage,Current,Power,PN") | |
for c in extra_fields: | |
f.write("," + c) | |
f.write('\n') | |
ordered = [] | |
for k in comp.keys(): | |
c = comp[k] | |
ref = ';'.join(c['Reference']) | |
l = [ref, str(len(c['Reference'])), c['Identifier'], c['Package'], c['Value'], | |
c['Tolerance'], c['Voltage'], c['Current'], c['Power'], c['PN']] | |
for y in extra_fields: | |
l.append(c[y]) | |
# for y in c.keys(): | |
# if y not in standard_fields: | |
# l.append | |
l = ['"' + x + '"' for x in l] | |
f.write(','.join(l)) | |
f.write('\n') |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment