-
Notifications
You must be signed in to change notification settings - Fork 116
/
censusdata.py
executable file
·198 lines (156 loc) · 6.21 KB
/
censusdata.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
#!/usr/bin/env python3
# Download Census data files for the 2000 Decadal Summary File 1:
# https://www2.census.gov/census_2000/datasets/Summary_File_1/STATENAME
# If you look for documentation, you'll see pointers to the 730-page
# PDF sf1.pdf. Don't bother: it's completely wrong and must be for
# some earlier dataset.
# Instead, the documentation is in the files inside:
# http://www.census.gov/support/2000/SF1/SF1SAS.zip
# inside which, SF101.Sas describes the fields in st00001.uf1
# where st is the state abbreviation.
import os, sys
import re
import argparse
import zipfile
from collections import OrderedDict
# While testing:
from pprint import pprint
# A dictionary: { fileno: dic } where fileno is an int from 1 to 39 or 'geo'
# and dic is another dictionary of 'censuscode': "long description"
# where censuscode is a 7-char string like P000001 or H016H018.
CensusCodes = {}
# Fields in the sf1geo file
GeoFields = {}
def codesFromZipFile(zipfilename):
zf = zipfile.ZipFile(zipfilename, 'r')
pat = re.compile(b" *([A-Z][0-9]{3}[0-9A-Z]{3,4})=' *(.*)'")
for name in zf.namelist():
if not name.lower().endswith('.sas'):
continue
# The sf1geo file is special, so parse it separately
if name == 'sf1geo.sas':
parse_geo_sas_lines(zf.read(name).split(b'\n'))
continue
filematch = re.match('sf([0-9]{3}).sas', name.lower())
if not filematch:
# print(name, "doesn't match filematch pattern")
continue
code_dict = OrderedDict()
fileno = int(filematch.group(1))
# basename = os.path.basename(name)
# root, ext = os.path.splitext(basename)
# Every file stars with these five, which don't have p-numbers
code_dict['FILEID'] = 'File Identification'
code_dict['STUSAB'] = 'State/U.S.-Abbreviation (USPS)'
code_dict['CHARITER'] = 'Characteristic Iteration'
code_dict['CIFSN'] = 'Characteristic Iteration File Sequence Number'
code_dict['LOGRECNO'] = 'Logical Record Number'
saslines = zf.read(name).split(b'\n')
for line in saslines:
m = re.match(pat, line)
if m:
pcode, desc = [ s.decode() for s in m.groups() ]
# print("%7s -- %s" % (code, desc))
code_dict[pcode] = desc
# else:
# print("No match on line:", line)
CensusCodes[fileno] = code_dict
def parse_geo_sas_lines(lines):
"""lines are read from the sf1geo.sas file.
Create a dictionary of fields:
{ 'CODE': { 'name':'long name', 'start': int, 'end': int }
{ 'name', 'code', 'start', 'end' }
"""
labelpat = re.compile(b"(LABEL )?([A-Z0-9]*)\=\'(.*)\'")
fieldspat = re.compile(b"([A-Z0-9]+) \$ ([0-9]+)\-([0-9]+)")
for line in lines:
line = line.strip()
m = re.match(labelpat, line)
if m:
sys.stdout.flush()
# Assume here that labelpats all come before fieldspats,
# so if we're seeing a labelpat, it doesn't already exist
# inside GeoFields.
code = m.group(2).decode()
GeoFields[code] = { 'name': m.group(3).decode() }
continue
m = re.match(fieldspat, line)
if m:
# If there's a fieldspat for this code, it should have
# had a long description already using a labelpat,
# so the code (group(1)) should already be in GeoFields.
# print("groups:", m.groups())
code = m.group(1).decode()
GeoFields[code]['start'] = int(m.group(2)) - 1
GeoFields[code]['end'] = int(m.group(3))
continue
# pprint(GeoFields)
def file_for_code(code):
for fileno in CensusCodes:
if code in CensusCodes[fileno]:
return fileno
return None
def codes_for_description(desc):
codes = []
desc = desc.lower()
for fileno in CensusCodes:
for pcode in CensusCodes[fileno]:
if desc in CensusCodes[fileno][pcode].lower():
codes.append((pcode, CensusCodes[fileno][pcode]))
return codes
counties = []
def parse_geo_file(filename):
with open(filename) as fp:
for line in fp:
geo = parse_geo_line(line)
c = geo['COUNTY'].strip()
if c:
c = int(c)
if c not in counties:
counties.append(c)
counties.sort()
print("Counties:", counties)
def parse_geo_line(line):
"""Parse the <st>geo.uf1 file according to the GeoFields.
"""
d = {}
for code in GeoFields:
try:
d[code] = line[GeoFields[code]['start']:GeoFields[code]['end']]
except KeyError:
print("Key error, GeoFields[%s] =" % code, GeoFields[code])
break
# print("Line:", line)
# for field in d:
# print(field, ":", d[field], ":", GeoFields[field]['name'])
# print()
# print(d['COUNTY'], d['TRACT'], d['BLKGRP'], d['BLOCK'])
return d
if __name__ == '__main__':
parser = argparse.ArgumentParser(
description="Parse US Decennial census data")
parser.add_argument('-c', action="store", dest="code",
help='Show filenumber containing 6-digit census code')
parser.add_argument('-d', action="store", dest="desc",
help='Show entries containing a long description')
parser.add_argument('-g', action="store", dest="geo",
help='Parse the <ST>geo.uf1 file')
parser.add_argument('zipfile', help="location of SF1SAS.zip file")
args = parser.parse_args(sys.argv[1:])
# print(args)
# Pass in the path to SF1SAS.zip
codesFromZipFile(args.zipfile)
if args.code:
print("Files with code %s:" % args.code, file_for_code(args.code))
elif args.desc:
codes = codes_for_description(args.desc)
print('Codes containing description "%s":' % args.desc)
for pair in codes:
print("%s: %s" % pair)
elif args.geo:
parse_geo_file(args.geo)
else:
for fileno in CensusCodes:
print("\n==== File", fileno)
for pcode in CensusCodes[fileno]:
print("%7s: %s" % (pcode, CensusCodes[fileno][pcode]))