src/SummaryFile1.py

   1 import os
   2
   3 import GPS
   4 import StringUtils
   5
   6
   7 class RecordError(StandardError):
   8     pass
   9
  10 class GeoRecord:
  11     """
  12     This class wraps one record in an SF1 geo file.
  13     """
  14
  15     MINIMUM_LINE_LENGTH = 400
  16
  17
  18 class Block:
  19     """
  20     Represents a block (which is a special case of a GeoRecord.
  21     All we care about here is the block number, population,
  22     area, and coordinates.
  23     """
  24
  25     def __init__(self, geo_record):
  26         """We initialize from a GeoRecord object"""
  27         # All of these int/float conversions will throw a ValueError
  28         # if the input string cannot be converted o the specified
  29         # type.
  30         self.block_number = int(geo_record.block)
  31         self.tract_number = int(geo_record.tract)
  32         self.population = int(geo_record.pop100)
  33         self.area_land = float(geo_record.arealand)
  34         self.area_water = float(geo_record.areawatr)
  35
  36         self.coordinates = GPS.Coordinates()
  37         self.coordinates.latitude = float(geo_record.intptlat)
  38         self.coordinates.longitude = float(geo_record.intptlon)
  39
  40         if (self.total_area() == 0):
  41             raise InvalidAreaError('A block may not have zero area.')
  42
  43
  44     def total_area(self):
  45         return (self.area_land + self.area_water)
  46
  47
  48     def population_density(self):
  49         return (self.population / self.total_area())
  50
  51
  52 class GeoRecordParser:
  53
  54     def parse_file(self, path):
  55         """
  56         Assuming that path refers to an SF1 (geo) file, parse the
  57         geographic header records contained within it. Return a list
  58         of GeoRecord objects.
  59         """
  60
  61         # Our list of GeoRecord objects to return. Empty at first.
  62         records = []
  63
  64         f = open(path, 'r')
  65
  66         for line in f:
  67             record = self.parse_line(line)
  68             records.append(record)
  69
  70         f.close()
  71
  72         return records
  73
  74
  75     def parse_blocks(self, path):
  76         """Parse only the blocks from a geo file."""
  77         blocks = []
  78         records = self.parse_file(path)
  79
  80         for record in records:
  81             try:
  82                 block = Block(record)
  83                 blocks.append(block)
  84             except ValueError:
  85                 # A value couldn't be converted to the appropriate type.
  86                 continue
  87
  88         return blocks
  89
  90
  91
  92     def parse_line(self, line):
  93         """
  94         Parse one line of an SF1 geo file. Hopefully, the input will
  95         match the specification. We can check the line length here, or
  96         allow the GeoRecord class to parse the data meaningfully and
  97         throw an error if something doesn't look right.
  98         """
  99         if (len(line) < GeoRecord.MINIMUM_LINE_LENGTH):
 100             raise RecordError("The input line is too short. The SF1 specification requires a line length of %d characters; this line contains only %d characters" % (GeoRecord.MINIMUM_LINE_LENGTH, len(line)))
 101
 102         record = GeoRecord()
 103
 104         # Note that Python list indexes are zero-based, whereas the SF1
 105         # specification gives the field offsets as one-based. For example,
 106         # the first field, "File Identification," is defined as beginning
 107         # at position 1, and having length 6. The following line corresponds
 108         # to this definition.
 109         record.fileid = line[0:6]
 110
 111         # State / US Abbreviation (USPS)
 112         record.stusab = line[6:8]
 113
 114         # Summary Level
 115         record.sumlev = line[8:11]
 116
 117         # Geographic Component
 118         record.geocomp = line[11:13]
 119
 120         # Characteristic Iteration
 121         record.chariter = line[13:16]
 122
 123         # Characteristic Iteration File Sequence Number
 124         record.cifsn = line[16:18]
 125
 126         # Logical Record Number
 127         record.logrecno = line[18:25]
 128
 129         # Region
 130         record.region = line[25]
 131
 132         # Division
 133         record.division = line[26]
 134
 135         # State (Census)
 136         record.statece = line[27:29]
 137
 138         # State (FIPS)
 139         record.state = line[29:31]
 140
 141         # County
 142         record.county =  line[31:34]
 143
 144         # County Size Code
 145         record.countysc = line[34:36]
 146
 147         # County Subdivision (FIPS)
 148         record.cousub = line[36:41]
 149
 150         # FIPS County Subdivision Class Code
 151         record.cousubcc = line[41:43]
 152
 153         # County Subdivision Size Code
 154         record.cousubsc = line[43:45]
 155
 156         # Place (FIPS)
 157         record.place = line[45:50]
 158
 159         # FIPS Place Class Code
 160         record.placecc = line[50:52]
 161
 162         # Place Description Code
 163         record.placedc = line[52]
 164
 165         # Place Size Code
 166         record.placesc = line[53:55]
 167
 168         # Census Tract
 169         record.tract = line[55:61]
 170
 171         # Block Group
 172         record.blkgrp = line[61]
 173
 174         # Block
 175         record.block = line[62:66]
 176
 177         # Internal Use Code
 178         record.iuc = line[66:68]
 179
 180         # Consolidated City (FIPS)
 181         record.concit = line[68:71]
 182
 183         # FIPS Consolidated City Class Code
 184         record.concitcc = line[73:75]
 185
 186         # Consolidated City Size Code
 187         record.concitsc = line[75:77]
 188
 189         # American Indian Area/Alaska Native Area/Hawaiian Home Land
 190         # (Census)
 191         record.aianhh = line[77:81]
 192
 193         # American Indian Area/Alaska Native Area/Hawaiian Home Land
 194         # (FIPS)
 195         record.aianhhfp = line[81:86]
 196
 197         # FIPS American Indian Area/Alaska Native Area/Hawaiian Home
 198         # Land Class Code
 199         record.aianhhcc = line[86:88]
 200
 201         # American Indian Trust Land/Hawaiian Home Land Indicator
 202         record.aihhtli = line[88]
 203
 204         # American Indian Tribal Subdivision (Census)
 205         record.aitsce = line[89:92]
 206
 207         # American Indian Tribal Subdivision (FIPS)
 208         record.aits = line[92:97]
 209
 210         # FIPS American Indian Tribal Subdivision Class Code
 211         record.aitscc = line[97:99]
 212
 213         # Alaska Native Regional Corporation (FIPS)
 214         record.anrc = line[99:104]
 215
 216         # FIPS Alaska Native Regional Corporation Class Code
 217         record.anrccc = line[104:106]
 218
 219         # Metropolitan Statistical Area/Consolidated Metropolitan
 220         # Statistical Area
 221         record.msacmsa = line[106:110]
 222
 223         # MSA/CMSA Size Code
 224         record.masc = line[110:112]
 225
 226         # Consolidated Metropolitan Statistical Area
 227         record.cmsa = line[112:114]
 228
 229         # Metropolitan Area Central City Indicator
 230         record.macci = line[114]
 231
 232         # Primary Metropolitan Statistical Area
 233         record.pmsa = line[115:119]
 234
 235         # New England County Metropolitan Area
 236         record.necma = line[119:123]
 237
 238         # New England County Metropolitan Area Central City Indicator
 239         record.necmacci = line[123]
 240
 241         # New England County Metropolitan Area Size Code
 242         record.necmasc = line[124:126]
 243
 244         # Extended Place Indicator
 245         record.exi = line[126]
 246
 247         # Urban Area
 248         record.ua = line[127:132]
 249
 250         # Urban Area Size Code
 251         record.uasc = line[132:134]
 252
 253         # Urban Area Type
 254         record.ustype = line[134]
 255
 256         # Urban/Rural
 257         record.ur = line[135]
 258
 259         # Congressional District (106th)
 260         record.cd106 = line[136:138]
 261
 262         # Congressional District (108th)
 263         record.cd108 = line[138:140]
 264
 265         # Congressional District (109th)
 266         record.cd109 = line[140:142]
 267
 268         # Congressional District (110th)
 269         record.cd110 = line[142:144]
 270
 271         # State Legislative District (Upper Chamber)
 272         record.sldu = line[144:147]
 273
 274         # State Legislative District (Lower Chamber)
 275         record.sldl = line[147:150]
 276
 277         # Voting District
 278         record.vtd = line[150:156]
 279
 280         # Voting District Indicator
 281         record.vtdi = line[156]
 282
 283         # ZIP Code Tabulation Area (3 digit)
 284         record.zcta3 = line[157:160]
 285
 286         # ZIP Code Tabulation Area (5 digit)
 287         record.zcta5 = line[160:165]
 288
 289         # Subbarrio (FIPS)
 290         record.submcd = line[165:170]
 291
 292         # FIPS Subbarrio Class Code
 293         record.submcdcc = line[170:172]
 294
 295         # Area (Land)
 296         record.arealand = line[172:186]
 297
 298         # Area (Water)
 299         record.areawatr = line[186:200]
 300
 301         # Area Name - Legal/Statistical
 302         # Area Description (LSAD)
 303         # Term - Part Indicator
 304         record.name = line[200:290]
 305
 306         # Functional Status Code
 307         record.funcstat = line[290]
 308
 309         # Geographic Change User Note Indicator
 310         record.gcuni = line[291]
 311
 312         # Population Count (100%)
 313         record.pop100 = line[292:301]
 314
 315         # Housing Unit Count (100%)
 316         record.hu100 = line[301:310]
 317
 318         # Internal Point (Latitude)
 319         record.intptlat = line[310:319]
 320
 321         # Internal Point (Longitude)
 322         record.intptlon = line[319:329]
 323
 324         # Legal/Statistical Area Description Code
 325         record.lsadc = line[329:331]
 326
 327         # Part Flag
 328         record.partflag = line[331]
 329
 330         # School District (Elementary)
 331         record.sdelm = line[332:337]
 332
 333         # School District (Secondary)
 334         record.sdsec = line[337:342]
 335
 336         # School District (Unified)
 337         record.sduni = line[342:347]
 338
 339         # Traffic Analysis Zone
 340         record.taz = line[347:353]
 341
 342         # Oregon Urban Growth Area
 343         record.uga = line[353:358]
 344
 345         # Public Use Microdata Area - 5% File
 346         record.puma5 = line[358:363]
 347
 348         # Public Use Microdata Area - 1% File
 349         record.puma1 = line[363:368]
 350
 351         # Reserved
 352         record.reserve2 = line[368:383]
 353
 354         # Metropolitan Area Central City
 355         record.macc = line[383:388]
 356
 357         # Urban Area Central Place
 358         record.uacp = line[388:393]
 359
 360         # Reserved
 361         record.reserved = line[393:400]
 362
 363
 364         return record
 365
 366
 367
 368 def FindClosestBlock(blocks, target_coords):
 369     """
 370     Find the closest block (from within blocks) to the GPS
 371     coordinates given by target_coords.
 372     """
 373
 374     # Empty by default. Hopefully we're passed some blocks.
 375     closest_block = None
 376     min_distance = 999999999.0 # Don't look at me like that.
 377
 378     for block in blocks:
 379         this_distance = GPS.CalculateDistance(target_coords, block.coordinates)
 380         if (this_distance < min_distance):
 381             closest_block = block
 382             min_distance = this_distance
 383
 384     return closest_block
 385
 386
 387
 388 def FindAveragePopulationDensity(coords, geo_file_path):
 389     grp = GeoRecordParser()
 390     blocks = grp.parse_blocks(geo_file_path)
 391     closest_block = FindClosestBlock(blocks, coords)
 392
 393     return closest_block.population_density()