src/SummaryFile1.py

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