| 1 |
1631
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Downloads REMIB data for all nodes
|
| 3 |
|
|
# Usage: self
|
| 4 |
|
|
|
| 5 |
1633
|
aaronmk
|
import csv
|
| 6 |
|
|
import itertools
|
| 7 |
1631
|
aaronmk
|
import os.path
|
| 8 |
|
|
import StringIO
|
| 9 |
|
|
import sys
|
| 10 |
|
|
import urllib2
|
| 11 |
|
|
|
| 12 |
|
|
sys.path.append(os.path.dirname(__file__)+"/../../../lib")
|
| 13 |
|
|
|
| 14 |
1637
|
aaronmk
|
import exc
|
| 15 |
1631
|
aaronmk
|
import profiling
|
| 16 |
|
|
import streams
|
| 17 |
|
|
import strings
|
| 18 |
1637
|
aaronmk
|
import timeout
|
| 19 |
1631
|
aaronmk
|
import util
|
| 20 |
|
|
|
| 21 |
1638
|
aaronmk
|
timeout_ = 20 # sec
|
| 22 |
1637
|
aaronmk
|
|
| 23 |
1633
|
aaronmk
|
alphabet = map(chr, xrange(ord('A'), ord('Z')+1))
|
| 24 |
|
|
|
| 25 |
1637
|
aaronmk
|
class InputException(Exception): pass
|
| 26 |
|
|
|
| 27 |
1631
|
aaronmk
|
def is_ignore(line):
|
| 28 |
|
|
line = strings.remove_line_ending(line)
|
| 29 |
|
|
return line == '' or line.startswith('\t') or line.find(',') < 0
|
| 30 |
|
|
|
| 31 |
|
|
def main():
|
| 32 |
1633
|
aaronmk
|
def clear_line(): sys.stderr.write('\n')
|
| 33 |
1631
|
aaronmk
|
log_indent = 0
|
| 34 |
|
|
def log(msg, line_ending='\n'):
|
| 35 |
|
|
sys.stderr.write((' '*log_indent)+msg+line_ending)
|
| 36 |
|
|
|
| 37 |
|
|
stdout = streams.LineCountOutputStream(sys.stdout)
|
| 38 |
|
|
|
| 39 |
|
|
# Get by family ('familia') because that is the most general level at which
|
| 40 |
|
|
# an identification can be made. This assumes all records have a family.
|
| 41 |
|
|
url_template = ('http://www.conabio.gob.mx/remib/cgi-bin/'
|
| 42 |
|
|
'remib_distribucion.cgi?lengua=EN&niveltax=familia&taxon=[prefix]%25&'
|
| 43 |
|
|
'pais=Todos&pais_otro=&estado=100&formato=csv&mapa=no&mapabase=estados'
|
| 44 |
|
|
'&coleccion=id%3D[node_id]')
|
| 45 |
1633
|
aaronmk
|
done = False
|
| 46 |
|
|
for node_id in itertools.count(1):
|
| 47 |
|
|
if done: break
|
| 48 |
1631
|
aaronmk
|
log('Processing node #'+str(node_id)+'...')
|
| 49 |
|
|
log_indent += 1
|
| 50 |
1633
|
aaronmk
|
start_line_num = stdout.line_num
|
| 51 |
1631
|
aaronmk
|
profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
|
| 52 |
|
|
|
| 53 |
|
|
node_url_template = url_template.replace('[node_id]', str(node_id))
|
| 54 |
1633
|
aaronmk
|
for prefix_chars in itertools.product(alphabet, repeat=2):
|
| 55 |
|
|
if done: break
|
| 56 |
|
|
prefix = ''.join(prefix_chars)
|
| 57 |
1631
|
aaronmk
|
log('Processing prefix '+prefix+'...')
|
| 58 |
1633
|
aaronmk
|
row_ct = 0
|
| 59 |
|
|
def print_status(line_ending='\n'):
|
| 60 |
|
|
log('Processed '+str(row_ct)+' row(s)', line_ending)
|
| 61 |
1631
|
aaronmk
|
log_indent += 1
|
| 62 |
|
|
|
| 63 |
|
|
url = node_url_template.replace('[prefix]', prefix)
|
| 64 |
1637
|
aaronmk
|
stream = streams.StreamIter(streams.TimeoutInputStream(
|
| 65 |
|
|
urllib2.urlopen(url), timeout_))
|
| 66 |
1631
|
aaronmk
|
|
| 67 |
|
|
# Copy lines
|
| 68 |
1637
|
aaronmk
|
try:
|
| 69 |
1641
|
aaronmk
|
util.skip(stream, is_ignore) # skip header
|
| 70 |
|
|
try:
|
| 71 |
|
|
metadata_row = csv.reader(stream).next()
|
| 72 |
|
|
assert metadata_row[0] == 'COLLECTION'
|
| 73 |
|
|
except StopIteration:
|
| 74 |
|
|
done = True # empty response means no more nodes
|
| 75 |
|
|
|
| 76 |
1637
|
aaronmk
|
for line in stream:
|
| 77 |
|
|
if is_ignore(line):
|
| 78 |
|
|
error = strings.remove_prefix('\t\t', line)
|
| 79 |
|
|
if len(error) != len(line): raise InputException(error)
|
| 80 |
|
|
break
|
| 81 |
|
|
if row_ct % 100 == 0: print_status('\r')
|
| 82 |
|
|
# CR at end so next print overwrites msg
|
| 83 |
|
|
stdout.write(line)
|
| 84 |
|
|
row_ct += 1
|
| 85 |
|
|
except (InputException, timeout.TimeoutException), e:
|
| 86 |
|
|
clear_line()
|
| 87 |
|
|
log('! Output line '+str(stdout.line_num)+': '+exc.str_(e))
|
| 88 |
1631
|
aaronmk
|
stream.close()
|
| 89 |
|
|
|
| 90 |
1633
|
aaronmk
|
profiler.add_iters(row_ct)
|
| 91 |
|
|
print_status()
|
| 92 |
1631
|
aaronmk
|
log_indent -= 1
|
| 93 |
|
|
|
| 94 |
1633
|
aaronmk
|
profiler.stop()
|
| 95 |
1631
|
aaronmk
|
log(profiler.msg())
|
| 96 |
|
|
|
| 97 |
|
|
if stdout.line_num == start_line_num: line_range = '<none>'
|
| 98 |
|
|
else: line_range = str(start_line_num)+'-'+str(stdout.line_num-1)
|
| 99 |
|
|
log('Used output lines '+line_range)
|
| 100 |
|
|
|
| 101 |
|
|
log_indent -= 1
|
| 102 |
|
|
|
| 103 |
|
|
main()
|
