Mercurial > dive4elements > river

annotate flys-backend/contrib/shpimporter/utils.py @ 4995:998b29c8d2fd dami

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Improve debug output for unsupported features and skipped shapefiles
author Andre Heinecke <aheinecke@intevation.de>
date Fri, 15 Feb 2013 11:51:58 +0100
parents a6ee62a070b0
children 769593a84606
rev   line source
2798
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
1 import os
4874
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
2 import sys
3654
59ca5dab2782 Shape importer: use python's OptionParse to read user specific configuration from command line.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents: 2798
diff changeset
3 from shpimporter import DEBUG, INFO, ERROR
4970
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
4 try:
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
5 from osgeo import ogr
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
6 except ImportErrror:
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
7 import ogr
2798
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
8
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
9 SHP='.shp'
4976
a6ee62a070b0 I'm learning how to use cursors \o/
Andre Heinecke <aheinecke@intevation.de>
parents: 4970
diff changeset
10 SQL_SELECT_RIVER_ID="SELECT id FROM rivers WHERE name = %s"
2798
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
11
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
12 def findShapefiles(path):
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
13 shapes = []
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
14
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
15 for root, dirs, files in os.walk(path):
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
16 if len(files) == 0:
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
17 continue
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
18
3654
59ca5dab2782 Shape importer: use python's OptionParse to read user specific configuration from command line.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents: 2798
diff changeset
19 DEBUG("Processing directory '%s' with %i files " % (root, len(files)))
2798
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
20
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
21 for f in files:
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
22 idx = f.find(SHP)
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
23 if (idx+len(SHP)) == len(f):
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
24 shapes.append((f.replace(SHP, ''), root + "/" + f))
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
25
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
26 return shapes
5a654f2e35bc Added a python tool to import shapefiles into database.
Ingo Weinzierl <ingo.weinzierl@intevation.de>
parents:
diff changeset
27
4970
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
28 def getRiverId(dbconn, name):
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
29 """
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
30 Returns the id of the river "name"
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
31 Dbconn must be a python database connection api compliant object
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
32 """
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
33 cur = dbconn.cursor()
4976
a6ee62a070b0 I'm learning how to use cursors \o/
Andre Heinecke <aheinecke@intevation.de>
parents: 4970
diff changeset
34 cur.execute(SQL_SELECT_RIVER_ID, (name,))
4970
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
35 row = cur.fetchone()
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
36 if row:
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
37 return row[0]
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
38 else:
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
39 return 0
174fbaa3d344 Add handling of River Names and remove target_src parameter
Andre Heinecke <aheinecke@intevation.de>
parents: 4935
diff changeset
40
4935
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
41 def getUTF8(string):
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
42 """
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
43 Tries to convert the string to a UTF-8 encoding by first checking if it
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
44 is UTF-8 and then trying cp1252
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
45 """
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
46 try:
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
47 return unicode.encode(unicode(string, "UTF-8"), "UTF-8")
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
48 except UnicodeDecodeError:
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
49 # Probably European Windows names so lets try again
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
50 return unicode.encode(unicode(string, "cp1252"), "UTF-8")
c0a58558b817 Importer: - Handle regular expressions for attribute names
Andre Heinecke <aheinecke@intevation.de>
parents: 4887
diff changeset
51
4874
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
52 def getUTF8Path(path):
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
53 """
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
54 Tries to convert path to utf-8 by first checking the filesystemencoding
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
55 and trying the default windows encoding afterwards.
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
56 Returns a valid UTF-8 encoded unicode object or throws a UnicodeDecodeError
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
57 """
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
58 try:
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
59 return unicode.encode(unicode(path, sys.getfilesystemencoding()), "UTF-8")
b1d7e600b43b (importer) Add utility function to convert paths to utf-8
Andre Heinecke <aheinecke@intevation.de>
parents: 3654
diff changeset
60 except UnicodeDecodeError:
4887
1f6e544f7a7f Importer: Use cp1252 instead of latin-9 to guess filename encodings
Andre Heinecke <aheinecke@intevation.de>
parents: 4884
diff changeset
61 # Probably European Windows names so lets try again
1f6e544f7a7f Importer: Use cp1252 instead of latin-9 to guess filename encodings
Andre Heinecke <aheinecke@intevation.de>
parents: 4884
diff changeset
62 return unicode.encode(unicode(path, "cp1252"), "UTF-8")
4995
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
63
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
64 WKB_MAP = {
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
65 ogr.wkb25Bit : 'wkb25Bit',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
66 ogr.wkbGeometryCollection : 'wkbGeometryCollection',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
67 ogr.wkbGeometryCollection25D :'wkbGeometryCollection25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
68 ogr.wkbLineString : 'wkbLineString',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
69 ogr.wkbLineString25D : 'wkbLineString25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
70 ogr.wkbLinearRing : 'wkbLinearRing',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
71 ogr.wkbMultiLineString : 'wkbMultiLineString',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
72 ogr.wkbMultiLineString25D : 'wkbMultiLineString25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
73 ogr.wkbMultiPoint : 'wkbMultiPoint',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
74 ogr.wkbMultiPoint25D : 'wkbMultiPoint25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
75 ogr.wkbMultiPolygon : 'wkbMultiPolygon',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
76 ogr.wkbMultiPolygon25D : 'wkbMultiPolygon25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
77 ogr.wkbNDR : 'wkbNDR',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
78 ogr.wkbNone : 'wkbNone',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
79 ogr.wkbPoint : 'wkbPoint',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
80 ogr.wkbPoint25D : 'wkbPoint25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
81 ogr.wkbPolygon : 'wkbPolygon',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
82 ogr.wkbPolygon25D : 'wkbPolygon25D',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
83 ogr.wkbUnknown : 'wkbUnknown',
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
84 ogr.wkbXDR : 'wkbXDR'
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
85 }
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
86
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
87 def getWkbString(type):
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
88 return WKB_MAP.get(type) or "Unknown"
998b29c8d2fd Improve debug output for unsupported features and skipped shapefiles
Andre Heinecke <aheinecke@intevation.de>
parents: 4976
diff changeset
89
http://dive4elements.wald.intevation.org