2536: Improve PlaceUtils.py a little bit

svn: r11621

src/PlaceUtils.py

@@ -1,9 +1,10 @@
-#                                                     -*- python -*-
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 #
 # Gramps - a GTK+/GNOME based genealogy program
 #
-# Copyright (C) 2000-2006  Donald N. Allingham
+# Copyright (C) 2007-2009  B. Malengier
+# Copyright (C) 2009  Swoon on bug tracker
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
@@ -22,7 +23,6 @@
 
 # $Id:PlaceUtils.py 9912 2008-01-22 09:17:46Z acraphae $
 
-# Written by Benny Malengier
 
 #-------------------------------------------------------------------------
 #
@@ -31,6 +31,12 @@
 #-------------------------------------------------------------------------
 from gettext import gettext as _
 
+#-------------------------------------------------------------------------
+#
+# GRAMPS modules
+#
+#-------------------------------------------------------------------------
+
 
 #-------------------------------------------------------------------------
 #
@@ -66,6 +72,193 @@ if 'N' == South or 'S' == North or 'E' == West or 'W' == East:
     translate_en_loc['W'] = 'W'
 # end localisation part
 
+
+#------------------
+#
+# helper functions
+#
+#------------------
+
+def __convert_structure_to_float(sign, degs, mins=0, secs=0.0) :
+    """helper function which converts a structure to a nice
+    representation
+    """
+    v = float(degs) 
+    if mins is not None:
+        v += float(mins) / 60.
+    if secs is not None:
+        v += secs / 3600.
+    if sign == "-":
+        v = v * -1.
+
+    return v
+
+def __convert_using_float_repr(stringValue):
+    """ helper function that tries to convert the string using the float
+    representation
+    """
+    try : 
+        v = float(stringValue)      
+        return v
+    except ValueError :
+        return None;
+
+def __convert_using_colon_repr(stringValue):
+    """ helper function that tries to convert the string using the colon
+    representation
+    """
+    if stringValue.find(r':') == -1 :
+        return None
+
+    l = stringValue.split(':')
+    if len(l) < 2 or len(l) > 3:
+        return None
+    l[0]=l[0].strip()
+    # if no characters before ':' nothing useful is input!
+    if len(l[0]) == 0:
+        return None
+    if l[0][0] == '-':
+        sign = '-'
+        l[0]=l[0][1:]
+    else:
+        sign = '+'
+    try:
+        degs = int(l[0])
+        if degs < 0:
+            return None
+    except:
+        return None
+    try:
+        mins = int(l[1])
+        if mins < 0 or mins >= 60:
+            return None
+    except:
+        return None
+    secs=0.
+    if len(l) == 3:
+        try:
+            secs = float(l[2])
+            if secs < 0. or secs >= 60.:
+                return None
+        except:
+            return None
+
+    return __convert_structure_to_float(sign, degs, mins, secs)
+
+def __convert_using_classic_repr(stringValue, typedeg):
+    """helper function that tries to convert the string using the colon
+    representation
+    """
+    if stringValue.find(r'_') != -1:
+        return None    # not a valid lat or lon
+
+    #exchange some characters
+    stringValue = stringValue.replace(u'°',r'_')
+    #allow to input ° as #
+    stringValue = stringValue.replace(r'#',r'_')
+    #allow to input " as ''
+    stringValue = stringValue.replace(r"''",r'"')
+    #allow some special unicode symbols
+    stringValue = stringValue.replace(u'\u2033',r'"')
+    stringValue = stringValue.replace(u'\u2032',r"'")
+    #ignore spaces, a regex with \s* would be better here...
+    stringValue = stringValue.replace(r' ', r'')
+    stringValue = stringValue.replace(r'\t', r'')
+
+    # get the degrees, must be present
+    if stringValue.find(r'_') == -1:
+        return None
+    l = stringValue.split(r'_')
+    if len(l) != 2:
+        return None
+
+    try: 
+        degs = int(l[0])  #degrees must be integer value
+        if degs < 0:
+            return None
+    except:
+        return None
+    # next: minutes might be present once
+    l2 = l[1].split(r"'")
+    l3 = l2
+    mins = 0
+    if len(l2) > 2:
+        return None
+    if len(l2) == 2:
+        l3 = [l2[1],]
+        try:
+            mins = int(l2[0]) #minutes must be integer value
+            if mins < 0 or mins >= 60:
+                return None
+        except:
+            return None
+    # next: seconds might be present once
+    l3 = l3[0].split(r'"')
+    last = l3[0]
+    secs = 0.
+    if len(l3) > 2:
+        return None 
+    if len(l3) == 2:
+        last = l3[1]
+        try:
+            secs = float(l3[0])
+            if secs < 0. or secs >= 60.:
+                return None
+        except:
+            return None
+    # last entry should be the direction
+    if typedeg == 'lat':
+        if last == 'N':
+            sign = '+'
+        elif last == 'S':
+            sign = '-'
+        else:
+            return None
+    elif typedeg == 'lon':
+        if last == 'E':
+            sign = '+'
+        elif last == 'W':
+            sign = '-'
+        else:
+            return None
+    else:
+        return None
+
+    return __convert_structure_to_float(sign, degs, mins, secs)
+
+def __convert_float_val(val, typedeg = "lat"):
+    # function converting input to float, recognizing decimal input, or 
+    # degree notation input. Only english input
+    # There is no check on maximum/minimum of degree
+    # In case of degree minutes seconds direction input,
+    # it is checked that degree >0, 0<= minutes <= 60,
+    # 0<= seconds <= 60, direction is in the directions dic.
+
+    #change , to . so that , input works in non , localization
+    #this is no problem, as a number like 100,000.20 cannot appear in 
+    #lat/lon
+    #change XX,YY into XX.YY 
+    if val.find(r'.') == -1 :
+        val = val.replace(u',', u'.')
+    
+    # format: XX.YYYY
+    v = __convert_using_float_repr(val) 
+    if v is not None:
+        return v
+
+    # format: XX:YY:ZZ
+    v = __convert_using_colon_repr(val) 
+    if v is not None : 
+        return v
+
+    # format: XX° YY' ZZ" [NSWE]
+    v = __convert_using_classic_repr(val, typedeg) 
+    if v is not None : 
+        return v
+    
+    # no format succeeded
+    return None
+
 #-------------------------------------------------------------------------
 #
 # conversion function
@@ -97,175 +290,6 @@ def conv_lat_lon(latitude, longitude, format="D.D4"):
                                   and -180   180th meridian
     """
 
-    #-------------------------------------------------------------------------
-    # begin internal function converting one input to float
-    #-------------------------------------------------------------------------
-    def convert_float_val(val, typedeg = "lat"):
-        # function converting input to float, recognizing decimal input, or 
-        # degree notation input. Only english input
-        # There is no check on maximum/minimum of degree
-        # In case of degree minutes seconds direction input,
-        # it is checked that degree >0, 0<= minutes <= 60,
-        # 0<= seconds <= 60, direction is in the directions dic.
-        v = None
-        sign = None
-        secs = None
-        mins = None
-        degs = None
-        error = False
-
-        #change , to . so that , input works in non , localization
-        #this is no problem, as a number like 100,000.20 cannot appear in 
-        #lat/lon
-        try : 
-            test = float('10,11')
-        except ValueError :
-            #change 10,11 into 10.11
-            #if point is already present in val, we can do nothing
-            if val.find(r'.') == -1 :
-                val = val.replace( r',',r'.')
-        try : 
-            test = float('10.11')
-        except ValueError :
-            #change 10.11 into 10,11
-            #if comma is already present in val, we can do nothing
-            if val.find(r',') == -1 :
-                val = val.replace( r'.',r',')
-        
-        try: 
-            v = float(val)      #decimal notation, now float
-        except ValueError:
-            # look for : notation
-            if val.find(r':') != -1 :
-                l = val.split(':')
-                if len(l) < 2 or  len(l) > 3:
-                    error = True
-                l[0]=l[0].strip()
-                # if no characters before ':' nothing useful is input!
-                if len(l[0]) == 0:
-                    return None
-                if l[0][0] == '-':
-                    sign = '-'
-                    l[0]=l[0][1:]
-                else:
-                    sign = '+'
-                try:
-                    degs = int(l[0])
-                    if degs < 0:
-                        error = True
-                except:
-                    error = True
-                try:
-                    mins = int(l[1])
-                    if mins < 0 or mins >= 60:
-                        error = True
-                except:
-                    error = True
-                secs=0.
-                if len(l) == 3:
-                    try:
-                        secs = float(l[2])
-                        if secs < 0. or secs >= 60.:
-                            error = True
-                    except:
-                        error = True
-
-            # if nothing found yet, look for classical notation
-            if val.find(r'_') != -1:
-                error = True   # not a valid lat or lon
-            val = val.replace( r'°',r'_')
-            #allow to input ° as #
-            val = val.replace( r'#',r'_')
-            #allow to input " as ''
-            val = val.replace( r"''",r'"')
-            #allow some special unicode symbols
-            val = val.replace( u'\u2033',r'"')
-            val = val.replace( u'\u2032',r"'")
-            #ignore spaces, a regex with \s* would be better here...
-            val = val.replace(r' ', r'')
-            val = val.replace(r'\t', r'')
-            # get the degrees, must be present to parse as old degree notation
-            if val.find(r'_') != -1:
-                l = val.split('_')
-                if len(l) != 2:
-                    error = True 
-                else:
-                    try: 
-                        degs = int(l[0])  #degrees must be integer value
-                        if degs < 0:
-                            error = True
-                    except:
-                        error = True
-                    # next: minutes might be present once
-                    l2 = l[1].split(r"'")
-                    l3 = l2
-                    mins = 0
-                    if len(l2) > 2:
-                        error = True 
-                    if len(l2) == 2:
-                        l3 = [l2[1],]
-                        try:
-                            mins = int(l2[0]) #minutes must be integer value
-                            if mins < 0 or mins >= 60:
-                                error = True
-                        except:
-                            error = True
-                    # next: seconds might be present once
-                    l3 = l3[0].split(r'"')
-                    last = l3[0]
-                    secs = 0.
-                    if len(l3) > 2:
-                        error = True 
-                    if len(l3) == 2:
-                        last = l3[1]
-                        try:
-                            secs = float(l3[0])
-                            if secs < 0. or secs >= 60.:
-                                error = True
-                        except:
-                            error = True
-                    # last entry should be the direction
-                    last = last.strip()    #remove leading/trailing spaces
-                    if typedeg == 'lat':
-                        if last == 'N':
-                            sign = '+'
-                        elif last == 'S':
-                            sign = '-'
-                        else:
-                            error = True
-                    if typedeg == 'lon':
-                        if last == 'E':
-                            sign = '+'
-                        elif last == 'W':
-                            sign = '-'
-                        else:
-                            error = True
-            # degs should have a value now
-            if degs is None:
-                error = True
-
-        if error:
-            return None
-        if v is not None:
-            return v
-        #we have a degree notation, convert to float
-        v = float(degs) 
-        if secs is not None:
-            v += secs / 3600.
-        if mins is not None:
-            v += float(mins) / 60.
-        if sign =="-":
-            v = v * -1.
-
-        return v
-    #-------------------------------------------------------------------------
-    # end internal function converting one input to float
-    #-------------------------------------------------------------------------
-    
-    #-------------------------------------------------------------------------
-    # begin convert function
-    #-------------------------------------------------------------------------
-    
     # we start the function changing latitude/longitude in english
     if latitude.find('N') == -1 and latitude.find('S') == -1:
         # entry is not in english, convert to english
@@ -277,8 +301,8 @@ def conv_lat_lon(latitude, longitude, format="D.D4"):
         longitude = longitude.replace(translate_en_loc['E'],'E')
 
     # convert to float
-    lat_float = convert_float_val(latitude,  'lat')
+    lat_float = __convert_float_val(latitude,  'lat')
-    lon_float = convert_float_val(longitude, 'lon')
+    lon_float = __convert_float_val(longitude, 'lon')
     
     # give output (localized if needed)
     if lat_float is None or lon_float is None:
@@ -365,7 +389,7 @@ def conv_lat_lon(latitude, longitude, format="D.D4"):
             else:
                 str_lon = ("%d°%02d'%05.2f\"" % (deg_lon, min_lon+1, 0.)) \
                           + dir_lon
-          
+
         return  (str_lat, str_lon)
 
     if format == "DEG-:":
@@ -441,9 +465,6 @@ def conv_lat_lon(latitude, longitude, format="D.D4"):
                 str_lon = sign_lon + \
                           ("%03d%02d%06.3f" % (deg_lon, min_lon+1, 0.))
         return str_lat + str_lon
-    #-------------------------------------------------------------------------
-    # end convert function
-    #-------------------------------------------------------------------------
 
 
 
@@ -486,7 +507,7 @@ if __name__ == '__main__':
     
     lat, lon = '50.849888888888', '2.885897222222'
     test_formats_success(lat,lon)
-    lat, lon = ' 50°50\'59.60"N', '  2°53\'9.23"E'
+    lat, lon = u' 50°50\'59.60"N', u'  2°53\'9.23"E'
     test_formats_success(lat,lon)
     lat, lon = ' 50 : 50 : 59.60 ', ' -2:53 : 9.23   '
     test_formats_success(lat,lon)
@@ -494,24 +515,24 @@ if __name__ == '__main__':
     test_formats_fail(lat,lon)
     lat, lon = ' 50:50: 59.60', '  d u m my'
     test_formats_fail(lat,lon)
-    lat, lon =  ' 50°59.60"N', '  2°53\'E'
+    lat, lon =  u' 50°59.60"N', u'  2°53\'E'
     test_formats_success(lat,lon)
-    lat, lon =  ' 11° 11\' 11" N, 11° 11\' 11" O', ' '
+    lat, lon =  u' 11° 11\' 11" N, 11° 11\' 11" O', ' '
     test_formats_fail(lat,lon)
     # very small negative
     lat, lon =  '-0.00006', '-0.00006'
     test_formats_success(lat,lon)
     # missing direction N/S
-    lat, lon =  ' 50°59.60"', '  2°53\'E'
+    lat, lon =  u' 50°59.60"', u'  2°53\'E'
     test_formats_fail(lat,lon)
     # wrong direction on latitude
-    lat, lon =  ' 50°59.60"E', '  2°53\'N'
+    lat, lon =  u' 50°59.60"E', u'  2°53\'N'
     test_formats_fail(lat,lon)
     # same as above
-    lat, lon =  ' 50°59.99"E', '  2°59\'59.99"N'
+    lat, lon =  u' 50°59.99"E', u'  2°59\'59.99"N'
     test_formats_fail(lat,lon)
     # test precision
-    lat, lon =  ' 50°59.99"S', '  2°59\'59.99"E'
+    lat, lon =  u' 50°59.99"S', u'  2°59\'59.99"E'
     test_formats_success(lat,lon)
     # to large value of lat
     lat, lon =  '90.849888888888', '2.885897222222'
@@ -520,55 +541,55 @@ if __name__ == '__main__':
     lat, lon =  '90', '-180'
     test_formats_success(lat,lon)
     # extreme values allowed
-    lat, lon =  '90° 00\' 00.00" S ', '179° 59\'59.99"W'
+    lat, lon =  u'90° 00\' 00.00" S ', u'179° 59\'59.99"W'
     test_formats_success(lat,lon)
     # extreme value not allowed
-    lat, lon =  '90° 00\' 00.00" N', '180° 00\'00.00" E'
+    lat, lon =  u'90° 00\' 00.00" N', u'180° 00\'00.00" E'
     test_formats_fail(lat,lon)
     # extreme values allowed
     lat, lon =  '90: 00: 00.00 ', '-179: 59:59.99'
     test_formats_success(lat,lon)
     # extreme value not allowed
-    lat, lon =  '90° 00\' 00.00" N', '180:00:00.00'
+    lat, lon =  u'90° 00\' 00.00" N', '180:00:00.00'
     test_formats_fail(lat,lon)
     # extreme values not allowed
     lat, lon =  '90', '180'
     test_formats_fail(lat,lon)
-    lat, lon =  ' 89°59\'60"N', '  2°53\'W'
+    lat, lon =  u' 89°59\'60"N', u'  2°53\'W'
     test_formats_fail(lat,lon)
-    lat, lon =  ' 89°60\'00"N', '  2°53\'W'
+    lat, lon =  u' 89°60\'00"N', u'  2°53\'W'
     test_formats_fail(lat,lon)
-    lat, lon =  ' 89.1°40\'00"N', '  2°53\'W'
+    lat, lon =  u' 89.1°40\'00"N', u'  2°53\'W'
     test_formats_fail(lat,lon)
-    lat, lon =  ' 89°40\'00"N', '  2°53.1\'W'
+    lat, lon =  u' 89°40\'00"N', u'  2°53.1\'W'
     test_formats_fail(lat,lon)
     lat, lon =  '0', '0'
     test_formats_success(lat,lon,
             "Special 0 value, crossing 0-meridian and equator")
     # small values close to equator
-    lat, lon =  ' 1°1"N', '  1°1\'E'
+    lat, lon =  u' 1°1"N', u'  1°1\'E'
     test_formats_success(lat,lon)
     # roundoff
-    lat, lon =  ' 1°59.999"N', '  1°59.999"E'
+    lat, lon =  u' 1°59.999"N', u'  1°59.999"E'
     test_formats_success(lat,lon,'Examples of round off and how it behaves')
-    lat, lon =  ' 1°59\'59.9999"N', '  1°59\'59.9999"E'
+    lat, lon =  u' 1°59\'59.9999"N', u'  1°59\'59.9999"E'
     test_formats_success(lat,lon,'Examples of round off and how it behaves')
-    lat, lon =  '89°59\'59.9999"S', '179°59\'59.9999"W'
+    lat, lon =  u'89°59\'59.9999"S', u'179°59\'59.9999"W'
     test_formats_success(lat,lon,'Examples of round off and how it behaves')
-    lat, lon =  '89°59\'59.9999"N', '179°59\'59.9999"E'
+    lat, lon =  u'89°59\'59.9999"N', u'179°59\'59.9999"E'
     test_formats_success(lat,lon,'Examples of round off and how it behaves')
     #insane number of decimals:
-    lat, lon =  '89°59\'59.99999999"N', '179°59\'59.99999999"E'
+    lat, lon =  u'89°59\'59.99999999"N', u'179°59\'59.99999999"E'
     test_formats_success(lat,lon,'Examples of round off and how it begaves')
     #recognise '' as seconds "
-    lat, lon =  '89°59\'59.99\'\' N', '179°59\'59.99\'\'E'
+    lat, lon =  u'89°59\'59.99\'\' N', u'179°59\'59.99\'\'E'
     test_formats_success(lat,lon, "input \" as ''")
     #test localisation of , and . as delimiter
     lat, lon = '50.849888888888', '2,885897222222'
     test_formats_success(lat,lon, 'localisation of . and , ')
-    lat, lon =  '89°59\'59.9999"S', '179°59\'59,9999"W'
+    lat, lon =  u'89°59\'59.9999"S', u'179°59\'59,9999"W'
     test_formats_success(lat,lon, 'localisation of . and , ')
-    lat, lon =  '89°59\'1.599,999"S', '179°59\'59,9999"W'
+    lat, lon =  u'89°59\'1.599,999"S', u'179°59\'59,9999"W'
     test_formats_fail(lat,lon, 'localisation of . and , ')
     #rest
     lat, lon =  '81.2', '-182.3'
@@ -591,8 +612,10 @@ if __name__ == '__main__':
     test_formats_success(lat,lon)
     lat, lon =  '+50: 0 : 1 : 1', '-2:1:2'
     test_formats_fail(lat,lon)
-    lat, lon = '+61° 43\' 60.00"', '+17° 7\' 60.00"'
+    lat, lon = u'+61° 43\' 60.00"', u'+17° 7\' 60.00"'
     test_formats_fail(lat,lon)
-    lat, lon = '+61° 44\' 00.00"N', '+17° 8\' 00.00"E'
+    lat, lon = u'+61° 44\' 00.00"N', u'+17° 8\' 00.00"E'
     test_formats_success(lat,lon)
+    lat, lon =  ': 0 : 1 : 1', ':1:2'
+    test_formats_fail(lat,lon)