# Ruby / rexml / xpath bug?

**URL:** <https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647>\
**Category:** Ruby\
**Created:** [September 15, 2008, 10:16pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647 "2008-09-15T22:16:02Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 10:16pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/1 "2008-09-15T22:16:02Z")

</div>

I’m doing something I would think is very straight-forward but am  
getting the strangest result. I’m wondering if anyone could help to  
let me know if this is a bug or if I’m doing something I can’t see…

I copied a Google Maps script from Google Maps with Rails and Ajax  
straight from code. It requests a geocoding from Google and parses  
the XML using rexml. When it gets to the address with 2 results, the  
“each” iterator is looping through, but on the second element is  
giving me info from the first. below is the script exactly.

require ‘open-uri’  
require ‘rexml/document’

# Retrieve geocode information for all records in the Stores table

task :google\_geocode =\> :environment do  
api\_key=GOOGLE\_MAPS\_KEY

```
(Store.find :all).each do |store|

  puts "\nStore: #{store.name}"
  puts "Source Address: #{store.full_address}"

  xml=open("http://maps.google.com/maps/geo?

```

q=#{CGI.escape(store.full\_address)}&output=xml&key=#{api\_key}").read  
doc=REXML::Document.new(xml)

```
  puts "Status: "+doc.elements['//kml/Response/Status/code'].text

  if doc.elements['//kml/Response/Status/code'].text != '200'
    puts "Unable to parse Google response for #{store.name}"
  else
    doc.root.each_element('//Response') do |response|
      response.each_element('//Placemark') do |place|
        #there are 2 "Placemarks" in one particular "Response"
        lng,lat=place.elements['Point/

```

coordinates’].text.split(’,’)  
puts " Result Address: " \<\< place.elements[’//  
address’].text  
# the address, lat and lng come from the 1st element twice  
puts " ID: " \<\< place.attributes[‘id’]  
# … but the ‘id’ is correct.  
puts " Latitude: #{lat}"  
puts " Longitude: #{lng}"  
end # end each place  
end # end each response  
end # end if result == 200  
end # end each store  
end # end rake task

##### 

What I don’t understand, is that the XML for each “place” is correct,  
but the place.elements[’//address’] keeps finding the previous  
iterations information. How can this be?

Thank you.

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 10:20pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/2 "2008-09-15T22:20:44Z")

</div>

On 15.09.2008 21:38, altinsel wrote:

> require ‘open-uri’
> 
> ```
> response.each_element('//Placemark') do |place|
> #there are 2 "Placemarks" in one particular "Response"
> lng,lat=place.elements['Point/
> 
> ```
> 
> coordinates’].text.split(’,’)

This seems weird: you search for potential multiple elements but use  
method “text” on the result. Not sure what this is going to yield but  
I’d rather iterate.

> ```
> puts " Result Address: " << place.elements['//
> 
> ```
> 
> address’].text

see above

> ##### 
> 
> What I don’t understand, is that the XML for each “place” is correct,  
> but the place.elements[’//address’] keeps finding the previous  
> iterations information. How can this be?

Hm, weird. Are you sure the XML is as expected?

Btw, you can combine searching for //Response and //Placemark in one  
since you do not do anything with “Response”.

Kind regards

robert

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 11:41pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/3 "2008-09-15T23:41:11Z")

</div>

The funny result is the Ron Jon store in Orange, CA (second from  
last)…

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 11:00pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/4 "2008-09-15T23:00:42Z")

</div>

Thank you for your reply, Robert. Here is the code with output and  
puts of the XML (the Stores are coming from a local database). Does  
this help?

##CODE##

require ‘open-uri’  
require ‘rexml/document’

# Retrieve geocode information for all records in the Stores table

task :google\_geocode =\> :environment do  
api\_key=GOOGLE\_MAPS\_KEY

```
(Store.find :all).each do |store|

  puts "\nStore: #{store.name}"
  puts "Source Address: #{store.full_address}"

  xml=open("http://maps.google.com/maps/geo?

```

q=#{CGI.escape(store.full\_address)}&output=xml&key=#{api\_key}").read  
doc=REXML::Document.new(xml)

```
  puts "Status: "+doc.elements['//kml/Response/Status/code'].text

  if doc.elements['//kml/Response/Status/code'].text != '200'
    puts "Unable to parse Google response for #{store.name}"
  else
    puts doc
    doc.root.each_element('//Response') do |response|
      puts " ****"
      puts response
      puts " ****"
      response.each_element('//Placemark') do |place|
        # puts " ****"
        # puts place.elements['//Point'];
        # puts " ****"
        lng,lat=place.elements['Point/

```

coordinates’].text.split(’,’)  
puts " Result Address: " \<\< place.elements[’//  
address’].text  
puts " ID: " \<\< place.attributes[‘id’]  
puts " Latitude: #{lat}"  
puts " Longitude: #{lng}"  
puts “^^^^^^^^^^^^^^^^^^^^^^^^^”  
end # end each place  
end # end each response  
end # end if result == 200  
end # end each store  
end # end rake task

##OUTPUT##

rake google\_geocode  
(in /Users/acar/Documents/Design/Code/magicmaponrails)

Store: The Original Ron Jon Surf Shop  
Source Address: 901 Central Avenue, Long Beach Island, NJ, 08008  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>901 Central Ave, Beach Haven, NJ 08008, USA\</  
address\>US\</  
CountryNameCode\>NJ\</  
AdministrativeAreaName\>Beach Haven\</  
LocalityName\>901 Central Ave\</  
ThoroughfareName\>08008\</  
PostalCodeNumber\>\</  
Country\>-74.107299,39.756819,0\</  
coordinates\>
* * *

901 Central Avenue, Long Beach Island, NJ, 08008\</  
name\>`200`geocode\</  
Status\>901 Central Ave, Beach Haven, NJ  
08008, USAUS\</  
CountryNameCode\>NJ\</  
AdministrativeAreaName\>Beach Haven\</  
LocalityName\>901 Central Ave\</  
ThoroughfareName\>08008\</  
PostalCodeNumber\>\</  
Country\>-74.107299,39.756819,0\</  
coordinates\>

* * *

Result Address: 901 Central Ave, Beach Haven, NJ 08008, USA  
ID: p1  
Latitude: 39.756819  
Longitude: -74.107299  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: One of aKind Ron Jon Surf Shop  
Source Address: 4151 North Atlantic Avenue, Cocoa Beach, FL, 32931  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>4151 N Atlantic Ave,  
Cocoa Beach, FL 32931, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Cocoa Beach\</  
LocalityName\>4151 N Atlantic Ave\</  
ThoroughfareName\>32931\</  
PostalCodeNumber\>\</  
Country\>-80.608106,28.356747,0\</  
coordinates\>
* * *

4151 North Atlantic Avenue, Cocoa Beach, FL, 32931\</  
name\>`200`geocode\</  
Status\>4151 N Atlantic Ave, Cocoa Beach,  
FL 32931, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Cocoa Beach\</  
LocalityName\>4151 N Atlantic Ave\</  
ThoroughfareName\>32931\</  
PostalCodeNumber\>\</  
Country\>-80.608106,28.356747,0\</  
coordinates\>

* * *

Result Address: 4151 N Atlantic Ave, Cocoa Beach, FL 32931, USA  
ID: p1  
Latitude: 28.356747  
Longitude: -80.608106  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: Ron Jon Surf Shop - Sunrise  
Source Address: 2610 Sawgrass Mills Circle, Sunrise, FL, 33323  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>2610 Sawgrass Mills Cir,  
Fort Lauderdale, FL 33323, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Fort Lauderdale\</  
LocalityName\>2610 Sawgrass Mills Cir\</  
ThoroughfareName\>33323\</  
PostalCodeNumber\>\</  
Country\>-80.315710,26.154207,0\</  
coordinates\>
* * *

2610 Sawgrass Mills Circle, Sunrise, FL, 33323\</  
name\>`200`geocode\</  
Status\>2610 Sawgrass Mills Cir, Fort  
Lauderdale, FL 33323, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Fort Lauderdale\</  
LocalityName\>2610 Sawgrass Mills Cir\</  
ThoroughfareName\>33323\</  
PostalCodeNumber\>\</  
Country\>-80.315710,26.154207,0\</  
coordinates\>

* * *

Result Address: 2610 Sawgrass Mills Cir, Fort Lauderdale, FL 33323,  
USA  
ID: p1  
Latitude: 26.154207  
Longitude: -80.315710  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: Ron Jon Surf Shop - Orlando  
Source Address: 5160 International Drive, Orlando, FL, 32819  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>5160 International Dr,  
Orlando, FL 32819, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Orlando\</  
LocalityName\>5160 International Dr\</  
ThoroughfareName\>32819\</  
PostalCodeNumber\>\</  
Country\>-81.450308,28.469598,0\</  
coordinates\>
* * *

5160 International Drive, Orlando, FL, 32819\</  
name\>`200`geocode\</  
Status\>5160 International Dr, Orlando, FL  
32819, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Orlando\</  
LocalityName\>5160 International Dr\</  
ThoroughfareName\>32819\</  
PostalCodeNumber\>\</  
Country\>-81.450308,28.469598,0\</  
coordinates\>

* * *

Result Address: 5160 International Dr, Orlando, FL 32819, USA  
ID: p1  
Latitude: 28.469598  
Longitude: -81.450308  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: Ron Jon Surf Shop - Key West  
Source Address: 503 Front Street`, Key West, FL, 33040  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>503 Front St, Key West, FL 33040,  
USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Key West\</  
LocalityName\>503 Front St\</  
ThoroughfareName\>33040\</  
PostalCodeNumber\>\</  
Country\>-81.805852,24.560319,0\</  
coordinates\>
* * *

503 Front Street`, Key West, FL, 33040\</  
name\>`200`geocode\</  
Status\>503 Front St, Key West, FL 33040,  
USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Key West\</  
LocalityName\>503 Front St\</  
ThoroughfareName\>33040\</  
PostalCodeNumber\>\</  
Country\>-81.805852,24.560319,0\</  
coordinates\>

* * *

Result Address: 503 Front St, Key West, FL 33040, USA  
ID: p1  
Latitude: 24.560319  
Longitude: -81.805852  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: Ron Jon Surf Shop - California  
Source Address: 20 City Blvd., Orange, CA, 92868  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>20 City Blvd W, Orange, CA 92868,  
USAUS\</  
CountryNameCode\>CA\</  
AdministrativeAreaName\>Orange\</  
LocalityName\>20 City Blvd W\</  
ThoroughfareName\>92868\</  
PostalCodeNumber\>\</  
Country\>-117.892764,33.782893,0\</  
coordinates\>20 City  
Blvd E, Orange, CA 92868, USAUS\</  
CountryNameCode\>CA\</  
AdministrativeAreaName\>Orange\</  
LocalityName\>20 City Blvd E\</  
ThoroughfareName\>92868\</  
PostalCodeNumber\>\</  
Country\>-117.891102,33.784368,0\</  
coordinates\>
* * *

20 City Blvd., Orange, CA, 92868\</  
name\>`200`geocode\</  
Status\>20 City Blvd W, Orange, CA 92868,  
USAUS\</  
CountryNameCode\>CA\</  
AdministrativeAreaName\>Orange\</  
LocalityName\>20 City Blvd W\</  
ThoroughfareName\>92868\</  
PostalCodeNumber\>\</  
Country\>-117.892764,33.782893,0\</  
coordinates\>20 City  
Blvd E, Orange, CA 92868, USAUS\</  
CountryNameCode\>CA\</  
AdministrativeAreaName\>Orange\</  
LocalityName\>20 City Blvd E\</  
ThoroughfareName\>92868\</  
PostalCodeNumber\>\</  
Country\>-117.891102,33.784368,0\</  
coordinates\>

* * *

Result Address: 20 City Blvd W, Orange, CA 92868, USA  
ID: p1  
Latitude: 33.782893  
Longitude: -117.892764  
^^^^^^^^^^^^^^^^^^^^^^^^^  
Result Address: 20 City Blvd W, Orange, CA 92868, USA  
ID: p2  
Latitude: 33.784368  
Longitude: -117.891102  
^^^^^^^^^^^^^^^^^^^^^^^^^

Store: Ron Jon Cape Caribe Resort  
Source Address: 1000 Shorewood Drive, Cape Canaveral, FL, 32920  
Status: 200

\<?xml version='1.0' encoding='UTF-8'?\>1000 Shorewood Dr, Cape  
Canaveral, FL 32920, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Cape Canaveral\</  
LocalityName\>1000 Shorewood Dr\</  
ThoroughfareName\>32920\</  
PostalCodeNumber\>\</  
Country\>-80.598021,28.405217,0\</  
coordinates\>
* * *

1000 Shorewood Drive, Cape Canaveral, FL, 32920\</  
name\>`200`geocode\</  
Status\>1000 Shorewood Dr, Cape Canaveral,  
FL 32920, USAUS\</  
CountryNameCode\>FL\</  
AdministrativeAreaName\>Cape Canaveral\</  
LocalityName\>1000 Shorewood Dr\</  
ThoroughfareName\>32920\</  
PostalCodeNumber\>\</  
Country\>-80.598021,28.405217,0\</  
coordinates\>

* * *

Result Address: 1000 Shorewood Dr, Cape Canaveral, FL 32920, USA  
ID: p1  
Latitude: 28.405217  
Longitude: -80.598021  
^^^^^^^^^^^^^^^^^^^^^^^^^

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 11:42pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/5 "2008-09-15T23:42:22Z")

</div>

I’m surprised it works at all with the XPaths all being absolute like  
that. An XPath expression starting with ‘//’ means start at the  
document root. My guess is that you are getting both addresses each  
time, and the call to text() just prints the first one.

Try ‘.//address’ or just ‘address’ instead.

– Mark.

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 16, 2008, 8:40am UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/6 "2008-09-16T08:40:46Z")

</div>

On 15.09.2008 23:48, altinsel wrote:

> Mark,
> 
> Thank you. The ‘.//’ syntax seems to work. (I’m very happy about  
> that).

I’d still rather combine the two into a single iteration.

> Now if you don’t mind I’d like to ask: as far as I understand it, the  
> “each\_element” method should only be plucking a specific branch  
> (“place”) off the tree, and that branch is a new, stand-alone tree,  
> and shouldn’t even have the first element’s data, let alone the rest  
> of the tree. I even So how was that information even getting in  
> there?

[http://www.w3schools.com/xpath/xpath\_syntax.asp](http://www.w3schools.com/xpath/xpath_syntax.asp)

Basically a node knows about the whole document (it knows its parent as  
well). So you cannot only look down but up as well. This is not too  
unusual (just think about filesystems).

Cheers

robert

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 16, 2008, 6:17pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/7 "2008-09-16T18:17:13Z")

</div>

On Sep 15, 5:48 pm, altinsel [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> there?  
> In XPath, the prefix ‘//’ means start from the top of the tree, not  
> the ‘current’ node or “branch”.

And I agree with Robert, the code is far from ideal; there is far too  
much iteration. The code could be much simpler and more efficient. See  
my diatribe on this here:  
[http://markthomas.org/2008/08/22/eschew-xpathic-iteration/](http://markthomas.org/2008/08/22/eschew-xpathic-iteration/)

– Mark.

---

<div class="post-metadata">

**Author:** ![altinsel](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/c0e974/32.png) [@altinsel](https://www.ruby-forum.com/u/altinsel)\
**Post date:** [September 15, 2008, 11:50pm UTC](https://www.ruby-forum.com/t/ruby-rexml-xpath-bug/148647/8 "2008-09-15T23:50:34Z")

</div>

Mark,

Thank you. The ‘.//’ syntax seems to work. (I’m very happy about  
that).

Now if you don’t mind I’d like to ask: as far as I understand it, the  
“each\_element” method should only be plucking a specific branch  
(“place”) off the tree, and that branch is a new, stand-alone tree,  
and shouldn’t even have the first element’s data, let alone the rest  
of the tree. I even So how was that information even getting in  
there?