SikuliX

No parsers found error when running from command line

Asked by Sunil

Hello there,

I've been trying to parse an xml file and write selected nodes to a text file. If I run the script from within Sikuli IDE, everything works fine. The run produces expected xmltest.txt file into the current directory.
The problem arises when I run the same script from command line., it throws No parsers found error.

I run it as,
D:\INSTALLER\sikuli-script.cmd -r D:\test.sikuli

The error is,
+++ trying to start Sikuli Script
[error] Script aborted with some error:
Traceback (most recent call last):
  File "D:\test.py", line 22, in <module>
    t1 = parse_xml(file_name)
  File "D:\test.py", line 15, in parse_xml
    dom = read_xml(file_name)
  File "D:\test.py", line 10, in read_xml
    pdata = parseString(data)
  File "D:\INSTALLER\sikuli-script.jar\Lib\xml\dom\minidom.py", line 1934, in parseString
  File "D:\INSTALLER\sikuli-script.jar\Lib\xml\dom\minidom.py", line 1907, in _do_pulldom_parse
  File "D:\INSTALLER\sikuli-script.jar\Lib\xml\dom\pulldom.py", line 351, in parseString
  File "D:\INSTALLER\sikuli-script.jar\Lib\xml\sax\__init__.py", line 94, in make_parser
xml.sax._exceptions.SAXReaderNotAvailable: No parsers found

Sikuli Script:
-------------
import os
from xml.dom.minidom import parseString
cd = os.path.dirname(os.path.abspath(sys.argv[0]))
file_name = cd+"\\test.xml"

# read xml
def read_xml(xml_file):
    f = open(xml_file, 'r')
    data = f.read()
    pdata = parseString(data)
    f.close()
    return pdata

def parse_xml(file_name):
    dom = read_xml(file_name)
    node_list = dom.getElementsByTagName('node')
    first_node = node_list[0].firstChild.nodeValue
    return first_node

fo = open(cd+'\\xmltest.txt', 'a')
fo.write("\nFirst node value\n")
t1 = parse_xml(file_name)
fo.write(str(t1))
fo.close()

XML file:
----------
<?xml version="1.0"?>
<test>
 <tag1>
  <node>VALUE</node>
 </tag1>
</test>

Output xmltest.txt file, if run through Sikuli IDE,
-----------------
First node value
VALUE

Occasionally, the IDE also throws same parser not found error that can be rid of by clearing all sikuli temp data.

Question information

Language:
English Edit question
Status:
Solved
For:
SikuliX Edit question
Assignee:
RaiMan Edit question
Solved by:
RaiMan
Solved:
Last query:
Last reply:
Revision history for this message
RaiMan (raimund-hocke) said : 		#1

This worked for me in all cases:

from xml.dom.minidom import parseString
xml = '<?xml version="1.0"?><test><tag1><node>VALUE</node></tag1></test>'
dom = parseString(xml)
node_list = dom.getElementsByTagName('node')
first_node = node_list[0].firstChild.nodeValue
print first_node

Revision history for this message
Sunil (snakum) said : 		#2

Thanks, RaiMan.
I hope, you also tried it with a separate test.xml file rather than defining xml tags into the sikuli script itself.
Also it's essential to run the script in command line mode, else the error may not come (though, occasionally it comes).
I've been having a hard time getting rid of this error, without any luck.

Revision history for this message
RaiMan (raimund-hocke) said : 		#3

--1. I surely ran it from command line using sikuli-script.jar.

--2. my test proves, that the problem is not the xml-parsing stuff as such. If the string to parse is "readable" it works.

So, yes, the problem might be with the file containing the xml.
Might be an encoding problem: take care, that your file is stored as utf-8 and read it using utf-8 encoding.

Revision history for this message
Sunil (snakum) said : 		#4

Hi Raimund,
I did found a workaround to avoid the parser issue in my case, but the original issue persists and I'm bit curious to find out the root-cause. If following information makes any sense to probe into,

1. While running the script in command-line mode as "script-name.sikuli" (i.e. without exclusively specifying the jython) the error comes.
example: D:\INSTALLER\sikuli-script.cmd -r D:\Scripts\script-name.sikuli

2. While running the same script but as python script with sikuli-script.jar and jython, it works.
example: java -cp sikuli-script.jar org.python.util.jython D:\Scripts\script-name.sikuli\script-name.py

I'm using a Win7 x64 machine with Java7 x64, with home variables set for both and included into system path too.

Thanks,

Revision history for this message
RaiMan (raimund-hocke) said : 		#5

linked this question to bug 1210474, to have it together

Revision history for this message
Sunil (snakum) said : 		#6

From the bug 1210474 discussion,

"a Sikuli script should not be named as a Python module, when this Python module should be imported."

This is exactly what I did, imported a module which is saved as .sikuli script for parsing an xml. I tried the "from modname import this" or "from modname import *" but that didn't help. Isn't it conventional to import python modules with the file name. Will you please suggest a better/standard way to avoid such issues?

Revision history for this message
Best RaiMan (raimund-hocke) said : 		#7

--- Isn't it conventional to import python modules with the file name?
Of course it is.
But that is not the question here.

Sikuli as a convenience feature allows to import other .sikuli scripts as "modules". Together with this convenience, Sikuli adds the .sikuli folder to the image path, so you do not need to handle the submodule images separately.

This is implemented internally as a hook into the Jython import process.

lets make an example:
You made submodule.sikuli, that contains some supporting functions to be used all over the place.

the usage would be e.g.
from submodule import *

This looks exactly like a normal Python import, that should find and import a file named submodule.py from sys.path

But since we are using Sikuli to run our scripts:
- Sikuli first searches for a submodule.sikuli on sys.path.
- then it tells the Jython import, to import the contained submodule.py
- and it adds submodule.sikuli to Sikuli's image path

If Sikuli does not find a submodule.sikuli, it delegates the further processing to Jython import, which might or might not find a submodule.py

In your case you did not even name a submodule like a Jython module to import (xml), but the main script was named xml.sikuli. But since the working folder (the folder where the script run started from) is always on sys.path for Sikuli imports, the xml.sikuli is found first (and of course does not contain the expected substructure) and hence hides the wanted Jython module xml.

So a bit more precise the rule should read:
"A Sikuli script, that can be found on sys.path at runtime, should not be named as a Python module, that is to be imported in the same script run."

To avoid such clashes, I use script names like testSomefeature.sikuli, when I want to try out a Python/Jython module somefeature.py.

Revision history for this message
RaiMan (raimund-hocke) said : 		#8

is solved

Revision history for this message
Sunil (snakum) said : 		#9

Hi Raiman,

Thank you for the detailed suggestion.
I really appreciate your efforts put on Sikuli project.

Though I'm yet to try it, one interesting fact in my case is I haven't named any script/module as xml.sikuli or xml.py. All I meant by "imported a 'module' which is saved as .sikuli script for parsing xml file" is any user defined module name (not native python module name, as you mentioned xml.sikuli).

I'll update my findings here once done with it.
Thanks,

Revision history for this message
Sunil (snakum) said : 		#10

Finally, the issue got resolved from the clues RaiMan posted on linked bug 1210474.

"a Sikuli script should not be named as a Python module, when this Python module should be imported."

I didn't had any script with name "xml.sikuli" but was not noticing the another script having name with another Python module "struct.sikuli". The xml parser script was running successfully if placed into some other location, but the moment it put into the library of all other scripts (where the culprit struct.sikuli was lying) it started throwing Parser not found error.

Even if you create a dummy directory with this name (struct.sikuli) the xml parsing script (as posted into initial post) will fail with parser not found error, which otherwise works fine!

Thanks for all your support, it got resolved finally.

Revision history for this message
Sunil (snakum) said : 		#11

Thanks RaiMan, that solved my question.