Q & A with wolframalpha and duckduckgo

[Hellow]

I have made a python script that uses a wolframalphas https://bitbucket.org/jaraco/wolframalpha and duckduckgos api https://pypi.python.org/pypi/duckduckgo2. The script first checks if wa has a answer (best solution), it not it uses duckduckgo, if that fails aswell vox tells your that is it couldnt find any answer to (Your question), Please rephrase.

I take no credit for making the python apis. All credit goes to the original authors. However you do need to download the files i provided, some are slightly changes in order to make it work with VC

1.Go register on the link below to get a apikey to wa (Please, do not use the one included. It's just for testing purpose)
https://developer.wolframalpha.com/portal/apisignup.html

2. Download duckduckgo.py and wolframalpha.py attached to the post. Move them to \plugins\PY\Lib

3. Download tellme.py and move it to \plugins\PY

4. Paste the group command to your lgb.

5. Do a full restart

6. Take it for a spin. Test it with, tell me "your question"

I have only tested with english

[jitterjames]

Looks like a good start.  Unfortunately I get a lot of errors for questions that Wolfram should be able to answer.

Here is a snippet of my log.  As you can see the question is simply: 'what is the distance from montreal to New York', which I know Wolfram can answer.  Even if it can't I think it should be passing the question on to duck duck rather than failing on the error.

How much did you actually test this?  Did you try a variety of questions?

Code: [Select]

1/1/2014 10:03:52 AM 746 [action] PY.ExecString:tellme('what is the distance from montreal to New York')

1/1/2014 10:03:56 AM 585 [plugin] Python Error:IronPython.Runtime.UnboundLocalException: Local variable 'failed_answer' referenced before assignment.
   at IronPython.Runtime.Operations.PythonOps.CheckUninitialized(Object value, String name)
   at Microsoft.Scripting.Interpreter.FuncCallInstruction`3.Run(InterpretedFrame frame)
   at Microsoft.Scripting.Interpreter.Interpreter.HandleException(InterpretedFrame frame, Exception exception)
   at Microsoft.Scripting.Interpreter.Interpreter.Run(InterpretedFrame frame)
   at Microsoft.Scripting.Interpreter.LightLambda.Run2[T0,T1,TRet](T0 arg0, T1 arg1)
   at IronPython.Runtime.FunctionCaller`1.Call1(CallSite site, CodeContext context, Object func, T0 arg0)
   at System.Dynamic.UpdateDelegates.UpdateAndExecute3[T0,T1,T2,TRet](CallSite site, T0 arg0, T1 arg1, T2 arg2)
   at Microsoft.Scripting.Interpreter.DynamicInstruction`4.Run(InterpretedFrame frame)
   at Microsoft.Scripting.Interpreter.Interpreter.Run(InterpretedFrame frame)
   at Microsoft.Scripting.Interpreter.LightLambda.Run2[T0,T1,TRet](T0 arg0, T1 arg1)
   at IronPython.Compiler.PythonScriptCode.RunWorker(CodeContext ctx)
   at IronPython.Compiler.PythonScriptCode.Run(Scope scope)
   at IronPython.Compiler.RuntimeScriptCode.InvokeTarget(Scope scope)
   at IronPython.Compiler.RuntimeScriptCode.Run(Scope scope)
   at Microsoft.Scripting.Hosting.CompiledCode.Execute(ScriptScope scope)
   at vcPlugin.iPy.execString(String strCode)


[jitterjames]

FYI, The same question works fine with the non-python version that we previously posted to the forum:

Code: [Select]

<?xml version="1.0" encoding="utf-16"?>
<command id="612" name="wolfram" enabled="true" alwaysOn="False" confirm="False" requiredConfidence="0" loop="False" loopDelay="0" loopMax="0" description="">
  <action>
    <cmdType>Scrape</cmdType>
    <cmdString>http://api.wolframalpha.com/v2/query?appid=[yourAPIkey]&amp;input={1}&amp;format=plaintext</cmdString>
    <cmdRepeat>1</cmdRepeat>
  </action>
  <action>
    <cmdType>Results.RegExSingle</cmdType>
    <cmdString>&lt;plaintext&gt;([^&lt;].*?)&lt;/plaintext&gt;</cmdString>
    <cmdRepeat>1</cmdRepeat>
  </action>
  <if ifBlockDisabled="False" ifNot="False">
    <ifType>LastActionSuccess</ifType>
    <ifParams>&amp;&amp;</ifParams>
    <then>
      <action>
        <cmdType>Results.MatchConcat</cmdType>
        <cmdString>{CR}&amp;&amp;3</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString>&amp;quot;&amp;&amp;"</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString>&amp;apos;&amp;&amp;'</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString> |</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>OSD.ShowText</cmdType>
        <cmdString>{LastResult}</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>TTS.Speak</cmdType>
        <cmdString>{LastResult}</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
    </then>
    <else>
      <action>
        <cmdType>TTS.Speak</cmdType>
        <cmdString>I'm sorry, I don't know the answer to: {1} Could you speak more clearly, or perhaps try to rephrase your question.</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>OSD.ShowText</cmdType>
        <cmdString>I'm sorry, I don't know the answer to: {CR}{1}{CR}Could you speak more clearly, or perhaps try to rephrase your question.</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
    </else>
  </if>
  <phrase>Wolfram, Winklethorpe, Hey Jeeves, Dearie, Can you tell me</phrase>
  <payloadDictation>payloadDictation: Regular</payloadDictation>
</command>

[Hellow]

I tested it alot, in cpython. the same question works with the same code in cpython(with print statements instead of voxcalls). This just makes me insane.. Ill see what i can do.

[nime5ter]

You can't define a variable in your 'try' block and then use it in the 'except' block.  The whole point of the code in the 'except' is that the 'try' failed so you can't assume that your variable was defined correctly.

[Hellow]

Yeah, i saw that from the log. The problem seems to be with the answer, Action:  OSD.ShowText - Error:: Error('Syntax error at line 82: illegal character in content',)&&10000&&-10. The answer is encoded from wa as utf-8. <plaintext>5.403×10^7 cm  (centimeters)</plaintext>. The code works perfect in cpython. I have no ide how to fix this error. Only thing i can find is this http://support.microsoft.com/kb/238833

[jitterjames]

The error has nothing to do with the TTS Action.  I believe it is being thrown based entirely on what is coming back from wolfram.  For some reason the wolframalpha.py is unable to process what it is getting back from wolfram.  It seems to have something to do with results that have numbers in them.

This simple code will throw the error :

Code: [Select]

import wolframalpha

def tellme(s=''):
    client = wolframalpha.Client('XXXXXX-XXXXXX')
    res = client.query(s)
    for p in res.pods:
        print p.text


tellme("what is 3 times 6")
The error occurs on the line:

Code: [Select]

res = client.query(s)Changing the question will result in either, getting a correct response, or getting an error on a different line number.  I think the line number actually refers to the line of text that is being returned by wolfram.

I have no idea how to fix it either.

[Hellow]

I was editing the post at the same time as you wrote your. So far it seems that the parser has problems with × (u'00D7'). Ill see if i can replace it with x before document gets parsed.

[Hellow]

The scripts now works thanks to Simon Opelt. It was a bug in ironpython, but he provided a workaround.

[jitterjames]

Thanks, but, provided it where?

[Hellow]

http://stackoverflow.com/questions/21586438/ironpython-elementtree-illegal-character-in-content

[Etrust]

Hi Guys,

I would like to use Wolframalpha in combination with voxcommand but I am a little bit confused..

I have allready an API Key and I would like to know the steps exactly to make wolframalpha work with voxcommand. The post from the author is old and I would like to know where I can find the steps to setup.

Can somebody help?

I am loving voxcommand. My wife can not hear my tests anymore :=)))))

[Hellow]

Have you tried the step by step instruction? I posted?

[Etrust]

Have you tried the step by step instruction? I posted?

Hi Hellow!!

not yet because james have reported some errors and I was not sure if this is the latest versoin and tutorial 

But if you can confirm that it will work using the version 2.1 than I will try

Are you able to make questions and are you able to have the answer with TTS? Or have you a video on youtube?

[Hellow]

Been a while since I have used it. But when it worked I could ask questions and get a reply from vox.
There is no video.

[jitterjames]

@Hellow:

You mentioned a fix near the middle/end of this thread.

Did you already apply the fix to the code you posted at the beginning of this thread?

@Etrust:

You can also try the non-Python version that I posted above.  I believe it also works fine: http://voxcommando.com/forum/index.php?topic=1319.msg11344#msg11344

[Etrust]

@Hellow:

You mentioned a fix near the middle/end of this thread.

Did you already apply the fix to the code you posted at the beginning of this thread?

@Etrust:

You can also try the non-Python version that I posted above.  I believe it also works fine: http://voxcommando.com/forum/index.php?topic=1319.msg11344#msg11344

Hi James,

Yes I will. have I understand that your non-Phyton version I have only to import the xml you have posted and modify my api key and that's it? Or are there more steps to do?

[nime5ter]

Hi Etrust,

Yes, you've understood correctly. James's command does not use the Python plugin. You just need to insert your own API key in his XML.

I think the forum has several Wolfram Alpha solutions posted in various threads, so it's confusing. But all of them probably work reasonably well, and most of them are similar.

In this post there's a link to a video where you can see one of the versions working: http://voxcommando.com/forum/index.php?topic=1463.msg12773#msg12773

The main limitation is that we have to use a payload dictation if we want to be able to ask any question. For this reason, I find it works better to have the following settings in my Options: 1. "Show OSD Alternates" selected, and 2. "Don't execute if #Alternates >=2". This way, VoxCommando will not automatically try to use the "most likely" result that it hears, and will instead show me a list of possibilities to choose from, as demonstrated in the video.

[Etrust]

Hi Etrust,

Yes, you've understood correctly. James's command does not use the Python plugin. You just need to insert your own API key in his XML.

I think the forum has several Wolfram Alpha solutions posted in various threads, so it's confusing. But all of them probably work reasonably well, and most of them are similar.

In this post there's a link to a video where you can see one of the versions working: http://voxcommando.com/forum/index.php?topic=1463.msg12773#msg12773

The main limitation is that we have to use a payload dictation if we want to be able to ask any question. For this reason, I find it works better to have the following settings in my Options: 1. "Show OSD Alternates" selected, and 2. "Don't execute if #Alternates >=2". This way, VoxCommando will not automatically try to use the "most likely" result that it hears, and will instead show me a list of possibilities to choose from, as demonstrated in the video.

Hi nime5ster!!

Yes I have found more posts and was a little bit confused 
Great! I will try your version and jame's version soon as I arrive home today. My idea is to setup voxcommando with wolframalpha using a very very small pc running win 8.1 in my livingroom without monitor, mouse and keyboard. So I will need a solution to have an answer without to choose one from the list because I will not have a monitor and will not have access to the list. It's possible to let vocommando choose the first answer automaticcaly for me?

[nime5ter]

Ha. Wow. I think you are being very optimistic, particularly if you are speaking English with a Swiss accent -- but it can't hurt to try!

Yes, by default, VoxCommando will execute the most likely phrase heard, even if it also interprets alternative possibilities.

It is only if you choose otherwise, that it will not do that.

I recommend reading through the "OSD Alternates options" section of the wiki: http://voxcommando.com/mediawiki/index.php?title=Options

[Etrust]

Ha. Wow. I think you are being very optimistic, particularly if you are speaking English with a Swiss accent -- but it can't hurt to try!

Yes, by default, VoxCommando will execute the most likely phrase heard, even if it also interprets alternative possibilities.

It is only if you choose otherwise, that it will not do that.

I recommend reading through the "OSD Alternates options" section of the wiki: http://voxcommando.com/mediawiki/index.php?title=Options

Hi nime5ster!!

  absolutely! I have definitly a swiss accent but for now voxcommando was able to understand me  James did a great accent recgonizer 

Yes the wiki confirmed that VC will choose automatically the most like answer: "Note that, by default -- even if it recognizes possible alternates -- VC will automatically execute the most likely command (i.e. Option 1) after several seconds"

I have not found a possibility to change the seconds to 0 or to ignore all options and choose emmadiately the most likely answer and not wait several seconds.

[jitterjames]

Actually the wiki was slightly incorrect and I just updated it.  It should just say:

Note that, by default -- even if it recognizes possible alternates -- VC will automatically execute the most likely command (i.e. Option 1)

That is, it will always execute option 1 immediately when you are using the default settings.

[jitterjames]

However, you will most likely find that free dictation is far less accurate than commands that do not use dictation.  Even with a relatively plain English accent it will often get my dictation slightly wrong, and I find I need to use alternates for Wolfram (or other dictation based) commands to be useful.

[Etrust]

FYI, The same question works fine with the non-python version that we previously posted to the forum:

Code: [Select]

<?xml version="1.0" encoding="utf-16"?>
<command id="612" name="wolfram" enabled="true" alwaysOn="False" confirm="False" requiredConfidence="0" loop="False" loopDelay="0" loopMax="0" description="">
  <action>
    <cmdType>Scrape</cmdType>
    <cmdString>http://api.wolframalpha.com/v2/query?appid=YOUR_API_KEY&amp;input={1}&amp;format=plaintext</cmdString>
    <cmdRepeat>1</cmdRepeat>
  </action>
  <action>
    <cmdType>Results.RegExSingle</cmdType>
    <cmdString>&lt;plaintext&gt;([^&lt;].*?)&lt;/plaintext&gt;</cmdString>
    <cmdRepeat>1</cmdRepeat>
  </action>
  <if ifBlockDisabled="False" ifNot="False">
    <ifType>LastActionSuccess</ifType>
    <ifParams>&amp;&amp;</ifParams>
    <then>
      <action>
        <cmdType>Results.MatchConcat</cmdType>
        <cmdString>{CR}&amp;&amp;3</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString>&amp;quot;&amp;&amp;"</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString>&amp;apos;&amp;&amp;'</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>Results.Replace</cmdType>
        <cmdString> |</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>OSD.ShowText</cmdType>
        <cmdString>{LastResult}</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>TTS.Speak</cmdType>
        <cmdString>{LastResult}</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
    </then>
    <else>
      <action>
        <cmdType>TTS.Speak</cmdType>
        <cmdString>I'm sorry, I don't know the answer to: {1} Could you speak more clearly, or perhaps try to rephrase your question.</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
      <action>
        <cmdType>OSD.ShowText</cmdType>
        <cmdString>I'm sorry, I don't know the answer to: {CR}{1}{CR}Could you speak more clearly, or perhaps try to rephrase your question.</cmdString>
        <cmdRepeat>1</cmdRepeat>
      </action>
    </else>
  </if>
  <phrase>Wolfram, Winklethorpe, Hey Jeeves, Dearie, Can you tell me</phrase>
  <payloadDictation>payloadDictation: Regular</payloadDictation>
</command>

Hi James and Guys,

I have imported this script to my VC. I can see the payloadDictation: Regular in red (please check screen). It seems that dictation is not supported in SP Modus. Is that true? Will I have to switch to the no SP mode?

[Kalle]

Yes, you can not use payload dictation with SP Version.

Each engine has advantages and disadvantages. Pay attention to features *and* to limitations (e.g., the new engine does not support open dictation or training, but works with bluetooth microphones).

[Etrust]

Hi kalle,

ok thanks for the confirmation. So there is no way to use wolfram with SP mode   

[Kalle]

No, not with dictation.

[nime5ter]

Hi kalle,

ok thanks for the confirmation. So there is no way to use wolfram with SP mode   

Just a reminder that our wiki documentation is the place to start, when looking for answers. http://voxcommando.com/mediawiki/index.php?title=Version_2#Choice_of_speech_engines

@eTrust -- Now that you mention it, I have been wondering why you are using SP rather than the other engine. Are you using a bluetooth microphone?

Generally, SP is an important option for users who don't speak one of the standard 8 languages that come with Windows, or users who want to use a bluetooth microphone. Otherwise, the SAPI 5 speech engine is preferable.

If you are running a Windows 8 machine with a German OS, it is possible to install both German and English language packs.

[jitterjames]

Why do you feel that you need to use SP mode anyway?  If you are using a good quality microphone and using English phrases in your commands then the regular mode with some training is probably a better choice all around.

[Etrust]

Hi nime5ter Hi James!

I speak portuguese (brazil) and would like to use it in VC. BUT for the moment I am speaking only english (with a swiss accent hehe) with VC AND Wolfram does not work in portuguese. So it is maybe "for the moment" a good idea to switch to the no sp mode. You are right.

[jitterjames]

I suggest you use two separate installations and configurations for each.

One using SP for your PT-BR configuration
One using non - SP / English configuration

[Etrust]

I suggest you use two separate installations and configurations for each.

One using SP for your PT-BR configuration
One using non - SP / English configuration

Ok I will do that. Thank you guys!