Re: [IronPython] InteractiveCode and function definitions
IncompleteStatement means that the user is allowed to type more code. If you want to know whether or not it's a valid (complete) string, just check for it not being Invalid. A function definition is never complete in Python because there's never a terminating curly brace :). On Thu, Apr 16, 2009 at 10:05 AM, Michael Foord fuzzy...@voidspace.org.ukwrote: Hello guys, We're trying to detect whether a section of code is complete (to mimic the behaviour of the interactive interpreter). First of all we tried using the Python standard library code module which provides interactive console classes. There are two outstanding bugs on codeplex (one reported by me today) which prevent this being an ideal solution: http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=22064 http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=21881 The second approach was to create a ScriptSource and looking at the code properties to tell if the statement is complete or not (using IronPython 2.0.1). However we can never get it to return a ScriptParseResult.Complete for function definitions. Code below shows using \n for newlines but we have also tried with \r\n. import clr clr.AddReference('IronPython') clr.AddReference('Microsoft.Scripting') from IronPython.Hosting import Python from Microsoft.Scripting import SourceCodeKind, ScriptCodeParseResult engine = Python.CreateEngine() s = engine.CreateScriptSourceFromString('def f():\n print 1\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x003F [IncompleteStatement] s = engine.CreateScriptSourceFromString('def f():\n print 1\n\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x0040 [IncompleteStatement] The DLR hosting spec has little helpful to say on the matter as far as I can tell. Looking at an example from Tomas it doesn't seem very different from what we're doing: http://blog.tomasm.net/2009/04/15/python-says-hello-to-ruby/ Any clues as to what we are doing wrong or how to procede? Thanks Michael -- http://www.ironpythoninaction.com/ http://www.voidspace.org.uk/blog ___ Users mailing list Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com ___ Users mailing list Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com
Re: [IronPython] InteractiveCode and function definitions
That's only because you've decided to arbitrarily[*] define \n\n as being a signal to mean complete. That's not part of the actual language specification. In fact, if I append \n\n print 2\n\n to that string, it's still a valid Python program. The key here is that the user has entered a complete thought is a property of the interpreter and not of the language. I might well decide that the commit key sequence is Control+E (as it is in SQL Server Management Studio) instead of enter enter. My point is that it's not correct for IronPython to dictate the semantics of your interpreter. [*] Okay, arbitrary is a bit strong in that it's what python.exe and ipy.exe defines. :) On Thu, Apr 16, 2009 at 10:36 AM, Michael Foord fuzzy...@voidspace.org.ukwrote: Curt Hagenlocher wrote: IncompleteStatement means that the user is allowed to type more code. If you want to know whether or not it's a valid (complete) string, just check for it not being Invalid. A function definition is never complete in Python because there's never a terminating curly brace :). But that isn't sufficient to implement an interactive interpreter on top of. This code conceptually is complete as far as an interactive interpreter is concerned: 'def f():\n print 1\n\n' It also means you can't distinguish between the previous kind of incomplete (which is incomplete because the user *could* type more code) and this kind of incomplete: 'a = ' or: 'a = (1 + 2 +' Which are both incomplete because the user *must* type more code. (Although the latter two give IncompleteToken - I wonder if that would be enough.) Because of the other IronPython bugs we can't use the code module and ScriptSource / ScriptParseResult doesn't give sufficient information. Any other ideas? Michael On Thu, Apr 16, 2009 at 10:05 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Hello guys, We're trying to detect whether a section of code is complete (to mimic the behaviour of the interactive interpreter). First of all we tried using the Python standard library code module which provides interactive console classes. There are two outstanding bugs on codeplex (one reported by me today) which prevent this being an ideal solution: http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=22064 http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=21881 The second approach was to create a ScriptSource and looking at the code properties to tell if the statement is complete or not (using IronPython 2.0.1). However we can never get it to return a ScriptParseResult.Complete for function definitions. Code below shows using \n for newlines but we have also tried with \r\n. import clr clr.AddReference('IronPython') clr.AddReference('Microsoft.Scripting') from IronPython.Hosting import Python from Microsoft.Scripting import SourceCodeKind, ScriptCodeParseResult engine = Python.CreateEngine() s = engine.CreateScriptSourceFromString('def f():\n print 1\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x003F [IncompleteStatement] s = engine.CreateScriptSourceFromString('def f():\n print 1\n\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x0040 [IncompleteStatement] The DLR hosting spec has little helpful to say on the matter as far as I can tell. Looking at an example from Tomas it doesn't seem very different from what we're doing: http://blog.tomasm.net/2009/04/15/python-says-hello-to-ruby/ Any clues as to what we are doing wrong or how to procede? Thanks Michael --http://www.ironpythoninaction.com/ http://www.voidspace.org.uk/blog ___ Users mailing list Users@lists.ironpython.com mailto:Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com ___ Users mailing list Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com -- http://www.ironpythoninaction.com/ http://www.voidspace.org.uk/blog ___ Users mailing list Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com ___ Users mailing list Users@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com
Re: [IronPython] InteractiveCode and function definitions
Curt Hagenlocher wrote: That's only because you've decided to arbitrarily[*] define \n\n as being a signal to mean complete. That's not part of the actual language specification. It's the behaviour of the interactive interpreter though - which specifies something. It's also the specification adhered to by the code module for implementing interactive interpreters. In fact, if I append \n\n print 2\n\n to that string, it's still a valid Python program. The key here is that the user has entered a complete thought is a property of the interpreter and not of the language. I might well decide that the commit key sequence is Control+E (as it is in SQL Server Management Studio) instead of enter enter. My point is that it's not correct for IronPython to dictate the semantics of your interpreter. Fine, so do you have any suggestions as to how to replicate the behaviour of the interactive interpreter - whether or not it counts as a specification? Michael [*] Okay, arbitrary is a bit strong in that it's what python.exe and ipy.exe defines. :) On Thu, Apr 16, 2009 at 10:36 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Curt Hagenlocher wrote: IncompleteStatement means that the user is allowed to type more code. If you want to know whether or not it's a valid (complete) string, just check for it not being Invalid. A function definition is never complete in Python because there's never a terminating curly brace :). But that isn't sufficient to implement an interactive interpreter on top of. This code conceptually is complete as far as an interactive interpreter is concerned: 'def f():\n print 1\n\n' It also means you can't distinguish between the previous kind of incomplete (which is incomplete because the user *could* type more code) and this kind of incomplete: 'a = ' or: 'a = (1 + 2 +' Which are both incomplete because the user *must* type more code. (Although the latter two give IncompleteToken - I wonder if that would be enough.) Because of the other IronPython bugs we can't use the code module and ScriptSource / ScriptParseResult doesn't give sufficient information. Any other ideas? Michael On Thu, Apr 16, 2009 at 10:05 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Hello guys, We're trying to detect whether a section of code is complete (to mimic the behaviour of the interactive interpreter). First of all we tried using the Python standard library code module which provides interactive console classes. There are two outstanding bugs on codeplex (one reported by me today) which prevent this being an ideal solution: http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=22064 http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=21881 The second approach was to create a ScriptSource and looking at the code properties to tell if the statement is complete or not (using IronPython 2.0.1). However we can never get it to return a ScriptParseResult.Complete for function definitions. Code below shows using \n for newlines but we have also tried with \r\n. import clr clr.AddReference('IronPython') clr.AddReference('Microsoft.Scripting') from IronPython.Hosting import Python from Microsoft.Scripting import SourceCodeKind, ScriptCodeParseResult engine = Python.CreateEngine() s = engine.CreateScriptSourceFromString('def f():\n print 1\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x003F [IncompleteStatement] s = engine.CreateScriptSourceFromString('def f():\n print 1\n\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x0040 [IncompleteStatement] The DLR hosting spec has little helpful to say on the matter as far as I can tell. Looking at an example from Tomas it doesn't seem very different from what we're doing: http://blog.tomasm.net/2009/04/15/python-says-hello-to-ruby/ Any clues as to what we are doing wrong or how to procede? Thanks Michael --http://www.ironpythoninaction.com/ http://www.voidspace.org.uk/blog ___
Re: [IronPython] InteractiveCode and function definitions
What I've done is to test explicitly for the blank line at the end in conjuction with the test for ScriptCodeParseResult.IncompleteStatement: completeThought = (result == ScriptCodeParseResult.Complete or result == ScriptCodeParseResult.Invalid or (result == ScriptCodeParseResult.IncompleteStatement and text.endswith('\n\n'))) On Thu, Apr 16, 2009 at 10:54 AM, Michael Foord fuzzy...@voidspace.org.ukwrote: Curt Hagenlocher wrote: That's only because you've decided to arbitrarily[*] define \n\n as being a signal to mean complete. That's not part of the actual language specification. It's the behaviour of the interactive interpreter though - which specifies something. It's also the specification adhered to by the code module for implementing interactive interpreters. In fact, if I append \n\n print 2\n\n to that string, it's still a valid Python program. The key here is that the user has entered a complete thought is a property of the interpreter and not of the language. I might well decide that the commit key sequence is Control+E (as it is in SQL Server Management Studio) instead of enter enter. My point is that it's not correct for IronPython to dictate the semantics of your interpreter. Fine, so do you have any suggestions as to how to replicate the behaviour of the interactive interpreter - whether or not it counts as a specification? Michael [*] Okay, arbitrary is a bit strong in that it's what python.exe and ipy.exe defines. :) On Thu, Apr 16, 2009 at 10:36 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Curt Hagenlocher wrote: IncompleteStatement means that the user is allowed to type more code. If you want to know whether or not it's a valid (complete) string, just check for it not being Invalid. A function definition is never complete in Python because there's never a terminating curly brace :). But that isn't sufficient to implement an interactive interpreter on top of. This code conceptually is complete as far as an interactive interpreter is concerned: 'def f():\n print 1\n\n' It also means you can't distinguish between the previous kind of incomplete (which is incomplete because the user *could* type more code) and this kind of incomplete: 'a = ' or: 'a = (1 + 2 +' Which are both incomplete because the user *must* type more code. (Although the latter two give IncompleteToken - I wonder if that would be enough.) Because of the other IronPython bugs we can't use the code module and ScriptSource / ScriptParseResult doesn't give sufficient information. Any other ideas? Michael On Thu, Apr 16, 2009 at 10:05 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Hello guys, We're trying to detect whether a section of code is complete (to mimic the behaviour of the interactive interpreter). First of all we tried using the Python standard library code module which provides interactive console classes. There are two outstanding bugs on codeplex (one reported by me today) which prevent this being an ideal solution: http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=22064 http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=21881 The second approach was to create a ScriptSource and looking at the code properties to tell if the statement is complete or not (using IronPython 2.0.1). However we can never get it to return a ScriptParseResult.Complete for function definitions. Code below shows using \n for newlines but we have also tried with \r\n. import clr clr.AddReference('IronPython') clr.AddReference('Microsoft.Scripting') from IronPython.Hosting import Python from Microsoft.Scripting import SourceCodeKind, ScriptCodeParseResult engine = Python.CreateEngine() s = engine.CreateScriptSourceFromString('def f():\n print 1\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x003F [IncompleteStatement] s = engine.CreateScriptSourceFromString('def f():\n print 1\n\n', 'foo', SourceCodeKind.InteractiveCode) s.GetCodeProperties() Microsoft.Scripting.ScriptCodeParseResult object at 0x0040 [IncompleteStatement] The DLR hosting spec has little helpful to say on the matter as far as I can tell. Looking at an example from Tomas it doesn't seem very
Re: [IronPython] InteractiveCode and function definitions
Curt Hagenlocher wrote: What I've done is to test explicitly for the blank line at the end in conjuction with the test for ScriptCodeParseResult.IncompleteStatement: completeThought = (result == ScriptCodeParseResult.Complete or result == ScriptCodeParseResult.Invalid or (result == ScriptCodeParseResult.IncompleteStatement and text.endswith('\n\n'))) Thanks guys - I've got a working implementation that seems to handle all the cases we need. Nice one. Michael On Thu, Apr 16, 2009 at 10:54 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Curt Hagenlocher wrote: That's only because you've decided to arbitrarily[*] define \n\n as being a signal to mean complete. That's not part of the actual language specification. It's the behaviour of the interactive interpreter though - which specifies something. It's also the specification adhered to by the code module for implementing interactive interpreters. In fact, if I append \n\n print 2\n\n to that string, it's still a valid Python program. The key here is that the user has entered a complete thought is a property of the interpreter and not of the language. I might well decide that the commit key sequence is Control+E (as it is in SQL Server Management Studio) instead of enter enter. My point is that it's not correct for IronPython to dictate the semantics of your interpreter. Fine, so do you have any suggestions as to how to replicate the behaviour of the interactive interpreter - whether or not it counts as a specification? Michael [*] Okay, arbitrary is a bit strong in that it's what python.exe and ipy.exe defines. :) On Thu, Apr 16, 2009 at 10:36 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Curt Hagenlocher wrote: IncompleteStatement means that the user is allowed to type more code. If you want to know whether or not it's a valid (complete) string, just check for it not being Invalid. A function definition is never complete in Python because there's never a terminating curly brace :). But that isn't sufficient to implement an interactive interpreter on top of. This code conceptually is complete as far as an interactive interpreter is concerned: 'def f():\n print 1\n\n' It also means you can't distinguish between the previous kind of incomplete (which is incomplete because the user *could* type more code) and this kind of incomplete: 'a = ' or: 'a = (1 + 2 +' Which are both incomplete because the user *must* type more code. (Although the latter two give IncompleteToken - I wonder if that would be enough.) Because of the other IronPython bugs we can't use the code module and ScriptSource / ScriptParseResult doesn't give sufficient information. Any other ideas? Michael On Thu, Apr 16, 2009 at 10:05 AM, Michael Foord fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk mailto:fuzzy...@voidspace.org.uk wrote: Hello guys, We're trying to detect whether a section of code is complete (to mimic the behaviour of the interactive interpreter). First of all we tried using the Python standard library code module which provides interactive console classes. There are two outstanding bugs on codeplex (one reported by me today) which prevent this being an ideal solution: http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=22064 http://ironpython.codeplex.com/WorkItem/View.aspx?WorkItemId=21881 The second approach was to create a ScriptSource and looking at the code properties to tell if the statement is complete or not (using IronPython 2.0.1). However we can never get it to return a ScriptParseResult.Complete for function definitions. Code below shows using \n for newlines but we have also tried with \r\n. import clr