Problem with an huge dictionary

[Peter Otten]

keobox wrote:

> On 20 Mag, 12:58, Thomas Lehmann <t.lehm... at rtsgroup.net> wrote:
>> > The question is:
>> > Is there a limit on the number of entries a dictionary can have i
>> > jython?
>>
>> > I wrote a little app where my data is stored in a huge dictionary
>> > (11746 entries) generated with a python script.
>> > When I try to import the dictionary, jython complains with the
>> > following message:
>>
>> 1) You did not say what you have saved (content of your dictionary).
> 
> - As you can see I received the error when I tried to import
> jmoco_event_data.py module from the jython interpreter prompt.
> - The "import" will cause the module compilation, jython will produce
> a .class file instead of .pyc.
> - jmoco_event_data.py contains 4 dictionaries:
> 
> typesValToName={
> 220:'EMS_EVENT_EMS_INTERNAL_BASE',
> 221:'EMS_EVENT_INTERNAL_TYPE_BASE',
> 222:'EMS_EVENT_INTERNAL_N_E_P_M_EVENT',
> 223:'EMS_EVENT_INTERNAL_N_E_ALARM_RCVD',
> 224:'EMS_EVENT_NE_SPECIFIC_BASE',
> ... 11746 entries int key: string value
> }
> 
> typesNameToVal={
> 'EMS_EVENT_EMS_INTERNAL_BASE':220,
> 'EMS_EVENT_INTERNAL_TYPE_BASE':221,
> 'EMS_EVENT_INTERNAL_N_E_P_M_EVENT':222,
> 'EMS_EVENT_INTERNAL_N_E_ALARM_RCVD':223,
> 'EMS_EVENT_NE_SPECIFIC_BASE':224,
> ... total 11746 entries string key: int value
> }
> 
> causesValToName={
> 0:'NOT_APPLICABLE_UNKNOWN',
> 1:'SOFTWARE_CAUSE_UNKNOWN',
> 2:'ABSENT_MODULE',
> 3:'FAULTY_MODULE',
> ... 483 entries int key: string value
> }
> 
> causesNameToVal={
> 'NOT_APPLICABLE_UNKNOWN':0,
> 'SOFTWARE_CAUSE_UNKNOWN':1,
> 'ABSENT_MODULE':2,
> 'FAULTY_MODULE':3,
> ... 483 entries string key: int value
> }
> 
>> 2) You did not say how you have saved.
> 
> The dictionaries are in the jmoco_event_data.py module.
> 
>> From the callstack - also I have never used jython - it looks like
>> that
>> there is a try to create a class. It looks like - I may be wrong -
>> that
>> you have saved user objects in your dictionary - have you?
> 
> Nope, the dictionaries are only int to string mappings and string to
> int mappings
> 
>> If so you might fail on loading those objects - especially when your
>> program
>> does not have the code for it.
> 
> The failure happens during module's import, so my question is: Is
> jython able to handle such big dictionaries?

The problem seems to be the size of the code, i. e. the dict literal rather 
than the size of the dictionary. With a slightly smaller dict I get

$ cat make_dict_source.py
from itertools import chain, count, islice

def main():
    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument("n", type=int, default=100)
    parser.add_argument("py")

    args = parser.parse_args()

    prefix = ["data = {\n"]
    suffix = ["}\n"]
    pairs = ("'enum_whatever_{0}': {0},\n".format(i) for i in count())
    pairs = islice(pairs, args.n)
    with open(args.py, "w") as outstream:
        lines = chain(prefix, pairs, suffix)
        outstream.writelines(lines)

if __name__ == "__main__":
    main()
$ python make_dict_source.py 4000 tmp.py
$ jython tmp.py
Traceback (innermost last):
  (no code object) at line 0
java.lang.ClassFormatError: Invalid method Code length 71912 in class file 
org/python/pycode/_pyx0
        at java.lang.ClassLoader.defineClass1(Native Method)
        at java.lang.ClassLoader.defineClass(ClassLoader.java:632)
        at org.python.core.BytecodeLoader2.loadClassFromBytes(Unknown 
Source)
        at org.python.core.BytecodeLoader.makeClass(Unknown Source)
        at org.python.core.BytecodeLoader.makeCode(Unknown Source)
        at org.python.core.Py.compile_flags(Unknown Source)
        at org.python.core.Py.compile_flags(Unknown Source)
        at org.python.core.__builtin__.execfile_flags(Unknown Source)
        at org.python.util.PythonInterpreter.execfile(Unknown Source)
        at org.python.util.jython.main(Unknown Source)

java.lang.ClassFormatError: java.lang.ClassFormatError: Invalid method Code 
length 71912 in class file org/python/pycode/_pyx0
$ jython --version
Jython 2.2.1 on java1.6.0_0

Jython seems to generate a method that is bigger than the 64K allowed by 
Java. 

The simplest workaround is probably to let your module read the dict data 
from a separate file in e. g. csv format.

Peter