def getsplitinfo():
"""Return list of current split variables and splitmode
If no splits, return is [], None"""
splitvarlist = spss.GetSplitVariableNames()
if len(splitvarlist) == 0:
return [], None
else:
splittype = spssaux.getShow("split", olang="english")
if splittype.lower().startswith("layer"):
splittype="layered"
else:
splittype="separate"
return splitvarlist, splittype
def dooptbinex(target, binvars, suffix=["_bin"], minsize=10,
alpha=.05, overwrite=False, syntaxoutfile=None, contintervals=10,
treetable=True, recodetable=True, execute=True):
"""Execute STATS OPTBINEX command"""
# debugging
# makes debug apply only to the current thread
#try:
#import wingdbstub
#if wingdbstub.debugger != None:
#import time
#wingdbstub.debugger.StopDebug()
#time.sleep(2)
#wingdbstub.debugger.StartDebug()
#import thread
#wingdbstub.debugger.SetDebugThreads({thread.get_ident(): 1}, default_policy=0)
## for V19 use
#SpssClient._heartBeat(False)
#except:
#pass
# Check that TREES procedure is licensed if possible
treesavailable = True
try: # no client in external mode
SpssClient.StartClient()
treesavailable = SpssClient.IsOptionAvailable(SpssClient.LicenseOption.TREEVIEW)
except:
pass
finally:
SpssClient.StopClient()
if not treesavailable:
raise ValueError(_("""Error: This command requires the Decision Trees option, which is not licensed."""))
if syntaxoutfile is None and not execute:
raise ValueError(_("""No syntax output was specified, and execution was not requested.
There is nothing to do."""))
suffix = "".join(suffix) #tokenlist may come through with mult elements
outnames = [v + suffix for v in binvars]
toolong = [v for v in outnames if len(v) > 64] # wrong metric for Unicode
if toolong:
raise ValueError(_("""The following new names exceed 64 bytes.
Please choose a shorter suffix or rename the input variables:\n%s""") % " ".join(toolong))
vardict = spssaux.VariableDict()
if not overwrite:
existingnames = set([v.lower() for v in vardict.variables])
onames = existingnames.intersection(set([v.lower() for v in outnames]))
if onames:
raise ValueError(_("""Error: The following variables would be overwritten:\n%s""")\
% " ".join(onames))
xmlws = "x" + str(random.uniform(.1,1))
visible = (treetable and "yes") or "no"
tempdir = tempfile.gettempdir()
rulesfilespec = os.path.join(tempdir, "F" + str(random.uniform(.1,1)))
treetemplate = """TREE %(target)s BY %(indvar)s
/TREE DISPLAY=NONE /PRINT TREETABLE /GAIN SUMMARYTABLE=NO
/RULES OUTFILE="%(rulesfilespec)s"
/GROWTHLIMIT MAXDEPTH=1 MINPARENTSIZE=10 MINCHILDSIZE=%(minsize)s
/CHAID ALPHASPLIT=%(alpha)s INTERVALS=%(contintervals)s.
"""
if syntaxoutfile is None:
insertfile = os.path.join(tempdir, "F" + str(random.uniform(.1,1)))
else:
insertfile = syntaxoutfile
insertfilef = open(insertfile, "wb")
# run TREES for each independent variable and accumulate resulting tranformations
oattrs = spssaux.getShow("OATTRS")
empty = True
failedvars = []
outvars = []
outlabels = []
try:
for indvar in binvars:
# set OMS to use language-invariant text
spss.Submit("""set oattrs=eng.
oms select tables /if subtypes='TreeTable'
/destination format=oxml xmlworkspace="%(xmlws)s" viewer=%(visible)s
/tag = "%(xmlws)s".
""" % locals())
spss.Submit(treetemplate % locals())
spss.Submit("""omsend tag="%(xmlws)s".""" % locals())
labels = spss.EvaluateXPath(xmlws, "/",
"""//pivotTable//group[@text_eng="Primary Independent Variable"]/category[@text_eng="Split Values"]/cell/@text""")
if not labels:
failedvars.append(indvar)
continue
empty = False
outputname = indvar + suffix
if recodetable:
outvars.append(outputname)
outlabels.append(labels)
definitions = getrules(rulesfilespec, outputname, labels,
vardict[indvar].VariableLabel) # also removes temporary file
insertfilef.writelines([line + "\n" for line in definitions])
finally:
insertfilef.close()
spss.Submit("SET OATTRS=%s" % oattrs) # restore setting
spss.DeleteXPathHandle(xmlws)
if execute and not empty:
spss.Submit("""INSERT FILE="%s".""" % insertfile)
if syntaxoutfile is None:
os.remove(insertfile)
if failedvars or recodetable:
from spss import CellText
StartProcedure(_("Extended Optimal Binning"), "STATSOPTBINEX")
if failedvars:
wtable = spss.BasePivotTable("Warnings ", "Warnings")
wtable.Append(spss.Dimension.Place.row, "rowdim", hideLabels=True)
rowLabel = CellText.String("1")
wtable[(rowLabel,)] = \
spss.CellText.String(_("""These variables could not be binned. New variables were not created for them.\n%s""") % " ".join(failedvars))
if recodetable and not empty:
table = spss.BasePivotTable(_("Variable Binning"), "OPTBIN")
table.Append(spss.Dimension.Place.row, _("Variable"))
table.Append(spss.Dimension.Place.row, _("Value"))
table.Append(spss.Dimension.Place.column, _("Definition"),
hideName=True)
reccoldef = CellText.String("Definition")
for i in range(len(outvars)):
var = CellText.String(outvars[i])
for j, val in enumerate(outlabels[i]):
rec = CellText.String(j)
recval = CellText.String(val)
table[(var, rec, reccoldef)] = recval
spss.EndProcedure()
def arguments(name, dir, mode, proc):
msg = None
if not name: # name
msg = "Missing run name."
else:
name = name.strip()
m = re.match(r"^[\w -]+$", name) # check name
if not m:
msg = "Invalid name."
if not msg:
cwd = spssaux.getShow("DIRECTORY") # spss current working directory
if not dir:
path = os.path.join(cwd, name) # spss current working directory
else: # specified directory
dir = dir.strip()
m = re.match(r'^"([^"]+)"$', dir) # remove double quotes
if m:
dir = m.group(1).strip()
else:
m = re.match(r"^'([^']+)'$", dir) # remove single quotes
if m:
dir = m.group(1).strip()
m = re.match(r"^[~\.\\/:\w -]+$", dir) # check dir
if not m:
msg = "Invalid directory."
if not msg:
path = os.path.join(os.path.normpath(dir), name)
if not msg:
if mode:
mode = mode.strip()
if not re.match(r"^(debug|test)$", mode,
flags=re.IGNORECASE): # mode
msg = "Mode error."
if not msg:
setup = "" # get the setup
with open(path + ".set", "r") as f:
for line in f:
setup += line
if not setup:
msg = "Missing setup."
if not msg:
m = re.search(r"(^|;)\s*datain\s+([^;]+);", setup,
flags=re.IGNORECASE) # datain
if m:
datain = m.group(2).strip()
if proc != "combine":
m = re.match(r'^"([^"]+)"', datain) # remove double quotes
if m:
datain = m.group(1).strip()
else:
m = re.match(r"^'([^']+)'", datain) # remove single quotes
if m:
datain = m.group(1).strip()
else:
m = re.match(r"^([^ ]+)",
datain) # remove subsequent file names
if m:
datain = m.group(1).strip()
m = re.match(r"^[~\.\\/:\w -]+$", datain) # check name
if not m:
msg = "Invalid datain."
else:
msg = "Missing datain."
if not msg:
m = re.search(r"(^|;)\s*dataout\s+([^;]+);",
setup,
flags=re.IGNORECASE) # dataout
if m:
dataout = m.group(2).strip()
m = re.match(r"^(.+)\s+(all|con|concat|concatenate)$",
dataout,
flags=re.IGNORECASE) # rmove all or concatenate
if m:
dataout = m.group(1).strip()
m = re.match(r'^"([^"]+)"$', dataout) # remove double quotes
if m:
dataout = m.group(1).strip()
else:
m = re.match(r"^'([^']+)'$", dataout) # remove single quotes
if m:
dataout = m.group(1).strip()
m = re.match(r"^[~\.\\/:\w -]+$", dataout) # check name
if not m:
msg = "Invalid dataout."
if not msg:
if sys.platform.startswith("win"): # windows
m = re.match(r"^([a-zA-Z]:|[\\/])", dataout)
else: # linux
m = re.match("^(~|/)", dataout)
if not m: # not full path
dataout = os.path.join(
cwd, dataout) # prefix the current working directory
dataout = os.path.normpath(dataout) # normalize the path
else:
dataout = None # no dataout
if proc == "combine":
msg = "Missing dataout."
if msg:
print msg # print the error message
return None
return [path, datain, dataout, setup]
def putdata(name=None,
dir=None,
mode=None,
dataout=None,
impl=None,
mult=None):
msg = None
if not name: # name
msg = "Missing run name."
else:
m = re.match(r"^\s*(\w[\w\-]*)\s*$", name)
if m:
name = m.group(1)
else:
msg = "Run name error."
if not msg:
cwd = spssaux.getShow("DIRECTORY") # spss current working directory
if not dir:
path = os.path.join(cwd, name) # spss current working directory
else: # specified directory
m = re.match(r'^\s*"([~\.\\/:\w\- ]*)"\s*$', dir) # double quotes
if m:
dir = m.group(1).strip()
else:
m = re.match(r"^\s*'([~\.\\/:\w\- ]*)'\s*$",
dir) # single quotes
if m:
dir = m.group(1).strip()
else:
m = re.match(r"^\s*([~\.\\/:\w\-]*)\s*", dir) # no quotes
if m:
dir = m.group(1)
else:
msg = "Invalid directory."
if not msg:
path = os.path.join(os.path.normpath(dir), name)
if not msg: # mode
if mode:
m = re.match(r"^\s*(debug|test)\s*$", mode, flags=re.IGNORECASE)
if m:
mode = m.group(1).lower()
else:
msg = "Mode error."
if not msg: # dataout
if not dataout:
msg = "Missing dataout."
else:
m = re.match(r'^\s*"([~\.\\/:\w\- ]*)"\s*$',
dataout) # double quotes
if m:
dataout = m.group(1).strip()
else:
m = re.match(r"^\s*'([~\.\\/:\w\- ]*)'\s*$",
dataout) # single quotes
if m:
dataout = m.group(1).strip()
else:
m = re.match(r"^\s*([~\.\\/:\w\-]*)\s*",
dataout) # no quotes
if m:
dataout = m.group(1)
else:
msg = "Invalid dataout."
if not msg:
if sys.platform.startswith("win"): # windows
m = re.match(r"^([a-zA-Z]:|[\\/])", dataout)
else: # linux
m = re.match("^(~|/)", dataout)
if not m: # not full path
dataout = os.path.join(
cwd, dataout) # prefix the current working directory
dataout = os.path.normpath(dataout) # normalize the path
if not msg: # impl
if impl:
m = re.match(r"^\s*(all|\d+)\s*$", impl, flags=re.IGNORECASE)
if m:
impl = m.group(1).lower()
else:
msg = "Invalid implicate."
if not msg: # mult
if mult:
m = re.match(r"^\s*(all|\d+)\s*$", mult, flags=re.IGNORECASE)
if m:
mult = m.group(1).lower()
else:
msg = "Invalid multiple."
if not msg:
args = [] # execute putdata
args.append("/dataout={0}".format(dataout))
if impl:
args.append("/impl={0}".format(impl))
if mult:
args.append("/mult={0}".format(mult))
rc = srclib.execute("putdata", path, mode, args)
if rc != 0:
msg = "Abnormal termination of putdata"
else:
if not os.path.exists(dataout + ".out"): # write the imputed data
msg = "Missing " + dataout + ".out file"
else:
f = open(dataout + ".out", "r")
cmd = f.read()
f.close()
spss.Submit(cmd) # execute the command
if not mode:
os.remove(dataout + ".imp")
os.remove(dataout + ".out")
if msg:
print msg # print the error message
