# -*- encoding: utf-8 -*-

 '''

 See also:

 '''

 from __future__ import with_statement

 import sys, time

 import urllib, urlparse

 import tempfile

 import os

 import re

 import csv

 import cgi

 from cStringIO import StringIO

 from gettext import gettext as _

 from itertools import *

 from functools import *

 from subprocess import *

 import amara

 from amara.xslt import transform

 from amara.xpath.util import simplify

 from amara.bindery import html

 from amara.lib.util import *

 from akara.services import simple_service, response

 VAR_PAT = re.compile('VARIABLE\s+LABELS\s+(((\w+)\s+"([^"]+)"\s*)+)\.')

 VAR_DEF_PAT = re.compile('(\w+)\s+"([^"]+)"')

 VALUE_PAT = re.compile('VALUE\s+LABELS\s+((/(\w+)\s+(\'(\w+)\'\s+"([^"]+)"\s*)+)+)\.')

 VALUE_DEF_SET_PAT = re.compile('/(\w+)\s+((\'(\w+)\'\s+"([^"]+)"\s*)+)')

 VALUE_DEF_PAT = re.compile('\'(\w+)\'\s+"([^"]+)"')

 VALUE_SET_TYPE = 'value_set'

 VARIABLE_LABELS_TYPE = 'variable_labels'

 VALUE_LABELS_TYPE = 'value_labels'

 #R_SCRIPT = '''library(foreign)

 #mydata <- read.spss(file='%s')

 #write.csv2(mydata)

 #'''

 R_SCRIPT = '''library(Hmisc)

 mydata <- spss.get(file='%s')

 write.csv2(mydata)

 '''

 try:

     R_FILE_CMD = AKARA_MODULE_CONFIG.get('r_command', 'r')

 except NameError:

     #Not running from Akara

     R_FILE_CMD = 'r'

 POR_REQUIRED = _("The 'POR' POST parameter is mandatory.")

 SERVICE_ID = 'http://purl.org/akara/services/builtin/spss.json'

 @simple_service('POST', SERVICE_ID, 'spss.json', 'application/json')

 def spss2json(body, ctype, **params):

     '''

     Uses GNU R to convert SPSS to JSON

     Optionally tries to guess long labels from an original .SPS file

     Requires POST body of multipart/form-data

     Sample request:

     curl -F "POR=@foo.por" http://localhost:8880/spss.json

     curl -F "POR=@foo.por" -F "SPSS=@foo.sps" http://localhost:8880/spss.json

     '''

     #curl --request POST -F "POR=@lat506.por" -F "SPSS=@LAT506.SPS" http://labs.zepheira.com:8880/spss.json

     #Useful:

     # * [[http://wiki.math.yorku.ca/index.php/R:_Data_conversion_from_SPSS|R: Data conversion from SPSS]]

     import simplejson

     body = StringIO(body)

     form = cgi.FieldStorage(fp=body, environ=WSGI_ENVIRON)

     #for k in form:

     #    print >> sys.stderr, (k, form[k][:100])

     por = form.getvalue('POR')

     assert_not_equal(por, None, msg=POR_REQUIRED)

     spss = form.getvalue('SPSS')

     (items, varlabels, valuelabels) = parse_spss(por, spss)

     for count, item in enumerate(items):

         #print >> sys.stderr, row

         item['id'] = item['label'] = '_' + str(count)

         item['type'] = VALUE_SET_TYPE

     return simplejson.dumps({'items': items, VARIABLE_LABELS_TYPE: varlabels, VALUE_LABELS_TYPE: valuelabels}, indent=4)

 def parse_spss(spss_por, spss_syntax=None):

     '''

     Uses GNU R to convert SPSS to a simple Python data structure

     Optionally tries to guess long labels from an original .SPS file

     '''

     varlabels = {}

     valuelabels = {}

     if spss_syntax:

         matchinfo = VAR_PAT.search(spss_syntax)

         if matchinfo:

             #print >> sys.stderr, matchinfo.groups

             defns = matchinfo.group(1)

             for defn in VAR_DEF_PAT.finditer(defns):

                 varlabels[defn.group(1)] = defn.group(2)

         matchinfo = VALUE_PAT.search(spss_syntax)

         defsets = matchinfo.group(1)

         for defset in VALUE_DEF_SET_PAT.finditer(defsets):

             valuelabelset = {}

             for defn in VALUE_DEF_PAT.finditer(defset.group(2)):

                 valuelabelset[defn.group(1)] = defn.group(2)

             valuelabels[defset.group(1)] = valuelabelset

     #print >> sys.stderr, varlabels

     #print >> sys.stderr, valuelabels

     #print >> sys.stderr, por[:100]

     #print >> sys.stderr, spss[:100]

     temp = tempfile.mkstemp(suffix=".por")

     os.write(temp[0], spss_por)

     cmdline = R_FILE_CMD

     process = Popen(cmdline, stdin=PIPE, stdout=PIPE, stderr=PIPE, universal_newlines=True, shell=True)

     csvdata, perr = process.communicate(input=R_SCRIPT%temp[1])

     os.close(temp[0])

     os.remove(temp[1])

     if not csvdata:

         print >> sys.stderr, R_SCRIPT%temp[1]

         print >> sys.stderr, perr

         #FIXME: L10N

         raise ValueError('Empty output from the command line.  Probably a failure.  Command line: "%s"'%cmdline)

     def value(k, v):

         if k in valuelabels and v in valuelabels[k]:

             return valuelabels[k][v]

         else:

             return v

     r_reader = csv.DictReader(csvdata.splitlines(), delimiter=';')

     rows = [

         dict(((k, value(k, v.strip())) for (k, v) in row.iteritems()))

         for row in r_reader

     ]

     return (rows, varlabels, valuelabels)