w.c.s. : ré-écriture de wcs-extract.py pour plus de clarté

[progfou.git] / wcs / wcs-extract.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import os
import os.path
from time import gmtime, strftime
import simplejson as json

from wcs import publisher
from wcs.formdef import FormDef
from wcs.fields import TitleField, CommentField, TextField, \
                       StringField, ItemField, EmailField, \
                       DateField, FileField, BoolField

COPY_COMMAND = '/bin/cp'
COPY_ARGS = '-af'


def reduce_to_alnum(s, replacement_char='-'):
    """réduction d'une chaîne de caractères à de l'alpha-numérique"""

    avec_accent = u'çÇáàâÁÀÂéèêëÉÈÊËíìîïÍÌÎÏóòôöÓÒÔÖúùûüÚÙÛÜýỳyÿÝỲYŸ'
    sans_accent = u'cCaaaAAAeeeeEEEEiiiiIIIIooooOOOOuuuuUUUUyyyyYYYY'
    if type(s) is not unicode:
        s = unicode(s, 'utf-8')
        u  = False
    r = ''
    for c in s:
        index = avec_accent.find(c)
        if index >= 0:
            r += sans_accent[index]
        elif ('a' <= c.lower() <= 'z') or ('0' <= c <= '9'):
            r += c
        elif len(r) > 0 and r[-1] != replacement_char:
            r += replacement_char
        else: # r == '' or r[-1] == replacement_char
            pass
    r = r.strip(replacement_char)
    if not u:
        r = r.encode('utf-8')
    return r


def extract_fields(formdef, output_directory):
    """nommage des champs de façon unique"""
    # TODO: devrait retourner un résultat, qui serait alors sauvé en dehors

    f = open(os.path.join(output_directory, 'field-names.txt'), 'w')

    field_names = {}
    field_names_duplicates = {}
    for field in formdef.fields:
        if isinstance(field, TitleField) or isinstance(field, CommentField):
            continue
        name = reduce_to_alnum(field.label,'_').lower()
        if name in field_names.values(): # duplicat
            field_names_duplicates[name] = field_names_duplicates.get(name, 1) + 1
            name = '%s_%d' % (name, field_names_duplicates[name])
        field_names.update({field.id: name})
        print >>f, "%s:%s:%s" % (field.id, field_names[field.id], field.label)

    f.close()

    f = open(os.path.join(output_directory, 'field-names.json'), 'wb')
    f.write(json.dumps(field_names, ensure_ascii=False))
    f.close()


def extract_data(formdef, output_directory):
    """extraction des données du formulaire"""
    # TODO: devrait retourner un résultat, qui serait alors sauvé en dehors

    for object in formdef.data_class().select():
        result = {
            'num_dossier': object.id,
            'wcs_status': object.status,
            'wcs_workflow_status': object.get_workflow_status().name,
            'wcs_user_email': object.user.email,
            'wcs_user_display_name': object.user.display_name,
           #'wcs_last_modified': strftime('%Y-%m-%d %H:%M:%S', gmtime(object.last_modified())),
        }
        qfiles = { }
        for field in formdef.fields:
            field_id = str(field.id)
            if not field_id in object.data:
                continue
            if isinstance(field, TitleField) or isinstance(field, CommentField):
                continue
            field_name = field_names[field_id]
            data = object.data.get(field_id)
            if isinstance(field, StringField) or isinstance(field, TextField) \
            or isinstance(field, EmailField) or isinstance(field, ItemField):
                result[field_name] = data
            elif isinstance(field, BoolField):
                result[field_name] = (data == 'True')
            elif isinstance(field, DateField):
                result[field_name] = strftime('%Y-%m-%d', data)
            elif isinstance(field, FileField):
                extension = data.orig_filename.rpartition('.')[2].lower()
                result[field_name] = "%s.%s" % (field_name, extension)
                qfiles[field_name] = data.qfilename
            else:
                print "WARNING: unknown field type '%s' for '%s'" % \
                                        (field.__class__.__name__, field.label)
                raise RuntimeError

        num_dossier = result['num_dossier']
        nom = reduce_to_alnum(result['nom']).upper()
        prenom = reduce_to_alnum(result['prenom']).upper()
        adel = result['adresse_electronique'].replace('@','-').lower()

        filename = "%04d-%s-%s-%s" % (num_dossier, nom, prenom, adel)

        print "Dossier '%s'..." % filename,

        # copie des fichiers joints
        for f in qfiles:
            result[f] = filename + '_' + result[f]
            src = os.path.join(pub.app_dir, 'uploads', qfiles[f])
            dst = os.path.join(output_directory, result[f])
            if not os.path.exists(dst) or os.path.getmtime(src) > os.path.getmtime(dst):
                os.spawnl(os.P_WAIT, COPY_COMMAND, COPY_ARGS, src, dst)
                os.chmod(dst, 0644)

        # génération du fichier JSON
        f = open(os.path.join(output_directory, filename + '.json'), 'wb')
        f.write(json.dumps(result, ensure_ascii=False))
        f.close()

        print "OK."


if __name__ == '__main__':
    import sys

    if len(sys.argv) != 4:
        print >>sys.stderr, "Usage : %s <dossier-destination> <site> <formulaire>" % sys.argv[0]
        sys.exit(1)

    VHOST = sys.argv[2]
    FORM_NAME = sys.argv[3]
    OUTPUT_DIRECTORY = os.path.join(sys.argv[1], VHOST, FORM_NAME)

    os.umask(0022)
    # création du dossier d'extraction, au besoin
    if not os.path.isdir(OUTPUT_DIRECTORY):
        os.makedirs(OUTPUT_DIRECTORY, 0755)

    pub = publisher.WcsPublisher.create_publisher()
    pub.app_dir = os.path.join(pub.app_dir, VHOST)

    formdef = FormDef.get_by_urlname(FORM_NAME)

    extract_fields(formdef, OUTPUT_DIRECTORY)

    extract_data(formdef, OUTPUT_DIRECTORY)

    f = open(os.path.join(OUTPUT_DIRECTORY, 'last-run.txt'), 'w')
    f.write(strftime('%Y-%m-%d %H:%M:%S GMT', gmtime()))
    f.close()