web/views/magicsearch.py
author Sylvain Thénault <sylvain.thenault@logilab.fr>
Tue, 28 Jul 2009 16:33:59 +0200
changeset 2530 3cd117ecc3a8
parent 2476 1294a6bdf3bf
child 2567 961aa959f07a
permissions -rw-r--r--
bad merge?
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     1
"""a query preprocesser to handle quick search shortcuts for cubicweb
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     2
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     3
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     4
:organization: Logilab
1977
606923dff11b big bunch of copyright / docstring update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1433
diff changeset
     5
:copyright: 2001-2009 LOGILAB S.A. (Paris, FRANCE), license is LGPL v2.
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     6
:contact: http://www.logilab.fr/ -- mailto:contact@logilab.fr
1977
606923dff11b big bunch of copyright / docstring update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1433
diff changeset
     7
:license: GNU Lesser General Public License, v2.1 - http://www.gnu.org/licenses
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     8
"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
     9
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    10
__docformat__ = "restructuredtext en"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    11
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    12
import re
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    13
from logging import getLogger
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    14
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    15
from rql import RQLSyntaxError, BadRQLQuery, parse
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    16
from rql.nodes import Relation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    17
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    18
from cubicweb import Unauthorized
984
536e421b082b import updates
sylvain.thenault@logilab.fr
parents: 680
diff changeset
    19
from cubicweb.view import Component
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    20
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    21
LOGGER = getLogger('cubicweb.magicsearch')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    22
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    23
def _get_approriate_translation(translations_found, eschema):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    24
    """return the first (should be the only one) possible translation according
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    25
    to the given entity type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    26
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    27
    # get the list of all attributes / relations for this kind of entity
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    28
    existing_relations = set(eschema.subject_relations())
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    29
    consistent_translations = translations_found & existing_relations
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    30
    if len(consistent_translations) == 0:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    31
        return None
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    32
    return consistent_translations.pop()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    33
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    34
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    35
def translate_rql_tree(rqlst, translations, schema):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    36
    """Try to translate each relation in the RQL syntax tree
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    37
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    38
    :type rqlst: `rql.stmts.Statement`
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    39
    :param rqlst: the RQL syntax tree
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    40
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    41
    :type translations: dict
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    42
    :param translations: the reverted l10n dict
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    43
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    44
    :type schema: `cubicweb.schema.Schema`
2476
1294a6bdf3bf application -> instance where it makes sense
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1977
diff changeset
    45
    :param schema: the instance's schema
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    46
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    47
    # var_types is used as a map : var_name / var_type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    48
    vartypes = {}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    49
    # ambiguous_nodes is used as a map : relation_node / (var_name, available_translations)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    50
    ambiguous_nodes = {}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    51
    # For each relation node, check if it's a localized relation name
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    52
    # If it's a localized name, then use the original relation name, else
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    53
    # keep the existing relation name
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    54
    for relation in rqlst.get_nodes(Relation):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    55
        rtype = relation.r_type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    56
        lhs, rhs = relation.get_variable_parts()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    57
        if rtype == 'is':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    58
            try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    59
                etype = translations[rhs.value]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    60
                rhs.value = etype
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    61
            except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    62
                # If no translation found, leave the entity type as is
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    63
                etype = rhs.value
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    64
            # Memorize variable's type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    65
            vartypes[lhs.name] = etype
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    66
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    67
            try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    68
                translation_set = translations[rtype]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    69
            except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    70
                pass # If no translation found, leave the relation type as is
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    71
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    72
                # Only one possible translation, no ambiguity
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    73
                if len(translation_set) == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    74
                    relation.r_type = iter(translations[rtype]).next()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    75
                # More than 1 possible translation => resolve it later
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    76
                else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    77
                    ambiguous_nodes[relation] = (lhs.name, translation_set)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    78
    if ambiguous_nodes:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    79
        resolve_ambiguities(vartypes, ambiguous_nodes, schema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    80
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    81
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    82
def resolve_ambiguities(var_types, ambiguous_nodes, schema):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    83
    """Tries to resolve remaining ambiguities for translation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    84
    /!\ An ambiguity is when two different string can be localized with
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    85
        the same string
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    86
    A simple example:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    87
      - 'name' in a company context will be localized as 'nom' in French
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    88
      - but ... 'surname' will also be localized as 'nom'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    89
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    90
    :type var_types: dict
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    91
    :param var_types: a map : var_name / var_type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    92
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    93
    :type ambiguous_nodes: dict
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    94
    :param ambiguous_nodes: a map : relation_node / (var_name, available_translations)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    95
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    96
    :type schema: `cubicweb.schema.Schema`
2476
1294a6bdf3bf application -> instance where it makes sense
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1977
diff changeset
    97
    :param schema: the instance's schema
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    98
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    99
    # Now, try to resolve ambiguous translations
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   100
    for relation, (var_name, translations_found) in ambiguous_nodes.items():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   101
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   102
            vartype = var_types[var_name]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   103
        except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   104
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   105
        # Get schema for this entity type
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   106
        eschema = schema.eschema(vartype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   107
        rtype = _get_approriate_translation(translations_found, eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   108
        if rtype is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   109
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   110
        relation.r_type = rtype
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   111
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   112
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   113
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   114
QUOTED_SRE = re.compile(r'(.*?)(["\'])(.+?)\2')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   115
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   116
TRANSLATION_MAPS = {}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   117
def trmap(config, schema, lang):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   118
    try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   119
        return TRANSLATION_MAPS[lang]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   120
    except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   121
        assert lang in config.translations, '%s %s' % (lang, config.translations)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   122
        tr = config.translations[lang]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   123
        langmap = {}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   124
        for etype in schema.entities():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   125
            etype = str(etype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   126
            langmap[tr(etype).capitalize()] = etype
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   127
            langmap[etype.capitalize()] = etype
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   128
        for rtype in schema.relations():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   129
            rtype = str(rtype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   130
            langmap.setdefault(tr(rtype).lower(), set()).add(rtype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   131
            langmap.setdefault(rtype, set()).add(rtype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   132
        TRANSLATION_MAPS[lang] = langmap
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   133
        return langmap
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   134
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   135
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   136
class BaseQueryProcessor(Component):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   137
    __abstract__ = True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   138
    id = 'magicsearch_processor'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   139
    # set something if you want explicit component search facility for the
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   140
    # component
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   141
    name = None
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   142
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   143
    def process_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   144
        args = self.preprocess_query(uquery, req)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   145
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   146
            return req.execute(*args)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   147
        finally:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   148
            # rollback necessary to avoid leaving the connection in a bad state
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   149
            req.cnx.rollback()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   150
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   151
    def preprocess_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   152
        raise NotImplementedError()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   153
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   154
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   155
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   156
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   157
class DoNotPreprocess(BaseQueryProcessor):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   158
    """this one returns the raw query and should be placed in first position
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   159
    of the chain
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   160
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   161
    name = 'rql'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   162
    priority = 0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   163
    def preprocess_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   164
        return uquery,
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   165
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   166
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   167
class QueryTranslator(BaseQueryProcessor):
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   168
    """ parses through rql and translates into schema language entity names
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   169
    and attributes
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   170
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   171
    priority = 2
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   172
    def preprocess_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   173
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   174
            rqlst = parse(uquery, print_errors=False)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   175
        except (RQLSyntaxError, BadRQLQuery), err:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   176
            return uquery,
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   177
        schema = self.vreg.schema
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   178
        # rql syntax tree will be modified in place if necessary
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   179
        translate_rql_tree(rqlst, trmap(self.config, schema, req.lang), schema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   180
        return rqlst.as_string(),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   181
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   182
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   183
class QSPreProcessor(BaseQueryProcessor):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   184
    """Quick search preprocessor
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   185
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   186
    preprocessing query in shortcut form to their RQL form
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   187
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   188
    priority = 4
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   189
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   190
    def preprocess_query(self, uquery, req):
1138
22f634977c95 make pylint happy, fix some bugs on the way
sylvain.thenault@logilab.fr
parents: 984
diff changeset
   191
        """try to get rql from an unicode query string"""
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   192
        args = None
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   193
        self.req = req
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   194
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   195
            # Process as if there was a quoted part
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   196
            args = self._quoted_words_query(uquery)
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   197
        ## No quoted part
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   198
        except BadRQLQuery:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   199
            words = uquery.split()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   200
            if len(words) == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   201
                args = self._one_word_query(*words)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   202
            elif len(words) == 2:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   203
                args = self._two_words_query(*words)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   204
            elif len(words) == 3:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   205
                args = self._three_words_query(*words)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   206
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   207
                args = self._multiple_words_query(words)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   208
        return args
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   209
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   210
    def _get_entity_type(self, word):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   211
        """check if the given word is matching an entity type, return it if
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   212
        it's the case or raise BadRQLQuery if not
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   213
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   214
        etype = word.capitalize()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   215
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   216
            return trmap(self.config, self.vreg.schema, self.req.lang)[etype]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   217
        except KeyError:
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   218
            raise BadRQLQuery('%s is not a valid entity name' % etype)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   219
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   220
    def _get_attribute_name(self, word, eschema):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   221
        """check if the given word is matching an attribute of the given entity type,
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   222
        return it normalized if found or return it untransformed else
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   223
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   224
        """Returns the attributes's name as stored in the DB"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   225
        # Need to convert from unicode to string (could be whatever)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   226
        rtype = word.lower()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   227
        # Find the entity name as stored in the DB
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   228
        translations = trmap(self.config, self.vreg.schema, self.req.lang)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   229
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   230
            translations = translations[rtype]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   231
        except KeyError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   232
            raise BadRQLQuery('%s is not a valid attribute for %s entity type'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   233
                              % (word, eschema))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   234
        rtype = _get_approriate_translation(translations, eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   235
        if rtype is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   236
            raise BadRQLQuery('%s is not a valid attribute for %s entity type'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   237
                              % (word, eschema))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   238
        return rtype
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   239
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   240
    def _one_word_query(self, word):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   241
        """Specific process for one word query (case (1) of preprocess_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   242
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   243
        # if this is an integer, then directly go to eid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   244
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   245
            eid = int(word)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   246
            return 'Any X WHERE X eid %(x)s', {'x': eid}, 'x'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   247
        except ValueError:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   248
            etype = self._get_entity_type(word)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   249
            return '%s %s' % (etype, etype[0]),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   250
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   251
    def _complete_rql(self, searchstr, etype, rtype=None, var=None, searchattr=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   252
        searchop = ''
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   253
        if '%' in searchstr:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   254
            if rtype:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   255
                possible_etypes = self.schema.rschema(rtype).objects(etype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   256
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   257
                possible_etypes = [self.schema.eschema(etype)]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   258
            if searchattr or len(possible_etypes) == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   259
                searchattr = searchattr or possible_etypes[0].main_attribute()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   260
                searchop = 'LIKE '
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   261
        searchattr = searchattr or 'has_text'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   262
        if var is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   263
            var = etype[0]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   264
        return '%s %s %s%%(text)s' % (var, searchattr, searchop)
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   265
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   266
    def _two_words_query(self, word1, word2):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   267
        """Specific process for two words query (case (2) of preprocess_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   268
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   269
        etype = self._get_entity_type(word1)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   270
        # this is a valid RQL query : ("Person X", or "Person TMP1")
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   271
        if len(word2) == 1 and word2.isupper():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   272
            return '%s %s' % (etype, word2),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   273
        # else, suppose it's a shortcut like : Person Smith
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   274
        rql = '%s %s WHERE %s' % (etype, etype[0], self._complete_rql(word2, etype))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   275
        return rql, {'text': word2}
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   276
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   277
    def _three_words_query(self, word1, word2, word3):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   278
        """Specific process for three words query (case (3) of preprocess_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   279
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   280
        etype = self._get_entity_type(word1)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   281
        eschema = self.schema.eschema(etype)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   282
        rtype = self._get_attribute_name(word2, eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   283
        # expand shortcut if rtype is a non final relation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   284
        if not self.schema.rschema(rtype).is_final():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   285
            return self._expand_shortcut(etype, rtype, word3)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   286
        if '%' in word3:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   287
            searchop = 'LIKE '
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   288
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   289
            searchop = ''
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   290
        rql = '%s %s WHERE %s' % (etype, etype[0],
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   291
                                  self._complete_rql(word3, etype, searchattr=rtype))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   292
        return rql, {'text': word3}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   293
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   294
    def _multiple_words_query(self, words):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   295
        """specific process for more than 3 words query"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   296
        return ' '.join(words),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   297
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   298
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   299
    def _expand_shortcut(self, etype, rtype, searchstr):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   300
        """Expands shortcut queries on a non final relation to use has_text or
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   301
        the main attribute (according to possible entity type) if '%' is used in the
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   302
        search word
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   303
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   304
        Transforms : 'person worksat IBM' into
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   305
        'Personne P WHERE P worksAt C, C has_text "IBM"'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   306
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   307
        # check out all possilbe entity types for the relation represented
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   308
        # by 'rtype'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   309
        mainvar = etype[0]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   310
        searchvar = mainvar  + '1'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   311
        rql =  '%s %s WHERE %s %s %s, %s' % (etype, mainvar,  # Person P
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   312
                                             mainvar, rtype, searchvar, # P worksAt C
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   313
                                             self._complete_rql(searchstr, etype,
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   314
                                                                rtype=rtype, var=searchvar))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   315
        return rql, {'text': searchstr}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   316
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   317
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   318
    def _quoted_words_query(self, ori_rql):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   319
        """Specific process when there's a "quoted" part
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   320
        """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   321
        m = QUOTED_SRE.match(ori_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   322
        # if there's no quoted part, then no special pre-processing to do
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   323
        if m is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   324
            raise BadRQLQuery("unable to handle request %r" % ori_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   325
        left_words = m.group(1).split()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   326
        quoted_part = m.group(3)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   327
        # Case (1) : Company "My own company"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   328
        if len(left_words) == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   329
            try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   330
                word1 = left_words[0]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   331
                return self._two_words_query(word1, quoted_part)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   332
            except BadRQLQuery, error:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   333
                raise BadRQLQuery("unable to handle request %r" % ori_rql)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   334
        # Case (2) : Company name "My own company";
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   335
        elif len(left_words) == 2:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   336
            word1, word2 = left_words
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   337
            return self._three_words_query(word1, word2, quoted_part)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   338
            # return ori_rql
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   339
        raise BadRQLQuery("unable to handle request %r" % ori_rql)
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   340
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   341
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   342
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   343
class FullTextTranslator(BaseQueryProcessor):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   344
    priority = 10
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   345
    name = 'text'
1433
091ac3ba5d51 remove trailing white spaces
sylvain.thenault@logilab.fr
parents: 1172
diff changeset
   346
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   347
    def preprocess_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   348
        """suppose it's a plain text query"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   349
        return 'Any X WHERE X has_text %(text)s', {'text': uquery}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   350
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   351
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   352
661
4f61eb8a96b7 properly kill/depreciate component base class, only keep Component
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   353
class MagicSearchComponent(Component):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   354
    id  = 'magicsearch'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   355
    def __init__(self, req, rset=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   356
        super(MagicSearchComponent, self).__init__(req, rset)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   357
        processors = []
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   358
        self.by_name = {}
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   359
        for processorcls in self.vreg.registry_objects('components',
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   360
                                                       'magicsearch_processor'):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   361
            # instantiation needed
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   362
            processor = processorcls()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   363
            processors.append(processor)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   364
            if processor.name is not None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   365
                assert not processor.name in self.by_name
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   366
                self.by_name[processor.name.lower()] = processor
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   367
        self.processors = sorted(processors, key=lambda x: x.priority)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   368
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   369
    def process_query(self, uquery, req):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   370
        assert isinstance(uquery, unicode)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   371
        try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   372
            procname, query = uquery.split(':', 1)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   373
            proc = self.by_name[procname.strip().lower()]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   374
            uquery = query.strip()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   375
        except:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   376
            # use processor chain
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   377
            unauthorized = None
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   378
            for proc in self.processors:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   379
                try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   380
                    return proc.process_query(uquery, req)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   381
                # FIXME : we don't want to catch any exception type here !
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   382
                except (RQLSyntaxError, BadRQLQuery):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   383
                    pass
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   384
                except Unauthorized, ex:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   385
                    unauthorized = ex
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   386
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   387
                except Exception, ex:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   388
                    LOGGER.debug('%s: %s', ex.__class__.__name__, ex)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   389
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   390
            if unauthorized:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   391
                raise unauthorized
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   392
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   393
            # let exception propagate
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   394
            return proc.process_query(uquery, req)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   395
        raise BadRQLQuery(req._('sorry, the server is unable to handle this query'))