Пример #1
0
def train_dedup(req, dbname):
    db = TamerDB(dbname)
    if not db.dedup_model_exists():
        return HttpResponseRedirect(reverse('tamer:init_dedup_console', args=(dbname,)))
    else:
        db.dedup_all()
        return HttpResponseRedirect(reverse('tamer:compare_entities', args=(dbname,)))
Пример #2
0
def schema_map_source(req, dbname, sid):
    db = TamerDB(dbname)
    db.rebuild_schema_mapping_models()
    db.schema_map_source(sid)
    redirect_url = '/doit/' + dbname + '/sources/' + sid + '/map'
    return HttpResponse(simplejson.dumps({'redirect': redirect_url}),
                        mimetype='application/json')
Пример #3
0
def run_import(req, dbname):
    db = TamerDB(dbname)

    rclean = re.compile(r'[a-zA-Z][a-zA-Z0-9_-]*')
    def sanitize_db_constant(s):
        return rclean.match(str(s)).group() if s is not None else None

    # Clean table and attribute names
    schemaname = source_schema
    try:
        tablename = sanitize_db_constant(req.POST['tablename'])
        dataattr = [sanitize_db_constant(a)
                    for a in re.split(',', req.POST['dataattr'])]
    except:
	    return HttpResponse(simplejson.dumps({'status': 'fail'}),
                            mimetype='application/json')

    try:
        eidattr = sanitize_db_constant(req.POST['eidattr'])
    except:
        eidattr = None

    try:
        sidattr = sanitize_db_constant(req.POST['sidattr'])
    except:
        sidattr = None

    #return HttpResponse(simplejson.dumps({'e': eidattr, 's': sidattr, 't': tablename, 'd': dataattr, 'n': schemaname}), mimetype='application/json')

    db.import_from_pg_table(schemaname, tablename, eidattr, sidattr, dataattr)

    return HttpResponse(simplejson.dumps({'status': 'success'}), mimetype='application/json')
Пример #4
0
def import_auxiliary(req, dbname):
    db = TamerDB(dbname)
    schema = source_schema
    #try:
    table = req.POST['tablename']
    obj = req.POST['object']
    #except:
    #    return HttpResponse(simplejson.dumps({'status': 'fail'}),
    #                        mimetype='application/json')

    att = req.POST['attribute'] if 'attribute' in req.POST else None
    column = req.POST['columnname'] if 'columnname' in req.POST else None
    columna = req.POST['columna'] if 'columna' in req.POST else None
    columnb = req.POST['columnb'] if 'columnb' in req.POST else None
    template = req.POST['templatename'] if 'templatename' in req.POST else None

    if obj == 'attdict':
        db.import_attribute_dictionary(att, schema, table, column)
    elif obj == 'syndict':
        db.import_synonym_dictionary(att, schema, table, columna, columnb)
    elif obj == 'schema':
        db.import_global_schema(schema, table, column)
    elif obj == 'template':
        db.import_attribute_template(template, schema, table, column)

    return HttpResponse(simplejson.dumps({'status': 'success'}),
                        mimetype='application/json')
Пример #5
0
def train_dedup(req, dbname):
    db = TamerDB(dbname)
    if not db.dedup_model_exists():
        return HttpResponseRedirect('/tamer/' + dbname + '/initialize-dedup')
    else:
        db.dedup_all()
        return HttpResponseRedirect('/tamer/' + dbname + '/evaluate-dedup')
Пример #6
0
def run_import(req, dbname):
    db = TamerDB(dbname)

    rclean = re.compile(r'[a-zA-Z][a-zA-Z0-9_-]*')

    def sanitize_db_constant(s):
        return rclean.match(str(s)).group() if s is not None else None

    # Clean table and attribute names
    schemaname = source_schema
    try:
        tablename = sanitize_db_constant(req.POST['tablename'])
        dataattr = [sanitize_db_constant(a)
                    for a in re.split(',', req.POST['dataattr'])]
    except:
        return HttpResponse(simplejson.dumps({'status': 'fail'}),
                            mimetype='application/json')

    try:
        eidattr = sanitize_db_constant(req.POST['eidattr'])
    except:
        eidattr = None

    try:
        sidattr = sanitize_db_constant(req.POST['sidattr'])
    except:
        sidattr = None
    #return HttpResponse(simplejson.dumps({'e': eidattr, 's': sidattr, 't': tablename, 'd': dataattr, 'n': schemaname}), mimetype='application/json')
    db.import_from_pg_table(schemaname, tablename, eidattr, sidattr, dataattr)
    return HttpResponse(simplejson.dumps({'status': 'success'}), mimetype='application/json')
Пример #7
0
def config_console(req, dbname, model_name):
    db = TamerDB(dbname)
    mname = 'Entity Resolution' if model_name == 'dedup' else 'Schema Mapping'
    return render_to_response('tamer/config-console.html', {
        'navitems': nav_items(db),
        'mod_name': mname,
        'params': db.config_params(model_name)})
Пример #8
0
def train_dedup(req, dbname):
    db = TamerDB(dbname)
    if not db.dedup_model_exists():
        return HttpResponseRedirect('/tamer/' + dbname + '/initialize-dedup')
    else:
        db.dedup_all()
        return HttpResponseRedirect('/tamer/' + dbname + '/evaluate-dedup')
Пример #9
0
def config_console(req, dbname, model_name):
    db = TamerDB(dbname)
    mname = 'Entity Resolution' if model_name == 'dedup' else 'Schema Mapping'
    return render(req, 'tamer/config-console.html', {
        'navitems': nav_items(db),
        'mod_name': mname,
        'params': db.config_params(model_name)})
Пример #10
0
def dedup_source(req, dbname, sid):
    db = TamerDB(dbname)
    if db.dedup_model_exists():
        db.dedup_source(sid)
        redirect_url = '/tamer/' + dbname + '/evaluate-dedup'
    else:
        redirect_url = '/tamer/' + dbname + '/initialize-dedup'
    return HttpResponse(simplejson.dumps({'redirect': redirect_url}),
                        mimetype='application/json')
Пример #11
0
def dedup_source(req, dbname, sid):
    db = TamerDB(dbname)
    if db.dedup_model_exists():
        db.dedup_source(sid)
        redirect_url = '/tamer/' + dbname + '/evaluate-dedup'
    else:
        redirect_url = '/tamer/' + dbname + '/initialize-dedup'
    return HttpResponse(simplejson.dumps({'redirect': redirect_url}),
                        mimetype='application/json')
Пример #12
0
def widget_attr_radio(req, dbname):
    db = TamerDB(dbname)
    tablename = req.GET['tablename'] if 'tablename' in req.GET else req.POST[
        'tablename']
    n = req.GET['name'] if 'name' in req.GET else req.POST['name']
    return render_to_response('tamer/widget-attribute-radio.html', {
        'attrs': db.table_attributes(tablename),
        'inputname': n
    })
Пример #13
0
def dedup_source(req, dbname, sid):
    db = TamerDB(dbname)
    # db.dedup_source(sid)
    c = {
        'source_name': db.sourcename(sid),
        'dbname': dbname,
        'source_id': sid,
        'limits': [.1 * x for x in range(11)[5:]],
        'navitems': nav_items(db), }
    return render(req, 'tamer/dedup-console.html', c)
Пример #14
0
def widget_attr_labeller(req, dbname):
    db = TamerDB(dbname)

    tablename = req.GET['tablename'] if 'tablename' in req.GET else req.POST['tablename']
    labels = ['import as data', 'use as entity key', 'use as source key', 'do not import']

    attrs = [{'name': a, 'labels': labels}
             for a in db.table_attributes(tablename)]
    return render(req, 'tamer/widget-attribute-labeller.html', {
        'attrs': attrs})
Пример #15
0
def init_dedup_submit(req, dbname):
    db = TamerDB(dbname)
    important = []
    irrelevant = []
    for k in req.POST:
        if 'important-' in k: important.append(req.POST[k])
        if 'irrelevant-' in k: irrelevant.append(req.POST[k])
    db.init_dedup(important, irrelevant)
    return HttpResponse(simplejson.dumps({'status': 'OK'}),
                        mimetype='application/json')
Пример #16
0
def import_console(req, dbname, import_object):
    db = TamerDB(dbname)
    return render_to_response(
        'tamer/import-console.html', {
            'navitems': nav_items(db),
            'tables': db.schema_tables(source_schema),
            'global_attributes': db.global_attribute_names(),
            'schemaname': source_schema,
            'object': import_object
        })
Пример #17
0
def init_dedup_submit(req, dbname):
    db = TamerDB(dbname)
    important = []
    irrelevant = []
    for k in req.POST:
        if 'important-' in k: important.append(req.POST[k])
        if 'irrelevant-' in k: irrelevant.append(req.POST[k])
    db.init_dedup(important, irrelevant)
    return HttpResponse(simplejson.dumps({'status': 'OK'}),
                        mimetype='application/json')
Пример #18
0
def widget_attr_labeller(req, dbname):
    db = TamerDB(dbname)

    tablename = req.GET['tablename'] if 'tablename' in req.GET else req.POST['tablename']
    labels = ['import as data', 'use as entity key', 'use as source key', 'do not import']

    attrs = [{'name': a, 'labels': labels}
             for a in db.table_attributes(tablename)]
    return render_to_response('tamer/widget-attribute-labeller.html', {
        'attrs': attrs})
Пример #19
0
def import_auxiliary(req, dbname):
    db = TamerDB(dbname)
    schema = source_schema
    #try:
    table = req.POST['tablename']
    obj = req.POST['object']
    #except:
	#    return HttpResponse(simplejson.dumps({'status': 'fail'}),
    #                        mimetype='application/json')
 
    att = req.POST['attribute'] if 'attribute' in req.POST else None
    column = req.POST['columnname'] if 'columnname' in req.POST else None
    columna = req.POST['columna'] if 'columna' in req.POST else None
    columnb = req.POST['columnb'] if 'columnb' in req.POST else None
    template = req.POST['templatename'] if 'templatename' in req.POST else None

    if obj == 'attdict':
        db.import_attribute_dictionary(att, schema, table, column)
    elif obj == 'syndict':
        db.import_synonym_dictionary(att, schema, table, columna, columnb)
    elif obj == 'schema':
        db.import_global_schema(schema, table, column)
    elif obj == 'template':
        db.import_attribute_template(template, schema, table, column)

    return HttpResponse(simplejson.dumps({'status': 'success'}),
                        mimetype='application/json')
Пример #20
0
def source_console(req, dbname, sid):
    db = TamerDB(dbname)    

    tab = req.GET['tab'] if 'tab' in req.GET else 0

    source = {'id': sid, 'name': db.source_name(sid),
              'stats': db.source_stats(sid)}

    return render_to_response('tamer/source-console.html', {
        'navitems': nav_items(db), 'source': source, 'tab': tab,
        'actionsets': action_sets(db, tab, sid)})
Пример #21
0
def source_console(req, dbname, sid):
    db = TamerDB(dbname)

    tab = req.GET['tab'] if 'tab' in req.GET else 0

    source = {'id': sid, 'name': db.source_name(sid),
              'stats': db.source_stats(sid)}

    return render(req, 'tamer/source-console.html', {
        'navitems': nav_items(db), 'source': source, 'tab': tab,
        'actionsets': action_sets(db, tab, sid)})
Пример #22
0
def compare_entities(req, dbname):
    db = TamerDB(dbname)
    save_entity_comparison_feedback(req, db)
    target_similarity = req.GET['sim'] if 'sim' in req.GET \
        else None
    eid1, eid2, sim = db.get_entities_to_compare(target_similarity, req.GET['sort'] if 'sort' in req.GET else None)
    e1 = {'id': eid1, 'data': db.entity_data(eid1)}
    e2 = {'id': eid2, 'data': db.entity_data(eid2)}
    guess = 'Yes' if sim > 0.6 else 'No'
    attr = pretty_order_entity_attributes(e1, e2)
    return render(req, 'doit/compare-entities.html', {
        'attributes': attr, 'similarity': sim, 'guess': guess,
        'e1id': eid1, 'e2id': eid2, })
Пример #23
0
def compare_entities(req, dbname):
    db = TamerDB(dbname)
    save_entity_comparison_feedback(req, db)
    target_similarity = req.GET['sim'] if 'sim' in req.GET \
        else None
    eid1, eid2, sim = db.get_entities_to_compare(target_similarity, req.GET['sort'] if 'sort' in req.GET else None)
    e1 = {'id': eid1, 'data': db.entity_data(eid1)}
    e2 = {'id': eid2, 'data': db.entity_data(eid2)}
    guess = 'Yes' if sim > 0.6 else 'No'
    attr = pretty_order_entity_attributes(e1, e2)
    return render_to_response('doit/compare-entities.html', {
        'attributes': attr, 'similarity': sim, 'guess': guess,
        'e1id': eid1, 'e2id': eid2,})
Пример #24
0
def schema_map_source_run(req, dbname, sid):
    db = TamerDB(dbname)
    db.rebuild_schema_mapping_models()
    db.schema_map_source(sid)
    field_mappings = db.get_field_mappings_by_source(sid)
    attr_list = sorted(field_mappings.values(), key=lambda f: f['match']['score'], reverse=True)
    return render(req, 'tamer/match-rows.html', {'attr_list': attr_list})
Пример #25
0
def schema_map_source(req, dbname, sid):
    db = TamerDB(dbname)
    redirect_url = '%s/import/schema-map/questions' % settings.EXPERTSRC_URL
    thresh_incr_num = map(lambda x: (x + 1) / 10.0, reversed(range(20)))
    greens = map(lambda x: f2c(x / 1.0), thresh_incr_num)
    reds = map(lambda x: f2c(1.0 - x / 2.0), thresh_incr_num)
    thresh_incr = []
    for x in range(len(thresh_incr_num)):
        point = {}
        point['val'] = thresh_incr_num[x]
        point['green'] = greens[x]
        point['red'] = reds[x]
        thresh_incr.append(point)
    c = {'redirect_url': redirect_url,
        'source_id': db.sourcename(sid),
        'dbname': dbname,
        'sid': sid,
        'navitems': nav_items(db),
        'source_name': 'test_source',
        'threshhold_increments': thresh_incr, }
    return render(req, 'tamer/schema-map-console.html', c)
Пример #26
0
def schema_map_source(req, dbname, sid):
    db = TamerDB(dbname)
    db.rebuild_schema_mapping_models()
    db.schema_map_source(sid)
    redirect_url = '/doit/' + dbname + '/sources/' + sid + '/map'
    return HttpResponse(simplejson.dumps({'redirect': redirect_url}),
                        mimetype='application/json')
Пример #27
0
def schema_map_source_schedule(req, dbname, sid):
    db = TamerDB(dbname)
    thresh = req.POST['thresh']
    db.answer_with_thresh(sid, float(thresh))
    db.register_schema_map(sid)
    return HttpResponse(simplejson.dumps({'status': 'success'}),
                        mimetype='application/json')
Пример #28
0
def get_entity_data(req, dbname):
    db = TamerDB(dbname)
    entity_id = req.GET['entity_id']
    data = db.get_entity_data(entity_id)
    c = {'data': data}
    return render(req, 'tamer/entity-data.html', c)
Пример #29
0
def dedup_source_clusters(req, dbname, sid):
    db = TamerDB(dbname)
    data = db.get_cluster_data(sid)
    return HttpResponse(simplejson.dumps(data),
                        mimetype='application/json')
Пример #30
0
def init_dedup_console(req, dbname):
    db = TamerDB(dbname)
    return render_to_response('tamer/init-dedup-console.html', {
        'navitems': nav_items(db),
        'attributes': db.global_attributes()})
Пример #31
0
def set_config(req, dbname, model_name):
    db = TamerDB(dbname)
    for n, v in req.GET.iteritems():
        db.set_config(n, v)
    return HttpResponse(simplejson.dumps({'status': 'OK'}),
                        mimetype='application/json')
Пример #32
0
def main_console(req, dbname):
    db = TamerDB(dbname)
    return render(req, 'tamer/console.html', {
        'navitems': nav_items(db), })
Пример #33
0
def set_config(req, dbname, model_name):
    db = TamerDB(dbname)
    for n, v in req.GET.iteritems():
        db.set_config(n, v)
    return HttpResponse(simplejson.dumps({'status': 'OK'}),
                        mimetype='application/json')
Пример #34
0
def init_dedup_console(req, dbname):
    db = TamerDB(dbname)
    return render(req, 'tamer/init-dedup-console.html', {
        'navitems': nav_items(db),
        'attributes': db.global_attributes()})
Пример #35
0
def dedup_sim_pairs(req, dbname, sid):
    db = TamerDB(dbname)
    data = db.get_simpairs(sid)
    return HttpResponse(simplejson.dumps(data),
                        mimetype='application/json')
Пример #36
0
def import_console(req, dbname, import_object):
    db = TamerDB(dbname)
    return render_to_response('tamer/import-console.html', {
        'navitems': nav_items(db), 'tables': db.schema_tables(source_schema),
        'global_attributes': db.global_attribute_names(),
        'schemaname': source_schema, 'object': import_object})
Пример #37
0
def widget_attr_radio(req, dbname):
    db = TamerDB(dbname)
    tablename = req.GET['tablename'] if 'tablename' in req.GET else req.POST['tablename']
    n = req.GET['name'] if 'name' in req.GET else req.POST['name']
    return render_to_response('tamer/widget-attribute-radio.html', {
        'attrs': db.table_attributes(tablename), 'inputname': n})