views.py 10.1 KB
Newer Older
1 2 3 4
#!/usr/bin/env python3
"""
    Routes
"""
5
import copy
6 7 8 9 10
import json
import os
from collections import OrderedDict
from pathlib import Path

11
from validata_validate import csv_helpers
12
from validata_ui_next import app
13
from validata_ui_next.util import flash_error, flash_info, flash_success, flash_warning, ValidataSource
14
from validata_ui_next.validate_helper import ValidatorHelper
15
from validata_ui_next import error_messages
16

Pierre Dittgen's avatar
Pierre Dittgen committed
17
from flask import Flask, jsonify, redirect, render_template, request, url_for
18 19
import tabulator

20
from io import BytesIO
21

22 23

def extract_source_data(source: ValidataSource, preview_rows_nb=5):
24
    """ Computes table preview """
25 26 27 28 29 30 31

    def stringify(val):
        """ Transform value into string """
        if val is None:
            return ''
        return str(val)

32 33
    header = None
    rows = []
Pierre Dittgen's avatar
Pierre Dittgen committed
34
    nb_rows = 0
35 36 37 38 39 40

    delimiter = None
    if source.format == "csv":
        delimiter = csv_helpers.detect_dialect(source.data, format=source.format,
                                               scheme=source.scheme, custom_loaders=source.custom_loaders).delimiter
    with tabulator.Stream(source.data, format=source.format, scheme=source.scheme, custom_loaders=source.custom_loaders, delimiter=delimiter) as stream:
41 42 43 44
        for row in stream:
            if header is None:
                header = row
            else:
45
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
46
                nb_rows += 1
47
    preview_rows_nb = min(preview_rows_nb, nb_rows)
48 49
    return {'header': header,
            'rows_nb': nb_rows,
50 51 52
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
53 54


Pierre Dittgen's avatar
Pierre Dittgen committed
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
ERR_CODE_TO_CONTEXT = dict([
    # TODO: gets it from spec.json
    ('duplicate-header', 'head'),
    ('extra-value', 'body'),
    ('missing-value', 'body'),
    ('source-error', 'table'),
    ('schema-error', 'table'),
    ('non-matching-header', 'head'),
    ('blank-row', 'body'),
    ('blank-header', 'head'),
    ('enumerable-constraint', 'body'),
    ('http-error', 'table'),
    ('scheme-error', 'table'),
    ('type-or-format-error', 'body'),
    ('format-error', 'table'),
    ('extra-header', 'head'),
    ('pattern-constraint', 'body'),
    ('required-constraint', 'body'),
    ('missing-header', 'head'),
    ('maximum-length-constraint', 'body'),
    ('maximum-constraint', 'body'),
    ('minimum-length-constraint', 'body'),
    ('encoding-error', 'table'),
    ('io-error', 'table'),
    ('unique-constraint', 'body'),
    ('duplicate-row', 'body'),
    ('minimum-constraint', 'body'),

    # TODO: get it from validata_validate
84 85
    ('invalid-column-delimiter', 'table'),
    ('missing-headers', 'table'),
86 87
    ('wrong-headers-order', 'table'),
    ('extra-headers', 'table'),
Pierre Dittgen's avatar
Pierre Dittgen committed
88 89 90 91

    # Custom checks fall in default case: body
])

92
ERROR_MESSAGE_FUNC = {
Pierre Dittgen's avatar
Pierre Dittgen committed
93

94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
    # Core checks
    'blank-row': error_messages.blank_row,
    'duplicate-row': error_messages.duplicate_row,
    'enumerable-constraint': error_messages.enumerable_constraint,
    'maximum-constraint': error_messages.maximum_constraint,
    'required-constraint': error_messages.required_constraint,
    'type-or-format-error': error_messages.type_or_format_error,
    'pattern-constraint': error_messages.pattern_constraint,

    # Validata pre-checks
    'extra-headers': error_messages.extra_headers,
    'invalid-column-delimiter': error_messages.invalid_column_delimiter,
    'wrong-headers-order': error_messages.wrong_headers_order,
    'missing-headers': error_messages.missing_headers,

    # Validata custom checks
    'french-siret-value': error_messages.french_siret_value,
}


def improve_messages(errors, headers, schema):
115 116
    """ Translates and improve error messages """

117 118 119 120 121
    def error_message_default_func(error, headers, schema):
        """ Sets a new better error message """
        data_str = json.dumps(error)
        error['message'] = '[{}] {} ({})'.format(error['code'], error.get('message', ''), data_str)
        return error
122

123
    improved_errors = []
124 125 126

    for error in errors:

127 128
        improve_func = ERROR_MESSAGE_FUNC.get(error['code'], error_message_default_func)
        improved_errors.append(improve_func(error, headers, schema))
129

130
    return improved_errors
131 132


133
def contextualize(errors):
Pierre Dittgen's avatar
Pierre Dittgen committed
134 135
    """ add context to errors """

136 137 138
    return [{**err, 'context': ERR_CODE_TO_CONTEXT.get(err['code'], 'body')} for err in errors]


139
def create_validata_report(goodtables_report, schema):
140 141 142 143 144 145
    """ Creates an error report easier to handle and display in templates:
        - only one table
        - errors are contextualized
        - error-counts is ok
        - errors are grouped by lines
        - errors are separated into "structure" and "body"
146
        - error messages are improved
147 148 149 150 151 152 153 154 155 156 157
    """
    report = copy.deepcopy(goodtables_report)

    # One table is enough
    del report['table-count']
    report['table'] = report['tables'][0]
    del report['tables']
    del report['table']['error-count']
    del report['table']['time']
    del report['table']['valid']
    del report['valid']
158 159 160
    # use _ instead of - to ease information picking in jinja2 template
    report['table']['row_count'] = report['table']['row-count']

Pierre Dittgen's avatar
Pierre Dittgen committed
161 162 163 164 165
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

    # Headers title
166
    schema_fields = schema.get('fields', [])
Pierre Dittgen's avatar
Pierre Dittgen committed
167
    fields_dict = {f['name']: f['title'] for f in schema_fields}
168
    report['table']['headers_title'] = [fields_dict.get(h, r'/!\ colonne inconnue dans le schéma') for h in headers]
169 170 171 172 173

    # Add context to errors
    errors = contextualize(report['table']['errors'])
    del report['table']['errors']

174
    # Provide better (french) messages
175
    errors = improve_messages(errors, headers, schema)
176

177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192
    # Count errors
    report['error_count'] = len(errors)
    del report['error-count']

    # Then group them in 2 groups : structure and body
    report['table']['errors'] = {'structure': [], 'body': []}
    for err in errors:
        if err['context'] != 'body':
            report['table']['errors']['structure'].append(err)
        else:
            report['table']['errors']['body'].append(err)

    # and group body errors by row id
    rows = []
    current_row_id = 0
    for err in report['table']['errors']['body']:
193 194
        if not 'row-number' in err:
            print('ERR', err)
195 196 197 198 199 200
        row_id = err['row-number']
        del err['row-number']
        del err['context']
        if row_id != current_row_id:
            current_row_id = row_id
            rows.append({'row_id': current_row_id, 'errors': {}})
201 202 203 204 205 206 207

        column_id = err.get('column-number')
        if column_id is not None:
            del err['column-number']
            rows[-1]['errors'][column_id] = err
        else:
            rows[-1]['errors']['row'] = err
208
    report['table']['errors']['by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
209 210 211 212

    return report


213
def validate(schema_code, source: ValidataSource):
214 215
    """ Validate source and display report """

216 217 218
    goodtables_report = ValidatorHelper.validate(schema_code, **source.get_goodtables_source())

    source_data = extract_source_data(source)
219

220
    validata_report = create_validata_report(goodtables_report, ValidatorHelper.schema(schema_code))
221

222
    # return jsonify(validata_report)
Pierre Dittgen's avatar
Pierre Dittgen committed
223

Pierre Dittgen's avatar
Pierre Dittgen committed
224 225
    # Complete report
    val_info = ValidatorHelper.schema_info(schema_code)
226
    return render_template('validation_report.html', title='Rapport de validation',
227
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
228
                           source=source, source_type=source.type, source_data=source_data,
229
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
230 231
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
                                        {'url': url_for('scdl_validator', val_code=schema_code), 'title': val_info['title']}])
232 233


234 235 236 237 238 239 240 241
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


242 243 244 245 246 247
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
248
    validators = ValidatorHelper.schema_info_list()
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
    return render_template('home.html', title='Accueil', validators=validators)


@app.route('/about')
def about():
    """ Help -> About page """
    return render_template('about.html', title='À propos',
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'}, ])


@app.route('/validators')
def validators():
    """ No validators page """
    return redirect(url_for('home'))


@app.route('/validators/<val_code>', methods=['GET', 'POST'])
def scdl_validator(val_code):
    """ Validator page """

    if not ValidatorHelper.schema_exist(val_code):
        flash_error('Validateur [{}] inconnu'.format(val_code))
        return redirect(url_for('home'))

    if request.method == 'GET':

Pierre Dittgen's avatar
Pierre Dittgen committed
275
        val_info = ValidatorHelper.schema_info(val_code)
276 277 278 279 280 281 282 283 284 285 286 287 288 289
        input_param = request.args.get('input')

        # First form display
        if input_param is None or input_param not in ('url', 'example'):
            return render_template('validator.html', title=val_info['title'],
                                   val_info=val_info,
                                   breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'}, ])

        # Process URL
        else:
            url = request.args.get('url')
            if url is None or url == '':
                flash_error("Vous n'avez pas indiqué d'url à valider")
                return redirect(url_for('scdl_validator', val_code=val_code))
290
            return validate(val_code, ValidataSource(url, url, 'url'))
291 292 293 294 295 296 297 298 299 300 301 302 303

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
            flash_error('Source non définie')
            return redirect(url_for('scdl_validator', val_code=val_code))

        # File validation
        if input_param == 'file':
            f = request.files.get('file')
            if f is None:
                flash_warning("Vous n'avez pas indiqué de fichier à valider")
                return redirect(url_for('scdl_validator', val_code=val_code))
304 305

            return validate(val_code, ValidataSource(bytes_data(f), f.filename, 'file'))
306 307

        return 'Bizarre, vous avez dit bizarre ?'