views.py 10.2 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_validate.loaders import custom_loaders
13
14
15
16
from validata_ui import app
from validata_ui.util import flash_error, flash_info, flash_success, flash_warning, ValidataSource
from validata_ui.validate_helper import ValidatorHelper
from validata_ui import error_messages
17

Pierre Dittgen's avatar
Pierre Dittgen committed
18
from flask import Flask, jsonify, redirect, render_template, request, url_for
19
20
import tabulator

21
from io import BytesIO
22

23
24

def extract_source_data(source: ValidataSource, preview_rows_nb=5):
25
    """ Computes table preview """
26
27
28
29
30
31
32

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

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

    delimiter = None
    if source.format == "csv":
39
40
41
42
        delimiter = csv_helpers.detect_dialect(source.data, format=source.format, scheme=source.scheme,
                                               custom_loaders=custom_loaders).delimiter
    with tabulator.Stream(source.data, format=source.format, scheme=source.scheme, custom_loaders=custom_loaders,
                          delimiter=delimiter) as stream:
43
44
45
46
        for row in stream:
            if header is None:
                header = row
            else:
47
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
48
                nb_rows += 1
49
    preview_rows_nb = min(preview_rows_nb, nb_rows)
50
51
    return {'header': header,
            'rows_nb': nb_rows,
52
53
54
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
55
56


57
ERROR_MESSAGE_FUNC = {
Pierre Dittgen's avatar
Pierre Dittgen committed
58

59
    # Core checks
Pierre Dittgen's avatar
Pierre Dittgen committed
60
    # blank-header
61
    'blank-row': error_messages.blank_row,
Pierre Dittgen's avatar
Pierre Dittgen committed
62
    # duplicate-header
63
64
65
    'duplicate-row': error_messages.duplicate_row,
    'enumerable-constraint': error_messages.enumerable_constraint,
    'maximum-constraint': error_messages.maximum_constraint,
Pierre Dittgen's avatar
Pierre Dittgen committed
66
67
68
69
70
    'maximum-length-constraint': error_messages.maximum_length_constraint,
    'minimum-constraint': error_messages.minimum_constraint,
    'minimum-length-constraint': error_messages.minimum_length_constraint,
    # missing-value
    'pattern-constraint': error_messages.pattern_constraint,
71
72
    'required-constraint': error_messages.required_constraint,
    'type-or-format-error': error_messages.type_or_format_error,
Pierre Dittgen's avatar
Pierre Dittgen committed
73
    'unique-constraint': error_messages.unique_constraint,
74
75
76
77
78

    # Validata pre-checks
    'extra-headers': error_messages.extra_headers,
    'invalid-column-delimiter': error_messages.invalid_column_delimiter,
    'missing-headers': error_messages.missing_headers,
Pierre Dittgen's avatar
Pierre Dittgen committed
79
    'wrong-headers-order': error_messages.wrong_headers_order,
80
81
82
83
84
85
86

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


def improve_messages(errors, headers, schema):
87
88
    """ Translates and improve error messages """

89
90
    def error_message_default_func(error, headers, schema):
        """ Sets a new better error message """
Pierre Dittgen's avatar
Pierre Dittgen committed
91
92
        error['title'] = error['code']
        error['content'] = error.get('message', 'pas d\'information complémentaire')
93
        return error
94

95
    improved_errors = []
96
97
98

    for error in errors:

99
100
        improve_func = ERROR_MESSAGE_FUNC.get(error['code'], error_message_default_func)
        improved_errors.append(improve_func(error, headers, schema))
101

102
    return improved_errors
103
104


105
def contextualize(errors):
Pierre Dittgen's avatar
Pierre Dittgen committed
106
107
    """ add context to errors """

108
109
    def add_context(err):
        """ Adds context info based on row-nb presence """
110
        context = 'body' if 'row-number' in err and not err['row-number'] is None else 'table'
111
112
113
        return {**err, 'context': context}

    return list(map(add_context, errors))
114
115


116
def create_validata_report(goodtables_report, schema):
117
118
119
120
121
122
    """ 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"
123
        - error messages are improved
124
125
126
127
128
129
130
131
132
133
134
    """
    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']
135
136
137
    # 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
138
139
140
141
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
142
    # Computes column info
143
    schema_fields = schema.get('fields', [])
Pierre Dittgen's avatar
Pierre Dittgen committed
144
145
146
147
    fields_dict = {f['name']: (f.get('title', 'titre non défini'), f.get('description', '')) for f in schema_fields}
    report['table']['headers_title'] = [fields_dict[h][0] if h in fields_dict else 'colonne inconnue' for h in headers]
    report['table']['headers_description'] = [fields_dict[h][1]
                                              if h in fields_dict else 'Cette colonne n\'est pas définie dans le schema' for h in headers]
148
149
150
151
152

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

153
    # Provide better (french) messages
154
    errors = improve_messages(errors, headers, schema)
155

156
157
158
159
160
161
162
163
164
165
166
167
    # 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)

Pierre Dittgen's avatar
Pierre Dittgen committed
168
    # Checks if there are structure errors different to invalid-column-delimiter
169
170
171
    report['table']['display_body_errors'] = all(err['code'] == 'invalid-column-delimiter'
                                                 for err in report['table']['errors']['structure'])

Pierre Dittgen's avatar
Pierre Dittgen committed
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
    # Group body errors by row id
    rows = []
    current_row_id = 0
    for err in report['table']['errors']['body']:
        if not 'row-number' in err:
            print('ERR', err)
        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': {}})

        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
    report['table']['errors']['by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
192
193
194
195

    return report


196
def validate(schema_code, source: ValidataSource):
197
198
    """ Validate source and display report """

Pierre Dittgen's avatar
Pierre Dittgen committed
199
200
201
202
203
    try:
        goodtables_report = ValidatorHelper.validate(schema_code, **source.get_goodtables_source())
    except tabulator.exceptions.FormatError:
        flash_error('Erreur : format de fichier non supporté')
        return redirect(url_for('scdl_validator', val_code=schema_code))
204
205

    source_data = extract_source_data(source)
206

207
    validata_report = create_validata_report(goodtables_report, ValidatorHelper.schema(schema_code))
208

209
    # return jsonify(validata_report)
Pierre Dittgen's avatar
Pierre Dittgen committed
210

Pierre Dittgen's avatar
Pierre Dittgen committed
211
212
    # Complete report
    val_info = ValidatorHelper.schema_info(schema_code)
213
    return render_template('validation_report.html', title='Rapport de validation',
214
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
215
                           source=source, source_type=source.type, source_data=source_data,
216
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
217
218
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
                                        {'url': url_for('scdl_validator', val_code=schema_code), 'title': val_info['title']}])
219
220


221
222
223
224
225
226
227
228
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


229
230
231
232
233
234
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
235
    validators = ValidatorHelper.schema_info_list()
Pierre Dittgen's avatar
Pierre Dittgen committed
236
    flash_warning('Ce service est fourni en mode beta - certains problèmes peuvent subsister - nous mettons tout en œuvre pour améliorer son fonctionnement en continu')
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
    return render_template('home.html', title='Accueil', validators=validators)


@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
256
        val_info = ValidatorHelper.schema_info(val_code)
257
258
259
260
        input_param = request.args.get('input')

        # First form display
        if input_param is None or input_param not in ('url', 'example'):
261
            return render_template('validation_form.html', title=val_info['title'],
262
263
264
265
266
267
268
269
270
                                   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))
271
            return validate(val_code, ValidataSource(url, url, 'url'))
272
273
274
275

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
Pierre Dittgen's avatar
Pierre Dittgen committed
276
            flash_error('Aucun fichier à valider')
277
278
279
280
281
282
283
284
            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))
285
286

            return validate(val_code, ValidataSource(bytes_data(f), f.filename, 'file'))
287
288

        return 'Bizarre, vous avez dit bizarre ?'