views.py 8.59 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

    def stringify(val):
        """ Transform value into string """
Pierre Dittgen's avatar
Pierre Dittgen committed
29
        return '' if val is None else str(val)
30

31
32
    header = None
    rows = []
Pierre Dittgen's avatar
Pierre Dittgen committed
33
    nb_rows = 0
34

35
    options = {}
36
    if source.format == "csv":
37
38
        options['delimiter'] = csv_helpers.detect_dialect(source.data, format=source.format, scheme=source.scheme,
                                                          custom_loaders=custom_loaders).delimiter
39
    with tabulator.Stream(source.data, format=source.format, scheme=source.scheme, custom_loaders=custom_loaders,
40
                          **options) as stream:
41
42
        for row in stream:
            if header is None:
43
                header = ['' if v is None else v for v in row]
44
            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


55
def contextualize(errors):
Pierre Dittgen's avatar
Pierre Dittgen committed
56
57
    """ add context to errors """

58
59
    def add_context(err):
        """ Adds context info based on row-nb presence """
60
        context = 'body' if 'row-number' in err and not err['row-number'] is None else 'table'
61
62
63
        return {**err, 'context': context}

    return list(map(add_context, errors))
64
65


66
def create_validata_report(goodtables_report, schema):
67
68
69
70
71
72
    """ 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"
73
        - error messages are improved
74
75
76
77
78
79
80
81
82
83
84
    """
    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']
85
86
87
    # 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
88
89
90
91
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
92
    # Computes column info
93
    schema_fields = schema.get('fields', [])
Pierre Dittgen's avatar
Pierre Dittgen committed
94
95
96
97
    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]
98
99
100
101
102

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

103
    # Provide better (french) messages
104
    errors = validata_validate.improve_messages(errors, headers, schema)
105

106
107
108
109
110
111
112
113
114
115
116
117
    # 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
118
    # Checks if there are structure errors different to invalid-column-delimiter
119
120
121
    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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
    # 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
142
143
144
145

    return report


146
def validate(schema_code, source: ValidataSource):
147
148
    """ Validate source and display report """

Pierre Dittgen's avatar
Pierre Dittgen committed
149
150
151
152
153
    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))
154
155

    source_data = extract_source_data(source)
156

157
    validata_report = create_validata_report(goodtables_report, ValidatorHelper.schema(schema_code))
158

159
    # return jsonify(validata_report)
Pierre Dittgen's avatar
Pierre Dittgen committed
160

Pierre Dittgen's avatar
Pierre Dittgen committed
161
162
    # Complete report
    val_info = ValidatorHelper.schema_info(schema_code)
163
    return render_template('validation_report.html', title='Rapport de validation',
164
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
165
                           source=source, source_type=source.type, source_data=source_data,
166
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
167
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
Pierre Dittgen's avatar
Pierre Dittgen committed
168
169
                                        {'url': url_for('scdl_validator', val_code=schema_code),
                                         'title': val_info['title']}])
170
171


172
173
174
175
176
177
178
179
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


180
181
182
183
184
185
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
186
    validators = ValidatorHelper.schema_info_list()
Pierre Dittgen's avatar
Pierre Dittgen committed
187
    flash_warning('Ce service est fourni en mode beta - certains problèmes peuvent subsister - nous mettons tout en œuvre pour améliorer son fonctionnement en continu')
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
    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
207
        val_info = ValidatorHelper.schema_info(val_code)
208
209
210
211
        input_param = request.args.get('input')

        # First form display
        if input_param is None or input_param not in ('url', 'example'):
212
            return render_template('validation_form.html', title=val_info['title'],
213
214
215
216
217
218
219
220
221
                                   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))
222
            return validate(val_code, ValidataSource(url, url, 'url'))
223
224
225
226

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
Pierre Dittgen's avatar
Pierre Dittgen committed
227
            flash_error('Aucun fichier à valider')
228
229
230
231
232
233
234
235
            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))
236
237

            return validate(val_code, ValidataSource(bytes_data(f), f.filename, 'file'))
238
239

        return 'Bizarre, vous avez dit bizarre ?'