views.py 9.53 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

Pierre Dittgen's avatar
Pierre Dittgen committed
16
from flask import Flask, jsonify, redirect, render_template, request, url_for
17 18
import tabulator

19
from io import BytesIO
20

21 22

def extract_source_data(source: ValidataSource, preview_rows_nb=5):
23
    """ Computes table preview """
24 25 26 27 28 29 30

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

31 32
    header = None
    rows = []
Pierre Dittgen's avatar
Pierre Dittgen committed
33
    nb_rows = 0
34 35 36 37 38 39

    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:
40 41 42 43
        for row in stream:
            if header is None:
                header = row
            else:
44
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
45
                nb_rows += 1
46
    preview_rows_nb = min(preview_rows_nb, nb_rows)
47 48
    return {'header': header,
            'rows_nb': nb_rows,
49 50 51
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
52 53


Pierre Dittgen's avatar
Pierre Dittgen committed
54 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
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
83 84
    ('invalid-column-delimiter', 'table'),
    ('missing-headers', 'table'),
Pierre Dittgen's avatar
Pierre Dittgen committed
85 86 87 88 89

    # Custom checks fall in default case: body
])


90 91 92 93 94 95 96 97 98 99 100 101 102 103
def improve_messages(errors):
    """ Translates and improve error messages """

    def update_message(error, new_message):
        """ set a new error message """

        # Stores previous message if exists
        if 'message' in error:
            error['_original_message'] = error['message']
        error['message'] = new_message

    for error in errors:

        if error['code'] == 'blank-row':
Pierre Dittgen's avatar
Pierre Dittgen committed
104 105
            update_message(error, 'la ligne est vide')

106
        elif error['code'] == 'duplicate-row':
Pierre Dittgen's avatar
Pierre Dittgen committed
107 108 109 110 111 112 113 114 115
            msg_prefix = 'la ligne est identique '
            row_numbers = error['message-data']['row_numbers']
            if len(row_numbers) == 1:
                msg = msg_prefix + "à la ligne {}".format(row_numbers)
            else:
                idx = row_numbers.rfind(',')
                param_str = row_numbers[:idx] + ' et' + row_numbers[idx+1:]
                msg = msg_prefix + "aux lignes {}".format(param_str)
            update_message(error, msg)
116 117 118 119 120

    # Return updated error list
    return errors


121
def contextualize(errors):
Pierre Dittgen's avatar
Pierre Dittgen committed
122 123
    """ add context to errors """

124 125 126
    return [{**err, 'context': ERR_CODE_TO_CONTEXT.get(err['code'], 'body')} for err in errors]


Pierre Dittgen's avatar
Pierre Dittgen committed
127
def create_validata_report(goodtables_report, schema_fields=[]):
128 129 130 131 132 133
    """ 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"
134
        - error messages are improved
135 136 137 138 139 140 141 142 143 144 145
    """
    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']
146 147 148
    # 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
149 150 151 152 153 154 155
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

    # Headers title
    fields_dict = {f['name']: f['title'] for f in schema_fields}
    report['table']['headers_title'] = [fields_dict.get(h, '?? colonne inconnue dans le schéma ??') for h in headers]
156 157 158 159 160

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

161 162 163
    # Provide better (french) messages
    errors = improve_messages(errors)

164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
    # 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']:
        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': {}})
186 187 188 189 190 191 192

        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
193
    report['table']['errors']['by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
194 195 196 197

    return report


198
def validate(schema_code, source: ValidataSource):
199 200
    """ Validate source and display report """

201 202 203
    goodtables_report = ValidatorHelper.validate(schema_code, **source.get_goodtables_source())

    source_data = extract_source_data(source)
204

Pierre Dittgen's avatar
Pierre Dittgen committed
205
    validata_report = create_validata_report(goodtables_report, ValidatorHelper.schema_fields(schema_code))
206

207
    # return jsonify(validata_report)
Pierre Dittgen's avatar
Pierre Dittgen committed
208

Pierre Dittgen's avatar
Pierre Dittgen committed
209 210
    # Complete report
    val_info = ValidatorHelper.schema_info(schema_code)
211
    return render_template('validation_report.html', title='Rapport de validation',
212
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
213
                           source=source, source_type=source.type, source_data=source_data,
214
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
215 216
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
                                        {'url': url_for('scdl_validator', val_code=schema_code), 'title': val_info['title']}])
217 218


219 220 221 222 223 224 225 226
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


227 228 229 230 231 232
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
233
    validators = ValidatorHelper.schema_info_list()
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
    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
260
        val_info = ValidatorHelper.schema_info(val_code)
261 262 263 264 265 266 267 268 269 270 271 272 273 274
        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))
275
            return validate(val_code, ValidataSource(url, url, 'url'))
276 277 278 279 280 281 282 283 284 285 286 287 288

    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))
289 290

            return validate(val_code, ValidataSource(bytes_data(f), f.filename, 'file'))
291 292

        return 'Bizarre, vous avez dit bizarre ?'