views.py 17.5 KB
Newer Older
1
2
3
"""
    Routes
"""
4
import copy
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
5
import io
Pierre Dittgen's avatar
Pierre Dittgen committed
6
import itertools
7
import json
Christophe Benz's avatar
Christophe Benz committed
8
import logging
9
import subprocess
Christophe Benz's avatar
Christophe Benz committed
10
import tempfile
11
from datetime import datetime
12
from operator import itemgetter
13
14
from pathlib import Path
from urllib.parse import quote_plus
15

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
16
import requests
17
from backports.datetime_fromisoformat import MonkeyPatch
18
from commonmark import commonmark
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
19
from flask import make_response, redirect, render_template, request, url_for
Pierre Dittgen's avatar
Pierre Dittgen committed
20
21
from validata_core import compute_badge, csv_helpers, messages
from validata_core.loaders import custom_loaders
22

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
23
24
import tabulator

25
26
27
28
from . import app, config
from .ui_util import flash_error, flash_warning
from .validata_util import ValidataSource
from .validate_helper import ValidatorHelper
29

30
31
MonkeyPatch.patch_fromisoformat()

Christophe Benz's avatar
Christophe Benz committed
32
33
log = logging.getLogger(__name__)

34
35

def extract_source_data(source: ValidataSource, preview_rows_nb=5):
36
    """ Computes table preview """
37
38
39

    def stringify(val):
        """ Transform value into string """
Pierre Dittgen's avatar
Pierre Dittgen committed
40
        return '' if val is None else str(val)
41

42
43
    header = None
    rows = []
Pierre Dittgen's avatar
Pierre Dittgen committed
44
    nb_rows = 0
45

46
    options = {}
47
    if source.format == "csv":
48
        options['delimiter'] = csv_helpers.detect_dialect(source.source, format=source.format, scheme=source.scheme,
49
                                                          custom_loaders=custom_loaders).delimiter
50
    with tabulator.Stream(source.source, format=source.format, scheme=source.scheme, custom_loaders=custom_loaders,
51
                          **options) as stream:
52
53
        for row in stream:
            if header is None:
54
                header = ['' if v is None else v for v in row]
55
            else:
56
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
57
                nb_rows += 1
58
    preview_rows_nb = min(preview_rows_nb, nb_rows)
59
60
    return {'header': header,
            'rows_nb': nb_rows,
61
62
63
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
64
65


Pierre Dittgen's avatar
Pierre Dittgen committed
66
67
def improve_errors(errors):
    """Add context to errors, converts markdown content to HTML"""
68

Pierre Dittgen's avatar
Pierre Dittgen committed
69
70
    def improve_err(err):
        """Adds context info based on row-nb presence and converts content to HTML"""
71

Pierre Dittgen's avatar
Pierre Dittgen committed
72
73
74
75
        # Context
        update_keys = {
            'context': 'body' if 'row-number' in err and not err['row-number'] is None else 'table',
        }
Pierre Dittgen's avatar
Pierre Dittgen committed
76

Pierre Dittgen's avatar
Pierre Dittgen committed
77
        # markdown to HTML (with default values for 'title' and 'content')
Pierre Dittgen's avatar
Pierre Dittgen committed
78

Pierre Dittgen's avatar
Pierre Dittgen committed
79
80
81
        # Set default title if no title
        if not 'title' in err:
            update_keys['title'] = '[{}]'.format(err['code'])
Pierre Dittgen's avatar
Pierre Dittgen committed
82

Pierre Dittgen's avatar
Pierre Dittgen committed
83
84
85
86
        # Convert message to markdown only if no content
        # => for pre-checks errors
        if 'message' in err and not 'content' in err:
            update_keys['message'] = commonmark(err['message'])
Pierre Dittgen's avatar
Pierre Dittgen committed
87

Pierre Dittgen's avatar
Pierre Dittgen committed
88
89
90
        # Else, default message
        elif not 'message' in err or err['message'] is None:
            update_keys['message'] = '[{}]'.format(err['code'])
Pierre Dittgen's avatar
Pierre Dittgen committed
91

Pierre Dittgen's avatar
Pierre Dittgen committed
92
93
94
        # Message content
        md_content = '*content soon available*' if not 'content' in err else err['content']
        update_keys['content'] = commonmark(md_content)
95

Pierre Dittgen's avatar
Pierre Dittgen committed
96
        return {**err, **update_keys}
97

Pierre Dittgen's avatar
Pierre Dittgen committed
98
    return list(map(improve_err, errors))
99
100


Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
101
def create_validata_ui_report(validata_core_report, schema_dict):
102
103
104
105
106
107
    """ 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"
108
        - error messages are improved
109
    """
Pierre Dittgen's avatar
Pierre Dittgen committed
110
    report = copy.deepcopy(validata_core_report)
111
112
113
114
115
116
117
118
119

    # 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']
120
121
122
    # 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
123
124
125
126
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
127
    # Computes column info
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
128
129
    fields_dict = {f['name']: (f.get('title', 'titre non défini'), f.get('description', ''))
                   for f in schema_dict.get('fields', [])}
Pierre Dittgen's avatar
Pierre Dittgen committed
130
131
132
    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]
133

134
    # Provide better (french) messages
Pierre Dittgen's avatar
Pierre Dittgen committed
135
136
    errors = improve_errors(report['table']['errors'])
    del report['table']['errors']
137

138
139
140
141
142
143
144
    # 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:
145
        if err['tag'] == 'structure':
146
147
148
149
            report['table']['errors']['structure'].append(err)
        else:
            report['table']['errors']['body'].append(err)

Pierre Dittgen's avatar
Pierre Dittgen committed
150
    # Checks if there are structure errors different to invalid-column-delimiter
Pierre Dittgen's avatar
Pierre Dittgen committed
151
152
153
    structure_errors = report['table']['errors']['structure']
    report['table']['do_display_body_errors'] = len(structure_errors) == 0 or \
        all(err['code'] == 'invalid-column-delimiter' for err in structure_errors)
154

Pierre Dittgen's avatar
Pierre Dittgen committed
155
156
157
158
159
160
161
162
    # Checks if a column comparison is needed
    header_errors = ('missing-headers', 'extra-headers', 'wrong-headers-order')
    structure_errors = [{**err, 'in_column_comp': err['code'] in header_errors} for err in structure_errors]
    report['table']['errors']['structure'] = structure_errors
    column_comparison_needed = any(err['in_column_comp'] == True for err in structure_errors)
    column_comparison_table = []
    if column_comparison_needed:
        column_comparison_table = []
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
163
        field_names = [f['name'] for f in schema_dict.get('fields', [])]
164
        has_case_errors = False
Pierre Dittgen's avatar
Pierre Dittgen committed
165
166
        for t in itertools.zip_longest(headers, field_names, fillvalue=''):
            status = 'ok' if t[0] == t[1] else 'ko'
167
168
            if not has_case_errors and status == 'ko' and t[0].lower() == t[1].lower():
                has_case_errors = True
Pierre Dittgen's avatar
Pierre Dittgen committed
169
            column_comparison_table.append((*t, status))
170
171
172
173
174
        info = {}
        info['table'] = column_comparison_table
        info['has_missing'] = len(headers) < len(field_names)
        info['has_case_errors'] = has_case_errors
        report['table']['column_comparison_info'] = info
Pierre Dittgen's avatar
Pierre Dittgen committed
175
176
    report['table']['column_comparison_needed'] = column_comparison_needed

Pierre Dittgen's avatar
Pierre Dittgen committed
177
178
179
180
    # Group body errors by row id
    rows = []
    current_row_id = 0
    for err in report['table']['errors']['body']:
181
182
        if not 'row-number' in err:
            print('ERR', err)
Pierre Dittgen's avatar
Pierre Dittgen committed
183
184
185
186
187
188
189
190
191
192
193
194
195
        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
Pierre Dittgen's avatar
Pierre Dittgen committed
196
    report['table']['errors']['body_by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
197

198
199
200
201
202
203
204
205
206
    # Sort by error names in statistics
    stats = report['table']['error-stats']
    code_title_map = messages.ERROR_MESSAGE_DEFAULT_TITLE
    for key in ('structure-errors', 'value-errors'):
        # convert dict into tuples with french title instead of error code
        # and sorts by title
        stats[key]['count-by-code'] = sorted([(code_title_map.get(k, k), v) for k, v in stats[key]['count-by-code'].items()],
                                             key=itemgetter(0))

Pierre Dittgen's avatar
Pierre Dittgen committed
207
208
209
    return report


Pierre Dittgen's avatar
Pierre Dittgen committed
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
def compute_badge_message_and_color(badge):
    """Computes message and color from badge information"""
    structure = badge['structure']
    body = badge.get('body')

    # Bad structure, stop here
    if structure == 'KO':
        return (
            'structure invalide', 'red')

    # No body error
    if body == 'OK':
        return ('structure invalide', 'orange') if structure == 'WARN' else ('valide', 'green')

    # else compute quality ratio percent
Christophe Benz's avatar
Christophe Benz committed
225
    p = (1 - badge['error-ratio']) * 100.0
Pierre Dittgen's avatar
Pierre Dittgen committed
226
227
228
229
230
231
232
233
    msg = 'cellules valides : {:.1f}%'.format(p)
    return (msg, 'red') if body == 'KO' else (msg, 'orange')


def get_badge_url_and_message(badge):
    """Gets badge url from badge information"""

    msg, color = compute_badge_message_and_color(badge)
234
235
    return ('{}static/v1.svg?label=Validata&message={}&color={}'.format(
        config.SHIELDS_IO_BASE_URL, quote_plus(msg), color), msg)
Pierre Dittgen's avatar
Pierre Dittgen committed
236
237


Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
238
def validate(schema_url, source: ValidataSource):
239
240
    """ Validate source and display report """

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
241
242
243
244
245
246
    if config.API_VALIDATE_ENDPOINT is None:
        flash_error("No Validate endpoint defined :-(")
        return redirect(url_for("custom_validator"))
    api_url = config.API_VALIDATE_ENDPOINT
    headers = {"Accept": "application/json"}

Pierre Dittgen's avatar
Pierre Dittgen committed
247
    try:
248

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
249
250
251
252
253
254
255
256
257
258
259
260
        if source.is_url():
            params = {
                "schema": schema_url,
                "url": source.get_url(),
            }
            req = requests.get(api_url, params=params, headers=headers)

        else:
            files = {'file': (source.name, io.BytesIO(source.source))}
            data = {'schema': schema_url}
            req = requests.post(api_url, data=data, files=files, headers=headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
261
262
263
264
265
266
267
        # 400
        if req.status_code == 400:
            json_response = req.json()
            flash_error("Une erreur est survenue durant la validation: {}"
                        .format(json_response.get('message')))
            return redirect(url_for("home"))

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
268
        if not req.ok:
Pierre Dittgen's avatar
Pierre Dittgen committed
269
            flash_error("Un erreur s'est produite côté serveur :-(")
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
270
            return redirect(url_for("home"))
271

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
272
273
274
        json_response = req.json()
        validata_core_report = json_response['report']
        schema_dict = json_response['schema']
275

Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
276
277
278
279
    except requests.ConnectionError as err:
        logging.exception(err)
        flash_error(str(err))
        return redirect(url_for('home'))
280

Pierre Dittgen's avatar
Pierre Dittgen committed
281
    # Computes badge from report and badge configuration
282
    badge = compute_badge(validata_core_report, config.BADGE_CONFIG)
Pierre Dittgen's avatar
Pierre Dittgen committed
283
284
    badge_url, badge_msg = get_badge_url_and_message(badge)

285
286
287
288
289
290
    source_errors = [err for err in validata_core_report['tables'][0]['errors'] if err['code'] == 'source-error']
    if source_errors:
        err = source_errors[0]
        msg = "l'encodage du fichier est invalide. Veuillez le corriger" if 'charmap' in err[
            'message'] else err['message']
        flash_error('Erreur de source : {}'.format(msg))
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
291
        return redirect(url_for('custom_validator'))
292

293
    source_data = extract_source_data(source)
294

Pierre Dittgen's avatar
Pierre Dittgen committed
295
    # handle report date
296
    report_datetime = datetime.fromisoformat(validata_core_report['date']).astimezone()
297

Pierre Dittgen's avatar
Pierre Dittgen committed
298
    # Enhance validata_core_report
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
299
    validata_report = create_validata_ui_report(validata_core_report, schema_dict)
Pierre Dittgen's avatar
Pierre Dittgen committed
300

Pierre Dittgen's avatar
Pierre Dittgen committed
301
    # Display report to the user
302
    return render_template('validation_report.html', title='Rapport de validation',
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
303
                           val_info={}, report=validata_report,
Pierre Dittgen's avatar
Pierre Dittgen committed
304
                           schema_url=schema_url,
305
                           validation_date=report_datetime.strftime('le %d/%m/%Y à %Hh%M'),
306
                           source=source, source_type=source.type, source_data=source_data,
307
                           print_mode=request.args.get('print', 'false') == 'true',
Pierre Dittgen's avatar
Pierre Dittgen committed
308
                           badge_url=badge_url, badge_msg=badge_msg,
309
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
310
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
Pierre Dittgen's avatar
Pierre Dittgen committed
311
312
                                        {'url': url_for('custom_validator'), 'title': 'Schéma personnalisé'},
                                        ])
313
314


315
316
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
317
    iob = io.BytesIO()
318
319
320
321
322
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


323
324
325
326
327
328
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
329
    validators = ValidatorHelper.schema_info_list()
Christophe Benz's avatar
Christophe Benz committed
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
    external_validators = [
        {
            "title": "INSPIRE",
            "description": "proposé par la Commission Européenne pour tester des ressources géographiques (données, services ou métadonnées)",
            "url": "http://inspire-sandbox.jrc.ec.europa.eu/validator/"
        },
        {
            "title": "BAL",
            "description": "proposé par la mission Etalab pour tester des données Adresse produites localement (format BAL 1.1)",
            "url": "https://adresse.data.gouv.fr/bases-locales/validateur"
        },
        {
            "title": "CVDTC",
            "description": "proposé par l'AFIMB dans le cadre du projet Chouette pour tester des données de transport collectif (GTFS ou NEPTUNE)",
            "url": "http://www.conversion-validation-donnees-tc.org/"
        },
    ]
Christophe Benz's avatar
Christophe Benz committed
347
    flash_warning('Ce service est fourni en mode beta - certains problèmes peuvent subsister - nous mettons tout en œuvre pour améliorer son fonctionnement en continu.')
Christophe Benz's avatar
Christophe Benz committed
348
    return render_template('home.html', title='Accueil', validators=validators, external_validators=external_validators)
349
350
351
352
353
354
355
356


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


Pierre Dittgen's avatar
Pierre Dittgen committed
357
358
@app.route('/pdf')
def pdf_report():
359
    """PDF report generation"""
360
    err_prefix = 'Erreur de génération du rapport PDF'
361
362
363

    url_param = request.args.get('url')
    if not url_param:
364
        flash_error(err_prefix + ': URL non fournie')
Pierre Dittgen's avatar
Pierre Dittgen committed
365
        return redirect(url_for('home'))
366

Pierre Dittgen's avatar
Pierre Dittgen committed
367
368
369
370
371
372
373
    schema_param = request.args.get('schema')
    if not schema_param:
        flash_error(err_prefix + ': URL de schema non fournie')
        return redirect(url_for('home'))

    validation_url = '{}?input=url&print=true&url={}&schema={}'.format(url_for('custom_validator', _external=True),
                                                                       quote_plus(url_param), quote_plus(schema_param))
374

Pierre Dittgen's avatar
Pierre Dittgen committed
375
    with tempfile.NamedTemporaryFile(prefix='validata_{}_report_'.format(datetime.now().timestamp()), suffix='.pdf') as tmpfile:
Christophe Benz's avatar
Christophe Benz committed
376
        tmp_pdf_report = Path(tmpfile.name)
377
378
379
380
381

    cmd = ['chromium', '--headless', '--disable-gpu',
           '--print-to-pdf={}'.format(str(tmp_pdf_report)), validation_url]
    result = subprocess.run(cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    if result.returncode != 0:
382
383
        flash_error(err_prefix)
        log.error("Command %r returned an error: %r", cmd, result.stdout.decode('utf-8'))
384
385
        if tmp_pdf_report.exists():
            tmp_pdf_report.unlink()
Pierre Dittgen's avatar
Pierre Dittgen committed
386
        return redirect(url_for('home'))
387

Pierre Dittgen's avatar
Pierre Dittgen committed
388
    pdf_filename = 'Rapport de validation {}.pdf'.format(datetime.now().strftime('%d-%m-%Y %Hh%M'))
Christophe Benz's avatar
Christophe Benz committed
389
    response = make_response(tmp_pdf_report.read_bytes())
Pierre Dittgen's avatar
Pierre Dittgen committed
390
    response.headers.set('Content-disposition', 'attachment', filename=pdf_filename)
391
392
393
394
395
396
397
398
    response.headers.set('Content-type', 'application/pdf')
    response.headers.set('Content-length', tmp_pdf_report.stat().st_size)

    tmp_pdf_report.unlink()

    return response


Pierre Dittgen's avatar
Pierre Dittgen committed
399
@app.route('/validators/form', methods=['GET', 'POST'])
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
def custom_validator():
    """ Validator page """

    if config.API_VALIDATE_ENDPOINT is None:
        flash_error("URL de connexion à l'API non indiquée :-(")
        return redirect(url_for('home'))

    if request.method == 'GET':

        input_param = request.args.get('input')
        url_param = request.args.get("url")
        schema_param = request.args.get("schema")

        # First form display
        if input_param is None:
Pierre Dittgen's avatar
Pierre Dittgen committed
415
            return render_template('validation_form.html', title="Formulaire de validation",
Pierre Dittgen's avatar
wip  
Pierre Dittgen committed
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
                                   val_info=None,
                                   breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'}, ])

        # Process URL
        else:
            if url_param is None or url_param == '':
                flash_error("Vous n'avez pas indiqué d'url à valider")
                return redirect(url_for('custom_validator'))
            if schema_param is None or schema_param == '':
                flash_error("Vous n'avez pas indiqué d'url de schéma")
                return redirect(url_for('custom_validator'))
            try:
                return validate(schema_param, ValidataSource('url', url_param, url_param))
            except tabulator.exceptions.FormatError as e:
                flash_error('Erreur : Format de ressource non supporté')
                log.info(e)
                return redirect(url_for('custom_validator'))
            except tabulator.exceptions.HTTPError as e:
                flash_error('Erreur : impossible d\'accéder au fichier source en ligne')
                log.info(e)
                return redirect(url_for('custom_validator'))

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
            flash_error('Aucun fichier à valider')
            return redirect(url_for('custom_validator'))

        schema_param = request.form.get('schema')
        if schema_param is None:
            flash_error('Aucun schéma défini')
            return redirect(url_for('custom_validator'))

        # 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('custom_validator'))

            b_content = bytes_data(f)
            return validate(schema_param, ValidataSource('file', f.filename, b_content))

        return 'Bizarre, vous avez dit bizarre ?'