views.py 11.9 KB
Newer Older
1
2
3
4
#!/usr/bin/env python3
"""
    Routes
"""
5
import copy
6
import json
7
8
import subprocess
import time
9
from datetime import datetime
10
from io import BytesIO
11
from operator import itemgetter
12
13
from pathlib import Path
from urllib.parse import quote_plus
14

Pierre Dittgen's avatar
Pierre Dittgen committed
15
from commonmark import commonmark
16
from flask import make_response, redirect, render_template, request, url_for
17

18
import tabulator
19
from validata_core import csv_helpers, messages
Pierre Dittgen's avatar
Pierre Dittgen committed
20
from validata_core.loaders import custom_loaders
21
from validata_ui import app
Pierre Dittgen's avatar
Pierre Dittgen committed
22
from validata_ui.ui_util import flash_error, flash_warning
23
from validata_ui.validata_util import ValidataSource
24
from validata_ui.validate_helper import ValidatorHelper
25

26
27

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

    def stringify(val):
        """ Transform value into string """
Pierre Dittgen's avatar
Pierre Dittgen committed
32
        return '' if val is None else str(val)
33

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

38
    options = {}
39
    if source.format == "csv":
40
        options['delimiter'] = csv_helpers.detect_dialect(source.source, format=source.format, scheme=source.scheme,
41
                                                          custom_loaders=custom_loaders).delimiter
42
    with tabulator.Stream(source.source, format=source.format, scheme=source.scheme, custom_loaders=custom_loaders,
43
                          **options) as stream:
44
45
        for row in stream:
            if header is None:
46
                header = ['' if v is None else v for v in row]
47
            else:
48
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
49
                nb_rows += 1
50
    preview_rows_nb = min(preview_rows_nb, nb_rows)
51
52
    return {'header': header,
            'rows_nb': nb_rows,
53
54
55
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
56
57


Pierre Dittgen's avatar
Pierre Dittgen committed
58
59
def improve_errors(errors):
    """Add context to errors, converts markdown content to HTML"""
60

Pierre Dittgen's avatar
Pierre Dittgen committed
61
62
    def improve_err(err):
        """Adds context info based on row-nb presence and converts content to HTML"""
63

Pierre Dittgen's avatar
Pierre Dittgen committed
64
65
66
67
        # 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
68

Pierre Dittgen's avatar
Pierre Dittgen committed
69
        # markdown to HTML (with default values for 'title' and 'content')
Pierre Dittgen's avatar
Pierre Dittgen committed
70

Pierre Dittgen's avatar
Pierre Dittgen committed
71
72
73
        # Set default title if no title
        if not 'title' in err:
            update_keys['title'] = '[{}]'.format(err['code'])
Pierre Dittgen's avatar
Pierre Dittgen committed
74

Pierre Dittgen's avatar
Pierre Dittgen committed
75
76
77
78
        # 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
79

Pierre Dittgen's avatar
Pierre Dittgen committed
80
81
82
        # 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
83

Pierre Dittgen's avatar
Pierre Dittgen committed
84
85
86
        # Message content
        md_content = '*content soon available*' if not 'content' in err else err['content']
        update_keys['content'] = commonmark(md_content)
87

Pierre Dittgen's avatar
Pierre Dittgen committed
88
        return {**err, **update_keys}
89

Pierre Dittgen's avatar
Pierre Dittgen committed
90
    return list(map(improve_err, errors))
91
92


93
def create_validata_report(goodtables_report, schema):
94
95
96
97
98
99
    """ 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"
100
        - error messages are improved
101
102
103
104
105
106
107
108
109
110
111
    """
    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']
112
113
114
    # 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
115
116
117
118
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
119
    # Computes column info
120
    schema_fields = schema.get('fields', [])
Pierre Dittgen's avatar
Pierre Dittgen committed
121
122
123
124
    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]
125

126
    # Provide better (french) messages
Pierre Dittgen's avatar
Pierre Dittgen committed
127
128
    errors = improve_errors(report['table']['errors'])
    del report['table']['errors']
129

130
131
132
133
134
135
136
    # 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:
137
        if err['tag'] == 'structure':
138
139
140
141
            report['table']['errors']['structure'].append(err)
        else:
            report['table']['errors']['body'].append(err)

Pierre Dittgen's avatar
Pierre Dittgen committed
142
    # Checks if there are structure errors different to invalid-column-delimiter
Pierre Dittgen's avatar
Pierre Dittgen committed
143
144
145
    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)
146

Pierre Dittgen's avatar
Pierre Dittgen committed
147
148
149
150
    # Group body errors by row id
    rows = []
    current_row_id = 0
    for err in report['table']['errors']['body']:
151
152
        if not 'row-number' in err:
            print('ERR', err)
Pierre Dittgen's avatar
Pierre Dittgen committed
153
154
155
156
157
158
159
160
161
162
163
164
165
        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
166
    report['table']['errors']['body_by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
167

168
169
170
171
172
173
174
175
176
    # 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
177
178
179
    return report


180
def validate(schema_code, source: ValidataSource):
181
182
    """ Validate source and display report """

Pierre Dittgen's avatar
Pierre Dittgen committed
183
    try:
184
185
186
187
188
        goodtables_report = ValidatorHelper.validate(
            schema_code=schema_code,
            force_strings=True,
            **source.get_tabulator_params(),
        )
Pierre Dittgen's avatar
Pierre Dittgen committed
189
190
191
    except tabulator.exceptions.FormatError:
        flash_error('Erreur : format de fichier non supporté')
        return redirect(url_for('scdl_validator', val_code=schema_code))
192
193

    source_data = extract_source_data(source)
194

Pierre Dittgen's avatar
Pierre Dittgen committed
195
196
197
    # handle report date
    date_str = goodtables_report['date']
    report_date = datetime.strptime(date_str[:date_str.find('.')], '%Y-%m-%dT%H:%M:%S')
198

Pierre Dittgen's avatar
Pierre Dittgen committed
199
    # Enhance goodtables_report
200
    validata_report = create_validata_report(goodtables_report, ValidatorHelper.schema(schema_code).descriptor)
Pierre Dittgen's avatar
Pierre Dittgen committed
201

Pierre Dittgen's avatar
Pierre Dittgen committed
202
    # Display report to the user
Pierre Dittgen's avatar
Pierre Dittgen committed
203
    val_info = ValidatorHelper.schema_info(schema_code)
204
    return render_template('validation_report.html', title='Rapport de validation',
205
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
Pierre Dittgen's avatar
Pierre Dittgen committed
206
                           validation_date=report_date.strftime('le %d/%m/%Y à %Hh%M'),
207
                           source=source, source_type=source.type, source_data=source_data,
208
                           print_mode=request.args.get('print', 'false') == 'true',
209
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
210
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
Pierre Dittgen's avatar
Pierre Dittgen committed
211
212
                                        {'url': url_for('scdl_validator', val_code=schema_code),
                                         'title': val_info['title']}])
213
214


215
216
217
218
219
220
221
222
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


223
224
225
226
227
228
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
229
    validators = ValidatorHelper.schema_info_list()
Pierre Dittgen's avatar
Pierre Dittgen committed
230
    flash_warning('Ce service est fourni en mode beta - certains problèmes peuvent subsister - nous mettons tout en œuvre pour améliorer son fonctionnement en continu')
231
232
233
234
235
236
237
238
239
    return render_template('home.html', title='Accueil', validators=validators)


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


240
241
242
243
244
245
246
247
248
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
275
276
277
278
@app.route('/pdf/<val_code>')
def pdf_report(val_code):
    """PDF report generation"""
    err_prefix = 'Erreur de génération du rapport PDF: '

    if not ValidatorHelper.schema_exist(val_code):
        flash_error(err_prefix + 'schéma inconnu')
        return redirect(url_for('home'))

    url_param = request.args.get('url')
    if not url_param:
        flash_error(err_prefix+'url non fournie')
        return redirect(url_for('home'))

    validation_url = '{}?input=url&print=true&url={}'.format(url_for('scdl_validator', val_code=val_code, _external=True),
                                                             quote_plus(url_param))

    # temporary pdf_report.pdf filename
    tmp_pdf_report = Path('/tmp') / 'validata_{}_{}_pdf_report.pdf'.format(val_code, str(time.time()))

    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:
        flash_error(err_prefix + result.stdout)
        if tmp_pdf_report.exists():
            tmp_pdf_report.unlink()
        return redirect(url_for('home'))

    response = make_response(tmp_pdf_report.open('rb').read())
    response.headers.set('Content-disposition', 'attachment', filename='report.pdf')
    response.headers.set('Content-type', 'application/pdf')
    response.headers.set('Content-length', tmp_pdf_report.stat().st_size)

    tmp_pdf_report.unlink()

    return response


279
280
281
282
283
284
285
286
287
288
@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
289
        val_info = ValidatorHelper.schema_info(val_code)
290
291
292
293
        input_param = request.args.get('input')

        # First form display
        if input_param is None or input_param not in ('url', 'example'):
294
            return render_template('validation_form.html', title=val_info['title'],
295
296
297
298
299
                                   val_info=val_info,
                                   breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'}, ])

        # Process URL
        else:
300
301
            url_param = request.args.get('url')
            if url_param is None or url_param == '':
302
303
                flash_error("Vous n'avez pas indiqué d'url à valider")
                return redirect(url_for('scdl_validator', val_code=val_code))
304
            return validate(val_code, ValidataSource('url', url_param, url_param))
305
306
307
308

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
Pierre Dittgen's avatar
Pierre Dittgen committed
309
            flash_error('Aucun fichier à valider')
310
311
312
313
314
315
316
317
            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))
318

319
            return validate(val_code, ValidataSource('file', f.filename, bytes_data(f)))
320
321

        return 'Bizarre, vous avez dit bizarre ?'