views.py 12.8 KB
Newer Older
1 2 3 4
#!/usr/bin/env python3
"""
    Routes
"""
5
import copy
Pierre Dittgen's avatar
Pierre Dittgen committed
6
import itertools
7
import json
8 9
import subprocess
import time
10
from datetime import datetime
11
from io import BytesIO
12
from operator import itemgetter
13 14
from pathlib import Path
from urllib.parse import quote_plus
15

Pierre Dittgen's avatar
Pierre Dittgen committed
16
from commonmark import commonmark
17
from flask import make_response, redirect, render_template, request, url_for
18

19
import tabulator
20
from validata_core import csv_helpers, messages
Pierre Dittgen's avatar
Pierre Dittgen committed
21
from validata_core.loaders import custom_loaders
22
from validata_ui import app
Pierre Dittgen's avatar
Pierre Dittgen committed
23
from validata_ui.ui_util import flash_error, flash_warning
24
from validata_ui.validata_util import ValidataSource
25
from validata_ui.validate_helper import ValidatorHelper
26

27 28

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

    def stringify(val):
        """ Transform value into string """
Pierre Dittgen's avatar
Pierre Dittgen committed
33
        return '' if val is None else str(val)
34

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

39
    options = {}
40
    if source.format == "csv":
41
        options['delimiter'] = csv_helpers.detect_dialect(source.source, format=source.format, scheme=source.scheme,
42
                                                          custom_loaders=custom_loaders).delimiter
43
    with tabulator.Stream(source.source, format=source.format, scheme=source.scheme, custom_loaders=custom_loaders,
44
                          **options) as stream:
45 46
        for row in stream:
            if header is None:
47
                header = ['' if v is None else v for v in row]
48
            else:
49
                rows.append(list(map(stringify, row)))
Pierre Dittgen's avatar
Pierre Dittgen committed
50
                nb_rows += 1
51
    preview_rows_nb = min(preview_rows_nb, nb_rows)
52 53
    return {'header': header,
            'rows_nb': nb_rows,
54 55 56
            'data_rows': rows,
            'preview_rows_nb': preview_rows_nb,
            'preview_rows': rows[:preview_rows_nb]}
57 58


Pierre Dittgen's avatar
Pierre Dittgen committed
59 60
def improve_errors(errors):
    """Add context to errors, converts markdown content to HTML"""
61

Pierre Dittgen's avatar
Pierre Dittgen committed
62 63
    def improve_err(err):
        """Adds context info based on row-nb presence and converts content to HTML"""
64

Pierre Dittgen's avatar
Pierre Dittgen committed
65 66 67 68
        # 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
69

Pierre Dittgen's avatar
Pierre Dittgen committed
70
        # markdown to HTML (with default values for 'title' and 'content')
Pierre Dittgen's avatar
Pierre Dittgen committed
71

Pierre Dittgen's avatar
Pierre Dittgen committed
72 73 74
        # Set default title if no title
        if not 'title' in err:
            update_keys['title'] = '[{}]'.format(err['code'])
Pierre Dittgen's avatar
Pierre Dittgen committed
75

Pierre Dittgen's avatar
Pierre Dittgen committed
76 77 78 79
        # 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
80

Pierre Dittgen's avatar
Pierre Dittgen committed
81 82 83
        # 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
84

Pierre Dittgen's avatar
Pierre Dittgen committed
85 86 87
        # Message content
        md_content = '*content soon available*' if not 'content' in err else err['content']
        update_keys['content'] = commonmark(md_content)
88

Pierre Dittgen's avatar
Pierre Dittgen committed
89
        return {**err, **update_keys}
90

Pierre Dittgen's avatar
Pierre Dittgen committed
91
    return list(map(improve_err, errors))
92 93


Pierre Dittgen's avatar
Pierre Dittgen committed
94
def create_validata_ui_report(validata_core_report, schema):
95 96 97 98 99 100
    """ 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"
101
        - error messages are improved
102
    """
Pierre Dittgen's avatar
Pierre Dittgen committed
103
    report = copy.deepcopy(validata_core_report)
104 105 106 107 108 109 110 111 112

    # 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']
113 114 115
    # 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
116 117 118 119
    # Handy col_count info
    headers = report['table'].get('headers', [])
    report['table']['col_count'] = len(headers)

Pierre Dittgen's avatar
Pierre Dittgen committed
120
    # Computes column info
121
    schema_fields = schema.get('fields', [])
Pierre Dittgen's avatar
Pierre Dittgen committed
122 123 124 125
    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]
126

127
    # Provide better (french) messages
Pierre Dittgen's avatar
Pierre Dittgen committed
128 129
    errors = improve_errors(report['table']['errors'])
    del report['table']['errors']
130

131 132 133 134 135 136 137
    # 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:
138
        if err['tag'] == 'structure':
139 140 141 142
            report['table']['errors']['structure'].append(err)
        else:
            report['table']['errors']['body'].append(err)

Pierre Dittgen's avatar
Pierre Dittgen committed
143
    # Checks if there are structure errors different to invalid-column-delimiter
Pierre Dittgen's avatar
Pierre Dittgen committed
144 145 146
    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)
147

Pierre Dittgen's avatar
Pierre Dittgen committed
148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
    # 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 = []
        field_names = [f['name'] for f in schema_fields]
        for t in itertools.zip_longest(headers, field_names, fillvalue=''):
            status = 'ok' if t[0] == t[1] else 'ko'
            column_comparison_table.append((*t, status))
    report['table']['column_comparison_needed'] = column_comparison_needed
    report['table']['column_comparison_info'] = column_comparison_table

Pierre Dittgen's avatar
Pierre Dittgen committed
163 164 165 166
    # Group body errors by row id
    rows = []
    current_row_id = 0
    for err in report['table']['errors']['body']:
167 168
        if not 'row-number' in err:
            print('ERR', err)
Pierre Dittgen's avatar
Pierre Dittgen committed
169 170 171 172 173 174 175 176 177 178 179 180 181
        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
182
    report['table']['errors']['body_by_rows'] = rows
Pierre Dittgen's avatar
Pierre Dittgen committed
183

184 185 186 187 188 189 190 191 192
    # 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
193 194 195
    return report


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

199
    try:
Pierre Dittgen's avatar
Pierre Dittgen committed
200
        validata_core_report = ValidatorHelper.validate(
201 202 203 204
            schema_code=schema_code,
            force_strings=True,
            **source.get_tabulator_params(),
        )
205 206 207
    except tabulator.exceptions.FormatError:
        flash_error('Erreur : format de fichier non supporté')
        return redirect(url_for('scdl_validator', val_code=schema_code))
208 209

    source_data = extract_source_data(source)
210

Pierre Dittgen's avatar
Pierre Dittgen committed
211
    # handle report date
Pierre Dittgen's avatar
Pierre Dittgen committed
212
    date_str = validata_core_report['date']
Pierre Dittgen's avatar
Pierre Dittgen committed
213
    report_date = datetime.strptime(date_str[:date_str.find('.')], '%Y-%m-%dT%H:%M:%S')
214

Pierre Dittgen's avatar
Pierre Dittgen committed
215 216
    # Enhance validata_core_report
    validata_report = create_validata_ui_report(validata_core_report, ValidatorHelper.schema(schema_code).descriptor)
Pierre Dittgen's avatar
Pierre Dittgen committed
217

Pierre Dittgen's avatar
Pierre Dittgen committed
218
    # Display report to the user
Pierre Dittgen's avatar
Pierre Dittgen committed
219
    val_info = ValidatorHelper.schema_info(schema_code)
220
    return render_template('validation_report.html', title='Rapport de validation',
221
                           val_info=ValidatorHelper.schema_info(schema_code), report=validata_report,
Pierre Dittgen's avatar
Pierre Dittgen committed
222
                           validation_date=report_date.strftime('le %d/%m/%Y à %Hh%M'),
223
                           source=source, source_type=source.type, source_data=source_data,
224
                           print_mode=request.args.get('print', 'false') == 'true',
225
                           report_str=json.dumps(validata_report, sort_keys=True, indent=2),
Pierre Dittgen's avatar
Pierre Dittgen committed
226
                           breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'},
Pierre Dittgen's avatar
Pierre Dittgen committed
227 228
                                        {'url': url_for('scdl_validator', val_code=schema_code),
                                         'title': val_info['title']}])
229 230


231 232 233 234 235 236 237 238
def bytes_data(f):
    """ Gets bytes data from Werkzeug FileStorage instance """
    iob = BytesIO()
    f.save(iob)
    iob.seek(0)
    return iob.getvalue()


239 240 241 242 243 244
# Routes


@app.route('/')
def home():
    """ Home page """
Pierre Dittgen's avatar
Pierre Dittgen committed
245
    validators = ValidatorHelper.schema_info_list()
Pierre Dittgen's avatar
Pierre Dittgen committed
246
    flash_warning('Ce service est fourni en mode beta - certains problèmes peuvent subsister - nous mettons tout en œuvre pour améliorer son fonctionnement en continu')
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'))


256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
@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


295 296 297 298 299 300 301 302 303 304
@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
305
        val_info = ValidatorHelper.schema_info(val_code)
306 307 308 309
        input_param = request.args.get('input')

        # First form display
        if input_param is None or input_param not in ('url', 'example'):
310
            return render_template('validation_form.html', title=val_info['title'],
311 312 313 314 315
                                   val_info=val_info,
                                   breadcrumbs=[{'url': url_for('home'), 'title': 'Accueil'}, ])

        # Process URL
        else:
316 317
            url_param = request.args.get('url')
            if url_param is None or url_param == '':
318 319
                flash_error("Vous n'avez pas indiqué d'url à valider")
                return redirect(url_for('scdl_validator', val_code=val_code))
320
            return validate(val_code, ValidataSource('url', url_param, url_param))
321 322 323 324

    else:  # POST
        input_param = request.form.get('input')
        if input_param is None:
325
            flash_error('Aucun fichier à valider')
326 327 328 329 330 331 332 333
            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))
334

335
            return validate(val_code, ValidataSource('file', f.filename, bytes_data(f)))
336 337

        return 'Bizarre, vous avez dit bizarre ?'