Blame view

web/run.py 59.8 KB
bde97e4d   Goutte   Add more changes ...
1
2
# coding=utf-8

b33d5f62   hitier   Change StringIO i...
3
from io import StringIO
bc18b96c   Goutte   Implement first (...
4
import datetime
8644387c   Goutte   Use real data.
5
import gzip
bc18b96c   Goutte   Implement first (...
6
7
8
import json
import logging
import random
2fedd73b   Goutte   Initial implement...
9
import tarfile
bde97e4d   Goutte   Add more changes ...
10
import time
3f9c36da   hitier   Change urllib import
11
import urllib.request as urllib_request
9f57dceb   Goutte   Add half the brid...
12
import requests
11d86851   Goutte   Add support for s...
13
import re  # regex
45700883   Myriam Bouchemit   temperature for e...
14
import numpy
9f57dceb   Goutte   Add half the brid...
15

11d86851   Goutte   Add support for s...
16
from csv import writer as csv_writer, DictWriter as csv_dict_writer
596da00d   Goutte   Add more exceptio...
17
from math import sqrt, isnan
bc18b96c   Goutte   Implement first (...
18
19
20
from os import environ, remove as removefile
from os.path import isfile, join, abspath, dirname

3309d7c3   Goutte   Improve the date ...
21
# (i) Just ignore the requirements warning, we already require `dateutils`.
fb383448   Goutte   Implement the cac...
22
from dateutil.relativedelta import relativedelta
3309d7c3   Goutte   Improve the date ...
23
24
from dateutil import parser as dateparser

9390ec89   Goutte   Initial experimen...
25
from flask import Flask
9390ec89   Goutte   Initial experimen...
26
from flask import request
bc18b96c   Goutte   Implement first (...
27
from flask import url_for, send_from_directory, abort as abort_flask
a3e57903   hitier   Change jinja imports
28
29
from jinja2 import Environment, FileSystemLoader
from jinja2.utils import markupsafe
29b428ad   hitier   Change yaml imports
30
31
32
33
34
from yaml import load as yaml_load, dump
try:
    from yaml import CLoader as Loader
except ImportError:
    from yaml import Loader
f10f34d1   Goutte   More logic.
35
from netCDF4 import Dataset, date2num
9390ec89   Goutte   Initial experimen...
36

dabb9d5f   Goutte   Fix the layers' c...
37

9390ec89   Goutte   Initial experimen...
38
39
40
41
42
43
# PATH RELATIVITY #############################################################

THIS_DIRECTORY = dirname(abspath(__file__))


def get_path(relative_path):
a4a9ef03   Goutte   Cache generated C...
44
    """Get an absolute path from the relative path to this script directory."""
9390ec89   Goutte   Initial experimen...
45
46
47
48
49
50
51
52
53
54
55
    return abspath(join(THIS_DIRECTORY, relative_path))


# COLLECT GLOBAL INFORMATION FROM SOURCES #####################################

# VERSION
with open(get_path('../VERSION'), 'r') as version_file:
    version = version_file.read().strip()

# CONFIG
with open(get_path('../config.yml'), 'r') as config_file:
29b428ad   hitier   Change yaml imports
56
    config = yaml_load(config_file.read(), Loader=Loader)
9390ec89   Goutte   Initial experimen...
57

c0df94bc   Goutte   Adding more logs.
58
FILE_DATE_FMT = "%Y-%m-%dT%H:%M:%S"
1185f353   Goutte   Fix the CME Catal...
59
MOMENT_DATE_FMT = "%Y-%m-%dT%H:%M:%SZ"
54bb1311   Goutte   Bring back the CM...
60
CME_DATE_FMT = "%Y-%m-%dT%H:%MZ"
c0df94bc   Goutte   Adding more logs.
61

6288347a   Goutte   Add the HELP page...
62
63
# Are we on the SSA instance for ESA?
SSA = environ.get('SSA') == 'true'
20fdc1a4   Goutte   Set log level fro...
64
DEBUG = environ.get('DEBUG') == 'true'
6288347a   Goutte   Add the HELP page...
65
# SSA = True
9390ec89   Goutte   Initial experimen...
66

f75faf5f   Goutte   WIP
67
68
# LOGGING #####################################################################

1324cc91   Goutte   Make the footer i...
69
70
LOG_FILE = get_path('run.log')

f75faf5f   Goutte   WIP
71
log = logging.getLogger("HelioPropa")
20fdc1a4   Goutte   Set log level fro...
72
73
74
75
if DEBUG:
    log.setLevel(logging.DEBUG)
else:
    log.setLevel(logging.ERROR)
1324cc91   Goutte   Make the footer i...
76
logHandler = logging.FileHandler(LOG_FILE)
b2837a08   Goutte   Add three retries...
77
78
79
80
logHandler.setFormatter(logging.Formatter(
    "%(asctime)s - %(levelname)s - %(message)s"
))
log.addHandler(logHandler)
f75faf5f   Goutte   WIP
81
82


e18701b6   Goutte   Cache clear (remo...
83
84
# HARDCODED CONFIGURATION #####################################################

a2034dd9   Goutte   Convert from kilo...
85
86
ASTRONOMICAL_UNIT_IN_KM = 1.496e8

952e3d8f   Goutte   Move to another s...
87
88
89
# Absolute path to the installed CDF library from https://cdf.gsfc.nasa.gov/
CDF_LIB = '/usr/local/lib/libcdf'

e18701b6   Goutte   Cache clear (remo...
90
91
92
93
# Absolute path to the data cache directory
CACHE_DIR = get_path('../cache')

# These two configs are not in the YAML config because adding a new parameter
ea45ebf9   Goutte   Add the mocks of ...
94
# will not work as-is, you'd have to edit some netcdf-related code.
e18701b6   Goutte   Cache clear (remo...
95
96
97
98
99
100

# The slugs of the available parameters in the generated CSV files.
# The order matters. If you change this you also need to change the
# innermost loop of `get_data_for_target`.
# The javascript knows the targets' properties under these names.
PROPERTIES = ('time', 'vrad', 'vtan', 'vtot', 'btan', 'temp', 'pdyn', 'dens',
d1c44c51   Goutte   Enable Earth
101
              'atse', 'xhee', 'yhee')
e18701b6   Goutte   Cache clear (remo...
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132

# The parameters that the users can handle.
# The slug MUST be one of the properties above.
PARAMETERS = {
    'pdyn': {
        'slug': 'pdyn',
        'name': 'Dyn. Pressure',
        'title': 'The dynamic pressure.',
        'units': 'nPa',
        'active': True,
        'position': 10,
    },
    'vtot': {
        'slug': 'vtot',
        'name': 'Velocity',
        'title': 'The velocity of the particles.',
        'units': 'km/s',
        'active': False,
        'position': 20,
    },
    'btan': {
        'slug': 'btan',
        'name': 'B Tangential',
        'title': 'B Tangential.',
        'units': 'nT',
        'active': False,
        'position': 30,
    },
    'temp': {
        'slug': 'temp',
        'name': 'Temperature',
60b73eb1   Goutte   Change temperatur...
133
134
        'title': 'The temperature.',
        'units': 'eV',
e18701b6   Goutte   Cache clear (remo...
135
136
137
138
139
140
141
        'active': False,
        'position': 40,
    },
    'dens': {
        'slug': 'dens',
        'name': 'Density',
        'title': 'The density N.',
aa7247d6   Goutte   Generate a CDF fi...
142
        'units': 'cm^-3',
e18701b6   Goutte   Cache clear (remo...
143
144
145
        'active': False,
        'position': 50,
    },
d1c44c51   Goutte   Enable Earth
146
147
    'atse': {
        'slug': 'atse',
e18701b6   Goutte   Cache clear (remo...
148
149
150
151
152
153
154
155
        'name': 'Angle T-S-E',
        'title': 'Angle Target-Sun-Earth.',
        'units': 'deg',
        'active': False,
        'position': 60,
    },
}

48fa6323   Goutte   Try to fix the he...
156
# SETUP ENVIRONMENT ###########################################################
2fe06b17   Goutte   Move the ENV dire...
157
158
159
160

environ['SPACEPY'] = CACHE_DIR
environ['CDF_LIB'] = CDF_LIB

9390ec89   Goutte   Initial experimen...
161
162
163
# SETUP FLASK ENGINE ##########################################################

app = Flask(__name__, root_path=THIS_DIRECTORY)
20fdc1a4   Goutte   Set log level fro...
164
app.debug = DEBUG
b2837a08   Goutte   Add three retries...
165
if app.debug:
2fedd73b   Goutte   Initial implement...
166
    log.info("Starting Flask app IN DEBUG MODE...")
b2837a08   Goutte   Add three retries...
167
168
else:
    log.info("Starting Flask app...")
9390ec89   Goutte   Initial experimen...
169
170


48fa6323   Goutte   Try to fix the he...
171
172
173
174
def handle_error(e):
    log.error(e)
    return str(e)  # wish we could use the default error renderer here

befd6269   Goutte   Add Chris' changes.
175

48fa6323   Goutte   Try to fix the he...
176
177
178
app.register_error_handler(Exception, handle_error)


9390ec89   Goutte   Initial experimen...
179
180
181
182
183
184
185
186
187
# SETUP JINJA2 TEMPLATE ENGINE ################################################

def static_global(filename):
    return url_for('static', filename=filename)


def shuffle_filter(seq):
    """
    This shuffles the sequence it is applied to.
2fedd73b   Goutte   Initial implement...
188
    Jinja2 _should_ provide this.
9390ec89   Goutte   Initial experimen...
189
190
191
192
193
194
195
196
197
198
199
    """
    try:
        result = list(seq)
        random.shuffle(result)
        return result
    except:
        return seq


def markdown_filter(value, nl2br=False, p=True):
    """
2fedd73b   Goutte   Initial implement...
200
    Converts markdown into html.
9390ec89   Goutte   Initial experimen...
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
    nl2br: set to True to replace line breaks with <br> tags
    p: set to False to remove the enclosing <p></p> tags
    """
    from markdown import markdown
    from markdown.extensions.nl2br import Nl2BrExtension
    from markdown.extensions.abbr import AbbrExtension
    extensions = [AbbrExtension()]
    if nl2br is True:
        extensions.append(Nl2BrExtension())
    markdowned = markdown(value, output_format='html5', extensions=extensions)
    if p is False:
        markdowned = markdowned.replace(r"<p>", "").replace(r"</p>", "")
    return markdowned


bde97e4d   Goutte   Add more changes ...
216
_js_escapes = {
34e94ae3   hitier   Code reformating
217
218
219
220
221
222
223
224
225
226
227
    '\\': '\\u005C',
    '\'': '\\u0027',
    '"': '\\u0022',
    '>': '\\u003E',
    '<': '\\u003C',
    '&': '\\u0026',
    '=': '\\u003D',
    '-': '\\u002D',
    ';': '\\u003B',
    u'\u2028': '\\u2028',
    u'\u2029': '\\u2029'
bde97e4d   Goutte   Add more changes ...
228
229
}
# Escape every ASCII character with a value less than 32.
d4b3d381   hitier   Tiny python3 synt...
230
_js_escapes.update(('%c' % z, '\\u%04X' % z) for z in range(32))
bde97e4d   Goutte   Add more changes ...
231
232
233
234
235
236
237
238
239
240


def escapejs_filter(value):
    escaped = []
    for letter in value:
        if letter in _js_escapes:
            escaped.append(_js_escapes[letter])
        else:
            escaped.append(letter)

a3e57903   hitier   Change jinja imports
241
    return markupsafe.Markup("".join(escaped))
bde97e4d   Goutte   Add more changes ...
242

9390ec89   Goutte   Initial experimen...
243
244
245
246
247
248
249
250
251
252
253
254
tpl_engine = Environment(loader=FileSystemLoader([get_path('view')]),
                         trim_blocks=True,
                         lstrip_blocks=True)

tpl_engine.globals.update(
    url_for=url_for,
    static=static_global,
)

tpl_engine.filters['markdown'] = markdown_filter
tpl_engine.filters['md'] = markdown_filter
tpl_engine.filters['shuffle'] = shuffle_filter
bde97e4d   Goutte   Add more changes ...
255
tpl_engine.filters['escapejs'] = escapejs_filter
9390ec89   Goutte   Initial experimen...
256
257
258
259
260
261

tpl_global_vars = {
    'request': request,
    'version': version,
    'config': config,
    'now': datetime.datetime.now(),
fac54a01   Goutte   Use SSA instead o...
262
    'is_esa': SSA,
9390ec89   Goutte   Initial experimen...
263
264
265
266
267
}


# HELPERS #####################################################################

57f42bd7   Goutte   Log the abortions.
268
def abort(code, message):
b52b494b   Goutte   Add even more logs.
269
    log.error("Abort: " + message)
57f42bd7   Goutte   Log the abortions.
270
271
272
    abort_flask(code, message)


9390ec89   Goutte   Initial experimen...
273
274
275
276
277
278
279
280
def render_view(view, context=None):
    """
    A simple helper to render [view] template with [context] vars.
    It automatically adds the global template vars defined above, too.
    It returns a string, usually the HTML contents to display.
    """
    context = {} if context is None else context
    return tpl_engine.get_template(view).render(
d4b3d381   hitier   Tiny python3 synt...
281
        dict(list(tpl_global_vars.items()) + list(context.items()))
9390ec89   Goutte   Initial experimen...
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
    )


# def render_page(page, title="My Page", context=None):
#     """
#     A simple helper to render the md_page.html template with [context] vars &
#     the additional contents of `page/[page].md` in the `md_page` variable.
#     It automagically adds the global template vars defined above, too.
#     It returns a string, usually the HTML contents to display.
#     """
#     if context is None:
#         context = {}
#     context['title'] = title
#     context['md_page'] = ''
#     with file(get_path('page/%s.md' % page)) as f:
#         context['md_page'] = f.read()
#     return tpl_engine.get_template('md_page.html').render(
#         dict(tpl_global_vars.items() + context.items())
#     )

077980eb   Goutte   Improve availabil...
302

bc18b96c   Goutte   Implement first (...
303
304
305
306
307
308
309
def is_list_in_list(needle, haystack):
    for n in needle:
        if n not in haystack:
            return False
    return True


1324cc91   Goutte   Make the footer i...
310
311
312
313
314
315
316
317
318
319
def round_time(dt=None, round_to=60):
    """
    Round a datetime object to any time laps in seconds
    dt : datetime.datetime object, default now.
    roundTo : Closest number of seconds to round to, default 1 minute.
    """
    if dt is None:
        dt = datetime.datetime.now()
    seconds = (dt.replace(tzinfo=None) - dt.min).seconds
    rounding = (seconds + round_to / 2) // round_to * round_to
34e94ae3   hitier   Code reformating
320
    return dt + datetime.timedelta(0, rounding - seconds, -dt.microsecond)
1324cc91   Goutte   Make the footer i...
321
322


2d2af24b   Goutte   Add a basic orbit...
323
def datetime_from_list(time_list):
0b9821dd   Goutte   Clean up.
324
    """
2fedd73b   Goutte   Initial implement...
325
    Datetimes in retrieved CDFs are stored as lists of numbers,
80352490   Goutte   Multi model suppo...
326
327
    with DayOfYear starting at 0. We want it starting at 1 because it's what
    vendor parsers use, both in python and javascript.
0b9821dd   Goutte   Clean up.
328
    """
2d2af24b   Goutte   Add a basic orbit...
329
330
331
    # Day Of Year starts at 0, but for our datetime parser it starts at 1
    doy = '{:03d}'.format(int(''.join(time_list[4:7])) + 1)
    return datetime.datetime.strptime(
50d4f638   Goutte   Hotfix for a very...
332
        "%s%s%s" % (''.join(time_list[0:4]), doy, ''.join(time_list[7:-1])),
2d2af24b   Goutte   Add a basic orbit...
333
334
        "%Y%j%H%M%S%f"
    )
9390ec89   Goutte   Initial experimen...
335

ce8af118   Goutte   Fix the favicon.
336

927c69c3   Goutte   Make the local ca...
337
338
339
340
341
342
343
344
345
346
347
348
349
350
def get_local_filename(url):
    """
    Build the local cache filename for the distant file
    :param url: string
    :return: string
    """
    from slugify import slugify
    n = len('http://')
    if url.startswith('https'):
        n += 1
    s = url[n:]
    return slugify(s)


180d7d97   Goutte   Refactor heavily.
351
def get_target_config(slug):
2fedd73b   Goutte   Initial implement...
352
    for s in config['targets']:  # dumb
8644387c   Goutte   Use real data.
353
354
        if s['slug'] == slug:
            return s
180d7d97   Goutte   Refactor heavily.
355
    raise Exception("No target found in configuration for '%s'." % slug)
8644387c   Goutte   Use real data.
356
357


180d7d97   Goutte   Refactor heavily.
358
359
360
361
def check_target_config(slug):
    get_target_config(slug)


fb383448   Goutte   Implement the cac...
362
363
364
365
366
def get_active_targets():
    all_targets = config['targets']
    return [t for t in all_targets if not ('locked' in t and t['locked'])]


11d86851   Goutte   Add support for s...
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
def validate_tap_target_config(target):
    tc = get_target_config(target)
    if 'tap' not in tc:
        raise Exception("No `tap` configuration for target `%s`." % target)
    if 'target_name' not in tc['tap']:
        raise Exception("No `target_name` in the `tap` configuration for target `%s`." % target)
    return tc


# Using pyvo would be best.
# def retrieve_auroral_emissions_vopy(target_name):
#     api_url = "http://voparis-tap.obspm.fr/__system__/tap/run/tap/sync"
#     import pyvo as vo
#     service = vo.dal.TAPService(api_url)
#     # … can't figure out how to install pyvo and spacepy alongside (forking?)


def retrieve_auroral_emissions(target_name, d_started_at=None, d_stopped_at=None):
9f57dceb   Goutte   Add half the brid...
385
386
    """
    Work In Progress.
11d86851   Goutte   Add support for s...
387
    :param target_name: You should probably not let users define this value,
9f57dceb   Goutte   Add half the brid...
388
389
390
391
                        as our sanitizing for ADQL may not be 100% safe.
                        Use values from YAML configuration, instead.
                        Below is a list of the ids we found to be existing.
                        > SELECT DISTINCT target_name FROM apis.epn_core
7176ae9b   Goutte   Fix Myriam's cont...
392
393
                        - Jupiter
                        - Saturn
9f57dceb   Goutte   Add half the brid...
394
395
                        - Mars
                        - MERCURY
9f57dceb   Goutte   Add half the brid...
396
397
398
399
400
401
402
                        - Titan
                        - Io
                        - VENUS
                        - Ganymede
                        - Uranus
                        - Callisto
                        - Europa
7176ae9b   Goutte   Fix Myriam's cont...
403
404
405
406
407
408
                        
                        https://pdssbn.astro.umd.edu/holdings/nh-p-rex-2-pluto-v1.0/document/codmac_level_definitions.pdf
                        processing_level == 3
                        
                        API doc and sandbox:
                        http://voparis-tap-planeto.obspm.fr/__system__/tap/run/info
9f57dceb   Goutte   Add half the brid...
409
410
    :return: 
    """
11d86851   Goutte   Add support for s...
411
412
413
414

    # Try out the form
    # http://voparis-tap-planeto.obspm.fr/__system__/adql/query/form

9f57dceb   Goutte   Add half the brid...
415
    api_url = "http://voparis-tap.obspm.fr/__system__/tap/run/tap/sync"
11d86851   Goutte   Add support for s...
416
417
418
419
420
421
422
423
424
    if d_started_at is None:
        d_started_at = datetime.datetime.now()
        t_started_at = time.mktime(d_started_at.timetuple()) - 3600 * 24 * 365 * 2
        # t_started_at = 1
    else:
        t_started_at = time.mktime(d_started_at.timetuple())

    if d_stopped_at is None:
        d_stopped_at = datetime.datetime.now()
9f57dceb   Goutte   Add half the brid...
425
426
    t_stopped_at = time.mktime(d_stopped_at.timetuple())

11d86851   Goutte   Add support for s...
427
    def timestamp_to_jday(timestamp):
9f57dceb   Goutte   Add half the brid...
428
429
        return timestamp / 86400.0 + 2440587.5

11d86851   Goutte   Add support for s...
430
431
432
433
434
435
436
437
438
    def jday_to_timestamp(jday):
        return (jday - 2440587.5) * 86400.0

    def jday_to_datetime(jday):
        return datetime.datetime.utcfromtimestamp(jday_to_timestamp(jday))

    # SELECT DISTINCT dataproduct_type FROM apis.epn_core
    # > im sp sc

9f57dceb   Goutte   Add half the brid...
439
440
441
442
443
444
445
446
447
    query = """
SELECT
  time_min,
  time_max,
  thumbnail_url,
  external_link
FROM apis.epn_core
WHERE target_name='{target_name}'
AND   dataproduct_type='im'
11d86851   Goutte   Add support for s...
448
449
AND   time_min >= {jday_start}
AND   time_min <= {jday_stop}
9f57dceb   Goutte   Add half the brid...
450
451
452
ORDER BY time_min, granule_gid
""".format(
        target_name=target_name.replace("'", "\\'"),
11d86851   Goutte   Add support for s...
453
454
        jday_start=timestamp_to_jday(t_started_at),
        jday_stop=timestamp_to_jday(t_stopped_at)
9f57dceb   Goutte   Add half the brid...
455
    )
7176ae9b   Goutte   Fix Myriam's cont...
456
# AND   processing_level = 4
9f57dceb   Goutte   Add half the brid...
457
458
459
460
461
462
463
464

#     query = """
# SELECT DISTINCT target_name FROM apis.epn_core
# """

    try:
        response = requests.post(api_url, {
            'REQUEST': 'doQuery',
34e94ae3   hitier   Code reformating
465
466
            'LANG': 'ADQL',
            'QUERY': query,
9f57dceb   Goutte   Add half the brid...
467
            'TIMEOUT': '30',
34e94ae3   hitier   Code reformating
468
            'FORMAT': 'VOTable/td'
9f57dceb   Goutte   Add half the brid...
469
470
471
472
473
474
475
476
477
478
479
        })

        response_xml = response.text

        import xml.etree.ElementTree as ET
        root = ET.fromstring(response_xml)
        namespaces = {'vo': 'http://www.ivoa.net/xml/VOTable/v1.3'}
        rows_xpath = "./vo:RESOURCE/vo:TABLE/vo:DATA/vo:TABLEDATA/vo:TR"
        rows = []
        for row in root.findall(rows_xpath, namespaces):
            rows.append({
11d86851   Goutte   Add support for s...
480
481
                'time_min': jday_to_datetime(float(row[0].text)),
                'time_max': jday_to_datetime(float(row[1].text)),
9f57dceb   Goutte   Add half the brid...
482
483
484
485
                'thumbnail_url': row[2].text,
                'external_link': row[3].text,
            })

81960a4f   Goutte   Fix an issue repo...
486
487
488
        # print("Auroral emission query")
        # print(query)
        # print("Auroral emission rows")
9f57dceb   Goutte   Add half the brid...
489
        # print(rows)
81960a4f   Goutte   Fix an issue repo...
490

9f57dceb   Goutte   Add half the brid...
491
492
493
494
495
496
497
498
        return rows
    except Exception as e:
        print("Failed to retrieve auroral emissions :")
        print(e)

    # print(query)


180d7d97   Goutte   Refactor heavily.
499
def retrieve_amda_netcdf(orbiter, what, started_at, stopped_at):
8644387c   Goutte   Use real data.
500
    """
91c3d52d   Goutte   Add more logs.
501
502
    Handles remote querying AMDA's API for URLs, and then downloading,
    extracting and caching the netCDF files.
8644387c   Goutte   Use real data.
503
504
505
506
507
508
509
510
511
512
513
514
    :param orbiter: key of the source in the YAML config
    :param what: either 'model' or 'orbit', a key in the config of the source
    :param started_at:
    :param stopped_at:
    :return: a list of local file paths to netCDF (.nc) files
    """

    url = config['amda'].format(
        dataSet=what,
        startTime=started_at.isoformat(),
        stopTime=stopped_at.isoformat()
    )
c50cc9d8   Goutte   Continue fixing.
515
    log.info("Fetching remote gzip files list at '%s'." % url)
b2837a08   Goutte   Add three retries...
516
517
    retries = 0
    success = False
92abc15b   Goutte   Mistrust the API ...
518
    errors = []
b2837a08   Goutte   Add three retries...
519
520
521
    remote_gzip_files = []
    while not success and retries < 3:
        try:
3f9c36da   hitier   Change urllib import
522
            response = urllib_request.urlopen(url)
b2837a08   Goutte   Add three retries...
523
524
525
526
            remote_gzip_files = json.loads(response.read())
            if not remote_gzip_files:
                raise Exception("Failed to fetch data at '%s'." % url)
            if remote_gzip_files == 'NODATASET':
92abc15b   Goutte   Mistrust the API ...
527
528
529
                raise Exception("API says there's no dataset at '%s'." % url)
            if remote_gzip_files == 'ERROR':
                raise Exception("API returned an error at '%s'." % url)
077980eb   Goutte   Improve availabil...
530
            if remote_gzip_files == ['OUTOFTIME']:  # it happens
80352490   Goutte   Multi model suppo...
531
532
                return []
                # raise Exception("API says it's out of time at '%s'." % url)
b2837a08   Goutte   Add three retries...
533
534
            success = True
        except Exception as e:
d4b3d381   hitier   Tiny python3 synt...
535
            log.warning("Failed (%d/3) '%s' : %s" % (retries + 1, url, e.message))
92abc15b   Goutte   Mistrust the API ...
536
537
            remote_gzip_files = []
            errors.append(e)
b2837a08   Goutte   Add three retries...
538
539
540
        finally:
            retries += 1
    if not remote_gzip_files:
b52b494b   Goutte   Add even more logs.
541
        log.error("Failed to retrieve data from AMDA.")
91c3d52d   Goutte   Add more logs.
542
        log.error("Failed to fetch gzip files list for %s at '%s' : %s" %
34e94ae3   hitier   Code reformating
543
                  (orbiter, url, errors))
08abc2d4   Goutte   Remove duplicate ...
544
        abort(400, "Failed to fetch gzip files list for %s at '%s' : %s" %
34e94ae3   hitier   Code reformating
545
              (orbiter, url, errors))
08abc2d4   Goutte   Remove duplicate ...
546
547
    else:
        remote_gzip_files = list(set(remote_gzip_files))
9bfa6c42   Goutte   More bug hunting.
548
549

    log.debug("Fetched remote gzip files list : %s." % str(remote_gzip_files))
8644387c   Goutte   Use real data.
550

8644387c   Goutte   Use real data.
551
552
    local_gzip_files = []
    for remote_gzip_file in remote_gzip_files:
077980eb   Goutte   Improve availabil...
553
554
555
        # hotfixes to remove when fixed upstream @Myriam
        if remote_gzip_file in ['OUTOFTIME', 'ERROR']:
            continue  # sometimes half the response is okay, the other not
8644387c   Goutte   Use real data.
556
        if remote_gzip_file.endswith('/.gz'):
80352490   Goutte   Multi model suppo...
557
            continue  # this is just a plain bug
8644387c   Goutte   Use real data.
558
        remote_gzip_file = remote_gzip_file.replace('cdpp1', 'cdpp', 1)
077980eb   Goutte   Improve availabil...
559
        ################################################
e18701b6   Goutte   Cache clear (remo...
560
        local_gzip_file = join(CACHE_DIR, get_local_filename(remote_gzip_file))
8644387c   Goutte   Use real data.
561
562
        local_gzip_files.append(local_gzip_file)
        if not isfile(local_gzip_file):
9bfa6c42   Goutte   More bug hunting.
563
            log.debug("Retrieving '%s'..." % local_gzip_file)
3f9c36da   hitier   Change urllib import
564
            urllib_request.urlretrieve(remote_gzip_file, local_gzip_file)
9bfa6c42   Goutte   More bug hunting.
565
            log.debug("Retrieved '%s'." % local_gzip_file)
dc0be992   Goutte   Support having no...
566
567
        else:
            log.debug("Found '%s' in the cache." % local_gzip_file)
8644387c   Goutte   Use real data.
568
569
570
571

    local_netc_files = []
    for local_gzip_file in local_gzip_files:
        local_netc_file = local_gzip_file[0:-3]
9bfa6c42   Goutte   More bug hunting.
572
        log.debug("Unzipping '%s'..." % local_gzip_file)
3c064b17   Goutte   Ignore failures w...
573
574
        success = True
        try:
5ef50583   Goutte   Clean up.
575
            with gzip.open(local_gzip_file) as f:
3c064b17   Goutte   Ignore failures w...
576
577
578
579
580
                file_content = f.read()
                with open(local_netc_file, 'w+b') as g:
                    g.write(file_content)
        except Exception as e:
            success = False
dc0be992   Goutte   Support having no...
581
582
583
584
585
            log.error("Cannot process gz file '%s' from '%s' : %s" %
                      (local_gzip_file, url, e))
            # Sometimes, the downloaded gz is corrupted, and CRC checks fail.
            # We want to delete the local gz file and try again next time.
            removefile(local_gzip_file)
3c064b17   Goutte   Ignore failures w...
586
        if success:
dc0be992   Goutte   Support having no...
587
            local_netc_files.append(local_netc_file)
3c064b17   Goutte   Ignore failures w...
588
            log.debug("Unzipped '%s'." % local_gzip_file)
8644387c   Goutte   Use real data.
589

ea6c8d5d   Goutte   Add interval cons...
590
    return list(set(local_netc_files))  # remove possible dupes
8644387c   Goutte   Use real data.
591
592


596da00d   Goutte   Add more exceptio...
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
# class DataParser:
#     """
#     Default data parser
#     A wip to try to handle code exeptions sanely.
#     """
#
#     # Override these using the model configuration
#     default_nc_keys = {
#         'hee': 'HEE',
#         'vtot': 'V',
#         'magn': 'B',
#         'temp': 'T',
#         'dens': 'N',
#         'pdyn': 'P_dyn',
#         'atse': 'Delta_angle',
#     }
#
#     def __init__(self, target, model):
#         self.target = target
#         self.model = model
#         pass
#
#     def _read_var(self, nc, _keys, _key, mandatory=False):
#         try:
#             return nc.variables[_keys[_key]]
#         except KeyError:
#             pass
#         if mandatory:
#             raise Exception("No variable '%s' found in NetCDF." % _keys[_key])
#         return [None] * len(nc.variables['Time'])  # slow -- use numpy?
#
#     def parse(self, cdf_handle):
#         nc_keys = self.default_nc_keys.copy()
#
#         times = cdf_handle.variables['Time']  # YYYY DOY HH MM SS .ms
#         data_v = self._read_var(cdf_handle, nc_keys, 'vtot')
#         data_b = self._read_var(cdf_handle, nc_keys, 'magn')
#         data_t = self._read_var(cdf_handle, nc_keys, 'temp')
#         data_n = self._read_var(cdf_handle, nc_keys, 'dens')
#         data_p = self._read_var(cdf_handle, nc_keys, 'pdyn')
#         data_a = self._read_var(cdf_handle, nc_keys, 'atse')
#
#         return zip()


de97d643   Goutte   Fix more bugs.
638
639
def get_data_for_target(target_config, input_slug,
                        started_at, stopped_at):
180d7d97   Goutte   Refactor heavily.
640
641
642
643
    """
    :return: dict whose keys are datetime as str, values tuples of data
    """
    log.debug("Grabbing data for '%s'..." % target_config['slug'])
80352490   Goutte   Multi model suppo...
644

8644387c   Goutte   Use real data.
645
    try:
297a7dfc   Goutte   Add support for i...
646
        models = target_config['models'][input_slug]
077980eb   Goutte   Improve availabil...
647
648
    except Exception as e:
        abort(500, "Invalid model configuration for '%s' : %s"
180d7d97   Goutte   Refactor heavily.
649
650
              % (target_config['slug'], str(e)))
    try:
80352490   Goutte   Multi model suppo...
651
        orbits = target_config['orbit']['models']
d1c44c51   Goutte   Enable Earth
652
653
654
655
    except KeyError as e:
        orbits = []
        # abort(500, "Invalid orbit configuration for '%s' : %s"
        #       % (target_config['slug'], str(e)))
28ef3790   Goutte   Clean up.
656

58bfe281   Goutte   Handle start and ...
657
658
659
660
661
662
663
664
665
666
667
    def _sta_sto(_cnf, _sta, _sto):
        if 'started_at' in _cnf:
            _s0 = datetime.datetime.strptime(_cnf['started_at'], FILE_DATE_FMT)
            _s0 = max(_s0, _sta)
        else:
            _s0 = _sta
        if 'stopped_at' in _cnf:
            _s1 = datetime.datetime.strptime(_cnf['stopped_at'], FILE_DATE_FMT)
            _s1 = min(_s1, _sto)
        else:
            _s1 = _sto
aa7247d6   Goutte   Generate a CDF fi...
668
        return _s0, _s1
80352490   Goutte   Multi model suppo...
669

d1c44c51   Goutte   Enable Earth
670
671
672
673
674
675
676
677
678
    def _read_var(_nc, _keys, _key, mandatory=False):
        try:
            return _nc.variables[_keys[_key]]
        except KeyError:
            pass
        if mandatory:
            raise Exception("No variable '%s' found in NetCDF." % _keys[_key])
        return [None] * len(_nc.variables['Time'])  # slow -- use numpy!

fd1829c3   Goutte   Use the new AMDA API
679
    # Override these using the model configuration in config.yml
d1c44c51   Goutte   Enable Earth
680
681
682
683
684
685
686
687
688
689
    default_nc_keys = {
        'hee': 'HEE',
        'vtot': 'V',
        'magn': 'B',
        'temp': 'T',
        'dens': 'N',
        'pdyn': 'P_dyn',
        'atse': 'Delta_angle',
    }

80352490   Goutte   Multi model suppo...
690
    precision = "%Y-%m-%dT%H"  # model and orbits times are only equal-ish
180d7d97   Goutte   Refactor heavily.
691
    orbit_data = {}  # keys are datetime as str, values arrays of XY
ea6c8d5d   Goutte   Add interval cons...
692
693

    for orbit in orbits:
58bfe281   Goutte   Handle start and ...
694
        s0, s1 = _sta_sto(orbit, started_at, stopped_at)
ea6c8d5d   Goutte   Add interval cons...
695

d1c44c51   Goutte   Enable Earth
696
697
698
699
        nc_keys = default_nc_keys.copy()
        if 'parameters' in orbit:
            nc_keys.update(orbit['parameters'])

ea6c8d5d   Goutte   Add interval cons...
700
701
702
703
704
705
706
707
        orbit_files = retrieve_amda_netcdf(
            target_config['slug'], orbit['slug'], s0, s1
        )
        for orbit_file in orbit_files:
            log.debug("%s: opening orbit NETCDF4 '%s'..." %
                      (target_config['name'], orbit_file))
            cdf_handle = Dataset(orbit_file, "r", format="NETCDF4")
            times = cdf_handle.variables['Time']  # YYYY DOY HH MM SS .ms
d1c44c51   Goutte   Enable Earth
708
            data_hee = _read_var(cdf_handle, nc_keys, 'hee', mandatory=True)
ea6c8d5d   Goutte   Add interval cons...
709
710
711

            log.debug("%s: aggregating data from '%s'..." %
                      (target_config['name'], orbit_file))
d1c44c51   Goutte   Enable Earth
712
            for ltime, datum_hee in zip(times, data_hee):
1324cc91   Goutte   Make the footer i...
713
                try:
c269db3f   hitier   Fix array format ...
714
715
716
717
                    try:
                        ltime = [str(i, 'UTF8') for i in ltime]
                    except Exception as e:
                        print(e)
d1c44c51   Goutte   Enable Earth
718
719
                    dtime = datetime_from_list(ltime)
                except Exception:
34e94ae3   hitier   Code reformating
720
                    log.error("Failed to parse time from get__data_for_target %s." % ltime)
d1c44c51   Goutte   Enable Earth
721
                    raise
81cb3ba6   Goutte   Fix an issue with...
722
                # Keep only what's in the interval
ea6c8d5d   Goutte   Add interval cons...
723
                if s0 <= dtime <= s1:
34e94ae3   hitier   Code reformating
724
                    dkey = round_time(dtime, 60 * 60).strftime(precision)
ea6c8d5d   Goutte   Add interval cons...
725
726
                    orbit_data[dkey] = datum_hee
            cdf_handle.close()
180d7d97   Goutte   Refactor heavily.
727

8644387c   Goutte   Use real data.
728
    all_data = {}  # keys are datetime as str, values tuples of data
81cb3ba6   Goutte   Fix an issue with...
729

58bfe281   Goutte   Handle start and ...
730
731
732
733
734
    for model in models:
        s0, s1 = _sta_sto(model, started_at, stopped_at)
        model_files = retrieve_amda_netcdf(
            target_config['slug'], model['slug'], s0, s1
        )
d1c44c51   Goutte   Enable Earth
735
736
737
738
        nc_keys = default_nc_keys.copy()
        if 'parameters' in model:
            nc_keys.update(model['parameters'])

129181a6   Goutte   Add more logs to ...
739
        if len(model_files) == 0:
d4b3d381   hitier   Tiny python3 synt...
740
741
            log.warning("No model data for '%s' '%s'."
                        % (target_config['slug'], model['slug']))
129181a6   Goutte   Add more logs to ...
742

58bfe281   Goutte   Handle start and ...
743
        for model_file in model_files:
58bfe281   Goutte   Handle start and ...
744
745
746
            log.debug("%s: opening model NETCDF4 '%s'..." %
                      (target_config['name'], model_file))
            cdf_handle = Dataset(model_file, "r", format="NETCDF4")
596da00d   Goutte   Add more exceptio...
747

d1c44c51   Goutte   Enable Earth
748
            # log.debug(cdf_handle.variables.keys())
596da00d   Goutte   Add more exceptio...
749

58bfe281   Goutte   Handle start and ...
750
            times = cdf_handle.variables['Time']  # YYYY DOY HH MM SS .ms
d1c44c51   Goutte   Enable Earth
751
752
753
754
755
756
757
            data_v = _read_var(cdf_handle, nc_keys, 'vtot')
            data_b = _read_var(cdf_handle, nc_keys, 'magn')
            data_t = _read_var(cdf_handle, nc_keys, 'temp')
            data_n = _read_var(cdf_handle, nc_keys, 'dens')
            data_p = _read_var(cdf_handle, nc_keys, 'pdyn')
            data_a = _read_var(cdf_handle, nc_keys, 'atse')

d1c44c51   Goutte   Enable Earth
758
759
760
761
762
763
764
765
766
            # Usually:
            # Time, StartTime, StopTime, V, B, N, T, Delta_angle, P_dyn
            # Earth:
            # Time, BartelsNumber, ImfID, SwID, ImfPoints,
            # SwPoints, B_M_av, B_Vec_av, B_Theta_av,
            # B_Phi_av, B, T, N, V, Vlat, Vlon,
            # Alpha, RamP, E, Beta, Ma, Kp, R, DST,
            # AE, Flux, Flag, F10_Index, StartTime, StopTime

81cb3ba6   Goutte   Fix an issue with...
767
768
769
770
771
772
            # Don't ignore, but instead, compute a mean ?
            # Look for discretisation ? temporal drizzling ? 1D drizzling ?
            # The easy linear mean feels a bit rough too. Running mean too.
            # FIXME
            ignored_count = 0

58bfe281   Goutte   Handle start and ...
773
774
            log.debug("%s: aggregating data from '%s'..." %
                      (target_config['name'], model_file))
d1c44c51   Goutte   Enable Earth
775
776
777
            for ltime, datum_v, datum_b, datum_t, datum_n, datum_p, datum_a \
                    in zip(times, data_v, data_b, data_t, data_n, data_p, data_a):

58bfe281   Goutte   Handle start and ...
778
                try:
c269db3f   hitier   Fix array format ...
779
780
781
782
                    try:
                        ltime = [str(i, 'UTF8') for i in ltime]
                    except Exception as e:
                        print(e)
d1c44c51   Goutte   Enable Earth
783
784
785
786
                    dtime = datetime_from_list(ltime)
                except Exception:
                    log.error("Failed to parse time from %s." % ltime)
                    raise
596da00d   Goutte   Add more exceptio...
787

81cb3ba6   Goutte   Fix an issue with...
788
789
790
                if not (s0 <= dtime <= s1):
                    continue  # Cull what's out of the interval

34e94ae3   hitier   Code reformating
791
                droundtime = round_time(dtime, 60 * 60)
81cb3ba6   Goutte   Fix an issue with...
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
                dkey = droundtime.strftime(precision)

                x_hee = None
                y_hee = None
                if dkey in orbit_data:
                    x_hee = orbit_data[dkey][0] / ASTRONOMICAL_UNIT_IN_KM
                    y_hee = orbit_data[dkey][1] / ASTRONOMICAL_UNIT_IN_KM

                # First exception: V may be a vector instead of a scalar
                if hasattr(datum_v, '__len__'):
                    vrad = datum_v[0]
                    vtan = datum_v[1]
                    vtot = sqrt(vrad * vrad + vtan * vtan)
                else:  # eg: Earth
                    vrad = None
                    vtan = None
                    vtot = datum_v

                # Second exception: Earth is always at (1, 0)
                if target_config['slug'] == 'earth':
                    x_hee = 1
                    y_hee = 0

                # Third exception: B is a Vector3 or a Vector2 for Earth
                if target_config['slug'] == 'earth':
                    if model['slug'] == 'omni_hour_all':  # Vector3
                        datum_b = datum_b[0]
                    # if model['slug'] == 'ace_swepam_real':  # Vector2
                    #     datum_b = datum_b[0]
56028f6d   Myriam Bouchemit   bug for dyn press...
821
                    if model['slug'] == 'ace_swepam_real_1h':
206d8d14   Goutte   Clean up.
822
                        datum_p = datum_n * vtot * vtot * 1.6726e-6
81cb3ba6   Goutte   Fix an issue with...
823
824
825
                    if vtot is None or isnan(vtot):
                        continue

7176ae9b   Goutte   Fix Myriam's cont...
826
827
828
829
830
                # Fourth exception: Earth temp is in K, not eV
                # @nandre: à vérifier ! Where is la constante de Boltzmann ?
                if target_config['slug'] == 'earth' and datum_t:
                    datum_t = datum_t / 11605.0

81cb3ba6   Goutte   Fix an issue with...
831
832
                # Keep adding exceptions here until you can't or become mad

2be8e8a0   Goutte   Ignore bad data.
833
                # Ignore bad data
ff11e121   Goutte   Don't ignore bad ...
834
835
                # if numpy.isnan(datum_t):
                #     continue
2be8e8a0   Goutte   Ignore bad data.
836

81cb3ba6   Goutte   Fix an issue with...
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
                # Crude/Bad drizzling condition : first found datum in the hour
                # gets to set the data. Improve this.
                if dkey not in all_data:
                    # /!. The Set value MUST be in the same order as PROPERTIES
                    all_data[dkey] = (
                        dtime.strftime("%Y-%m-%dT%H:%M:%S+00:00"),
                        vrad, vtan, vtot,
                        datum_b, datum_t, datum_p, datum_n, datum_a,
                        x_hee, y_hee
                    )
                else:
                    ignored_count += 1

            # Improve this loop so as to remove this stinky debug log
            if ignored_count > 0:
                log.debug("    Ignored %d datum(s) during ~\"drizzling\"."
                          % ignored_count)

58bfe281   Goutte   Handle start and ...
855
            cdf_handle.close()
8644387c   Goutte   Use real data.
856

180d7d97   Goutte   Refactor heavily.
857
858
859
    return all_data


297a7dfc   Goutte   Add support for i...
860
def generate_csv_contents(target_slug, input_slug, started_at, stopped_at):
180d7d97   Goutte   Refactor heavily.
861
862
    target_config = get_target_config(target_slug)
    log.debug("Crunching CSV contents for '%s'..." % target_config['name'])
b33d5f62   hitier   Change StringIO i...
863
    si = StringIO()
180d7d97   Goutte   Refactor heavily.
864
865
866
    cw = csv_writer(si)
    cw.writerow(PROPERTIES)

297a7dfc   Goutte   Add support for i...
867
868
869
870
    all_data = get_data_for_target(
        target_config=target_config, input_slug=input_slug,
        started_at=started_at, stopped_at=stopped_at
    )
180d7d97   Goutte   Refactor heavily.
871
872

    log.debug("Writing and sorting CSV for '%s'..." % target_config['slug'])
8644387c   Goutte   Use real data.
873
874
    for dkey in sorted(all_data):
        cw.writerow(all_data[dkey])
2d2af24b   Goutte   Add a basic orbit...
875

180d7d97   Goutte   Refactor heavily.
876
    log.info("Generated CSV contents for '%s'." % target_config['slug'])
2d2af24b   Goutte   Add a basic orbit...
877
878
    return si.getvalue()

8644387c   Goutte   Use real data.
879

de97d643   Goutte   Fix more bugs.
880
881
def generate_csv_file_if_needed(target_slug, input_slug,
                                started_at, stopped_at):
297a7dfc   Goutte   Add support for i...
882
883
884
    filename = "%s_%s_%s_%s.csv" % (target_slug, input_slug,
                                    started_at.strftime(FILE_DATE_FMT),
                                    stopped_at.strftime(FILE_DATE_FMT))
e18701b6   Goutte   Cache clear (remo...
885
    local_csv_file = join(CACHE_DIR, filename)
80352490   Goutte   Multi model suppo...
886
887
888

    generate = True
    if isfile(local_csv_file):
297a7dfc   Goutte   Add support for i...
889
        # It needs to have more than one line to not be empty (headers)
80352490   Goutte   Multi model suppo...
890
891
892
893
894
895
896
897
898
        with open(local_csv_file) as f:
            cnt = 0
            for _ in f:
                cnt += 1
                if cnt > 1:
                    generate = False
                    break

    if generate:
c0df94bc   Goutte   Adding more logs.
899
900
901
        log.info("Generating CSV '%s'..." % local_csv_file)
        try:
            with open(local_csv_file, mode="w+") as f:
de97d643   Goutte   Fix more bugs.
902
903
904
905
                f.write(generate_csv_contents(
                    target_slug=target_slug, input_slug=input_slug,
                    started_at=started_at, stopped_at=stopped_at
                ))
c0df94bc   Goutte   Adding more logs.
906
907
            log.info("Generation of '%s' done." % filename)
        except Exception as e:
d1c44c51   Goutte   Enable Earth
908
909
910
            from sys import exc_info
            from traceback import extract_tb
            exc_type, exc_value, exc_traceback = exc_info()
dc0be992   Goutte   Support having no...
911
            log.error(e)
d1c44c51   Goutte   Enable Earth
912
913
            for trace in extract_tb(exc_traceback):
                log.error(trace)
5ede388f   Goutte   Make sure failed ...
914
            if isfile(local_csv_file):
d4b3d381   hitier   Tiny python3 synt...
915
                log.warning("Removing failed CSV '%s'..." % local_csv_file)
5ede388f   Goutte   Make sure failed ...
916
                removefile(local_csv_file)
34e94ae3   hitier   Code reformating
917
            # pprint(e)
9bfa6c42   Goutte   More bug hunting.
918
            abort(500, "Failed creating CSV '%s' : %s" % (filename, e))
c0df94bc   Goutte   Adding more logs.
919
920


e18701b6   Goutte   Cache clear (remo...
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
def remove_all_files(in_directory):
    """
    Will throw if something horrible happens.
    Does not remove recursively (could be done with os.walk if needed).
    Does not remove directories either.
    :param in_directory: absolute path to directory
    :return:
    """
    import os

    if not os.path.isdir(in_directory):
        raise ValueError("No directory to clean at '%s'.")

    removed_files = []
    for file_name in os.listdir(in_directory):
        file_path = os.path.join(in_directory, file_name)
        if os.path.isfile(file_path):
            os.remove(file_path)
            removed_files.append(file_path)

    return removed_files


28bb4b28   Goutte   API for the cache...
944
945
def remove_files_created_before(date, in_directory):
    """
077980eb   Goutte   Improve availabil...
946
947
948
    Will throw if something horrible happens.
    Does not remove recursively (could be done with os.walk if needed).
    Does not remove directories either.
28bb4b28   Goutte   API for the cache...
949
    :param date: datetime object
077980eb   Goutte   Improve availabil...
950
    :param in_directory: absolute path to directory
28bb4b28   Goutte   API for the cache...
951
952
953
954
955
956
957
    :return:
    """
    import os
    import time

    secs = time.mktime(date.timetuple())

077980eb   Goutte   Improve availabil...
958
959
    if not os.path.isdir(in_directory):
        raise ValueError("No directory to clean at '%s'.")
28bb4b28   Goutte   API for the cache...
960
961
962
963

    removed_files = []
    for file_name in os.listdir(in_directory):
        file_path = os.path.join(in_directory, file_name)
077980eb   Goutte   Improve availabil...
964
965
966
967
968
        if os.path.isfile(file_path):
            t = os.stat(file_path)
            if t.st_ctime < secs:
                os.remove(file_path)
                removed_files.append(file_path)
28bb4b28   Goutte   API for the cache...
969
970
971
972

    return removed_files


297a7dfc   Goutte   Add support for i...
973
974
def get_input_slug_from_query(inp=None):
    if inp is None:
b662fcc3   hitier   Change input name...
975
        input_slug = request.args.get('input_slug', config['defaults']['input_slug'])
297a7dfc   Goutte   Add support for i...
976
977
    else:
        input_slug = inp
de97d643   Goutte   Fix more bugs.
978
    if input_slug not in [i['slug'] for i in config['inputs']]:
b662fcc3   hitier   Change input name...
979
        input_slug = config['defaults']['input_slug']  # be tolerant instead of yelling loudly
297a7dfc   Goutte   Add support for i...
980
981
982
    return input_slug


284f4688   Goutte   Continue layers i...
983
984
985
def get_interval_from_query():
    """
    Get the interval from the query, or from defaults.
3309d7c3   Goutte   Improve the date ...
986
    Returns ISO date strings.
284f4688   Goutte   Continue layers i...
987
    """
65b27161   hitier   Change default ti...
988
    before = relativedelta(months=1)
284f4688   Goutte   Continue layers i...
989
990
991
992
993
994
995
996
997
998
    after = relativedelta(months=1)
    today = datetime.datetime.now().replace(hour=0, minute=0, second=0)
    started_at = today - before
    stopped_at = today + after
    default_started_at = started_at.strftime(FILE_DATE_FMT)
    default_stopped_at = stopped_at.strftime(FILE_DATE_FMT)

    started_at = request.args.get('started_at', default_started_at)
    stopped_at = request.args.get('stopped_at', default_stopped_at)

3309d7c3   Goutte   Improve the date ...
999
1000
1001
1002
1003
    d_started_at = dateparser.isoparse(started_at)
    d_stopped_at = dateparser.isoparse(stopped_at)
    started_at = d_started_at.strftime(FILE_DATE_FMT)
    stopped_at = d_stopped_at.strftime(FILE_DATE_FMT)

284f4688   Goutte   Continue layers i...
1004
1005
1006
    return started_at, stopped_at


1185f353   Goutte   Fix the CME Catal...
1007
def get_catalog_layers(input_slug, target_slug, started_at, stopped_at):
0332f168   Goutte   Initial support f...
1008
    """
284f4688   Goutte   Continue layers i...
1009
1010
    In the JSON file we have "columns" and "data".
    Of course, each JSON file has its own columns, with different conventions.
0332f168   Goutte   Initial support f...
1011
1012
1013
1014
1015
    
    :param input_slug: 
    :param target_slug: 
    :return: 
    """
0332f168   Goutte   Initial support f...
1016
    import json
2204c3f7   Goutte   Improve layers co...
1017
1018

    def _get_index_of_key(_data, _key):
2204c3f7   Goutte   Improve layers co...
1019
1020
1021
1022
1023
1024
1025
        try:
            index = _data['columns'].index(_key)
        except ValueError:
            log.error("Key %s not found in columns of %s" % (_key, _data))
            raise
        return index

1185f353   Goutte   Fix the CME Catal...
1026
1027
1028
1029
1030
1031
1032
1033
1034
    try:
        started_at = datetime.datetime.strptime(started_at, FILE_DATE_FMT)
    except:
        abort(400, "Invalid started_at parameter : '%s'." % started_at)
    try:
        stopped_at = datetime.datetime.strptime(stopped_at, FILE_DATE_FMT)
    except:
        abort(400, "Invalid stopped_at parameter : '%s'." % stopped_at)

0332f168   Goutte   Initial support f...
1035
    catalog_layers = {}
1ab47144   Goutte   Add new menus, up...
1036
    for config_layer in config['layers']['catalogs']:
0332f168   Goutte   Initial support f...
1037
1038
        if 'data' not in config_layer:
            continue
0332f168   Goutte   Initial support f...
1039
1040
1041
1042
1043
1044
        catalog_layers[config_layer['slug']] = []
        for cl_datum in config_layer['data']:
            if input_slug not in cl_datum:
                continue
            if target_slug not in cl_datum[input_slug]:
                continue
b03d5eb1   Goutte   Add the last CME ...
1045
1046
1047
1048
1049
            if cl_datum[input_slug][target_slug] is None:
                # We used ~ in the config, there are no constraints
                constraints = []
            else:
                constraints = cl_datum[input_slug][target_slug]['constraints']
1185f353   Goutte   Fix the CME Catal...
1050

0332f168   Goutte   Initial support f...
1051
1052
            with open(get_path("../data/catalog/%s" % cl_datum['file'])) as f:
                json_data = json.load(f)
284f4688   Goutte   Continue layers i...
1053
1054
1055
                if 'start' not in cl_datum:
                    log.error("Invalid configuration: 'start' is missing.")
                    continue  # skip this
1185f353   Goutte   Fix the CME Catal...
1056
                if 'format' not in cl_datum:
54bb1311   Goutte   Bring back the CM...
1057
1058
1059
                    cl_datum['format'] = CME_DATE_FMT
                    # log.error("Invalid configuration: 'format' is missing.")
                    # continue  # skip this
284f4688   Goutte   Continue layers i...
1060
1061
1062
1063
1064
                start_index = _get_index_of_key(json_data, cl_datum['start'])
                if 'stop' not in cl_datum:
                    stop_index = start_index
                else:
                    stop_index = _get_index_of_key(json_data, cl_datum['stop'])
2204c3f7   Goutte   Improve layers co...
1065

0332f168   Goutte   Initial support f...
1066
                for json_datum in json_data['data']:
2204c3f7   Goutte   Improve layers co...
1067
                    validates_any_constraint = False
b03d5eb1   Goutte   Add the last CME ...
1068
1069
                    if 0 == len(constraints):
                        validates_any_constraint = True
2204c3f7   Goutte   Improve layers co...
1070
1071
                    for constraint in constraints:
                        validates_constraint = True
d4b3d381   hitier   Tiny python3 synt...
1072
                        for key, possible_values in iter(constraint.items()):
2204c3f7   Goutte   Improve layers co...
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
                            actual_value = json_datum[_get_index_of_key(
                                json_data, key
                            )]
                            if actual_value not in possible_values:
                                validates_constraint = False
                                break
                        if validates_constraint:
                            validates_any_constraint = True
                            break
                    if not validates_any_constraint:
0332f168   Goutte   Initial support f...
1083
                        continue
284f4688   Goutte   Continue layers i...
1084
1085
                    start_time = json_datum[start_index]
                    stop_time = json_datum[stop_index]
1185f353   Goutte   Fix the CME Catal...
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096

                    start_time = datetime.datetime.strptime(
                        start_time, cl_datum['format']
                    )
                    stop_time = datetime.datetime.strptime(
                        stop_time, cl_datum['format']
                    )

                    if start_time < started_at:
                        continue

0332f168   Goutte   Initial support f...
1097
                    catalog_layers[config_layer['slug']].append({
1185f353   Goutte   Fix the CME Catal...
1098
                        'start': start_time.strftime(MOMENT_DATE_FMT),
34e94ae3   hitier   Code reformating
1099
                        'stop': stop_time.strftime(MOMENT_DATE_FMT),
0332f168   Goutte   Initial support f...
1100
                    })
0332f168   Goutte   Initial support f...
1101
1102
1103
1104

    return catalog_layers


077980eb   Goutte   Improve availabil...
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
def get_hit_counter():
    hit_count_path = get_path("../VISITS")

    if isfile(hit_count_path):
        hit_count = int(open(hit_count_path).read())
    else:
        hit_count = 1

    return hit_count


a4a9ef03   Goutte   Cache generated C...
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
def increment_hit_counter():
    hit_count_path = get_path("../VISITS")

    if isfile(hit_count_path):
        hit_count = int(open(hit_count_path).read())
        hit_count += 1
    else:
        hit_count = 1

    hit_counter_file = open(hit_count_path, 'w')
    hit_counter_file.write(str(hit_count))
    hit_counter_file.close()

    return hit_count


4aaf6874   Goutte   Try fixing the ge...
1132
1133
def update_spacepy():
    """
11d86851   Goutte   Add support for s...
1134
    Importing pycdf will fail if the toolbox is not up to date.
4aaf6874   Goutte   Try fixing the ge...
1135
    """
4aaf6874   Goutte   Try fixing the ge...
1136
1137
1138
1139
1140
1141
1142
1143
1144
    try:
        log.info("Updating spacepy's toolbox…")
        import spacepy.toolbox

        spacepy.toolbox.update()
    except Exception as e:
        log.error("Failed to update spacepy : %s." % e)


077980eb   Goutte   Improve availabil...
1145
1146
1147
tpl_global_vars['visits'] = get_hit_counter()


a4a9ef03   Goutte   Cache generated C...
1148
1149
1150
# ROUTING #####################################################################

@app.route('/favicon.ico')
bde97e4d   Goutte   Add more changes ...
1151
def favicon():  # we want it served from the root, not from static/
a4a9ef03   Goutte   Cache generated C...
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
    return send_from_directory(
        join(app.root_path, 'static', 'img'),
        'favicon.ico', mimetype='image/vnd.microsoft.icon'
    )


@app.route("/")
@app.route("/home.html")
@app.route("/index.html")
def home():
077980eb   Goutte   Improve availabil...
1162
    increment_hit_counter()
bde97e4d   Goutte   Add more changes ...
1163
    parameters = PARAMETERS.values()
d4b3d381   hitier   Tiny python3 synt...
1164
    parameters = sorted(parameters, key=lambda x: x['position'])
297a7dfc   Goutte   Add support for i...
1165
    input_slug = get_input_slug_from_query()
0332f168   Goutte   Initial support f...
1166
    targets = [t for t in config['targets'] if not t['locked']]
284f4688   Goutte   Continue layers i...
1167
    started_at, stopped_at = get_interval_from_query()
0332f168   Goutte   Initial support f...
1168
1169
    for i, target in enumerate(targets):
        targets[i]['catalog_layers'] = get_catalog_layers(
1185f353   Goutte   Fix the CME Catal...
1170
            input_slug, target['slug'], started_at, stopped_at
0332f168   Goutte   Initial support f...
1171
        )
a4a9ef03   Goutte   Cache generated C...
1172
    return render_view('home.html.jinja2', {
0332f168   Goutte   Initial support f...
1173
1174
        # 'targets': config['targets'],
        'targets': targets,
bde97e4d   Goutte   Add more changes ...
1175
        'parameters': parameters,
297a7dfc   Goutte   Add support for i...
1176
        'input_slug': input_slug,
284f4688   Goutte   Continue layers i...
1177
1178
        'started_at': started_at,
        'stopped_at': stopped_at,
a4a9ef03   Goutte   Cache generated C...
1179
        'planets': [s for s in config['targets'] if s['type'] == 'planet'],
34e94ae3   hitier   Code reformating
1180
1181
1182
        'probes': [s for s in config['targets'] if s['type'] == 'probe'],
        'comets': [s for s in config['targets'] if s['type'] == 'comet'],
        'visits': get_hit_counter(),
a4a9ef03   Goutte   Cache generated C...
1183
1184
1185
    })


64a671cd   Goutte   Add an About page.
1186
1187
@app.route("/about.html")
def about():
36961d6f   Goutte   Slightly improve ...
1188
    import uuid
64a671cd   Goutte   Add an About page.
1189
1190
    increment_hit_counter()
    return render_view('about.html.jinja2', {
36961d6f   Goutte   Slightly improve ...
1191
1192
1193
        'authors_emails': [a['mail'] for a in config['authors']],
        'uuid4': str(uuid.uuid4())[0:3],
        'visits': get_hit_counter(),
64a671cd   Goutte   Add an About page.
1194
1195
1196
    })


6288347a   Goutte   Add the HELP page...
1197
1198
1199
1200
1201
1202
1203
@app.route("/help.html")
def help():
    return render_view('help.html.jinja2', {
        'visits': get_hit_counter(),
    })


297a7dfc   Goutte   Add support for i...
1204
1205
@app.route("/<target>_<inp>_<started_at>_<stopped_at>.csv")
def download_target_csv(target, inp, started_at, stopped_at):
a4a9ef03   Goutte   Cache generated C...
1206
1207
1208
1209
    """
    Grab data and orbit data for the specified `target`,
    rearrange it and return it as a CSV file.
    `started_at` and `stopped_at` should be UTC.
b662fcc3   hitier   Change input name...
1210
    `inp` is the input slug, omni or sa or sb.
a4a9ef03   Goutte   Cache generated C...
1211
    """
180d7d97   Goutte   Refactor heavily.
1212
    check_target_config(target)
a4a9ef03   Goutte   Cache generated C...
1213
    try:
c0df94bc   Goutte   Adding more logs.
1214
        started_at = datetime.datetime.strptime(started_at, FILE_DATE_FMT)
a4a9ef03   Goutte   Cache generated C...
1215
1216
1217
    except:
        abort(400, "Invalid started_at parameter : '%s'." % started_at)
    try:
c0df94bc   Goutte   Adding more logs.
1218
        stopped_at = datetime.datetime.strptime(stopped_at, FILE_DATE_FMT)
a4a9ef03   Goutte   Cache generated C...
1219
1220
    except:
        abort(400, "Invalid stopped_at parameter : '%s'." % stopped_at)
297a7dfc   Goutte   Add support for i...
1221
    input_slug = get_input_slug_from_query(inp=inp)
a4a9ef03   Goutte   Cache generated C...
1222

297a7dfc   Goutte   Add support for i...
1223
1224
1225
    filename = "%s_%s_%s_%s.csv" % (target, input_slug,
                                    started_at.strftime(FILE_DATE_FMT),
                                    stopped_at.strftime(FILE_DATE_FMT))
e18701b6   Goutte   Cache clear (remo...
1226
    local_csv_file = join(CACHE_DIR, filename)
297a7dfc   Goutte   Add support for i...
1227
1228
1229
1230
    generate_csv_file_if_needed(
        target_slug=target, input_slug=input_slug,
        started_at=started_at, stopped_at=stopped_at
    )
a4a9ef03   Goutte   Cache generated C...
1231
1232
1233
    if not isfile(local_csv_file):
        abort(500, "Could not cache CSV file at '%s'." % local_csv_file)

e18701b6   Goutte   Cache clear (remo...
1234
    return send_from_directory(CACHE_DIR, filename)
a4a9ef03   Goutte   Cache generated C...
1235
1236


297a7dfc   Goutte   Add support for i...
1237
1238
@app.route("/<targets>_<inp>_<started_at>_<stopped_at>.tar.gz")
def download_targets_tarball(targets, inp, started_at, stopped_at):
b2837a08   Goutte   Add three retries...
1239
    """
bc18b96c   Goutte   Implement first (...
1240
1241
1242
    Grab data and orbit data for each of the specified `targets`,
    in their own CSV file, and make a tarball of them.
    `started_at` and `stopped_at` should be UTC strings.
b2837a08   Goutte   Add three retries...
1243

ea6c8d5d   Goutte   Add interval cons...
1244
1245
    Note: we do not use this route anymore, but let's keep it shelved for now.

2fedd73b   Goutte   Initial implement...
1246
    targets: string list of targets' slugs, separated by `-`.
b2837a08   Goutte   Add three retries...
1247
    """
2fedd73b   Goutte   Initial implement...
1248
    separator = '-'
0511eed7   Goutte   Tarball generatio...
1249
    targets = targets.split(separator)
d4b3d381   hitier   Tiny python3 synt...
1250
    targets.sorty()
2fedd73b   Goutte   Initial implement...
1251
1252
    targets_configs = []
    for target in targets:
b2837a08   Goutte   Add three retries...
1253
1254
        if not target:
            abort(400, "Invalid targets format : `%s`." % targets)
180d7d97   Goutte   Refactor heavily.
1255
        targets_configs.append(get_target_config(target))
2fedd73b   Goutte   Initial implement...
1256
    if 0 == len(targets_configs):
b2837a08   Goutte   Add three retries...
1257
1258
        abort(400, "No valid targets specified. What are you doing?")

b2837a08   Goutte   Add three retries...
1259
    try:
ea6c8d5d   Goutte   Add interval cons...
1260
        started_at = datetime.datetime.strptime(started_at, FILE_DATE_FMT)
b2837a08   Goutte   Add three retries...
1261
1262
1263
    except:
        abort(400, "Invalid started_at parameter : '%s'." % started_at)
    try:
ea6c8d5d   Goutte   Add interval cons...
1264
        stopped_at = datetime.datetime.strptime(stopped_at, FILE_DATE_FMT)
b2837a08   Goutte   Add three retries...
1265
1266
    except:
        abort(400, "Invalid stopped_at parameter : '%s'." % stopped_at)
ea6c8d5d   Goutte   Add interval cons...
1267
1268
    sta = started_at.strftime(FILE_DATE_FMT)
    sto = stopped_at.strftime(FILE_DATE_FMT)
b2837a08   Goutte   Add three retries...
1269

297a7dfc   Goutte   Add support for i...
1270
1271
1272
1273
1274
    input_slug = get_input_slug_from_query(inp=inp)

    gzip_filename = "%s_%s_%s_%s.tar.gz" % (
        separator.join(targets), input_slug, sta, sto
    )
e18701b6   Goutte   Cache clear (remo...
1275
    local_gzip_file = join(CACHE_DIR, gzip_filename)
2fedd73b   Goutte   Initial implement...
1276
1277

    if not isfile(local_gzip_file):
0511eed7   Goutte   Tarball generatio...
1278
        log.debug("Creating the CSV files for the tarball...")
2fedd73b   Goutte   Initial implement...
1279
        for target_config in targets_configs:
297a7dfc   Goutte   Add support for i...
1280
1281
1282
            filename = "%s_%s_%s_%s.csv" % (
                target_config['slug'], input_slug, sta, sto
            )
e18701b6   Goutte   Cache clear (remo...
1283
            local_csv_file = join(CACHE_DIR, filename)
2fedd73b   Goutte   Initial implement...
1284
1285
            if not isfile(local_csv_file):
                with open(local_csv_file, mode="w+") as f:
297a7dfc   Goutte   Add support for i...
1286
1287
1288
1289
1290
1291
                    f.write(generate_csv_contents(
                        target_slug=target_config['slug'],
                        started_at=started_at,
                        stopped_at=stopped_at,
                        input_slug=input_slug
                    ))
2fedd73b   Goutte   Initial implement...
1292

0511eed7   Goutte   Tarball generatio...
1293
        log.debug("Creating the tarball '%s'..." % local_gzip_file)
2fedd73b   Goutte   Initial implement...
1294
1295
        with tarfile.open(local_gzip_file, "w:gz") as tar:
            for target_config in targets_configs:
297a7dfc   Goutte   Add support for i...
1296
1297
1298
                filename = "%s_%s_%s_%s.csv" % (
                    target_config['slug'], input_slug, sta, sto
                )
e18701b6   Goutte   Cache clear (remo...
1299
                local_csv_file = join(CACHE_DIR, filename)
2fedd73b   Goutte   Initial implement...
1300
1301
1302
                tar.add(local_csv_file, arcname=filename)

    if not isfile(local_gzip_file):
0511eed7   Goutte   Tarball generatio...
1303
        abort(500, "No tarball to serve. Looked at '%s'." % local_gzip_file)
2fedd73b   Goutte   Initial implement...
1304

e18701b6   Goutte   Cache clear (remo...
1305
    return send_from_directory(CACHE_DIR, gzip_filename)
b2837a08   Goutte   Add three retries...
1306

28bb4b28   Goutte   API for the cache...
1307

297a7dfc   Goutte   Add support for i...
1308
1309
@app.route("/<targets>_<inp>_<params>_<started_at>_<stopped_at>.nc")
def download_targets_netcdf(targets, inp, params, started_at, stopped_at):
bc18b96c   Goutte   Implement first (...
1310
    """
4aaf6874   Goutte   Try fixing the ge...
1311
1312
    NOTE : This is not used anymore.
    
bc18b96c   Goutte   Implement first (...
1313
    Grab data and orbit data for the specified `target`,
aa7247d6   Goutte   Generate a CDF fi...
1314
    rearrange it and return it as a NetCDF file.
e18701b6   Goutte   Cache clear (remo...
1315
    `started_at` and `stopped_at` are expected to be UTC.
bc18b96c   Goutte   Implement first (...
1316
1317
1318
1319

    targets: string list of targets' slugs, separated by `-`.
    params: string list of targets' parameters, separated by `-`.
    """
e18701b6   Goutte   Cache clear (remo...
1320
    separator = '-'  # /!\ this char should never be in target's slugs
bc18b96c   Goutte   Implement first (...
1321
    targets = targets.split(separator)
d4b3d381   hitier   Tiny python3 synt...
1322
    targets.sorty()
bc18b96c   Goutte   Implement first (...
1323
1324
1325
1326
1327
1328
1329
    targets_configs = []
    for target in targets:
        if not target:
            abort(400, "Invalid targets format : `%s`." % targets)
        targets_configs.append(get_target_config(target))
    if 0 == len(targets_configs):
        abort(400, "No valid targets specified. What are you doing?")
4aaf6874   Goutte   Try fixing the ge...
1330

bc18b96c   Goutte   Implement first (...
1331
    params = params.split(separator)
d4b3d381   hitier   Tiny python3 synt...
1332
    params.sorty()
bc18b96c   Goutte   Implement first (...
1333
1334
1335
1336
1337
    if 0 == len(params):
        abort(400, "No valid parameters specified. What are you doing?")
    if not is_list_in_list(params, PARAMETERS.keys()):
        abort(400, "Some parameters are not recognized in '%s'." % str(params))

57493104   Goutte   Add the time to t...
1338
    date_fmt = FILE_DATE_FMT
bc18b96c   Goutte   Implement first (...
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
    try:
        started_at = datetime.datetime.strptime(started_at, date_fmt)
    except:
        abort(400, "Invalid started_at parameter : '%s'." % started_at)
    try:
        stopped_at = datetime.datetime.strptime(stopped_at, date_fmt)
    except:
        abort(400, "Invalid stopped_at parameter : '%s'." % stopped_at)
    sta = started_at.strftime(date_fmt)
    sto = stopped_at.strftime(date_fmt)

297a7dfc   Goutte   Add support for i...
1350
1351
1352
1353
1354
    input_slug = get_input_slug_from_query(inp=inp)

    nc_filename = "%s_%s_%s_%s_%s.nc" % (
        separator.join(targets), separator.join(params), input_slug, sta, sto
    )
e18701b6   Goutte   Cache clear (remo...
1355
    nc_path = join(CACHE_DIR, nc_filename)
bc18b96c   Goutte   Implement first (...
1356
1357
1358
1359
1360

    if not isfile(nc_path):
        log.debug("Creating the NetCDF file '%s'..." % nc_filename)
        nc_handle = Dataset(nc_path, "w", format="NETCDF4")
        try:
ea6c8d5d   Goutte   Add interval cons...
1361
            nc_handle.description = "Model and orbit data for targets"  # todo
bc18b96c   Goutte   Implement first (...
1362
            nc_handle.history = "Created " + time.ctime(time.time())
ea6c8d5d   Goutte   Add interval cons...
1363
            nc_handle.source = "Heliopropa (CDDP)"
bc18b96c   Goutte   Implement first (...
1364
1365
1366
1367
1368
            available_params = list(PROPERTIES)
            for target in targets_configs:
                target_slug = target['slug']
                log.debug("Adding group '%s' to the NetCDF..." % target_slug)
                nc_group = nc_handle.createGroup(target_slug)
297a7dfc   Goutte   Add support for i...
1369
1370
1371
1372
                data = get_data_for_target(
                    target_config=target, input_slug=input_slug,
                    started_at=started_at, stopped_at=stopped_at
                )
bc18b96c   Goutte   Implement first (...
1373
                dkeys = sorted(data)
34e94ae3   hitier   Code reformating
1374
                dimension = 'dim_' + target_slug
ceeb2f4a   Goutte   Add the target co...
1375
                nc_handle.createDimension(dimension, len(dkeys))
57493104   Goutte   Add the time to t...
1376
1377

                # TIME #
ceeb2f4a   Goutte   Add the target co...
1378
                nc_time = nc_group.createVariable('time', 'i8', (dimension,))
57493104   Goutte   Add the time to t...
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
                nc_time.units = "hours since 1970-01-01 00:00:00"
                nc_time.calendar = "standard"
                times = []
                for dkey in dkeys:
                    time_as_string = data[dkey][0][:-6]  # remove +00:00 tail
                    date = datetime.datetime.strptime(time_as_string, date_fmt)
                    times.append(date2num(
                        date, units=nc_time.units, calendar=nc_time.calendar
                    ))
                nc_time[:] = times

                # SELECTED PARAMETERS #
bc18b96c   Goutte   Implement first (...
1391
1392
1393
1394
                nc_vars = []
                indices = []
                for param in params:
                    indices.append(available_params.index(param))
ceeb2f4a   Goutte   Add the target co...
1395
                    nc_var = nc_group.createVariable(param, 'f8', (dimension,))
5a6d4498   Goutte   Add a title to ea...
1396
                    nc_var.units = PARAMETERS[param]['units']
bc18b96c   Goutte   Implement first (...
1397
1398
1399
1400
1401
1402
1403
1404
                    nc_vars.append(nc_var)
                for i, nc_var in enumerate(nc_vars):
                    index = indices[i]
                    values = []
                    for dkey in dkeys:
                        dval = data[dkey]
                        values.append(dval[index])
                    nc_var[:] = values
ceeb2f4a   Goutte   Add the target co...
1405
1406

                # ORBIT #
6491a1f1   Goutte   Fix up the bugs l...
1407
                nc_x = nc_group.createVariable('xhee', 'f8', (dimension,))
ceeb2f4a   Goutte   Add the target co...
1408
                nc_x.units = 'Au'
6491a1f1   Goutte   Fix up the bugs l...
1409
                nc_y = nc_group.createVariable('yhee', 'f8', (dimension,))
ceeb2f4a   Goutte   Add the target co...
1410
1411
1412
                nc_y.units = 'Au'
                values_x = []
                values_y = []
6491a1f1   Goutte   Fix up the bugs l...
1413
1414
                index_x = available_params.index('xhee')
                index_y = available_params.index('yhee')
ceeb2f4a   Goutte   Add the target co...
1415
1416
1417
1418
1419
1420
1421
1422
                for dkey in dkeys:
                    dval = data[dkey]
                    values_x.append(dval[index_x])
                    values_y.append(dval[index_y])
                nc_x[:] = values_x
                nc_y[:] = values_y
            log.debug("Writing NetCDF '%s'..." % nc_filename)

d1c44c51   Goutte   Enable Earth
1423
        except Exception:
57493104   Goutte   Add the time to t...
1424
            log.error("Failed to generate NetCDF '%s'." % nc_filename)
d1c44c51   Goutte   Enable Earth
1425
            raise
bc18b96c   Goutte   Implement first (...
1426
1427
1428
1429
1430
1431
        finally:
            nc_handle.close()

    if not isfile(nc_path):
        abort(500, "No NetCDF to serve. Looked at '%s'." % nc_path)

e18701b6   Goutte   Cache clear (remo...
1432
    return send_from_directory(CACHE_DIR, nc_filename)
bc18b96c   Goutte   Implement first (...
1433
1434


297a7dfc   Goutte   Add support for i...
1435
1436
@app.route("/<targets>_<inp>_<started_at>_<stopped_at>.cdf")
def download_targets_cdf(targets, inp, started_at, stopped_at):
aa7247d6   Goutte   Generate a CDF fi...
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
    """
    Grab data and orbit data for the specified `target`,
    rearrange it and return it as a CDF file.
    `started_at` and `stopped_at` are expected to be UTC.

    targets: string list of targets' slugs, separated by `-`.
    params: string list of targets' parameters, separated by `-`.
    """
    separator = '-'  # /!\ this char should never be in target's slugs
    targets = targets.split(separator)
d4b3d381   hitier   Tiny python3 synt...
1447
    targets.sorty()
aa7247d6   Goutte   Generate a CDF fi...
1448
1449
1450
1451
1452
1453
1454
1455
1456
    targets_configs = []
    for target in targets:
        if not target:
            abort(400, "Invalid targets format : `%s`." % targets)
        targets_configs.append(get_target_config(target))
    if 0 == len(targets_configs):
        abort(400, "No valid targets specified. What are you doing?")

    params = PARAMETERS.keys()
aa7247d6   Goutte   Generate a CDF fi...
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468

    try:
        started_at = datetime.datetime.strptime(started_at, FILE_DATE_FMT)
    except:
        abort(400, "Invalid started_at parameter : '%s'." % started_at)
    try:
        stopped_at = datetime.datetime.strptime(stopped_at, FILE_DATE_FMT)
    except:
        abort(400, "Invalid stopped_at parameter : '%s'." % stopped_at)
    sta = started_at.strftime(FILE_DATE_FMT)
    sto = stopped_at.strftime(FILE_DATE_FMT)

297a7dfc   Goutte   Add support for i...
1469
1470
1471
1472
1473
    input_slug = get_input_slug_from_query(inp=inp)

    cdf_filename = "%s_%s_%s_%s.cdf" % (
        separator.join(targets), input_slug, sta, sto
    )
aa7247d6   Goutte   Generate a CDF fi...
1474
1475
1476
1477
    cdf_path = join(CACHE_DIR, cdf_filename)

    if not isfile(cdf_path):
        log.debug("Creating the CDF file '%s'..." % cdf_filename)
604616e4   Goutte   Misc changes from...
1478
        try:
8a48d5fa   Goutte   Make sure spacepy...
1479
            from spacepy import pycdf
4aaf6874   Goutte   Try fixing the ge...
1480
1481
1482
1483
1484
1485
1486
1487
1488
        except ImportError:
            # If spacepy's toolbox is not up-to-date, importing will fail.
            # So, let's update and try again !
            update_spacepy()
            try:
                from spacepy import pycdf
            except ImportError as e:
                log.error("Failed to import pycdf from spacepy : %s" % e)
                raise
2fe06b17   Goutte   Move the ENV dire...
1489
1490
1491
        except Exception as e:
            log.error("Failed to import pycdf from spacepy : %s" % e)
            raise
4aaf6874   Goutte   Try fixing the ge...
1492

8a48d5fa   Goutte   Make sure spacepy...
1493
        try:
952e3d8f   Goutte   Move to another s...
1494
            cdf_handle = pycdf.CDF(cdf_path, masterpath='')
54bb1311   Goutte   Bring back the CM...
1495
1496
            targets_names = ', '.join([t['name'] for t in targets_configs])
            description = "Model and orbit data for %s." % targets_names
952e3d8f   Goutte   Move to another s...
1497
1498
1499
            cdf_handle.attrs['Description'] = description
            cdf_handle.attrs['Author'] = "Heliopropa.irap.omp.eu (CDPP)"
            cdf_handle.attrs['Created'] = str(time.ctime(time.time()))
aa7247d6   Goutte   Generate a CDF fi...
1500

54bb1311   Goutte   Bring back the CM...
1501
1502
1503
1504
            # fixme: Try changing the name from 00 to something relevant
            cdf_handle.attrs['Title'] = "Heliopropa - %s" % targets_names
            #######

aa7247d6   Goutte   Generate a CDF fi...
1505
1506
1507
            available_params = list(PROPERTIES)
            for target in targets_configs:
                target_slug = target['slug']
297a7dfc   Goutte   Add support for i...
1508
1509
1510
1511
                data = get_data_for_target(
                    target_config=target, input_slug=input_slug,
                    started_at=started_at, stopped_at=stopped_at
                )
aa7247d6   Goutte   Generate a CDF fi...
1512
1513
1514
                dkeys = sorted(data)

                values = []
aa7247d6   Goutte   Generate a CDF fi...
1515
                for dkey in dkeys:
952e3d8f   Goutte   Move to another s...
1516
1517
1518
1519
                    time_str = data[dkey][0][:-6]  # remove +00:00 tail
                    date = datetime.datetime.strptime(time_str, FILE_DATE_FMT)
                    values.append(date)
                kt = "%s_time" % target_slug
4aaf6874   Goutte   Try fixing the ge...
1520
                cdf_handle.new(kt, type=pycdf.const.CDF_EPOCH)
952e3d8f   Goutte   Move to another s...
1521
                cdf_handle[kt] = values
4aaf6874   Goutte   Try fixing the ge...
1522
                # cdf_handle[kt].attrs['FIELDNAM'] = "Time since 0 A.D"
aa7247d6   Goutte   Generate a CDF fi...
1523
1524
1525

                for param in params:
                    k = "%s_%s" % (target_slug, param)
4aaf6874   Goutte   Try fixing the ge...
1526
                    # print("PARAM %s" % k)
aa7247d6   Goutte   Generate a CDF fi...
1527
1528
                    values = []
                    i = available_params.index(param)
4aaf6874   Goutte   Try fixing the ge...
1529
                    has_nones = False
aa7247d6   Goutte   Generate a CDF fi...
1530
                    for dkey in dkeys:
4aaf6874   Goutte   Try fixing the ge...
1531
1532
1533
1534
1535
1536
1537
1538
1539
                        value = data[dkey][i]
                        if value is None:
                            has_nones = True
                        values.append(value)
                    if has_nones:
                        # PyCDF hates it when there are Nones.
                        # Since we don't know what value to set instead,
                        # let's skip the param altogether.
                        continue
aa7247d6   Goutte   Generate a CDF fi...
1540
                    cdf_handle[k] = values
952e3d8f   Goutte   Move to another s...
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
                    attrs = cdf_handle[k].attrs
                    attrs['UNITS'] = PARAMETERS[param]['units']
                    attrs['LABLAXIS'] = PARAMETERS[param]['name']
                    attrs['FIELDNAM'] = PARAMETERS[param]['title']
                    if values:
                        attrs['VALIDMIN'] = min(values)
                        attrs['VALIDMAX'] = max(values)

                kx = "%s_xhee" % target_slug
                ky = "%s_yhee" % target_slug
aa7247d6   Goutte   Generate a CDF fi...
1551
1552
1553
1554
1555
                values_xhee = []
                values_yhee = []
                index_x = available_params.index('xhee')
                index_y = available_params.index('yhee')
                for dkey in dkeys:
391d581c   Goutte   Make the CDF down...
1556
1557
1558
                    value_xhee = data[dkey][index_x]
                    value_yhee = data[dkey][index_y]
                    # We've got some `None`s cropping up in the data sometimes.
4aaf6874   Goutte   Try fixing the ge...
1559
                    # PyCDF does not digest Nones at all.
391d581c   Goutte   Make the CDF down...
1560
1561
1562
1563
1564
                    # While they solve this upstream, let's make an ugly fix!
                    if (value_xhee is not None) and (value_yhee is not None):
                        values_xhee.append(value_xhee)
                        values_yhee.append(value_yhee)
                    else:
4aaf6874   Goutte   Try fixing the ge...
1565
1566
                        values_xhee.append(0)
                        values_yhee.append(0)
d4b3d381   hitier   Tiny python3 synt...
1567
                        log.warning("Orbit data for %s has NaNs." % target_slug)
952e3d8f   Goutte   Move to another s...
1568
1569
1570
1571
                cdf_handle[kx] = values_xhee
                cdf_handle[ky] = values_yhee
                cdf_handle[kx].attrs['UNITS'] = 'Au'
                cdf_handle[ky].attrs['UNITS'] = 'Au'
aa7247d6   Goutte   Generate a CDF fi...
1572
1573

            log.debug("Writing CDF '%s'..." % cdf_filename)
952e3d8f   Goutte   Move to another s...
1574
1575
            cdf_handle.close()
            log.debug("Wrote CDF '%s'." % cdf_filename)
aa7247d6   Goutte   Generate a CDF fi...
1576
1577
1578

        except Exception as e:
            log.error("Failed to generate CDF '%s'." % cdf_filename)
952e3d8f   Goutte   Move to another s...
1579
1580
            if isfile(cdf_path):
                removefile(cdf_path)
aa7247d6   Goutte   Generate a CDF fi...
1581
1582
            raise

aa7247d6   Goutte   Generate a CDF fi...
1583
1584
1585
1586
1587
1588
    if not isfile(cdf_path):
        abort(500, "No CDF to serve. Looked at '%s'." % cdf_path)

    return send_from_directory(CACHE_DIR, cdf_filename)


11d86851   Goutte   Add support for s...
1589
1590
1591
1592
1593
1594
1595
1596
@app.route("/<target>_auroral_catalog.csv")
def download_auroral_catalog_csv(target):
    tc = validate_tap_target_config(target)
    log.debug("Requesting auroral emissions CSV for %s..." % tc['name'])

    filename = "%s_auroral_catalog.csv" % (target)
    local_csv_file = join(CACHE_DIR, filename)

11d86851   Goutte   Add support for s...
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
    target_name = tc['tap']['target_name']
    emissions = retrieve_auroral_emissions(target_name)

    # Be careful with regexes in python 2 ; best always use the ^$
    thumbnail_url_filter = re.compile("^.*proc(?:_small)?\\.(?:jpe?g|png|webp|gif|bmp|tiff)$")

    # Filter the emissions
    def _keep_emission(emission):
        ok = thumbnail_url_filter.match(emission['thumbnail_url'])
        # print("ok", ok, emission['thumbnail_url'])
        return bool(ok)

    emissions = [e for e in emissions if _keep_emission(e)]

    header = ('time_min', 'time_max', 'thumbnail_url', 'external_link')
    if len(emissions):
        header = emissions[0].keys()
b33d5f62   hitier   Change StringIO i...
1614
    si = StringIO()
11d86851   Goutte   Add support for s...
1615
1616
1617
    cw = csv_dict_writer(si, fieldnames=header)
    cw.writeheader()
    # 'time_min', 'time_max', 'thumbnail_url', 'external_link'
34e94ae3   hitier   Code reformating
1618
    # cw.writerow(head)
11d86851   Goutte   Add support for s...
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631

    log.debug("Writing auroral emissions CSV for %s..." % tc['name'])
    cw.writerows(emissions)

    log.info("Generated auroral emissions CSV contents for %s." % tc['name'])
    return si.getvalue()

    # if not isfile(local_csv_file):
    #     abort(500, "Could not cache CSV file at '%s'." % local_csv_file)
    #
    # return send_from_directory(CACHE_DIR, filename)


11d86851   Goutte   Add support for s...
1632
1633
1634
1635
1636
1637
1638
1639
1640
@app.route("/test/auroral/<target>")
def test_auroral_emissions(target):
    tc = validate_tap_target_config(target)
    target_name = tc['tap']['target_name']
    retrieved = retrieve_auroral_emissions(target_name)

    return "%d results:\n%s" % (len(retrieved), str(retrieved))


28bb4b28   Goutte   API for the cache...
1641
1642
# API #########################################################################

e18701b6   Goutte   Cache clear (remo...
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
@app.route("/cache/clear")
def cache_clear():
    """
    Removes all files from the cache.
    Note: It also removes the .gitkeep file. Not a problem for prod.
    """
    removed_files = remove_all_files(CACHE_DIR)
    count = len(removed_files)
    return "Cache cleared! Removed %d file%s." \
           % (count, 's' if count != 1 else '')


d9710a98   Goutte   Rename the cleanu...
1655
1656
@app.route("/cache/cleanup")
def cache_cleanup():
28bb4b28   Goutte   API for the cache...
1657
1658
    """
    Removes all files from the cache that are older than roughly one month.
e18701b6   Goutte   Cache clear (remo...
1659
    Note: It also removes the .gitkeep file. Maybe it should not, but hey.
28bb4b28   Goutte   API for the cache...
1660
1661
    """
    a_month_ago = datetime.datetime.now() - datetime.timedelta(days=32)
e18701b6   Goutte   Cache clear (remo...
1662
    removed_files = remove_files_created_before(a_month_ago, CACHE_DIR)
d9710a98   Goutte   Rename the cleanu...
1663
1664
1665
    count = len(removed_files)
    return "Cache cleaned! Removed %d old file%s." \
           % (count, 's' if count != 1 else '')
28bb4b28   Goutte   API for the cache...
1666
1667


b500e561   Goutte   Invert the orbits...
1668
1669
1670
1671
@app.route("/cache/warmup")
def cache_warmup():
    """
    Warms up the cache for the current day.
b500e561   Goutte   Invert the orbits...
1672
    """
390a3587   Goutte   Add a CRON config...
1673
    warmup_started_at = datetime.datetime.now()
284f4688   Goutte   Continue layers i...
1674
    sta, sto = get_interval_from_query()
b662fcc3   hitier   Change input name...
1675
    inp = config['defaults']['input_slug']  # default input, maybe warm them all up ?
fb383448   Goutte   Implement the cac...
1676
1677

    targets = get_active_targets()
fb383448   Goutte   Implement the cac...
1678
    targets_slugs = [target['slug'] for target in targets]
d4b3d381   hitier   Tiny python3 synt...
1679
    targets_slugs.sorty()
4aaf6874   Goutte   Try fixing the ge...
1680
1681
1682
1683

    update_spacepy()
    for target in targets:
        download_target_csv(target['slug'], inp, sta, sto)
297a7dfc   Goutte   Add support for i...
1684
    download_targets_cdf('-'.join(targets_slugs), inp, sta, sto)
fb383448   Goutte   Implement the cac...
1685

390a3587   Goutte   Add a CRON config...
1686
1687
1688
1689
    warmup_ended_at = datetime.datetime.now()
    warmup_timedelta = warmup_ended_at - warmup_started_at

    return "Done in %s." % str(warmup_timedelta)
b500e561   Goutte   Invert the orbits...
1690
1691


1324cc91   Goutte   Make the footer i...
1692
@app.route("/log")
596da00d   Goutte   Add more exceptio...
1693
@app.route("/log.html")
1324cc91   Goutte   Make the footer i...
1694
1695
def log_show():
    with open(LOG_FILE, 'r') as f:
bde97e4d   Goutte   Add more changes ...
1696
        contents = f.read()
fb383448   Goutte   Implement the cac...
1697
    return "<pre>" + contents + "</pre>"
bde97e4d   Goutte   Add more changes ...
1698
1699


1324cc91   Goutte   Make the footer i...
1700
1701
1702
1703
1704
1705
1706
@app.route("/log/clear")
def log_clear():
    with open(LOG_FILE, 'w') as f:
        f.truncate()
    return "Log cleared successfully."


1754789b   Goutte   Decorate and clea...
1707
1708
1709
1710
# DEV TOOLS ###################################################################

# @app.route("/inspect")
# def analyze_cdf():
a4a9ef03   Goutte   Cache generated C...
1711
#     """
1754789b   Goutte   Decorate and clea...
1712
#     For debug purposes.
a4a9ef03   Goutte   Cache generated C...
1713
#     """
1754789b   Goutte   Decorate and clea...
1714
1715
#     cdf_to_inspect = get_path("../res/dummy.nc")
#     cdf_to_inspect = get_path("../res/dummy_jupiter_coordinates.nc")
a4a9ef03   Goutte   Cache generated C...
1716
#
b33d5f62   hitier   Change StringIO i...
1717
#     si = StringIO()
1754789b   Goutte   Decorate and clea...
1718
1719
#     cw = csv.DictWriter(si, fieldnames=['Name', 'Shape', 'Length'])
#     cw.writeheader()
a4a9ef03   Goutte   Cache generated C...
1720
#
1754789b   Goutte   Decorate and clea...
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
#     # Time, StartTime, StopTime, V, B, N, T, Delta_angle, P_dyn, QualityFlag
#     cdf_handle = Dataset(cdf_to_inspect, "r", format="NETCDF4")
#     for variable in cdf_handle.variables:
#         v = cdf_handle.variables[variable]
#         cw.writerow({
#             'Name': variable,
#             'Shape': v.shape,
#             'Length': v.size,
#         })
#     cdf_handle.close()
a4a9ef03   Goutte   Cache generated C...
1731
1732
1733
1734
#
#     return si.getvalue()


9390ec89   Goutte   Initial experimen...
1735
1736
1737
# MAIN ########################################################################

if __name__ == "__main__":
952e3d8f   Goutte   Move to another s...
1738
    # Debug mode is on, as the production server does not use this but run.wsgi
9390ec89   Goutte   Initial experimen...
1739
1740
    extra_files = [get_path('../config.yml')]
    app.run(debug=True, extra_files=extra_files)