/
twexo.py
897 lines (815 loc) · 36.1 KB
/
twexo.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
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
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
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
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
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
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
"""
twexo.py - TESS Web Explode
TESS Input Catalog web browser data exploder.
WARNING: This routines opens tabs on your browser!
This routine saves a local file for the html page!
Always sanity check the target is correctly resolved!
High proper motion targets may not be resolved correctly.
Creates a local html page tailored for the target of interest
and loads the page in your web browser with the links to the major
catalogs to get information about the target. Targets can be specified
by TIC id, TOI number, common name, and ra/dec coordinates.
USAGE:
-to display command line arguments:
python twexo.py -h
-query by Name
python twexo.py -n 'Pi Mensae'
-For the impatient and the EXPLODE experience add -E
python twexo.py -n 'Pi Mensae' -E
-query by TOI number
python twexo.py -toi 144.01
-query by TIC number
python twexo.py -t 261136679
-query by Coordinates in decimal degrees
python twexo.py -c 84.291198 -80.469143
AUTHORS: Christopher J. Burke (MIT)
Testing and Advice from Susan Mullally (STScI) and Jennifer Burt (MIT)
VERSION: 0.5
NOTES: This routine opens tabs on your browser!
This routine saves a local file for the html!
DEPENDENCIES:
python 3+
astropy
astroquery
numpy
SPECIAL THANKS TO:
Includes code from the python MAST query examples
https://mast.stsci.edu/api/v0/pyex.html
Brett Morris (UW) for help with the name resolving from another the tess-point project
"""
import webbrowser
import numpy as np
import os
import argparse
from astropy.coordinates import SkyCoord
from astropy.time import Time, TimezoneInfo
from astroquery.gaia import Gaia
import astropy.units as u
import sys
import datetime
import json
try: # Python 3.x
from urllib.parse import quote as urlencode
from urllib.request import urlretrieve
from urllib.parse import urlencode as dict_urlencode
from urllib.request import urlopen
except ImportError: # Python 2.x
from urllib import pathname2url as urlencode
from urllib import urlretrieve
from urllib import urlendcode as dict_urlencode
from urllib import urlopen
try: # Python 3.x
import http.client as httplib
except ImportError: # Python 2.x
import httplib
import csv
import time
# Look for optional tess-point for predicting which TESS sectors data is avail
foo_have_tp = True
try:
from tess_stars2px import tess_stars2px_function_entry
except:
foo_have_tp = False
def idx_filter(idx, *array_list):
new_array_list = []
for array in array_list:
new_array_list.append(array[idx])
return new_array_list
## [Mast Query]
def mastQuery(request):
server='mast.stsci.edu'
# Grab Python Version
version = ".".join(map(str, sys.version_info[:3]))
# Create Http Header Variables
headers = {"Content-type": "application/x-www-form-urlencoded",
"Accept": "text/plain",
"User-agent":"python-requests/"+version}
# Encoding the request as a json string
requestString = json.dumps(request)
requestString = urlencode(requestString)
# opening the https connection
conn = httplib.HTTPSConnection(server)
# Making the query
conn.request("POST", "/api/v0/invoke", "request="+requestString, headers)
# Getting the response
resp = conn.getresponse()
head = resp.getheaders()
content = resp.read().decode('utf-8')
# Close the https connection
conn.close()
return head,content
## [Mast Query]
def safe_char(input):
""" https://stackoverflow.com/questions/5843518/remove-all-special-characters-punctuation-and-spaces-from-string
"""
if input:
import re
input = re.sub('\W+', '', input)
return input
if __name__ == '__main__':
# Parse the command line arguments
parser = argparse.ArgumentParser()
parser.add_argument("-t", "--ticId", type=int, \
help="TIC Id [int] ")
parser.add_argument("-c", "--coord", type=float, nargs=2, \
help="RA and Dec of target [deg]")
parser.add_argument("-n", "--name", nargs=1, type=str, \
help="Search for a target by resolving its name with SESAME")
parser.add_argument("-toi", "--toi", type=float, \
help="TOI number of target")
parser.add_argument("-E", "--explode", action='store_true', \
help="Pre-load all URLs into tabs of browser rather than just the link page")
parser.add_argument("-nw", "--noweb", action='store_true', \
help="Do not open in web. Only generate html that plays well with wkhtmltopdf conversion tool")
parser.add_argument("-of", "--outfile", nargs=1, type=str,\
help="Override default html output filename [default (without this argument) is based on query]. DO NOT include .html in this argument")
args = parser.parse_args()
# DEBUG BLOCK for hard coding input parameters and testing
# class test_arg:
# def __init__(self):
# self.ticId = 366443426
# self.ticId = None
# self.coord = None
# self.name = None
# self.name = 'Pi Mensae'
# self.toi = None
# self.explode = False
# self.toi = 383.01
# args = test_arg()
#Search radius for reporting nearby TIC targets
SEARCH_RAD = 60.0 # arcsec
# At least one Mode -t -c -n -toi must have been specified
if (args.ticId is None) and (args.coord is None) and (args.name is None) and (args.toi is None):
print('You must specify one and only one mode -t, -c, -n, -toi')
print('`python twex.py -h\' for help')
sys.exit(1)
useTIC = 0
# If TIC specified assign it to useTIC
# and query MAST for RA and Dec and other catalog identifiers
if args.ticId is not None:
useTIC = int(args.ticId)
starTics = np.array([useTIC], dtype=np.int32)
ticStringList = ['{0:d}'.format(x) for x in starTics]
# Setup mast query
request = {'service':'Mast.Catalogs.Filtered.Tic', \
'params':{'columns':'*', 'filters':[{ \
'paramName':'ID', 'values':ticStringList}]}, \
'format':'json', 'removenullcolumns':True}
headers, outString = mastQuery(request)
outObject = json.loads(outString)
if len(outObject['data']) > 0:
oo = outObject['data'][0]
starRa = outObject['data'][0]['ra']
starDec = outObject['data'][0]['dec']
star2mass = outObject['data'][0]['TWOMASS']
stargaia = outObject['data'][0]['GAIA']
if 'pmRA' in oo:
starPmRa = outObject['data'][0]['pmRA']
else:
starPmRa = 0.0
if 'pmDEC' in oo:
starPmDec = outObject['data'][0]['pmDEC']
else:
starPmDec = 0.0
starPmTot = np.sqrt(starPmRa*starPmRa + starPmDec*starPmDec)
if 'e_pmRA' in oo:
starPmRaE = outObject['data'][0]['e_pmRA']
starPmDecE = outObject['data'][0]['e_pmDEC']
starPmTotE = np.sqrt(starPmRaE*starPmRaE + starPmDecE*starPmDecE)
else:
starPmRaE = 100.0
starPmDecE = 100.0
starPmTotE = 1000.0
if 'Teff' in oo:
starTeff = outObject['data'][0]['Teff']
else:
starTeff = 0.0
if 'e_Teff' in oo:
starTeffE = outObject['data'][0]['e_Teff']
else:
starTeffE = 0.0
if 'logg' in oo:
starLogg = outObject['data'][0]['logg']
else:
starLogg = 0.0
if 'e_logg' in oo:
starLoggE = outObject['data'][0]['e_logg']
else:
starLoggE = 0.0
if 'rad' in oo:
starRad = outObject['data'][0]['rad']
else:
starRad = 0.0
if 'e_rad' in oo:
starRadE = outObject['data'][0]['e_rad']
else:
starRadE = 0.0
if 'mass' in oo:
starMass = outObject['data'][0]['mass']
else:
starMass = 0.0
if 'e_mass' in oo:
starMassE = outObject['data'][0]['e_mass']
else:
starMassE = 0.0
starTmag = outObject['data'][0]['Tmag']
#print(outObject['data'][0])
else:
print('MAST search of TIC: {0} did not return an entry'.format(args.ticId))
sys.exit(1)
# TOI specified get toi -> tic mapping from table at exofop
if args.toi is not None:
toiURL = 'https://exofop.ipac.caltech.edu/tess/download_toi.php?sort=toi&output=csv'
tmp = []
with urlopen(toiURL) as response:
for line in response:
line = line.decode('utf-8')
for x in csv.reader([line]):
tmp.append("|".join(x))
dtypeseq = ['i4','f8']
dtypeseq.extend(['i4']*7)
dtypeseq.extend(['U3','U3'])
dtypeseq.extend(['f8','f8'])
dtypeseq.extend(['U40','i4','U40','U40','U40'])
dtypeseq.extend(['f8']*27)
dtypeseq.extend(['U80']*4)
dataBlock = np.genfromtxt(tmp, \
dtype=dtypeseq, delimiter='|',skip_header=1)
exoTic = dataBlock['f0']
exoToi = dataBlock['f1']
idx = np.where(exoToi == args.toi)[0]
if len(idx) == 0:
print('Could not find TOI {0} at ExoFop'.format(args.toi))
sys.exit(1)
useTIC = int(exoTic[idx])
# Now that we have the TIC do a MAST query to get TIC values
starTics = np.array([useTIC], dtype=np.int32)
ticStringList = ['{0:d}'.format(x) for x in starTics]
# Setup mast query
request = {'service':'Mast.Catalogs.Filtered.Tic', \
'params':{'columns':'*', 'filters':[{ \
'paramName':'ID', 'values':ticStringList}]}, \
'format':'json', 'removenullcolumns':True}
headers, outString = mastQuery(request)
outObject = json.loads(outString)
oo = outObject['data'][0]
starRa = outObject['data'][0]['ra']
starDec = outObject['data'][0]['dec']
star2mass = outObject['data'][0]['TWOMASS']
stargaia = outObject['data'][0]['GAIA']
if 'pmRA' in oo:
starPmRa = outObject['data'][0]['pmRA']
else:
starPmRa = 0.0
if 'pmDEC' in oo:
starPmDec = outObject['data'][0]['pmDEC']
else:
starPmDec = 0.0
starPmTot = np.sqrt(starPmRa*starPmRa + starPmDec*starPmDec)
if 'e_pmRA' in oo:
starPmRaE = outObject['data'][0]['e_pmRA']
starPmDecE = outObject['data'][0]['e_pmDEC']
starPmTotE = np.sqrt(starPmRaE*starPmRaE + starPmDecE*starPmDecE)
else:
starPmRaE = 100.0
starPmDecE = 100.0
starPmTotE = 1000.0
if 'Teff' in oo:
starTeff = outObject['data'][0]['Teff']
else:
starTeff = 0.0
if 'e_Teff' in oo:
starTeffE = outObject['data'][0]['e_Teff']
else:
starTeffE = 0.0
if 'logg' in oo:
starLogg = outObject['data'][0]['logg']
else:
starLogg = 0.0
if 'e_logg' in oo:
starLoggE = outObject['data'][0]['e_logg']
else:
starLoggE = 0.0
if 'rad' in oo:
starRad = outObject['data'][0]['rad']
else:
starRad = 0.0
if 'e_rad' in oo:
starRadE = outObject['data'][0]['e_rad']
else:
starRadE = 0.0
if 'mass' in oo:
starMass = outObject['data'][0]['mass']
else:
starMass = 0.0
if 'e_mass' in oo:
starMassE = outObject['data'][0]['e_mass']
else:
starMassE = 0.0
starTmag = outObject['data'][0]['Tmag']
#print('TOI: {0} is associated with TIC: {1} at ExoFop'.format(args.toi, useTIC))
# Do name resolving to get ra and dec
if args.name is not None:
# Name resolve in try except for detecting problem
try:
coordinate = SkyCoord.from_name(args.name[0])
print("Coordinates for {0}: ({1}, {2})"
.format(args.name[0], coordinate.ra.degree,
coordinate.dec.degree))
starRa = coordinate.ra.degree
starDec = coordinate.dec.degree
except:
print("Could not resolve: {0}".format(args.name[0]))
sys.exit(1)
if args.coord is not None:
starRa = args.coord[0]
starDec = args.coord[1]
# Do a MAST cone search around the ra and dec to get the nearby stars
# for all runs and to find the closest TIC for the coordinate input
# and name query
# Protect against missing TICs
try:
# Do cone search around this position
startTime = time.time()
request = {'service':'Mast.Catalogs.Filtered.Tic.Position', \
'params':{'columns':'c.*', \
'filters':[ \
{'paramName':'Tmag',\
'values':[{'min':0, 'max':20.0}]}], \
'ra':'{:10.5f}'.format(starRa),\
'dec':'{:10.5f}'.format(starDec),\
'radius':'{:10.7f}'.format(SEARCH_RAD/3600.0) \
}, \
'format':'json', 'removenullcolumns':False}
while True:
headers, outString = mastQuery(request)
outObject = json.loads(outString)
if outObject['status'] != 'EXECUTING':
break
if time.time() - startTime > 30:
print('Working...')
startTime = time.time()
time.sleep(5)
ticList = np.array([x['ID'] for x in outObject['data']], dtype=np.int32)
ticRas = np.array([x['ra'] for x in outObject['data']], dtype=np.float)
ticDecs = np.array([x['dec'] for x in outObject['data']], dtype=np.float)
ticTeffs = np.array([x['Teff'] for x in outObject['data']], dtype=np.float)
ticTeffEs = np.array([x['e_Teff'] for x in outObject['data']], dtype=np.float)
ticLoggs = np.array([x['logg'] for x in outObject['data']], dtype=np.float)
ticLoggEs = np.array([x['e_logg'] for x in outObject['data']], dtype=np.float)
ticRads = np.array([x['rad'] for x in outObject['data']], dtype=np.float)
ticRadEs = np.array([x['e_rad'] for x in outObject['data']], dtype=np.float)
ticMasss = np.array([x['mass'] for x in outObject['data']], dtype=np.float)
ticMassEs = np.array([x['e_mass'] for x in outObject['data']], dtype=np.float)
ticTmags = np.array([x['Tmag'] for x in outObject['data']], dtype=np.float)
except:
# Cone search failed
print('TIC Search Fail at position {0} {1}'.format(starRa, starDec))
sys.exit(1)
# Check if no targets were returned in cone search
if len(ticList) == 0:
print('MAST cone search around TIC returned no objects')
sys.exit(1)
targCoord = SkyCoord(starRa, starDec, unit='deg')
ticCoords = SkyCoord(ticRas, ticDecs, unit='deg')
seps = targCoord.separation(ticCoords)
ia = np.argsort(seps)
ticList, ticRas, ticDes, seps, ticTeffs, ticTeffEs, ticLoggs, ticLoggEs, \
ticRads, ticRadEs, ticMasss, ticMassEs, ticTmags = idx_filter(ia , \
ticList, ticRas, ticDecs, seps, ticTeffs, ticTeffEs, ticLoggs, ticLoggEs, \
ticRads, ticRadEs, ticMasss, ticMassEs, ticTmags)
# If we hadn't defined which target is useTIC yet do it now
if useTIC == 0:
useTIC = ticList[0]
starTics = np.array([useTIC], dtype=np.int32)
ticStringList = ['{0:d}'.format(x) for x in starTics]
# Setup mast query
request = {'service':'Mast.Catalogs.Filtered.Tic', \
'params':{'columns':'*', 'filters':[{ \
'paramName':'ID', 'values':ticStringList}]}, \
'format':'json', 'removenullcolumns':True}
headers, outString = mastQuery(request)
outObject = json.loads(outString)
#print(outObject['data'][0])
oo = outObject['data'][0]
starRa = outObject['data'][0]['ra']
starDec = outObject['data'][0]['dec']
star2mass = outObject['data'][0]['TWOMASS']
stargaia = outObject['data'][0]['GAIA']
if 'pmRA' in oo:
starPmRa = outObject['data'][0]['pmRA']
else:
starPmRa = 0.0
if 'pmDEC' in oo:
starPmDec = outObject['data'][0]['pmDEC']
else:
starPmDec = 0.0
starPmTot = np.sqrt(starPmRa*starPmRa + starPmDec*starPmDec)
if 'e_pmRA' in oo:
starPmRaE = outObject['data'][0]['e_pmRA']
starPmDecE = outObject['data'][0]['e_pmDEC']
starPmTotE = np.sqrt(starPmRaE*starPmRaE + starPmDecE*starPmDecE)
else:
starPmRaE = 100.0
starPmDecE = 100.0
starPmTotE = 1000.0
if 'Teff' in oo:
starTeff = outObject['data'][0]['Teff']
else:
starTeff = 0.0
if 'e_Teff' in oo:
starTeffE = outObject['data'][0]['e_Teff']
else:
starTeffE = 0.0
if 'logg' in oo:
starLogg = outObject['data'][0]['logg']
else:
starLogg = 0.0
if 'e_logg' in oo:
starLoggE = outObject['data'][0]['e_logg']
else:
starLoggE = 0.0
if 'rad' in oo:
starRad = outObject['data'][0]['rad']
else:
starRad = 0.0
if 'e_rad' in oo:
starRadE = outObject['data'][0]['e_rad']
else:
starRadE = 0.0
if 'mass' in oo:
starMass = outObject['data'][0]['mass']
else:
starMass = 0.0
if 'e_mass' in oo:
starMassE = outObject['data'][0]['e_mass']
else:
starMassE = 0.0
starTmag = outObject['data'][0]['Tmag']
# If we weren't given TOI number check exofop to see if it has TOIs
if args.toi is None:
toiURL = 'https://exofop.ipac.caltech.edu/tess/download_toi.php?sort=toi&output=csv'
tmp = []
with urlopen(toiURL) as response:
for line in response:
line = line.decode('utf-8')
for x in csv.reader([line]):
tmp.append("|".join(x))
dtypeseq = ['i4','f8']
dtypeseq.extend(['i4']*7)
dtypeseq.extend(['U3','U3'])
dtypeseq.extend(['f8','f8'])
dtypeseq.extend(['U40','i4','U40','U40','U40'])
dtypeseq.extend(['f8']*27)
dtypeseq.extend(['U80']*4)
dataBlock = np.genfromtxt(tmp, \
dtype=dtypeseq, delimiter='|',skip_header=1)
exoTic = dataBlock['f0']
exoToi = dataBlock['f1']
idx = np.where(exoTic == useTIC)[0]
if len(idx) == 0:
toihdr = 'No TOIs associated with TIC at ExoFop'
toicheckstr = '<br>'
else:
toihdr = 'TIC Hosts TOIs'
toilist = []
for j in idx:
toilist.append('{0:.2f}<br>'.format(exoToi[j]))
toicheckstr = ' '.join(toilist)
else:
toihdr = 'Using TOI {:.2f}'.format(args.toi)
toicheckstr = ''
# FORM THE URLS
gaiaURLPart1 = 'https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html?searchQuery={"service":"GAIADR2","inputText":"'
gaiaURLPart2 = '{0:f} {1:f} r{2:f}",'.format(starRa,starDec,1.0/60.0)
gaiaURLPart3 = '"paramsService":"Mast.Catalogs.GaiaDR2.Cone","title":"Gaia (DR2)","columns":"*"}'
gaiaURL = gaiaURLPart1 + gaiaURLPart2 + gaiaURLPart3
# TESScut Target pixel file
tcutURLPart1 = 'https://mast.stsci.edu/tesscut/api/v0.1/astrocut?'
tcutURLPart2 = 'ra={0}&dec={1}'.format(starRa, starDec)
tcutURLPart3 = '&y=15&x=15&units=px§or=All'
tcutURL = tcutURLPart1 + tcutURLPart2 + tcutURLPart3
# ESO archive
DATE_STR = datetime.date.today().strftime('%Y-%m-%d')
esoURLPart1 = 'https://archive.eso.org/scienceportal/home?data_release_date=*:{0}&'.format(DATE_STR)
esoURLPart2 = 'pos={0},{1}'.format(starRa,starDec)
# esoURLPart3 = '&r=0.008333&sort=dist,-fov,-obs_date&s=P%2fDSS2%2fcolor&f=0.064387&fc=84.485552,-80.451816&cs=J2000&av=true&ac=false&c=8,9,10,11,12,13,14,15,16,17,18&mt=true&dts=true'
esoURLPart3 = '&r=0.008333&sort=dist,-fov,-obs_date&s=P%2fDSS2%2fcolor&f=0.064387&cs=J2000&av=true&ac=false&c=8,9,10,11,12,13,14,15,16,17,18&mt=true&dts=true'
esoURL = esoURLPart1 + esoURLPart2 + esoURLPart3
# IRSA finderchart
irsaURLPart1 = 'https://irsa.ipac.caltech.edu/applications/finderchart/servlet/api?locstr='
#irsaURLPart1 = 'https://irsa.ipac.caltech.edu/applications/finderchart/?__action=table.search&request=%7B%22startIdx%22%3A0%2C%22pageSize%22%3A100%2C%22id%22%3A%22QueryFinderChartWeb%22%2C%22tbl_id%22%3A%22results%22%2C%22UserTargetWorldPt%22%3A%22126.61572%3B10.08046%3BEQ_J2000%3B'
irsaURLPart2 = urlencode('2MASS J{0}'.format(star2mass))
# irsaURLPart2 = '2MASS J{0}'.format(star2mass)
irsaURLPart3 = '&mode=getResult&subsetsize=4.0&searchCatalog=no&survey='
irsaURLPart4 = 'DSS,SDSS,2MASS'
# irsaURLPart4 = urlencode('DSS,SDSS,2MASS')
#irsaURLPart3 = '%3Bned%22%2C%22imageSizeAndUnit%22%3A%220.042777777777777776%22%2C%22thumbnail_size%22%3A%22256%22%2C%22selectImage%22%3A%22dss%2Csdss%2C2mass%22%2C%22searchCatalog%22%3A%22no%22%2C%22ckgDSS%22%3A%22dss1Blue%2Cdss1Red%2Cdss2Blue%2Cdss2Red%2Cdss2IR%22%2C%22ckgSDSS%22%3A%22u%2Cg%2Cr%2Cz%22%2C%22ckg2MASS%22%3A%22j%2Ch%2Ck%22%2C%22imageSearchOptions%22%3A%22closed%22%2C%22META_INFO%22%3A%7B%22title%22%3A%22QueryFinderChartWeb%22%2C%22tbl_id%22%3A%22results%22%7D%7D&options=%7B%22tbl_group%22%3A%22results%22%2C%22removable%22%3Afalse%2C%22showTitle%22%3Afalse%2C%22pageSize%22%3A100%7D'
irsaURL = irsaURLPart1 + irsaURLPart2 + irsaURLPart3 + irsaURLPart4
# MAST Data portal
#mstURLPart1 = 'https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html?searchQuery=%7B%22service%22%3A%22CAOMDB%22%2C%22inputText%22%3A%22'
#mstURLPart2 = 'TIC%20{0:d}'.format(useTIC)
#mstURLPart3 = '%22%2C%22paramsService%22%3A%22Mast.Caom.Cone%22%2C%22columns%22%3A%22*%22%2C%22caomVersion%22%3Anull%7D'
mstURLPart1 = 'https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html?searchQuery={"service":"CAOMFILTERED","inputText":[{"paramName":"obs_collection","niceName":"obs_collection","values":[],"valString":"TESS","isDate":false,"separator":";","freeText":"TESS","displayString":"TESS"}],"position":"'
mstURLPart2 = '{0:f}, {1:f}, {2:f}",'.format(starRa,starDec,2.0/60.0/60.0)
mstURLPart3 = '"paramsService":"Mast.Caom.Filtered.Position","columns":"*"}'
mstURL = mstURLPart1 + mstURLPart2 + mstURLPart3
#print(mstURL)
#Vizier search
vizPOSTDict = {'-c': '2MASS J{0}'.format(star2mass)}
vizURLPart1 = 'http://vizier.u-strasbg.fr/viz-bin/VizieR-4?-out.max=50&-out.form=HTML+Table&-out.add=_r&-out.add=_RAJ+_DEJ&outaddvalue=default&-sort=_r&-order=I&-oc.form=sexa&'
vizURLPart2 = dict_urlencode(vizPOSTDict)
vizURLPart3 = '&-c.eq=J2000&-c.r=+30&-c.u=arcsec&-c.geom=r'
vizURL = vizURLPart1 + vizURLPart2 + vizURLPart3
#print(vizURL)
# exofop
exofopURL = 'https://exofop.ipac.caltech.edu/tess/target.php?id={0}'.format(useTIC)
# Simbad page
simbadPOSTDict = {'Ident': '2MASS J{0}'.format(star2mass)}
simbadURL = 'http://simbad.u-strasbg.fr/simbad/sim-basic?{0}'.format(dict_urlencode(simbadPOSTDict))
#print(simbadURL)
# Prepare strings for the HTML page
DATE_STR = datetime.date.today().strftime('%Y%m%d')
if args.toi is not None:
path = os.path.abspath('twexo_temp_{0}_{1:.2f}.html'.format(DATE_STR,args.toi))
ModeHeader = 'TOI: {0} is associated with TIC: {1} at ExoFop'.format(args.toi, useTIC)
if args.ticId is not None:
path = os.path.abspath('twexo_temp_{0}_{1:016d}.html'.format(DATE_STR,args.ticId))
ModeHeader = 'Using TIC: {0}'.format(useTIC)
if args.name is not None:
# strip white space and only keep regular characters to protect against
# special characters that would be bad to put into filenames also clip lenght
useName = safe_char(args.name[0])
if (len(useName)>30):
useName = useName[0:11]
path = os.path.abspath('twexo_temp_{0}_{1}.html'.format(DATE_STR,useName))
ModeHeader = 'From {0} Using TIC: {1}'.format(args.name[0], useTIC)
if args.coord is not None:
path = os.path.abspath('twexo_temp_{0}_{1:016d}.html'.format(DATE_STR,useTIC))
ModeHeader = 'From input coordinates Using TIC: {0}'.format(useTIC)
if args.outfile is not None:
path = os.path.relpath(args.outfile[0]+'.html')
print(path)
ticPos = 'Using TIC catalog position {0:.6f} {1:.6f} [J2000.0; epoch 2000.0]'.format(starRa, starDec)
twoMassId = '2MASS J{0} From TIC'.format(star2mass)
closeTICN = '{0} TIC entries within {1} arcsec of target {2}'.format(len(ticList)-1, SEARCH_RAD, useTIC)
neighTIC = []
for i, curTIC in enumerate(ticList[1:21]):
if np.isfinite(ticTeffs[i+1]):
curTeff = ticTeffs[i+1]
else:
curTeff = 0.0
if np.isfinite(ticLoggs[i+1]):
curLogg = ticLoggs[i+1]
else:
curLogg = 0.0
if np.isfinite(ticRads[i+1]):
curRad = ticRads[i+1]
else:
curRad = 0.0
neighTIC.append('{0:d} Sep [arcsec]: {1:8.3f} Tmag: {2:5.2f} Teff: {3:6.1f} Logg: {4:5.2f} Rs[Rsun]: {5:6.2f}<br>'.format(\
curTIC, seps[i+1].arcsecond, ticTmags[i+1], curTeff, \
curLogg, curRad))
neighTICStr = ' '.join(neighTIC)
# Since we got our positions from the TIC (which are in epoch 2000.0)
# propagate the proper motion to find coordinates in 2015.5 which is gaia
# First sanity check pm data from TIC
pmOK = True
if (not np.isfinite(starPmTot)) or (not np.isfinite(starPmTotE)):
pmOK = False
if np.abs(starPmTot)/starPmTotE < 1.5:
pmOK = False
if pmOK:
print('TIC has proper motion data. Getting 2015.5 position for GAIA')
ctic = SkyCoord(ra=starRa * u.deg,
dec=starDec * u.deg,
pm_ra_cosdec=starPmRa * u.mas/u.yr,
pm_dec=starPmDec * u.mas/u.yr,
obstime=Time('J2000.0'),
distance=1000.0*u.pc, # Use fake distance just for skycoord functionality
radial_velocity = 0.0 * u.km/u.s) # Use fake rv "
# convert position to GAIA DR2 2015.5
gaiaEpc = Time('J2015.5')
ctic_gaia_epoch = ctic.apply_space_motion(gaiaEpc)
gaiaPredRa = ctic_gaia_epoch.ra.degree
gaiaPredDec = ctic_gaia_epoch.dec.degree
else:
gaiaPredRa = starRa
gaiaPredDec = starDec
gaiaPos = 'Predicted GAIA position {0:.6f} {1:.6f} [J2000.0; epoch 2015.5]'.format(gaiaPredRa, gaiaPredDec)
# We are go for GAIA Cone search
ADSQL_Str = "SELECT \
DISTANCE( POINT('ICRS', ra, dec),\
POINT('ICRS', {0}, {1}) ) AS dist, phot_g_mean_mag, teff_val, teff_percentile_lower, \
teff_percentile_upper, radius_val, radius_percentile_lower, radius_percentile_upper,\
astrometric_gof_al, astrometric_excess_noise_sig, phot_bp_mean_mag, phot_rp_mean_mag, bp_rp, \
a_g_val, e_bp_min_rp_val, parallax, \
ra, dec , solution_id \
from gaiadr2.gaia_source WHERE 1 = CONTAINS( POINT('ICRS', ra, dec), \
CIRCLE('ICRS', {0}, {1}, 0.016666666666666666)) order by dist".format(gaiaPredRa, gaiaPredDec)
#print(ADSQL_Str)
job = Gaia.launch_job(ADSQL_Str)
r = job.get_results()
gaiaTeff = r['teff_val'][0]
gaiaTeffE = ((r['teff_percentile_upper'][0] - gaiaTeff) + (gaiaTeff - r['teff_percentile_lower'][0]))/2.0
gaiaRad = r['radius_val'][0]
gaiaRadE = ((r['radius_percentile_upper'][0] - gaiaRad) + (gaiaRad - r['radius_percentile_lower'][0]))/2.0
gaiaRpMag = r['phot_rp_mean_mag'][0]
gaiaGMag = r['phot_g_mean_mag'][0]
gaiaBpMag = r['phot_bp_mean_mag'][0]
gaiapar = r['parallax'][0]
gaiaag = float(r['a_g_val'][0])
if not np.isfinite(gaiaag):
gaiaag = 0.0
gaiaexclr = float(r['e_bp_min_rp_val'][0])
if not np.isfinite(gaiaexclr):
gaiaexclr = 0.0
gaiaastrogof = r['astrometric_gof_al'][0]
gaiaastrexsig = r['astrometric_excess_noise_sig'][0]
#print(gaiaGMag, gaiapar, gaiaag, gaiaBpMag, gaiaRpMag, gaiaexclr)
gaiaAbsGMag = gaiaGMag + 5.0 + 5.0*np.log10(gaiapar/1000.0)-gaiaag
gaiaClro = gaiaBpMag - gaiaRpMag - gaiaexclr
gaiaExStr = 'Other GAIA G: {0:5.2f} Bp: {1:5.2f} AbsG: {2:5.2f} (Bp-Rp)o: {3:5.3f} AstroGOF: {4:6.2f} AstroExNoiSig: {5:6.2f}<br>'.format(
gaiaGMag, gaiaBpMag, gaiaAbsGMag, gaiaClro, gaiaastrogof, gaiaastrexsig)
# print(r['teff_percentile_upper'])
# print(r['teff_percentile_lower'])
#print(r)
#LOOK for DV reports for this target available at MAST
# Setup mast query first step is to get obsid's for the target
print('Query MAST for DV reports')
dvStr = 'No DV Results for this target at MAST<br>'
request = {'service':'Mast.Caom.Filtered.Position', \
'params':{'columns':'*', \
'position':'{0:f}, {1:f}, {2:f}'.format(starRa,starDec,2.0/60.0/60.0), \
'filters':[
{'paramName':'project',
'values':['TESS']
},\
{'paramName':'dataproduct_type',
'values':['timeseries']
}]
}, \
'format':'json', 'removenullcolumns':True}
headers, outString = mastQuery(request)
outObject = json.loads(outString)
if not outObject['status'] == 'EXECUTING':
# Check if any objects returned
if len(outObject['data']) > 0:
# Now get list of obsids for the time series data
#print(len(outObject['data']))
obsidStr = []
for i, oo in enumerate(outObject['data']):
#print(i)
#print(oo)
#print(oo['obsid'])
obsidStr.append('{0:d}'.format(oo['obsid']))
# Have list of obs ids for this TIC
# Request the data products associated with this obs id
request = {'service':'Mast.Caom.Products', \
'params':{'obsid':','.join(obsidStr), \
}, \
'format':'json', 'removenullcolumns':True}
headers, outString = mastQuery(request)
outObject = json.loads(outString)
#print(len(outObject['data']))
dvLines = []
for i, oo in enumerate(outObject['data']):
#print(i)
dataURI = oo['dataURI']
suffix = dataURI[-3:]
if suffix == 'pdf':
useURL = 'https://mast.stsci.edu/api/v0/download/file?uri=' + dataURI
dvLines.append('<a href="{0}" target="_blank">{0}</a><br>'.format(useURL))
dvStr = ' '.join(dvLines)
else:
dvStr = 'No DV Results for this target at MAST<br>'
else:
print('Timeout on the DV report search at MAST')
dvStr = '***DV report search at MAST timed out. Try Later***<br>'
# Do an optional report that shows which sectors target is TESS observable
# One needs tess-point install for this to work
tesspStr = '<br>'
if foo_have_tp:
outID, outEclipLong, outEclipLat, outSec, outCam, outCcd, \
outColPix, outRowPix, scinfo = tess_stars2px_function_entry(
useTIC, gaiaPredRa, gaiaPredDec)
tpLines = ['<br>TESS Observe - Sec Cam Ccd Col Row<br>']
for i, curTIC in enumerate(outID):
tpLines.append('{0:d} {1:d} {2:d} {3:d} {4:.2f} {5:.2f}<br>'.format(outID[i], \
outSec[i],outCam[i], outCcd[i], outColPix[i], outRowPix[i]))
tesspStr = ' '.join(tpLines)
# Wkhtmltopdf is used to convert html to pdf
# On certain platforms URLs don't work unless they are one per line
# with args.noweb use a html line break
URLBreak = '|'
if args.noweb:
URLBreak = '<br>'
# We had all the HTML strings, put them in a dictionary that will
# will be used for replacement in HTML template below
htmlDict = {'neighTIC':neighTICStr, 'closeTICN':closeTICN, \
'twoMassId':twoMassId, 'ticPos':ticPos, \
'ModeHeader':ModeHeader, 'useTIC':useTIC, \
'exofopURL':exofopURL, 'simbadURL':simbadURL, 'vizURL':vizURL, \
'mstURL':mstURL, 'irsaURL':irsaURL, 'esoURL':esoURL, \
'toicheckstr':toicheckstr, 'toihdr':toihdr, 'gaiaPos':gaiaPos,\
'T_Tmag':starTmag, 'T_Teff':'{0:7.1f}±{1:5.1f}'.format(starTeff,starTeffE), \
'T_Logg':'{0:4.2f}±{1:4.2f}'.format(starLogg, starLoggE), \
'T_Rstar':'{0:6.2f}±{1:4.1f}'.format(starRad, starRadE), \
'T_Mstar':'{0:5.2f}±{1:5.2f}'.format(starMass, starMassE), \
'G_Rmag':'{0:6.3f}'.format(gaiaRpMag), \
'G_Teff':'{0:7.1f}±{1:5.1f}'.format(gaiaTeff, gaiaTeffE),\
'G_Rstar':'{0:6.2f}±{1:4.1f}'.format(gaiaRad, gaiaRadE), \
'tesspStr':tesspStr, 'dvStr':dvStr, 'gaiaExStr':gaiaExStr, \
'tcutURL':tcutURL, 'gaiaURL':gaiaURL, 'URLBreak':URLBreak}
#print(htmlDict['mstURL'])
#print('{mstURL}'.format(**htmlDict))
#print('hello')
#HTML TEMPLATE
template = """
<html>
<head>
<title>TESS Web Exploder {useTIC}</title>
<style>
table, th, td {{
border: 1px solid black;
border-collapse: collapse;
}}
th, td {{
padding: 5px;
}}
th {{
text-align: left;
}}
</style>
</head>
<body>
<h1>{ModeHeader}</h1>
<h3>{ticPos}</h3>
<h3>{gaiaPos}</h3>
<h3>{twoMassId}</h3>
<h3>{closeTICN}</h3>
{neighTIC}
<h3>{toihdr}</h3>
{toicheckstr}
<h3>Target Parameters</h3>
<table style="width:100%">
<tr>
<th>Catalog</th>
<th>Tmag/Rpmag</th>
<th>Teff</th>
<th>Logg</th>
<th>Rstar</th>
<th>Mstar</th>
</tr>
<tr>
<td>TIC</td>
<td>{T_Tmag}</td>
<td>{T_Teff}</td>
<td>{T_Logg}</td>
<td>{T_Rstar}</td>
<td>{T_Mstar}</td>
</tr>
<tr>
<td>GAIA DR2</td>
<td>{G_Rmag}</td>
<td>{G_Teff}</td>
<td>...</td>
<td>{G_Rstar}</td>
<td>...</td>
</tr>
</table>
{gaiaExStr}
<h2>Target Links</h2>
<a href="{exofopURL}" target="_blank">ExoFOP</a> {URLBreak}
<a href="{simbadURL}" target="_blank">Simbad</a> {URLBreak}
<a href="{vizURL}" target="_blank">Vizier</a> {URLBreak}
<a href='{mstURL}' target='_blank'>MAST TESS Data Holdings</a> {URLBreak}
<a href="{irsaURL}" target="_blank">IRSA Finderchart</a> {URLBreak}
<a href="{esoURL}" target="_blank">ESO Data Archive Holdings</a> {URLBreak}
<a href="{tcutURL}" target="_blank">TESScut TPF Download</a> {URLBreak}
<a href='{gaiaURL}' target='_blank'>GAIA DR2 60'' Cone Search @MAST</a>
<h2>NASA Ames SPOC DV Results Available at MAST</h2>
{dvStr}
<br>
{tesspStr}
</body>
</html>
"""
# Replace tags in HTML template with the strings in the htmlDict
page_string = template.format(**htmlDict)
url = 'file://' + path
# Write the html to a local file
with open(path, 'w') as f:
f.write(page_string)
if not args.noweb:
# If the user requested a web explode load em all!
if args.explode:
webbrowser.open(gaiaURL, new=2)
webbrowser.open(esoURL, new=2)
webbrowser.open(irsaURL, new=2)
webbrowser.open(mstURL, new=2)
webbrowser.open(vizURL, new=2)
webbrowser.open(simbadURL, new=2)
webbrowser.open(exofopURL, new=2)
# Load the webpage
webbrowser.open(url)