source: trunk/DataCheck/CheckRawData.sh@ 12648

Last change on this file since 12648 was 12645, checked in by Daniela Dorner, 13 years ago
changed option for automatic running
  • Property svn:executable set to *
File size: 14.7 KB
Line 
1#!/bin/bash
2
3today=`date +%F`
4logfile=/home/`whoami`/DataCheck/log/CheckRaw$today.log
5
6skipmd5sum="no" # fill md5 sums in any case
7skipmd5sum="iffilled" # fill md5 sum only if they are not yet in db
8#skipmd5sum="yes" # do not fill md5 sums in any case
9
10doupdate="yes" # update all entries
11doupdate="no" # fill only entries which are not yet existing
12
13fillonly="ifzipped" # fill only entries which are not yet existing + where zipped file is availabl
14
15echo "" >> $logfile 2>&1
16echo "" >> $logfile 2>&1
17echo "" >> $logfile 2>&1
18echo `date`"executing "$0"..." >> $logfile 2>&1
19echo "=====> doupdate: "$doupdate >> $logfile 2>&1
20echo "=====> skipmd5sum: "$skipmd5sum >> $logfile 2>&1
21
22password=`cat /home/fact/DataCheck/.pw`
23if [ "$password" == "" ]
24then
25 echo "please insert password in .pw file"
26fi
27
28# setup to use ftools
29export HEADAS=/opt/heasoft-6.11/x86_64-unknown-linux-gnu-libc2.13-0/
30source $HEADAS/headas-init.sh
31
32# get last 2 nights
33dates=( `date +%Y/%m/%d --date="-12hour"` `date +%Y/%m/%d --date="-36hour"` )
34
35# do rsync for rawfiles of these dates
36for date in ${dates[@]}
37do
38 echo "" >> $logfile 2>&1
39 echo "" >> $logfile 2>&1
40 echo "" >> $logfile 2>&1
41 rawdir=/loc_data/raw/$date
42 runnumber=`echo $date | sed -e 's/\///g'`
43 echo `date`": processing files in "$rawdir >> $logfile 2>&1
44 # check if data are available from that night
45 if ! [ -d $rawdir ]
46 then
47 echo `date`": no data available in "$rawdir >> $logfile 2>&1
48 continue
49 fi
50
51 # find all fits-files starting with the oldest file
52 echo `date`": finding files to be zipped in $rawdir..." >> $logfile 2>&1
53 fitsfiles=`find $rawdir -type f -name '*.fits'| sort `
54
55 # loop to zip files
56 echo `date`": checking files in $rawdir..." >> $logfile 2>&1
57 for file in $fitsfiles
58 do
59 echo "" >> $logfile 2>&1
60 echo "checking file "$file >> $logfile 2>&1
61 # check if raw file was changed in the last 30 minutes
62 isnew=`find $file -cmin -30`
63 if [ "$isnew" != "" ]
64 then
65 echo $file" is not older than 30 min => continue" >> $logfile 2>&1
66 continue
67 fi
68
69 # check if it is drs file
70 isnew=`ls $file | grep drs`
71 if [ "$isnew" != "" ]
72 then
73 echo $file" is a drs file => continue" >> $logfile 2>&1
74 continue
75 fi
76
77 # check if file is already finished
78 # original file on daq (if data was taken on daq
79 origfile=`echo $file | sed -e 's/loc_data/daq/'`
80 if [ -e $origfile ]
81 then
82 # get time of last modification as seconds since Epoch for both files
83 timeorig=`stat -c %Y $origfile`
84 timecopy=`stat -c %Y $file`
85 # compare times
86 if ! [ $timeorig -eq $timecopy ]
87 then
88 # if times are not the same, the file is still open => no zip
89 echo `date`": file "$file" not yet closed => continue" >> $logfile 2>&1
90 continue
91 fi
92 else
93 # if the origfile doesn't exist, the data was probably written not on daq but on data
94 echo `date`": file "$file" was probably taken on data and not daq " >> $logfile 2>&1
95 fi
96
97 # get run and file number form filename
98 runnumbererror="no"
99 numbererror="no"
100 numberfromname=`echo $file | grep -E -o '20[1-9][0-9][01][0-9][0-3][0-9]_[0-9]{3}'`
101 runnumberfromname=`echo $numberfromname | cut -d_ -f1`
102 filenumberfromname=`echo $numberfromname | cut -d_ -f2 | sed -e 's/^0//g' -e 's/^0//g'`
103 if [ "$runnumber" != "$runnumberfromname" ]
104 then
105 runnumbererror="yes"
106 echo "ERROR: for file "$file" runnumber from date and filename don't agree ("$runnumber" - "$runnumberfromname")"
107 echo "ERROR: for file "$file" runnumber from date and filename don't agree ("$runnumber" - "$runnumberfromname")" >> $logfile 2>&1
108 fi
109
110 # check if entry already exists
111 query3="SELECT fRunNumber FROM RunInfo WHERE Concat(fRunNumber, '_', LPAD(fFileNumber, 3, 0))='"$numberfromname"'"
112 echo "Q3:"$query3 >> $logfile 2>&1 2>&1
113 result3=`/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query3" 2>> $logfile`
114
115 # only proceed with file
116 # if information is not yet in database
117 # and no update is wished ($doupdate)
118 if [ "$result3" != "" ] && [ "$doupdate" == "no" ]
119 then
120 echo $file" has been inserted already => continue " >> $logfile 2>&1
121 continue
122 fi
123
124 # check if entry has already checksums
125 query5="SELECT fRunNumber FROM RunInfo WHERE Concat(fRunNumber, '_', LPAD(fFileNumber, 3, 0))='"$numberfromname"'"
126 query5=$query5" AND NOT ISNULL(fMd5sumRaw) AND NOT ISNULL(fMd5sumRawZip)"
127 echo "Q5:"$query5 >> $logfile 2>&1
128 result5=`/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query5" 2>> $logfile`
129
130 # get md5sums of raw and zip file
131 # to safe time for tests and update this can be skipped ($skipmd5sum)
132 md5sum=
133 md5sumzip=
134 if [ "$skipmd5sum" == "no" ] || [ "$skipmd5sum" == "iffilled" ]
135 then
136 #echo "skip: "$skipmd5sum >> $logfile 2>&1
137 #echo "res5: -"$result5"-" >> $logfile 2>&1
138 #if [ "$skipmd5sum" == "no" ] || [ "$result5$skipmd5sum" == "iffilled" ]
139 if [ "$skipmd5sum" == "no" ] || [ "$result5" == "" ]
140 then
141 zipfile=`echo $file | sed -e 's/raw/zipraw/' -e 's/fits/fits.gz/'`
142 # only do the md5sum if the zipfile is already available
143 if ls $zipfile >/dev/null >> $logfile 2>&1
144 then
145 echo "calculation md5sum for raw..." >> $logfile 2>&1
146 md5sum=`md5sum $file | cut -d' ' -f1`
147 echo "calculation md5sum for zipraw..." >> $logfile 2>&1
148 md5sumzip=`md5sum $zipfile | cut -d' ' -f1`
149 fi
150 fi
151 fi
152 if [ "$result3" == "" ] && [ "$md5sum" == "" ] && [ "$fillonly" == "ifzipped" ]
153 then
154 echo $file" is still missing zip => do not insert yet => continue " >> $logfile 2>&1
155 continue
156 fi
157
158 runtype=
159 # check if fits file is corrupted
160 numfitserrors=0
161 checkfitsfile=`fverify $file 2>> $logfile | grep '0 error(s)'`
162 if [ "$checkfitsfile" == "" ]
163 then
164 numfitserrors=1
165 fi
166
167 if [ $numfitserrors -eq 0 ]
168 then
169 # get run and file number from file
170 runnumberfromfile=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep NIGHT | grep -E -o '20[1-9][0-9][01][0-9][0-3][0-9]'`
171 filenumberfromfileorig=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep RUNID | grep -E -o '[0-9]{1,3}'`
172 numberfromfile=$runnumberfromfile"_"`printf %03d $filenumberfromfileorig`
173 # compare numbers
174 if [ "$numberfromfile" != "$numberfromname" ]
175 then
176 numbererror="yes"
177 echo "ERROR: for file "$file" number from filename and file don't agree ("$numberfromname" -"$numberfromfile")"
178 echo "ERROR: for file "$file" number from filename and file don't agree ("$numberfromname" -"$numberfromfile")" >> $logfile 2>&1
179 fi
180
181 # check if this run has drs file
182 drsfile=`echo $file | sed -e 's/fits/drs.fits/'`
183 numdrsfiles=`ls $drsfile 2>/dev/null | wc -l`
184
185 # get other variables from header
186 runtype=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep RUNTYPE | grep -E -o "['][a-z-]+[']" | sed -e "s/'//g"`
187 roi=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep NROI | grep -v NROITM | grep -E -o '[0-9]{1,4}'`
188 roitm=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep NROITM | grep -E -o '[0-9]{1,4}'`
189 numevents=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep Events | grep -E -o '[0-9]+'`
190 numphys=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRG ' | grep -E -o '[0-9]+'`
191 numext1=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGEXT1' | grep -E -o '[ ][0-9]+[ ]' | sed -e 's/\ //g'`
192 numext2=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGEXT2' | grep -E -o '[ ][0-9]+[ ]' | sed -e 's/\ //g'`
193 numelp=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGLPE' | grep -E -o '[0-9]+'`
194 numilp=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGLPI' | grep -E -o '[0-9]+'`
195 numoth=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGMISC' | grep -E -o '[0-9]+'`
196 numped=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGPED' | grep -E -o '[0-9]+'`
197 numtime=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep 'NTRGTIM' | grep -E -o '[0-9]+'`
198 # in newest data start time is in DATE-OBS
199 # in older data start time is in TSTART
200 # in the beginning TSTART was empty
201 runstart=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep DATE-OBS | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
202 runstart2=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep TSTART | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
203 if [ "$runstart" == "" ]
204 then
205 if [ "$runstart2" == "" ]
206 then
207 runstart=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep DATE | grep -v 'DATE-' | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
208 else
209 runstart=$runstart2
210 fi
211 fi
212 # in newest data start time is in DATE-END
213 # in older data start time is in TSTOP
214 # in the beginning TSTOP was empty
215 runstop=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep DATE-END | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
216 runstop2=`/home/fact/FACT++/fitsdump -h -t Events $file 2>/dev/null | grep TSTOP | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9]T[0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{6}'`
217 if [ "$runstop" == "" ]
218 then
219 if [ "$runstop2" == "" ]
220 then
221 runstop=`stat $file 2>/dev/null | grep Modify | grep -E -o '20[1-9][0-9][\-][01][0-9][\-][0-3][0-9][ ][0-2][0-9]:[0-6][0-9]:[0-6][0-9][.][0-9]{9}'`
222 else
223 runstop=$runstop2
224 fi
225 fi
226 fi
227
228 # set runtype to 'unknown' if no runtype could be queried
229 if [ "$runtype" == "" ]
230 then
231 runtype="n/a"
232 fi
233 # on 15.11.2011 the runtypes had different names
234 if [ "$date" == "2011/11/15" ]
235 then
236 if [ "$runtype" == "drs-calib" ]
237 then
238 runtype="drs-gain"
239 fi
240 if [ "$runtype" == "drs-time-calib" ]
241 then
242 runtype="drs-time"
243 fi
244 if [ "$runtype" == "pedestal" ]
245 then
246 runtype="drs-pedestal"
247 fi
248 if [ "$runtype" == "light-pulser" ]
249 then
250 runtype="light-pulser-ext"
251 fi
252 if [ "$runtype" == "pedestal-on" ]
253 then
254 runtype="pedestal"
255 fi
256 fi
257 # get runtype
258 query2="SELECT fRunTypeKEY FROM RunType WHERE fRunTypeName='"$runtype"'"
259 echo "Q2:"$query2 >> $logfile 2>&1
260 result2=( `/usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query2" 2>> $logfile` )
261 if [ ${#result2} -eq 0 ]
262 then
263 echo $numberfromname": Could not query fRunTypeKey for runtype "$runtype
264 echo $numberfromname": Could not query fRunTypeKey for runtype "$runtype >> $logfile 2>&1
265 exit
266 fi
267
268 # insert or update depending on whether run exists
269 if [ "$result3" == "" ]
270 then
271 query4="INSERT"
272 querymid=" fRunNumber="$runnumber", fFileNumber="$filenumberfromname", "
273 querystop=""
274 else
275 query4="UPDATE"
276 querystop=" WHERE fRunNumber="$runnumber" AND fFileNumber="$filenumberfromname
277 fi
278 query4=$query4" RunInfo SET "$querymid" fRunTypeKey="${result2[1]}
279 if [ $numfitserrors -eq 0 ]
280 then
281 query4=$query4", fNumEvents="$numevents", fROI="$roi", fROITimeMarker="$roitm
282 query4=$query4", fRunStart='"$runstart"', fRunStop='"$runstop"'"
283 if [ "$numphys" != "" ]
284 then
285 query4=$query4", fNumPhysicsTrigger="$numphys
286 fi
287 if [ "$numext1" != "" ]
288 then
289 query4=$query4", fNumExt1Trigger="$numext1
290 fi
291 if [ "$numext2" != "" ]
292 then
293 query4=$query4", fNumExt2Trigger="$numext2
294 fi
295 if [ "$numelp" != "" ]
296 then
297 query4=$query4", fNumELPTrigger="$numelp
298 fi
299 if [ "$numilp" != "" ]
300 then
301 query4=$query4", fNumILPTrigger="$numilp
302 fi
303 if [ "$numped" != "" ]
304 then
305 query4=$query4", fNumPedestalTrigger="$numped
306 fi
307 if [ "$numtime" != "" ]
308 then
309 query4=$query4", fNumTimeTrigger="$numtime
310 fi
311 if [ "$numoth" != "" ]
312 then
313 query4=$query4", fNumOtherTrigger="$numoth
314 fi
315 fi
316 if [ "$md5sum" != "" ]
317 then
318 query4=$query4", fMd5sumRaw='"$md5sum"', fMd5sumRawZip='"$md5sumzip"'"
319 query4=$query4", fHasDrsFile="$numdrsfiles
320 fi
321 query4=$query4", fFitsFileErrors="$numfitserrors
322 query4=$query4" "$querystop
323 echo "Q4:"$query4 >> $logfile 2>&1
324 if ! /usr/bin/mysql -u operator --host=fact01.fact.local --password=$password FactData -e "$query4" >> $logfile 2>&1
325 then
326 echo "insert/update of "$numfromfile" to mysql failed" >> $logfile 2>&1
327 fi
328
329 # print summary for debug
330 echo "*******" >> $logfile 2>&1
331 echo "summary for file "$file >> $logfile 2>&1
332 echo "*******" >> $logfile 2>&1
333 echo " errors: "$numfitserrors" (1 means errors exist)" >> $logfile 2>&1
334 echo " number of drs files: "$numdrsfiles >> $logfile 2>&1
335 echo " runnumber error: "$runnumbererror >> $logfile 2>&1
336 echo " number error: "$numbererror >> $logfile 2>&1
337 echo " roi: "$roi >> $logfile 2>&1
338 echo " roitm: "$roitm >> $logfile 2>&1
339 echo " runtype: "$runtype >> $logfile 2>&1
340 echo " numevents: "$numevents >> $logfile 2>&1
341 echo " md5sum: "$md5sum >> $logfile 2>&1
342 echo " md5sum(zip): "$md5sumzip >> $logfile 2>&1
343 echo " start: "$runstart >> $logfile 2>&1
344 echo " stop: "$runstop >> $logfile 2>&1
345
346 # missing
347 # check wether entry has been made (status table)
348 done
349done
350
351
Note: See TracBrowser for help on using the repository browser.