[u/mrichter/AliRoot.git] / PWGLF / FORWARD / analysis2 / proofAnalysis.sh

#!/bin/bash
# 
# BEGIN_MANUAL
# 	Script to help do PWGLF-Forward analsysis using ProofLite
#       =========================================================
# First, one need to figure out what to analyse.  We assume we have
# the ESDs from a real run in some directory - possibly in
# sub-directories, and similar for the MC data. 
# 
# Then, one needs to run this script in set-up mode e.g., 
# 
#   $0 --what=setup  \
#       --name=LHC10c \
#       --run=118560 \
#       --real-dir=/data/alice/data/pp/lhc10c/000118560/pass3 \
#       --real-pattern=AliESDs_*.root \
#       --mc-dir=/data/alice/data/pp/lhc10c/sim/lhc13d4/118560 \
#       --mc-pattern=root_archive.zip@AliESDs.root 
# 
# Note, all the settings are written to the file .config in the
# current directory, so you do not need to give the parameters at
# subsequent steps.  Note, you need a valid AliEn token to at this
# point to get the acceptance corrections.  The run number specified
# is only used for getting the acceptance correction.
# 
# Note, the use of the ZIP archives root_archive.zip and the sub-part
# specification @AliESDs.root for MC data. 
# 
# Next, we need to generate the corrections.  Do 
# 
#   $0 --what=corr 
# 
# and wait for the jobs to finish and terminate. Next, we need to
# extract and upload the corrections to our local corrections folder
# 
#   $0 --what=corr --step=upload 
#   $0 --what=corr --step=draw
# 
# If you already have the corrections, you can pass the option
# --corrections in the setup phase and skip this step.
# 
# Now we can submit our AOD generation jobs.  Do 
# 
#   $0 --what=aod 
# 
# and wait for the jobs to finish and terminate.  If you need to pass
# additional options to the train, one can do so after the special
# option -- e.g., to limit the number of events to 100000, do
# 
#   $0 --what=aod -- --events=100000
# 
# Next, we need to draw the summary results
# 
#   $0 --what=aod --step=draw 
# 
# Now, we should do the dN/deta analysis.  Do 
# 
#   $0 --what=dndeta
# 
# and wait for the jobs to finish and terminate.  Again, additional
# options to the train can be passed after --.  If you passed the
# option --sys=1 in the setup phase, then this will run 3 jobs for
# real and MC each - one for INEL, INEL>0, and NSD (V0-AND).  Next, we
# need to draw the summary and final plot
# 
#   $0 --what=dndeta --step=draw 
# 
# To generate the P(Nch) data, do 
# 
#   $0 --what=multdists
#   $0 --what=multdists --step=draw 
# 
# To collect all PDFs into a single directory do 
# 
#   $0 --what=collect
# 
# Enjoy.
# 
# Comments, questions, bugs, flames, suggestions, etc. should be sent
# to Christian Holm Christensen <cholm@nbi.dk>
# 
# END_MANUAL
if test ! -f $ALICE_ROOT/PWGLF/FORWARD/analysis2/baseAnalysis.sh  ; then 
    echo "baseAnalysis not found!" > /dev/stderr 
    exit 1
fi
. $ALICE_ROOT/PWGLF/FORWARD/analysis2/baseAnalysis.sh 

run=
serv=
real_ds=
mc_ds=
my_real_ds=
my_mc_ds=
nwrkrs=0
par=0
batch=0
# === Various functions ==============================================


# === Implement base functions =======================================
# --- Usage ----------------------------------------------------------
setup_usage()
{
    cat <<EOF
  -r,--run=NUMBER           Specify run number ($run)
  -d,--real-ds=NAME         Data set of real data ($real_ds)
  -D,--mc-ds=NAME           Data set of MC data ($mc_dir)
  -P,--server=ADDRESS	    Proof server ($serv)
  -W,--workers=N            Number of workers ($nwrks)
  -a,--par                  Use par files ($par)
  -b,--batch                Do not show GUI
EOF
}
# --- handle setup options -------------------------------------------
handle_setup_option()
{
    local arg="$1" 
    local opt="$2"
    # echo "Handle arg=$arg opt=$opt"
    case $arg in 
	-r|--run)          run=$opt      ;; 
	-d|--real-ds)      real_ds=$opt  ;;
	-D|--mc-ds)        mc_ds=$opt    ;; 
	-W|--workers)      nwrks=${opt}  ;;
	-P|--server)       serv=${opt}   ;;
	-a|--par)          par=1         ;;
	*) echo "$0: [SETUP] Unknown option $arg"  ; exit 1 ;; 
    esac
}

# --- Get the grid home dir ------------------------------------------
outputs()
{
    # We should get the PROOF group and user name here
    # l=`id -n -g`/`id -n -u`
    l=`aliroot -l -b <<EOF
gSystem->RedirectOutput("/dev/null");
TProof::Reset("$serv", false);
TProof::Open("$serv");
gSystem->RedirectOutput(0);
Printf("%s/%s", gProof->GetGroup(), gProof->GetUser());
EOF`
    my_real_ds="/$l/${name}_aod_${now}"
    my_mc_ds="/$l/${name}_mcaod_${now}"
}
# --- Dump the setup -------------------------------------------------
dump_setup()
{
    local out=$1 
    cat >> ${out} <<-EOF 
	# Run analysed 
	run=${run}	
	# Real data 
	real_ds=${real_ds}
	mc_ds=${mc_ds}
	# Output 
	my_real_ds=${my_real_ds}
	my_mc_ds=${my_mc_ds}
	serv=${serv}
	nwrks=${nwrks}
	par=${par}
	batch=${batch}
	EOF
}

# ---- Get run number to use for Acceptance map ----------------------
run_for_acc()
{
    if test x$run = "x" || test $run -lt 1; then 
	echo "No run for acceptance correction specified" > /dev/stderr 
	exit 1
    fi
    echo $run
}
# --- Run set-ups ----------------------------------------------------
setup()
{
    echo "Calling _setup with $@"
    _setup $@ 
}    

# --- Check settings -------------------------------------------------
check_setup()
{
    if test "x$run" = "x" || test $run -lt 1 ; then 
	echo "Run not specified, or invalid ($run)" > /dev/stderr 
	exit 1
    fi
    if test "x$real_ds" = "x" ; then 
	echo "No real data set specified" > /dev/stderr 
	exit 1
    fi
    if test "x$mc_ds" = "x" ; then 
	echo "No MC data set specified" > /dev/stderr 
	# exit 1
    fi
}


# --- Show the setup -------------------------------------------------
print_setup()
{
    cat <<-EOF
	Run:			${run}
	Real data set:
	  Data set:		${real_ds}
	  Output:		${my_real_ds}
	MC data:
	  Data set:		${mc_ds}
	  Output:		${my_mc_ds}
	Use PAR files:		${par}
	Number of workers:      ${nwrks}
	Server:                 ${serv}
	EOF
}

# --- Make URI -------------------------------------------------------
url_opts()
{
    local mc=$1 	; shift
    local type=$1 	; shift 
    local trig=$1 	; shift
    local uopt="mode=default"
    local tree=esdTree 

    local inp=${real_ds}
    local oup=`basename ${my_real_ds}`
    if test $mc -gt 0 ; then 
	inp=${mc_ds}
	oup=`basename ${my_mc_ds}`
    fi

    case $type in 
	*aod)
	    uopt="${uopt}&dsname=${oup}"
	    ;; 
	*dndeta|*multdists)
	    tree=aodTree 
	    # Modify for input dir for our files
	    inp=$my_real_ds
	    if test $mc -gt 0 ; then 
		inp=$my_mc_ds
	    fi
	    ;;
    esac
    if test x$inp = x; then 
	echo "No input for $nme, giving up" > /dev/stderr 
	return 
    fi
    if test $mc -gt 0; then 
	uopt="${uopt}&mc"
    fi
    if test $par -gt 0 ; then 
	uopt="${uopt}&par=tasks"
    fi
    if test x$uuopts != x ; then 
	uopt="${uopt}&${uuopts}"
    fi
    if test "x$nwrks" != "x" && test $nwrks -gt 0 ; then 
	uopt="${uopt}&workers=${nwrks}"
    fi
    case $server in 
	alice-caf.cern.ch) uopts="${uopts}&reset=hard" ;; 
    esac 
	    
    # PROOF-lite URL form:
    # 
    #  lite://<datadir_or_list>[?<options>][#<treeName]
    # 
    # Options:
    #  clear=PKGS                 Clear packages ','-separated
    #  mc                         Assume simulation input
    #  mode=default|rec|sim       AliROOT mode 
    #  par=tasks|all              Use par files 
    #  pattern=GLOB               File name pattern 
    #  recursive                  Recursive scan [true]
    #  reset=soft|hard            Reset cluster [hard]
    #  workers=N[x]               Number of workers to use [8]
    #  wrapper=CMD                Wrapper command []

    url="proof://${serv}/${inp}?${uopt}#${tree}"

}

# --- Run the train --------------------------------------------------
# Usage:
# 
allAboard()
{
    local type=$1 ; shift
    local trig=$1 ; shift
    local lopts=
    if test $batch -gt 0 ; then 
	lopts="--batch"
    fi
    _allAboard "$type" "$trig" $lopts $@ 
}

# === Procedual code =================================================
runIt $@

#
# EOF
#
Commit	Line	Data
b7a6aa91	1	#!/bin/bash
	2	#
	3	# BEGIN_MANUAL
	4	# Script to help do PWGLF-Forward analsysis using ProofLite
	5	# =========================================================
	6	# First, one need to figure out what to analyse. We assume we have
	7	# the ESDs from a real run in some directory - possibly in
	8	# sub-directories, and similar for the MC data.
	9	#
	10	# Then, one needs to run this script in set-up mode e.g.,
	11	#
	12	# $0 --what=setup \
	13	# --name=LHC10c \
	14	# --run=118560 \
	15	# --real-dir=/data/alice/data/pp/lhc10c/000118560/pass3 \
	16	# --real-pattern=AliESDs_*.root \
	17	# --mc-dir=/data/alice/data/pp/lhc10c/sim/lhc13d4/118560 \
	18	# --mc-pattern=root_archive.zip@AliESDs.root
	19	#
	20	# Note, all the settings are written to the file .config in the
	21	# current directory, so you do not need to give the parameters at
	22	# subsequent steps. Note, you need a valid AliEn token to at this
	23	# point to get the acceptance corrections. The run number specified
	24	# is only used for getting the acceptance correction.
	25	#
	26	# Note, the use of the ZIP archives root_archive.zip and the sub-part
	27	# specification @AliESDs.root for MC data.
	28	#
	29	# Next, we need to generate the corrections. Do
	30	#
	31	# $0 --what=corr
	32	#
	33	# and wait for the jobs to finish and terminate. Next, we need to
	34	# extract and upload the corrections to our local corrections folder
	35	#
	36	# $0 --what=corr --step=upload
	37	# $0 --what=corr --step=draw
	38	#
	39	# If you already have the corrections, you can pass the option
	40	# --corrections in the setup phase and skip this step.
	41	#
	42	# Now we can submit our AOD generation jobs. Do
	43	#
	44	# $0 --what=aod
	45	#
	46	# and wait for the jobs to finish and terminate. If you need to pass
	47	# additional options to the train, one can do so after the special
	48	# option -- e.g., to limit the number of events to 100000, do
	49	#
	50	# $0 --what=aod -- --events=100000
	51	#
	52	# Next, we need to draw the summary results
	53	#
	54	# $0 --what=aod --step=draw
	55	#
	56	# Now, we should do the dN/deta analysis. Do
	57	#
	58	# $0 --what=dndeta
	59	#
	60	# and wait for the jobs to finish and terminate. Again, additional
	61	# options to the train can be passed after --. If you passed the
	62	# option --sys=1 in the setup phase, then this will run 3 jobs for
	63	# real and MC each - one for INEL, INEL>0, and NSD (V0-AND). Next, we
	64	# need to draw the summary and final plot
65	#
66	# $0 --what=dndeta --step=draw
67	#
68	# To generate the P(Nch) data, do
69	#
70	# $0 --what=multdists
71	# $0 --what=multdists --step=draw
72	#
73	# To collect all PDFs into a single directory do
74	#
75	# $0 --what=collect
76	#
77	# Enjoy.
78	#
79	# Comments, questions, bugs, flames, suggestions, etc. should be sent
80	# to Christian Holm Christensen <cholm@nbi.dk>
81	#
82	# END_MANUAL
83	if test ! -f $ALICE_ROOT/PWGLF/FORWARD/analysis2/baseAnalysis.sh ; then
84	echo "baseAnalysis not found!" > /dev/stderr
85	exit 1
86	fi
87	. $ALICE_ROOT/PWGLF/FORWARD/analysis2/baseAnalysis.sh
88
89	run=
90	serv=
91	real_ds=
92	mc_ds=
93	my_real_ds=
94	my_mc_ds=
95	nwrkrs=0
96	par=0
97	batch=0
98	# === Various functions ==============================================
99
100
101	# === Implement base functions =======================================
102	# --- Usage ----------------------------------------------------------
103	setup_usage()
104	{
105	cat <<EOF
106	-r,--run=NUMBER Specify run number ($run)
107	-d,--real-ds=NAME Data set of real data ($real_ds)
108	-D,--mc-ds=NAME Data set of MC data ($mc_dir)
109	-P,--server=ADDRESS Proof server ($serv)
110	-W,--workers=N Number of workers ($nwrks)
111	-a,--par Use par files ($par)
112	-b,--batch Do not show GUI
113	EOF
114	}
115	# --- handle setup options -------------------------------------------
116	handle_setup_option()
117	{
118	local arg="$1"
119	local opt="$2"
120	# echo "Handle arg=$arg opt=$opt"
121	case $arg in
122	-r\|--run) run=$opt ;;
123	-d\|--real-ds) real_ds=$opt ;;
124	-D\|--mc-ds) mc_ds=$opt ;;
125	-W\|--workers) nwrks=${opt} ;;
126	-P\|--server) serv=${opt} ;;
127	-a\|--par) par=1 ;;
128	*) echo "$0: [SETUP] Unknown option $arg" ; exit 1 ;;
129	esac
130	}
131
132	# --- Get the grid home dir ------------------------------------------
133	outputs()
134	{
135	# We should get the PROOF group and user name here
136	# l=`id -n -g`/`id -n -u`
137	l=`aliroot -l -b <<EOF
138	gSystem->RedirectOutput("/dev/null");
139	TProof::Reset("$serv", false);
140	TProof::Open("$serv");
141	gSystem->RedirectOutput(0);
142	Printf("%s/%s", gProof->GetGroup(), gProof->GetUser());
143	EOF`
144	my_real_ds="/$l/${name}_aod_${now}"
145	my_mc_ds="/$l/${name}_mcaod_${now}"
146	}
147	# --- Dump the setup -------------------------------------------------
148	dump_setup()
149	{
150	local out=$1
151	cat >> ${out} <<-EOF
152	# Run analysed
153	run=${run}
154	# Real data
155	real_ds=${real_ds}
156	mc_ds=${mc_ds}
157	# Output
158	my_real_ds=${my_real_ds}
159	my_mc_ds=${my_mc_ds}
160	serv=${serv}
161	nwrks=${nwrks}
162	par=${par}
163	batch=${batch}
164	EOF
165	}
166
167	# ---- Get run number to use for Acceptance map ----------------------
168	run_for_acc()
169	{
170	if test x$run = "x" \|\| test $run -lt 1; then
171	echo "No run for acceptance correction specified" > /dev/stderr
172	exit 1
173	fi
174	echo $run
175	}
176	# --- Run set-ups ----------------------------------------------------
177	setup()
178	{
179	echo "Calling _setup with $@"
180	_setup $@
181	}
182
183	# --- Check settings -------------------------------------------------
184	check_setup()
185	{
186	if test "x$run" = "x" \|\| test $run -lt 1 ; then
187	echo "Run not specified, or invalid ($run)" > /dev/stderr
188	exit 1
189	fi
190	if test "x$real_ds" = "x" ; then
191	echo "No real data set specified" > /dev/stderr
192	exit 1
193	fi
194	if test "x$mc_ds" = "x" ; then
195	echo "No MC data set specified" > /dev/stderr
196	# exit 1
197	fi
198	}
199
200
201
202	# --- Show the setup -------------------------------------------------
203	print_setup()
204	{
205	cat <<-EOF
206	Run: ${run}
207	Real data set:
208	Data set: ${real_ds}
209	Output: ${my_real_ds}
210	MC data:
211	Data set: ${mc_ds}
212	Output: ${my_mc_ds}
213	Use PAR files: ${par}
214	Number of workers: ${nwrks}
215	Server: ${serv}
216	EOF
217	}
218
219	# --- Make URI -------------------------------------------------------
220	url_opts()
221	{
222	local mc=$1 ; shift
223	local type=$1 ; shift
224	local trig=$1 ; shift
225	local uopt="mode=default"
226	local tree=esdTree
227
228	local inp=${real_ds}
229	local oup=`basename ${my_real_ds}`
230	if test $mc -gt 0 ; then
231	inp=${mc_ds}
232	oup=`basename ${my_mc_ds}`
233	fi
234
235	case $type in
236	*aod)
237	uopt="${uopt}&dsname=${oup}"
238	;;
239	dndeta\|multdists)
240	tree=aodTree
241	# Modify for input dir for our files
242	inp=$my_real_ds
243	if test $mc -gt 0 ; then
244	inp=$my_mc_ds
245	fi
246	;;
247	esac
248	if test x$inp = x; then
249	echo "No input for $nme, giving up" > /dev/stderr
250	return
251	fi
252	if test $mc -gt 0; then
253	uopt="${uopt}&mc"
254	fi
255	if test $par -gt 0 ; then
256	uopt="${uopt}&par=tasks"
257	fi
258	if test x$uuopts != x ; then
259	uopt="${uopt}&${uuopts}"
260	fi
261	if test "x$nwrks" != "x" && test $nwrks -gt 0 ; then
262	uopt="${uopt}&workers=${nwrks}"
263	fi
264	case $server in
265	alice-caf.cern.ch) uopts="${uopts}&reset=hard" ;;
266	esac
267
268	# PROOF-lite URL form:
269	#
270	# lite://<datadir_or_list>[?<options>][#<treeName]
271	#
272	# Options:
273	# clear=PKGS Clear packages ','-separated
274	# mc Assume simulation input
275	# mode=default\|rec\|sim AliROOT mode
276	# par=tasks\|all Use par files
277	# pattern=GLOB File name pattern
278	# recursive Recursive scan [true]
279	# reset=soft\|hard Reset cluster [hard]
280	# workers=N[x] Number of workers to use [8]
281	# wrapper=CMD Wrapper command []
282
283	url="proof://${serv}/${inp}?${uopt}#${tree}"
284
285	}
286
287	# --- Run the train --------------------------------------------------
288	# Usage:
289	#
290	allAboard()
291	{
292	local type=$1 ; shift
293	local trig=$1 ; shift
294	local lopts=
295	if test $batch -gt 0 ; then
296	lopts="--batch"
297	fi
298	_allAboard "$type" "$trig" $lopts $@
299	}
300
301	# === Procedual code =================================================
302	runIt $@
303
304	#
305	# EOF
306	#
307