refactor: update input and dataset handling for all other scripts

JeffersonLab · Oct 5, 2023 · 12f6e06 · 12f6e06
1 parent 32ce6d0
commit 12f6e06
Show file tree

Hide file tree

Showing 4 changed files with 57 additions and 83 deletions.
diff --git a/bin/deploy-timelines.sh b/bin/deploy-timelines.sh
@@ -1,36 +1,37 @@
 #!/bin/bash
 # copy timelines to a target webserver directory
 
+set -e
+set -u
 source $(dirname $0)/environ.sh
 
 # timeline webserver directory
 TIMELINEDIR=/u/group/clas/www/clas12mon/html/hipo
 
+# input finding command
+inputCmd="$TIMELINESRC/bin/set-input-dir.sh -s timeline_web"
+inputCmdOpts=""
+
 usage() {
   echo """
-  USAGE: $0 [OPTIONS]... -d [DATASET] -t [TARGET_DIRECTORY]
-
-  =================================================================
-  CAUTION: READ ALL OF THIS BEFORE RUNNING !
-
-     If you really don't want to read, do a dry run using -D:
-       $0 -D [ARGS]...
-
-  WARNING: be careful not to overwrite anything you shouldn't
-  =================================================================
+  USAGE: $0 [OPTIONS]...
+
+  ========================================================================
+  | CAUTION: READ ALL OF THIS GUIDE BEFORE RUNNING!
+  |
+  |   If you want a quick start, try the most typical command:
+  |     $0 -d [DATASET] -t [TARGET_DIRECTORY] -D
+  |   This will only print what will be done; remove -D to actually run
+  |
+  | WARNING: be careful not to overwrite anything you shouldn't
+  ========================================================================
 
   REQUIRED OPTIONS: copy timelines for dataset [DATASET] to [TARGET_DIRECTORY]
-  -----------------
-
-  * Specify one or both of the following:
 
-    -d [DATASET_NAME]   unique dataset name, defined by the user
-                        default = based on [INPUT_DIR]
-
-    -i [INPUT_DIR]      directory containing run subdirectories of timeline histograms
-                        default = ./outfiles/[DATASET_NAME]/timeline_web
-
-  * Also, specify the target:
+  *** Specify at least one of the following:""" >&2
+  $inputCmd -h
+  echo """
+  *** Also, specify the target:
 
     -t [TARGET_DIRECTORY]   the top-level destination where the directory of
                             final timelines will be copied to
@@ -40,7 +41,6 @@ usage() {
         TIMELINEDIR = $TIMELINEDIR
     - use the -c option for a [TARGET_DIRECTORY] not relative to \$TIMELINEDIR
     - you must have write permission to the destination directory
-    - when in doubt, explore \$TIMELINEDIR
     - recommendations for [TARGET_DIRECTORY]:
       
       \$LOGNAME                  your personal directory for testing
@@ -71,24 +71,25 @@ dryRun=false
 customTarget=false
 subDir=""
 inputDir=""
-while getopts "d:t:Dcs:i:" opt; do
+while getopts "d:i:Ut:Dcs:" opt; do
   case $opt in
-    d) dataset=$OPTARG      ;;
+    d) inputCmdOpts+=" -d $OPTARG" ;;
+    i) inputCmdOpts+=" -i $OPTARG" ;;
+    U) inputCmdOpts+=" -U" ;;
     t) targetDirArg=$OPTARG ;;
-    D) dryRun=true          ;;
-    c) customTarget=true    ;;
-    s) subDir=$OPTARG       ;;
-    i) inputDir=$OPTARG     ;;
-    *) exit 100             ;;
+    D) dryRun=true ;;
+    c) customTarget=true ;;
+    s) subDir=$OPTARG ;;
+    *) exit 100 ;;
   esac
 done
 
-# check required options
-[ -z "$dataset" -o -z "$targetDirArg" ] && printError "missing required options -d and -t" && usage && exit 100
+# set input/output directories and dataset name
+dataset=$($inputCmd $inputCmdOpts -D)
+inputDir=$($inputCmd $inputCmdOpts -I)
 
-# specify input directory
-[ -z "$inputDir" ] && inputDir=$TIMELINESRC/outfiles/$dataset/timeline_web
-[ ! -d $inputDir ] && printError "input timelines directory $inputDir does not exist" && exit 100
+# check required options
+[ -z "$dataset" -o -z "$inputDir" -o -z "$targetDirArg" ] && printError "missing one or more required options" && usage && exit 100
 
 # specify target directory
 [ -z "$subDir" ] && subDir=$dataset

diff --git a/bin/run-detectors-timelines.sh b/bin/run-detectors-timelines.sh
@@ -99,7 +99,7 @@ if ${modes['list']}; then
   exit $?
 fi
 
-# set directories and dataset name
+# set input/output directories and dataset name
 dataset=$($inputCmd $inputCmdOpts -D)
 inputDir=$($inputCmd $inputCmdOpts -I)
 [ -z "$outputDir" ] && outputDir=$(pwd -P)/outfiles/$dataset

diff --git a/bin/run-physics-timelines.sh b/bin/run-physics-timelines.sh
@@ -9,6 +9,10 @@ inputDir=""
 dataset=""
 outputDir=""
 
+# input finding command
+inputCmd="$TIMELINESRC/bin/set-input-dir.sh -s timeline_physics"
+inputCmdOpts=""
+
 # usage
 sep="================================================================"
 if [ $# -eq 0 ]; then
@@ -18,14 +22,9 @@ if [ $# -eq 0 ]; then
   $sep
   Creates web-ready physics timelines locally
 
-  REQUIRED OPTIONS: specify one or both of the following:
-
-    -d [DATASET_NAME]   unique dataset name, defined by the user
-                        default = based on [INPUT_DIR]
-
-    -i [INPUT_DIR]      directory containing run subdirectories of timeline histograms
-                        default = ./outfiles/[DATASET_NAME]/timeline_physics
-
+  REQUIRED OPTIONS: specify at least one of the following:""" >&2
+  $inputCmd -h
+  echo """
   OPTIONAL OPTIONS:
 
     -o [OUTPUT_DIR]     output directory
@@ -36,49 +35,24 @@ if [ $# -eq 0 ]; then
 fi
 
 # parse options
-while getopts "i:d:o:" opt; do
+while getopts "d:i:Uo:" opt; do
   case $opt in
-    i) 
-      if [ -d $OPTARG ]; then
-        inputDir=$(realpath $OPTARG)
-      else
-        printError "input directory $OPTARG does not exist"
-        exit 100
-      fi
-      ;;
-    d) 
-      echo $OPTARG | grep -q "/" && printError "dataset name must not contain '/' " && exit 100
-      [ -z "$OPTARG" ] && printError "dataset name may not be empty" && exit 100
-      dataset=$OPTARG
-      ;;
-    o)
-      outputDir=$OPTARG
-      ;;
+    d) inputCmdOpts+=" -d $OPTARG" ;;
+    i) inputCmdOpts+=" -i $OPTARG" ;;
+    U) inputCmdOpts+=" -U" ;;
+    o) outputDir=$OPTARG ;;
   esac
 done
 
-# set directories and dataset name
-# FIXME: copied implementation from `run-detector-timelines.sh`
-if [ -z "$inputDir" -a -n "$dataset" ]; then
-  inputDir=$(pwd -P)/outfiles/$dataset/timeline_physics # default input directory is in ./outfiles/
-elif [ -n "$inputDir" -a -z "$dataset" ]; then
-  dataset=$(ruby -e "puts '$inputDir'.split('/')[-4..].join('_')") # set dataset using last few subdirectories in inputDir dirname
-elif [ -z "$inputDir" -a -z "$dataset" ]; then
-  printError "required options, either [INPUT_DIR] or [DATASET_NAME], have not been set"
-  exit 100
-fi
+# set input/output directories and dataset name
+dataset=$($inputCmd $inputCmdOpts -D)
+inputDir=$($inputCmd $inputCmdOpts -I)
 [ -z "$outputDir" ] && outputDir=$(pwd -P)/outfiles/$dataset
 
 # set subdirectories
 finalDir=$outputDir/timeline_web
 logDir=$outputDir/log
 
-# check input directory
-if [ ! -d $inputDir ]; then
-  printError "input directory $inputDir does not exist"
-  exit 100
-fi
-
 # print settings
 echo """
 Settings:

diff --git a/bin/set-input-dir.sh b/bin/set-input-dir.sh
@@ -7,8 +7,13 @@ set -u
 source $(dirname $0)/environ.sh
 
 # defaults
+inputDir=""
+subDir=""
+dataset=""
 datasetDefault="test_v0"
-subDirDefault=""
+useDefaults=false
+printDatasetOnly=false
+printInputDirOnly=false
 
 # usage guide
 usage() {
@@ -17,7 +22,7 @@ usage() {
                         default = $datasetDefault
 
     -i [INPUT_DIR]      directory containing run subdirectories of timeline histograms
-                        default = ./outfiles/[DATASET_NAME]/$subDirDefault
+                        default = ./outfiles/[DATASET_NAME]/$subDir
 
     -U                  use the above default [DATASET_NAME] and [INPUT_DIR]""" >&2
 }
@@ -39,12 +44,6 @@ if [ $# -eq 0 ]; then
 fi
 
 # parse options
-inputDir=""
-subDir=""
-dataset=""
-useDefaults=false
-printDatasetOnly=false
-printInputDirOnly=false
 while getopts "d:i:Us:hDI" opt; do
   case $opt in
     d)