rename(unitproc/bkt-replace_apostrophes):Was replace_contractions
[BK-2020-03.git] / unitproc / bklu
CommitLineData
0d22dfcd
SBS
1#!/bin/bash
2
3# Date: 2020-05-04T18:50Z
4# Author: Steven Baltakatei Sandoval
5# Description: Script to lookup file hash from existing hash list.
6# Note: Use hide-show-block function to aid readability. (ex: https://www.emacswiki.org/emacs/HideShow ).
7
8#== Variable Initialization ==
9
10#== Global constants ==
11SCRIPT_TTL=10 # Limit script life to this in seconds.
12PATH="/usr/local/bin/:$PATH" # Add user binary executable directory to PATH.
13PATH="/opt/bktei:$PATH" # Add 'optional' executable directory to PATH.
14SCRIPT_HOSTNAME=$(hostname) # Save hostname of system running this script.
15SCRIPT_VERSION="bktemplate.sh 0.0.0" # Define version of script. Used by function 'showVersion'.
16SCRIPT_TIME_SHORT="$(date +%Y%m%dT%H%M%S%z)" # Save current date & time in ISO-8601 format (YYYYmmddTHHMMSS+zzzz).
17SCRIPT_DATE_SHORT="$(date +%Y%m%d)" # Save current date in ISO-8601 format.
18
19#== Function Definitions ==
20echoerr() {
21 # Usage: echo [ arguments ]
22 # Description: Prints provided arguments to stderr.
23 # Input: unspecified
24 # Output: stderr
25 # Script function dependencies: none
26 # External function dependencies: echo
27 # Last modified: 2020-04-11T21:16Z
28 # Last modified by: Steven Baltakatei Sandoval
29 # License: GPLv3+
30 # Ref./Attrib:
31 # [1]: # Roth, James (2010-06-07). ["How to print text to stderr instead of stdout"](https://stackoverflow.com/a/2990533). Licensed CC BY-SA 4.0.
32
33 echo "$@" 1>&2; # Define stderr echo function. See [1].
34 return 0; # Function finished.
35} # Define stderr message function.
36vbm() {
37 # Usage: vbm "DEBUG:verbose message here"
38 # Description: Prints verbose message ("vbm") to stderr if OPTION_VERBOSE is set to "true".
39 # Input:
40 # - OPTION_VERBOSE variable set by processArguments function. (ex: "true", "false")
41 # - "$@" positional arguments fed to this function.
42 # Output: stderr
43 # Script function dependencies: echoerr
44 # External function dependencies: echo
45 # Last modified: 2020-04-11T23:57Z
46 # Last modified by: Steven Baltakatei Sandoval
47 # License: GPLv3+
48 # Ref./Attrib:
49
50 if [ "$OPTION_VERBOSE" == "true" ]; then
51 FUNCTION_TIME=$(date --iso-8601=ns); # Save current time in nano seconds.
52 echoerr "[$FUNCTION_TIME] ""$@"; # Display argument text.
53 fi
54
55 # End function
56 return 0; # Function finished.
57} # Verbose message display function.
58showUsage() {
59 # Usage: showUsage
60 # Description: Displays script usage information.
61 # Input: none
62 # Output: stderr
63 # Script function dependencies: echoerr
64 # External dependencies: bash (5.0.3), echo
65 # Last modified: 2020-05-04T16:58Z
66 # Last modified by: Steven Baltakatei Sandoval
67 # License: GPLv3+
68 # Ref./Attrib.:
69
70 echoerr "USAGE:"
71 echoerr " bklu [OPTIONS] [FILE]"
72 echoerr
73 echoerr " [FILE] hash is compared against file contents of directory"
74 echoerr " specified by --input-dir option or environment variable"
75 echoerr " BK_FILEHASHLIST_DIR ."
76 echoerr
77 echoerr "OPTIONS:"
78 echoerr " --digest-algo [ digest name ]"
79 echoerr " Specify digest command name. Value should be GNU"
80 echoerr " coreutils commands such as: sha256sum, md5sum, b2sum."
81 echoerr " Default value is: b2sum"
82 echoerr
83 echoerr " -h, --help"
84 echoerr " Display help information."
85 echoerr
86 echoerr " --version"
87 echoerr " Display script version."
88 echoerr
89 echoerr " -v, --verbose"
90 echoerr " Display debugging info."
91 echoerr
92 echoerr " -o, --output-file [ file ]"
93 echoerr " Specify output file."
94 echoerr
95 echoerr " -i, --input-file [ file ]"
96 echoerr " Specify input file."
97 echoerr
98 echoerr " -o, --output-dir [ directory ]"
99 echoerr " Specify output directory."
100 echoerr
101 echoerr " -i, --input-dir [ directory ]"
102 echoerr " Specify input directory."
103 echoerr
104
105
106 # End function
107 return 0; # Function finished.
108} # Display information on how to use this script.
109showVersion() {
110 # Usage: showVersion
111 # Descriptoin: Displays script version and license information.
112 # Input: unspecified
113 # Output: stderr
114 # Script function dependencies: echoerr
115 # External function dependencies: echo
116 # Last modified: 2020-04-11T23:57Z
117 # Last modified by: Steven Baltakatei Sandoval
118 # License: GPLv3+
119 # Ref./Attrib:
120
121 # Initialize function
122 vbm "DEBUG:showVersion function called."
123
124 # Perform work
125 OUTPUT="$SCRIPT_VERSION"
126
127 # Display results
128 echoerr "$OUTPUT";
129
130 # End function
131 vbm "DEBUG:showVersion function ended."
132 return 0; # Function finished.
133} # Display script version.
134processArguments() {
135 # Usage: processArguments "$@"
136 # Description: Processes provided arguments in order to set script option variables useful for
137 # changing how other functions behave. For example, it may:
138 # 1. Activate verbose mode
139 # Input: "$@" (list of arguments provided to the function)
140 # Output: Sets following variables used by other functions:
141 # OPTION_VERBOSE Indicates verbose mode enable status. (ex: "true", "false")
142 # DIROUT1 Path to output directory.
143 # FILEOUT1 Path to output file.
144 # DIRIN1 Path to input directory.
145 # FILEIN1 Path to input file.
146 # OPTION_FILEOUT1_OVERWRITE Indicates whether file FILEOUT1 should be overwritten (ex: "true", "false')
147 # DIGEST_ALGO Name of digest command (unverified).
148 # Script function dependencies:
149 # - echoerr Displays messages to stderr.
150 # - vbm Displays messsages to stderr if OPTION_VERBOSE set to "true".
151 # External dependencies: bash (5.0.3), echo
152 # Last modified: 2020-05-04T18:13Z
153 # Last modified by: Steven Baltakatei Sandoval
154 # License: GPLv3+
155 # Ref./Attrib.:
156 # [1]: Marco Aurelio (2014-05-08). "echo that outputs to stderr". https://stackoverflow.com/a/23550347
157
158 # Initialize function
159 #vbm "DEBUG:processArguments function called."
160
161 # Perform work
162 while [ ! $# -eq 0 ]; do # While number of arguments ($#) is not (!) equal to (-eq) zero (0).
163 #1>&2 echo "DEBUG:Starting processArguments while loop." # Debug stderr message. See [1].
164 #1>&2 echo "DEBUG:Provided arguments are:""$@" # Debug stderr message. See [1].
165 case "$1" in
166 -h | --help) showUsage; exit 1;; # Display usage.
167 --version) showVersion; exit 1;; # Show version
168 -v | --verbose) OPTION_VERBOSE="true"; vbm "DEBUG:Verbose mode enabled.";; # Enable verbose mode. See [1].
169 -i | --input-file) # Define input file path
170 if [ -f "$2" ]; then # If $2 is file that exists, set FILEIN1 to $2, pop $2.
171 FILEIN1="$2";
172 shift;
173 vbm "DEBUG:Input file FILEIN1 set to:""$2";
174 else
175 echoerr "ERROR: Specified input file does not exist:""$2";
176 echoerr "Exiting.";
177 exit 1;
178 fi ;;
179 -I | --input-dir) # Define input directory path.
180 if [ -d "$2" ]; then # If $2 is dir that exists, set DIRIN1 to $2, pop $2.
181 DIRIN1="$2";
182 shift;
183 vbm "DEBUG:Input directory DIRIN1 set to:""$2";
184 else # Display error if $2 is not a valid dir.
185 echoerr "ERROR:Specified input directory does not exist:""$2";
186 echoerr "Exiting.";
187 exit 1;
188 fi ;;
189 -o | --output-file) # Define output file path
190 if [ -f "$2" ]; then # If $2 is file that exists, prompt user to continue to overwrite, set FILEOUT1 to $2, pop $2.
191 echoerr "Specified output file $2 already exists. Overwrite? (y/n):"
192 read m; case $m in
193 y | Y | yes) OPTION_FILEOUT1_OVERWRITE="true";;
194 n | N | no) OPTION_FILEOUT1_OVERWRITE="false";;
195 *) echoerr "Invalid selection. Exiting."; exit 1;;
196 esac
197 if [ "$OPTION_FILEOUT1_OVERWRITE" == "true" ]; then
198 FILEOUT1="$2";
199 shift;
200 vbm "DEBUG:Output file FILEOUT1 set to:""$2";
201 else
202 echoerr "ERORR:Exiting in order to not overwrite output file:""$FILEOUT1";
203 exit 1;
204 fi
205 else
206 FILEOUT1="$2";
207 shift;
208 vbm "DEBUG:Output file FILEOUT1 set to:""$2";
209 fi ;;
210 -O | --output-dir) # Define output directory path
211 if [ -d "$2" ]; then # If $2 is dir that exists, set DIROUT1 to $2, pop $2
212 DIROUT1="$2";
213 shift;
214 vbm "DEBUG:Output directory DIROUT1 set to:""$2";
215 else
216 echoerr "ERROR:Specified output directory is not valid:""$2";
217 echoerr "Exiting.";
218 exit 1;
219 fi ;;
220 --digest-algo) # Define file digest algorithm command name.
221 if ! [ -z "$2" ]; then # If $2 is non-zero, set DIGEST_ALGO to $2, pop $2
222 DIGEST_ALGO="$2";
223 shift;
224 vbm "DEBUG:DIGEST_ALGO set to:""$DIGEST_ALGO";
225 else
226 echoerr "ERROR:Cannot set DIGEST_ALGO because no string specified."; echoerr "Exiting."; exit 1;
227 fi ;;
228 *) if [ -f "$1" ] && ! [ -v FILEIN1 ]; then # If $1 is file and FILEIN1 not set, then set FILEIN1 to $1, pop $1.
229 FILEIN1="$1";
230 shift;
231 vbm "DEBUG:FILEIN1 set to:""$FILEIN1";
232 else # Handle non-option (ex: [FILE] )
233 echoerr "ERROR: Unrecognized non-option argument."; echoerr "Exiting."; exit 1;
234 fi ;;
235 esac
236 shift;
237 done
238
239 # End function
240 vbm "DEBUG:processArguments function ended."
241 return 0; # Function finished.
242} # Evaluate script options from positional arguments (ex: $1, $2, $3, etc.).
243checkExecutables() {
244 # Usage: checkExecutables [ command1 ] [ command2 ] [...] [ commandN ]
245 # Description: Checks that provided commands exist. If one or more do not exist, then display missing commands and exit with error.
246 # Input:
247 # - command names (arguments)
248 # Output: commands that don't exist (stderr)
249 # Script function dependencies:
250 # - echoerr for displaying errors via stderr
251 # - processArguments for setting OPTION_VERBOSE
252 # - vbm for displaying verbose messages if OPTION_VERBOSE is "true"
253 # External dependencies: bash (5.0.3), command
254 # Last modified: 2020-05-04T17:12Z
255 # Last modified by: Steven Baltakatei Sandoval
256 # License: GPLv3+
257 # Ref./Attrib.:
258 # [1]: SiegeX (2010-12-12). ["Difference between return and exit in Bash functions."](https://stackoverflow.com/a/4419971). Licensed CC BY-SA 4.0.
259 # [2]: Darryl Hein (2009-12-23). ["Add a new element to an array without specifying the index in Bash"](https://stackoverflow.com/a/1951523). Licensed CC BY-SA 4.0.
260 # [3]: kojiro (2012-10-03). ["Convert command line arguments into an array in Bash"](https://stackoverflow.com/a/12711853)
261 # [4]: niieani (2016-01-12). ["How to copy an array in Bash?"](https://stackoverflow.com/a/34733375/10850071). Licensed CC BY-SA 4.0.
262
263 # Initialize function
264 vbm "DEBUG:checkExecutables function called."
265 declare -a candidateCommandsNames # Initialize array for storing positional arguments provided to this function.
266 candidateCommandsNames=("$@") # Save positional arguments to variable as string. See [3].
267 vbm "DEBUG:candidateCommandsNames:""$@"
268 vbm "DEBUG:candidateCommandsNames[0]:""${candidateCommandsNames[0]}"
269 vbm "DEBUG:candidateCommandsNames[1]:""${candidateCommandsNames[1]}"
270 vbm "DEBUG:candidateCommandsNames[2]:""${candidateCommandsNames[2]}"
271 declare -a outputInvalidCommandsArray # Initialize arary for storing names of invalid commands.
272 declare -a outputValidCommandsArray # Initialize array for storing names of valid commands.
273
274 # Perform work
275 for candidateCommandName in "${candidateCommandsNames[@]}"; do # Search through all space-delimited text for valid commands.
276 if command -v "$candidateCommandName" 1>/dev/null 2>/dev/null; then # Check if a command is valid or not.
277 outputValidCommandsArray+=("$candidateCommandName") ; # See [2].
278 vbm "DEBUG:Adding $candidateCommandName to outputValidCommandsArray."
279 else
280 outputInvalidCommandsArray+=("$candidateCommandName") ; # See [2].
281 vbm "DEBUG:Adding $candidateCommandName to outputInvalidCommandsArray."
282 fi
283 done
284
285 # Output results
286 if [ ${#outputInvalidCommandsArray[@]} -gt 0 ]; then # if number of elements in outputInvalidCommandsArray greater than 0, then display offending commands and exit 1.
287 echoerr "ERROR: Invalid commands found:""${outputInvalidCommandsArray[@]}"; # display invalid commands as error
288 exit 1; # See [1].
289 elif [ ${#outputInvalidCommandsArray[@]} -eq 0 ]; then # if number of elements in outputInvalidCommandsArray equals zero, then return 0.
290 vbm "DEBUG: Valid commands are:""${outputValidCommandsArray[@]}"; # display valid commands if verbose mode enabled
291 return 0; # See [1].
292 else
293 echoerr "ERROR: Check outputInvalidCommandsArray.";
294 fi
295
296 # End function
297 vbm "DEBUG:checkExecutables function ended."
298 return 0; # Function finished.
299} # Check that certain executables exist.
300updateTimeConstants() {
301 # Usage: updateTimeConstants
302 # Description: Updates time-related variables for use by other scripts.
303 # Input: (none)
304 # Output: Sets following variables:
305 # TIME_CURRENT Current time in long ISO-8601 format. (ex: YYYY-mm-ddTHH:MM:SS+ZZZZ)
306 # TIME_CURRENT_SHORT Current time in short ISO-8601 format. (ex: YYYYmmddTHHMMSS+ZZZZ)
307 # DATE_CURRENT Current date in ISO-8601 format. (ex: YYYY-mm-dd)
308 # DATE_CURRENT_SHORT Current date in short ISO-8601 format. (ex: YYYYmmdd)
309 # DATE_TOMORROW Tomorrow's date in ISO-8601 format. (ex: YYYY-mm-dd)
310 # TIME_NEXT_MIDNIGHT Time of tomorrow's midnight in long (ex: YYYY-mm-ddTHH:MM:SS+ZZZZ)
311 # ISO-861 format.
312 # SEC_TIL_MIDNIGHT Seconds until next midnight.
313 # Script function dependencies:
314 # - echoerr for displaying errors via stderr
315 # - processArguments for setting OPTION_VERBOSE
316 # - vbm for displaying verbose messages if OPTION_VERBOSE is "true"
317 # External dependencies: bash (5.0.3), date, echo
318 # Last modified: 2020-04-11T23:59Z
319 # Last modified by: Steven Baltakatei Sandoval
320 # License: GPLv3+
321 # Ref./Attrib.:
322
323 # Initialize function
324 vbm "DEBUG:updateTimeConstants function called."
325
326 # Perform work
327 TIME_CURRENT="$(date --iso-8601=seconds)" ;
328 TIME_CURRENT_SHORT="$(date -d "$TIME_CURRENT" +%Y%m%dT%H%M%S%z)"
329 DATE_CURRENT="$(date -d "$TIME_CURRENT" --iso-8601=date)" ;
330 DATE_CURRENT_SHORT="$(date -d "$TIME_CURRENT" +%Y%m%d)" ;
331 DATE_TOMORROW="$(date -d "$TIME_CURRENT next day" --iso-8601=date)" ;
332 TIME_NEXT_MIDNIGHT="$(date -d "$DATE_TOMORROW" --iso-8601=seconds)" ;
333 SECONDS_UNTIL_NEXT_MIDNIGHT="$(( $(date +%s -d "$TIME_NEXT_MIDNIGHT") - $(date +%s -d "$TIME_CURRENT") ))" ;
334
335 # End function
336 vbm "DEBUG:updateTimeConstants function ended."
337 return 0; # Function finished.
338} # Update time constants
339getGnucuDigest() {
340 # Usage: getGnucuDigest [algo] [file]
341 # Description: Calculates cryptographic digest using specified GnuCoreutils command.
342 # Input:
343 # $1: [algo] (ex: sha256sum, sha512sum, b2sum, md5sum)
344 # $2: [file]
345 # Output: stdout
346 # Script function dependencies:
347 # - echoerr for displaying errors via stderr
348 # - processArguments for setting OPTION_VERBOSE
349 # - vbm for displaying verbose messages if OPTION_VERBOSE is "true"
350 # - checkExecutables for checking that digest algo commands exist
351 # External dependencies: bash (5.0.3), echo, awk
352 # Last modified: 2020-05-04T18:51Z
353 # Last modified by: Steven Baltakatei Sandoval
354 # License: GPLv3+
355 # Ref./Attrib.:
356 # [1]: gnu.org (2000-11). [GNU Coreutils](https://www.gnu.org/software/coreutils/manual/coreutils.html). Licensed FDL v1.3+.
357 # [2]: linuxize.com (2019-07-19). [How to Check if a String Contains a Substring in Bash](https://linuxize.com/post/how-to-check-if-string-contains-substring-in-bash/).
358
359 # Initialize function
360 vbm "DEBUG:getGnucuDigest function called."
361 DIGEST_ALGO=$1 # Define candidate digest algorithm
362 TARGET_FILE=$2 # Define target file to be digested
363 GNUCU_DIGESTS="b2sum md5sum sha1sum sha224sum sha256sum sha384sum sha512sum" # Define space-delimited string with GNU Coreutils digest algorithms. See [1].
364 vbm "DEBUG:DIGEST_ALGO set to:""$DIGEST_ALGO"
365 vbm "TARGET_FILE set to:""$TARGET_FILE"
366 vbm "GNUCU_DIGESTS set to:""$GNUCU_DIGESTS"
367 checkExecutables "$DIGEST_ALGO"; # Check that specified alogirthm command available.
368 if [[ "$GNUCU_DIGESTS" =~ .*"$DIGEST_ALGO".* ]]; then
369 vbm "DEBUG:Confirmed to be a valid GNU Coreutils command:""$DIGEST_ALGO";
370 else
371 vbm "ERROR:Not a valid GNU coreutils command:""$DIGEST_ALGO";
372 exit 1;
373 fi # Check specified algo against GNUCU_DIGESTS. See [2].
374
375 # Perform work
376 OUTPUT="$("$DIGEST_ALGO" "$TARGET_FILE" | awk '{print $1}')"
377 vbm "DEBUG:Calculated digest using $DIGEST_ALGO is:""$OUTPUT"
378
379 # Output results
380 echo "$OUTPUT"
381
382 # End function
383 vbm "DEBUG:getGnucuDigest function ended."
384 return 0; # Function finished.
385} # Calculate GNUCoreutils digest
386
387
388#== Main Program Definition ==
389main() {
390 # Usage: main "$@" # See [1].
391 # Input: unspecified (note: all Global Constants declared at beginning of script assumed to be available)
392 # OPTION_VERBOSE (used by vbm, set by processArguments) (ex: "true", "false")
393 # DIGEST_ALGO (used by getGnucuDigest, set by processArguments) (ex: "md5sum")
394 # Output: unspecified
395 # Script function dependencies:
396 # - echoerr for displaying errors via stderr
397 # - vbm for displaying verbose messages if OPTION_VERBOSE is "true"
398 # - processArguments for setting OPTION_VERBOSE
399 # - checkExecutables for checking that specified commands are available
400 # External dependencies: bash (5.0.3), echo
401 # Last modified: 2020-05-04T18:44Z
402 # Last modified by: Steven Baltakatei Sandoval
403 # License: GPLv3+
404 # Ref./Attrib.:
405 # [1]: ErichBSchulz (2011-11-20). [How to correctly pass bash script arguments to functions](https://stackoverflow.com/a/8198970). Licensed CC BY-SA 4.0.
406
407 # Initialize function
408 processArguments "$@" # Process arguments.
409 vbm "DEBUG:main function called."
410 vbm "DEBUG:main function arguments are:""$@"
411 checkExecutables "echo" "date" "cut" "awk" "grep" # Confirm that executables necessary to run are available. Exit if any fail.
412
413 # Process inputs
414 # If DIRIN1 unset and global variable BK_FILEHASHLIST_DIR set and is dir, then set DIRIN1 to BK_FILEHASHLIST_DIR.
415 if ! [ -v DIRN1 ] && [ -v BK_FILEHASHLIST_DIR ] && [ -d "$BK_FILEHASHLIST_DIR" ]; then
416 DIRIN1="$BK_FILEHASHLIST_DIR";
417 vbm "DEBUG:DIRN1 not set, BK_FILEHASHLIST_DIR is dir, so setting DIRIN1 to BK_FILEHASHLIST_DIR:""$BK_FILEHASHLIST_DIR";
418 fi
419
420 # Set defaults
421 if ! [ -v DIGEST_ALGO ]; then # If DIGEST_ALGO is not set then set it to default value of "b2sum"
422 DIGEST_ALGO="b2sum";
423 vbm "DEBUG:DIGEST_ALGO not set. Setting to default value:""$DIGEST_ALGO"
424 fi
425
426
427 # if [ -v FILEIN1 ]; then # VERBOSE: Display contents of FILEIN1 if FILEIN1 has been set.
428 # vbm "DEBUG:main function detects input file:""$FILEIN1" ;
429 # vbm "DEBUG:contents of following file is displayed below:""$FILEIN1" ;
430 # if [ "$OPTION_VERBOSE" == "true" ]; then # display FILEIN1 contents via cat if verbose mode specified
431 # echoerr "========""$FILEIN1"" START""========" ;
432 # cat "$FILEIN1" 1>&2 ;
433 # echoerr "========""$FILEIN1"" END""========" ;
434 # fi
435 # fi
436
437
438 # Perform work
439 DIGEST="$(getGnucuDigest "$DIGEST_ALGO" "$FILEIN1")" # Calculate digest of FILEIN1
440 vbm "DEBUG:Hash of $FILEIN1 is:""$DIGEST"
441 vbm "DEBUG:Value of DIGEST is:""$DIGEST"
442 vbm "DEBUG:Value of DIRIN1 is:""$DIRIN1"
443 OUTPUT="$(grep -ri "$DIGEST" "$DIRIN1")" # Search contents of DIRIN1 for pattern DIGEST, save results to OUTPUT.
444
445 # Output results
446 echo -e "$OUTPUT"
447
448 if [ -v FILEOUT1 ]; then # if FILEOUT1 set, write to this file.
449 vbm "DEBUG:main function detects FILEOUT1 set to:""$FILEOUT1" ;
450 echo -e "$OUTPUT" > "$FILEOUT1" ;
451 vbm "DEBUG:main funtion wrote OUTPUT to:""$FILEOUT1" ;
452 elif [ -v DIROUT1 ]; then # else if DIROUT1 set, set FILEOUT1 to timestamped filename, combine into PATHOUT1, write output to PATHOUT1.
453 vbm "DEBUG:main function detects DIROUT1 set to:""$DIROUT1" ;
454 FILEOUT1="$SCRIPT_TIME_SHORT".."$(basename "$0")".txt
455 PATHOUT1="$DIROUT1"/"$FILEOUT1"
456 echo -e "$OUTPUT" > $PATHOUT1 ;
457 vbm "DEBUG:main function wrote output file to:""$PATHOUT1" ;
458 fi
459
460 # End function
461 vbm "DEBUG:main function ended."
462 return 0; # Function finished.
463}
464
465#== Perform work and exit ==
466main "$@" # Run main function.
467exit 0;