#!/usr/bin/python3 # author: leadingsuspect(npub1xl56l7zr2464c2h3vj34nzqamwcf7eatvy2x47u87pkyuh4csrcq45zcuk) # usage: lookup [-h] [-b BEGIN] [-x MAX] [--version] [-q | -v] infile [outfile] 'searchstring as multiple regexpatterns' # =========================================================== import argparse import sys import logging import re # import shutil from io import UnsupportedOperation from builtins import FileNotFoundError, FileExistsError, PermissionError from collections import defaultdict def priorities(): return ([],[],[],[]) result = defaultdict(priorities) # Errorcodes not yet used as return codes E_BADARGS = 85 E_NOFILE = 86 E_NOPERM = 2 E_FILEEXISTS = 3 E_UNSUPPORTEDOP = 4 E_OTHERERR = 99 BLUE = '94m' # ansi light blue def colored_text(color, text): colored_text = f"\033[{color}{text}\033[00m" return colored_text def highlight(text): return colored_text(BLUE, text) # logging.warning('I print to stderr by default') # logging.info('For this you must change the level and add a handler.') # =========================================================== # Parsing Script Arguments parser = argparse.ArgumentParser( description='This script allows you to do multiple Pearl-Regex searches given as one big sentence with a dictionary file') parser.add_argument( 'infile', help='path/to/inputfile for the operation (default: stdin)', type=str, default=sys.stdin) parser.add_argument( 'outfile', help='path/to/outputfile for the operation (default: stdout)', nargs='?', type=str, default=sys.stdout) parser.add_argument( 'searchstring', help='Multiple Words that can be Perl-Regex-formed to lookup in a dictionary', type=str) parser.add_argument( '-b', '--begin', help='b - begin from line b of the file (default= 83)', type=int, default=83) parser.add_argument( '-x', '--max', help='i - max number of items shown per pattern (default: 3)', type=int, default=3) parser.add_argument( '--version', action='version', version='%(prog)s 2.0') group = parser.add_mutually_exclusive_group() # group.add_argument("-v", "--verbose", action="store_true") group.add_argument( "-q", "--quiet", help="opposite of verbosity", action="store_true") group.add_argument( '-v', '--verbosity', help="increase output verbosity", action="count", default=0) args = parser.parse_args() # =========================================================== # Check if file argument is a file and set processingFile if true # if os.path.isfile(args.infile): # processingFile = args.infile # else: # processingFile = None if args.infile: processingFile = args.infile else: processingFile = '' # =========================================================== # Check if searchstring argument holds valid Perl Regex Patterns and set searchWords if true if args.searchstring: try: delimiters = ' |,|-' # patterns = list(args.searchstring.split(delimiter)) patterns = list(re.split(delimiters, args.searchstring)) except: print("could not split searchstring") else: print ("No search string provided") sys.exit(E_BADARGS) # Check if Lines range is valid if args.begin: try: begin = int(args.begin) except: print("Plz provide line beginning: -bBeginlinenumber") else: print ("No begin linenumber provided") sys.exit(E_BADARGS) # Check if max items per Searchpattern is valid if args.max: try: maxitems = int(args.max) except: print("Plz provide max number of items per searchstring: -xMaxItemnumber") else: print ("No max itemnumber provided") sys.exit(E_BADARGS) # =========================================================== # The process itself def openfile(filename, modus): # use it within a exception handler with open(filename, modus, encoding="latin-1") as file: lines = file.readlines() pairmatches = [] exactmatches = [] if args.verbosity >= 1: # print("processing open file {}".format(filename)) if args.verbosity >=2: print("There are %d lines in the dictionary %s" % (len(lines), filename)) print(f"beginnig at line -b{begin}") print(f"max items per Pattern: -i{maxitems}.") print(f"searching... {patterns}") # for line in lines[83:]: for line in lines[begin:]: if (len(patterns)>1) and (re.search(r'' + args.searchstring, line)): exactmatches.append(line) else: for i in range(len(patterns)): p = r'' + patterns[i] if i>=1: if re.search((patterns[i-1] + r' ' + p),line): pairmatches.insert(0, highlight((patterns[i-1] + ' ' + p + ': ')) + line) # result[p].append(line) result[p][0].append(line) continue if re.search(r'(^|[^a-zA-Z\'])' + p + r'($|[^a-zA-Z\'])', line): # result[p].append(line) result[p][1].append(line) continue if re.search(r'([a-zA-Z\'-]' + p + r'[^a-zA-Z\']|[^a-zA-Z\']' + p + '[a-zA-Z\'-])', line): result[p][2].append(line) continue if re.search(p, line): result[p][3].append(line) if pairmatches or result or exactmatches: if result: for p in patterns: print(highlight((p + ': '))) matches_per_priority = result.get(p) if matches_per_priority: print(''.join([match for matches in matches_per_priority for match in matches][:maxitems])) if pairmatches: print("paired matches: ") print(''.join(pairmatches[:maxitems])) if exactmatches: print(highlight(args.searchstring + ': ')) print(''.join(exactmatches[:maxitems])) else: print("no matches found") def process(filename, modus): try: openfile(filename, modus) except FileNotFoundError: logging.warning("I did not find filename '%s'!" % filename) sys.exit(E_NOFILE) except PermissionError: logging.warning( "Permission denied to open '%s' in '%s' mode." % (filename, modus)) sys.exit(E_NOPERM) except FileExistsError: logging.warning("File '%s' already exists" % filename) sys.exit(E_FILEEXISTS) except UnsupportedOperation as UsO: logging.warning( "UnsupportedOperation: %s in '%s' mode" % (str(UsO), modus)) sys.exit(E_UNSUPPORTEDOP) except IOError: logging.warning("I/O Error occurred: %s" % str(IOError)) sys.exit(E_OTHERERR) # finally: # =========================================================== # verbosity level influences processing messages... if args.verbosity >= 2: print("verbosity level is: %s" % args.verbosity) # Process file if set if processingFile: print("'%s' is a file and will be processed" % args.infile) process(processingFile, 'r') else: print("'%s' is not a file to be processed" % args.infile) elif args.verbosity >= 1: # print("verbose") # Process file if set if processingFile: print("Processing '%s'" % args.infile) process(processingFile, 'r') else: print("'%s' no file" % args.infile) else: # print("processing in silence") if processingFile: process(processingFile, 'r')