cgrep.py

#!/usr/bin/python
from __future__ import print_function

#-----------------------------------------------------------------------
#  Imports
#-----------------------------------------------------------------------
import sys
import getopt


#-----------------------------------------------------------------------
#  Functions
#-----------------------------------------------------------------------

def Usage(msg=None):
    print("""

    cgrep -iknq <str>        <file>
    cgrep -f    <str_file>   <file>
 
    Search for matching 'clauses' in file.  Clause is a block
    of text separated by blank lines.
 
        -f <str_file>  Read strings from file.
        -i             Ignore case.
        -k <key>       Match <str> to entire value of <key> field.       
        -n             Print line numbers.
        -q             Only print the first matching clause.
        -v             Print non-matching clauses.
        -S             Print output "KEY=VALUE\nKEY=VALUE" to set value in shell.
                          Run like: eval $(cgrep -k KEY VALSTR)

""")
    if msg: print(msg)
    raise SystemExit


#  Read clauses
def read_clauses(files,lower_case=False):
    line_no = 0
    for file in files:
        clause = []
        #  Open file
        if file=="-":
            fin = sys.stdin
        else:
            fin = open(file,"r")
        for line in fin.readlines():
            line_no += 1
            line = line.strip()
            #  End of clause
            if not line:
                if clause:
                    yield clause
                clause = []
            #  Add line to clause
            else:
                if lower_case:
                    clause.append((line_no, line.lower()))
                else:
                    clause.append((line_no, line))
        #  Close file
        if not file=="-":
            fin.close()
        #  return last clause in file
        if clause:
            yield clause
            clause = []

def in_search_strings(val, search_strings, ignore_case=False):
        if ignore_case:
                return val.lower() in [s.lower() for s in search_strings]
        else:
                return val in [s for s in search_strings]

#  Escape shell text strings so that we can still assign them to shell
#  variables even if they contain: single quotes, double quotes, and/or spaces
def escape_shell_line(line):
        #  Return escaped string if it contains: single quote, double quote, or
        #  space
        for c in line:
                if c in (""" '" """):
                        return '"' + line.replace('"','\\"') + '"'
        #  Return original string
        return line

#  Read line like "LABEL: value" and return "LABEL=value"
#  so string can be used by Unix shell to set shell variable
def get_kvformat(line):
        # Get first split char
        for pos, character in enumerate(line):
                found = character in (' ',':')
                if found: break
        if found:
                k = line[:pos]
                v = line[pos+1:].strip()
                return k + '=' + escape_shell_line(v)
        else:
                return line




#-----------------------------------------------------------------------
#  Main
#-----------------------------------------------------------------------


#  Initialize options
string_file = None
search_key  = None
ignore_case = False
print_line_numbers = False
quiet = False
exclude = False
versbose = False
setvars  = False


#  Read options
opts, args = getopt.getopt(sys.argv[1:],"f:ik:nqvS")
for key,val in opts:
    if key=='-f':
        string_file = val
    elif key=='-k':
        search_key = val
    elif key=='-i':
        ignore_case = True
    elif key=='-n':
        print_line_numbers = True
    elif key=='q':
        quiet = True
    elif key=='-v':
        exclude = True
    elif key=='-S':
        setvars = True

#  If not string_file, then read single search string from input line
if not string_file:
    if len(args)<2:
        Usage()
    search_string, files = args[0], args[1:]
else:
    if len(args)<1:
        Usage()
    files = args


#  Get search strings from file
if string_file:
    search_strings = []
    fin = open(string_file,"r")
    for line in fin.readlines():
        line = line.strip()
        if line: 
            search_strings.append(line)
else:
    search_strings = (search_string,)

#  If ignoring cause, convert all search strings to lowercase
if ignore_case:
    search_strings = [s.lower() for s in search_strings]
    if search_key:
        search_key = search_key.lower()

#  Look for a specific key and value - entire value must match
if search_key:
    key_len = len(search_key)+1
    keys = (search_key+":", search_key+" ")

#  Read clauses
for clause in read_clauses(files):

    found = False
    #  Read line numbers and lines from clause
    for line_no, line in clause:
        #  Search specific key (key of form "key:" or "key ")
        if search_key:
            if line[:key_len] in keys:
                search_val = line[key_len:].strip()
                if ignore_case: search_val.lower()
                found = in_search_strings(search_val,search_strings)
                if found:
                        break
        #  Search entire clause
        else:
            if ignore_case:
               line = line.lower()
            for search_string in search_strings:
                if line.find(search_string)>=0:
                    found = True
                    break
        #  Search string found, so stop searching this clause.
        if found:
            break

    #  Print clause if string found.
    if found ^ exclude:
        for line_no, line in clause:
            #  Print line numbers with lines
            if print_line_numbers:
                print(line_no, line)
            #  Print lines only (default)
            else:
                if setvars:
                        print(get_kvformat(line))
                else:
                        print(line)
        print()