1    #! /usr/bin/env python
   2    #                        Time-stamp: <2001-07-10 22:02:36 cymbala>
   3    # ~/bin/do_ma_un.py
   4    
   5    
   6    # SYNTAX EXAMPLE:
   7    #  python bin/do_ma_un.py `find /cdrom/archive/lenin/works/1901/ -name '*htm*'`
   8    
   9    
  10    #
  11    #    WARNING! WARNING! WARNING!
  12    #      This script breaks a fundamental rule.  Usually, UNIX text
  13    #      files use a linefeed to separate lines.  This script thinks
  14    #      either a linefeed or a carriage return is a line separator.
  15    #      Usually a carriage return is _NOT_ considered a line separator.
  16    #      You have been warned!
  17    #
  18    # This script will (illegally) read logical lines properly from all
  19    # three types of text files:
  20    #            DOS  - CR/LF between logical lines.
  21    #            MAC  - CR    between logical lines.
  22    #            UNIX - LF    between logical lines.
  23    
  24    import sys
  25    import re
  26    import string
  27    import fileinput
  28    
  29    
  30    #  To put this class into production:
  31    #   - rename "Program" to something more descriptive.
  32    #   - remove "program" lines from end.
  33    #   - put this file into a subdirectory that is seen by Python's path.
  34    #   - import this file.
  35    #   - use class "Something_more_descriptive" like any other class.
  36    
  37    class Do_ma_un:
  38        def __init__(self):
  39            """Called when class is instantiated."""
  40    
  41            self.re_http = re.compile('^http:', re.I)
  42            pass
  43    
  44    
  45        def __call__(self, dictionary):
  46            """Called when instance is called without an attribute."""
  47    
  48            for input in dictionary['inputs']:
  49                dictionary['input'] = input
  50                self.fileinput_line_handler(dictionary)
  51                pass
  52            pass
  53    
  54    
  55        def lineno(self):
  56            """Keep track of logical line number within a file."""
  57            return self.linen
  58            pass
  59    
  60        
  61        def fileinput_line_handler(self, dictionary):
  62            """Examine logical lines regardless of file-type (MAC, DOC or UNIX)."""
  63    
  64            input = dictionary['input']
  65    
  66            # Read from text file, whether it's a disk file or a Web page.
  67            if self.re_http.match(input):
  68                import urllib
  69                page = urllib.urlopen(input)
  70                raw_lines = page.readlines()
  71                pass
  72            else:
  73                raw_lines = fileinput.input(input)
  74                pass
  75            
  76            self.linen = 0
  77            for line in raw_lines:
  78                #
  79                # Handle DOS files (assume DOS file ends with CR-LF):
  80                if line[-2:] == '\r\n':
  81                    line = line[:-2] + line[-1]
  82                    pass
  83                #
  84                # Handle MAC and UNIX files (in case of UNIX files, lines
  85                # is simply line converted to a list because the above
  86                # fileinput has already split by linefeeds.
  87                lines = string.split(line, '\r')
  88                #
  89                #
  90                # Handle all logical lines, from DOC, MAC and UNIX files:
  91                self.logical_line_handler(dictionary, lines)
  92            pass
  93    
  94    
  95        def logical_line_handler(self, dictionary, list):
  96            """Process physical line read by normal fileinput.input() attribute."""
  97            for i in range(len(list)):
  98                logical_line = list[i]
  99                #
 100                if logical_line == '':
 101                    continue
 102                else:
 103                    self.linen = self.linen + 1
 104                    pass
 105                #
 106                # Lines from MAC files will not end in linefeed.
 107                # Lines from other files will end in linefeed.
 108                #
 109                if logical_line[-1] == '\n':
 110                    logical_line = logical_line[:-1]
 111                    pass
 112                #
 113                dictionary['lineno'] = self.lineno()
 114                dictionary['logical_line'] = logical_line
 115                dictionary['function_to_call'](dictionary)
 116                #
 117                pass
 118            pass
 119    
 120    
 121    
 122        def print_first_three(self, dictionary):
 123            """This attribute is fed a line no. and line to do something with."""
 124            if dictionary['lineno'] <= 3:
 125                print dictionary['input'], dictionary['logical_line']
 126                pass
 127            pass
 128    
 129        pass
 130    # ---------------------------------
 131    
 132    if __name__ == '__main__':
 133        Program = Do_ma_un()
 134    
 135        dictionary = {}
 136        dictionary['inputs'] = sys.argv[1:]
 137        dictionary['function_to_call'] = Program.print_first_three
 138        #
 139        Program(dictionary)
 140        pass
 141    
 142    ###
 143    #