AptRdepends2JSON.py

   1 """Go from Debian apt-rdepends output to JSON for a D3.js chart."""
   2
   3 class AptRdepends2JSON:
   4     """Convert Debian apt-rdepends output to JSON for a D3.js chart.
   5
   6     In its man page, we read that apt-rdepends "performs recursive dependency
   7     listings similar to apt-cache". For instance, in the following snippet
   8     apt-rdepends has printed out the recursive dependencies for the two Debian
   9     packages "tar" and "libustr-1.0-1":
  10
  11     tar
  12       PreDepends: libacl1 (>= 2.2.51-8)
  13       PreDepends: libc6 (>= 2.17)
  14       PreDepends: libselinux1 (>= 1.32)
  15     libustr-1.0-1
  16       Depends: libc6 (>= 2.4)
  17       PreDepends: multiarch-support
  18
  19     Based on this snippet, we can then produce the following JSON for the
  20     D3.js chart:
  21
  22     [
  23         {
  24             "imports": [
  25                 "libacl1",
  26                 "libc6",
  27                 "libselinux1"
  28             ],
  29             "name": "tar"
  30         },
  31         {
  32             "imports": [
  33                 "libc6",
  34                 "multiarch-support"
  35             ],
  36             "name": "libustr-1.0-1"
  37         }
  38     ]
  39     """
  40
  41     def __init__(self, aptrdepends_output, d3_json):
  42         """We set the following attributes:
  43
  44         1. aptrdepends_output. The name of a file generated by something like...
  45
  46         apt-rdepends aptitude > aptrdepends_output
  47
  48         2. d3_json. The name of some JSON file that we will generate based on
  49         the contents of aptrdepends_output file.
  50
  51         3. package_list. json.dump generates d3_json file by serializing the
  52         package_list list.
  53         """
  54
  55         self.aptrdepends_output = aptrdepends_output
  56         self.d3_json = d3_json
  57         self.package_list = []
  58
  59     def trim_dependency_line(self, my_string):
  60         """Remove leading and trailing characters that apt-rdepends inserts
  61         into the my_string argument.
  62
  63         apt-rdepends specifies "Depends", "Pre-Depends", "strictly earlier",
  64         "exactly equal", etc relations between packages.
  65
  66         For now, we're only interested in package names...
  67
  68         >>> my_object = AptRdepends2JSON('apt-rdepends_aptitude',\
  69 'apt-rdepends_aptitude.json')
  70         >>> f = open(my_object.aptrdepends_output)
  71         >>> lines = f.readlines()
  72         >>> print(lines[1].rstrip('\\n'))
  73           Depends: aptitude-common (= 0.8.11-7)
  74         >>> my_package_name = my_object.trim_dependency_line(lines[1])
  75         >>> print(my_package_name)
  76         aptitude-common
  77         >>> print(lines[43].rstrip('\\n'))
  78           PreDepends: libc6 (>= 2.28)
  79         >>> my_package_name = my_object.trim_dependency_line(lines[43])
  80         >>> print(my_package_name)
  81         libc6
  82         """
  83
  84         trim_dependency_line_regex = re.compile('^\\s+\\S*Depends\\S\\s(\\S+)')
  85         my_match = trim_dependency_line_regex.match(my_string)
  86         return my_match.group(1)
  87
  88     def check_for_dependency_lines(self, my_string):
  89         """Return True if my_string argument matches ^\\s+\\S*Depends regex.
  90
  91         Among other things, in the following snippet apt-rdepends is saying
  92         that the recursive dependencies of libgcc1 are gcc-4.9-base, libc6, and
  93         multiarch-support...
  94
  95         libgcc1
  96           Depends: gcc-4.9-base (= 4.9.0-7)
  97           Depends: libc6 (>= 2.14)
  98           PreDepends: multiarch-support
  99
 100         >>> my_object = AptRdepends2JSON('apt-rdepends_aptitude',\
 101 'apt-rdepends_aptitude.json')
 102         >>> f = open(my_object.aptrdepends_output)
 103         >>> lines = f.readlines()
 104         >>> print(lines[0].rstrip('\\n'))
 105         aptitude
 106         >>> my_object.check_for_dependency_lines(lines[0]) == None
 107         True
 108         >>> print(lines[1].rstrip('\\n'))
 109           Depends: aptitude-common (= 0.8.11-7)
 110         >>> bool(my_object.check_for_dependency_lines(lines[1])) == True
 111         True
 112         >>> print(lines[43].rstrip('\\n'))
 113           PreDepends: libc6 (>= 2.28)
 114         >>> bool(my_object.check_for_dependency_lines(lines[21])) == True
 115         True
 116         """
 117
 118         this_is_a_dependency_line = re.compile('^\\s+\\S*Depends')
 119         return this_is_a_dependency_line.match(my_string)
 120
 121     def read_aptrdepends_output(self):
 122         """Loop over lines of self.aptrdepends_output file, and construct
 123         self.d3_json list for json.dump serialization.
 124
 125         This method takes apt-rdepends idiosyncracies into account. We're doing
 126         the following...
 127
 128         1. self.d3_json list was initialized in __init__
 129
 130         2. We loop over every line in self.aptrdepends_output file
 131
 132         3. If line matches ^\\s+\\S*Depends regex, do the following...
 133
 134         3A. Chop off extraneous "  (Pre)Depends: " substring
 135
 136         3B. Chop off extraneous " (...something...) " substring
 137
 138         3C. Append resulting string to imports list of current dictionary in
 139         loop
 140
 141         4. If line does not match ^\\s+\\S*Depends regex, append new dictionary
 142         to self.d3_json. New dictionary is initialized like so...
 143         {'name':current_line_in_loop,'imports':[]}
 144
 145         >>> my_aptrdepends = 'apt-rdepends_aptitude'
 146         >>> my_json = 'apt-rdepends_aptitude.json'
 147         >>> my_object = AptRdepends2JSON(my_aptrdepends, my_json)
 148         >>> my_object.read_aptrdepends_output()
 149         >>> print(my_object.package_list['children'][0])
 150         {'name': 'aptitude', 'imports': ['aptitude-common', 'libapt-pkg5.0', 'libboost-iostreams1.67.0', 'libboost-system1.67.0', 'libc6', 'libcwidget3v5', 'libgcc1', 'libncursesw6', 'libsigc++-2.0-0v5', 'libsqlite3-0', 'libstdc++6', 'libtinfo6', 'libxapian30']}
 151         """
 152
 153         aptrdepends_file = open(self.aptrdepends_output, 'r')
 154
 155         for rdepends_line in aptrdepends_file:
 156
 157             if self.check_for_dependency_lines(rdepends_line):
 158
 159                 dependency = self.trim_dependency_line(rdepends_line)
 160                 self.package_list[-1]['imports'].append(dependency)
 161
 162             else:
 163
 164                 chomp = rdepends_line.rstrip('\n')
 165                 self.package_list.append({'name':chomp, 'imports':[]})
 166
 167         self.package_list = {'name': 'recursive_aptitude_dependencies',\
 168             'children': self.package_list}
 169         json_file = open(self.d3_json, 'w')
 170         json.dump(self.package_list, json_file, indent=4)
 171         json_file.close()
 172
 173 if __name__ == "__main__":
 174     import re
 175     import json
 176
 177     import doctest
 178     doctest.testmod()