Download as
Browse Code »

Commit 19f1de42244a512b3f9de2c90fd93b5578a002e8

Authored by Tássia Camões Araújo
1 parent caf1024a
Exists in master and in 1 other branch add_vagrant

Minor changes, mainly documentation

Showing 8 changed files with 25 additions and 8 deletions   Show diff stats
src/bin/get_desktop.sh
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env bash
  2 +#
  3 +# get_desktop.sh - get packages which have desktop files
  4 +
2 5 cd /usr/share/app-install/desktop
3 6 sed -ne 's/X-AppInstall-Package=//p' * | sort -u
... ...
src/bin/get_pkgs_inst.py
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env python
  2 +#
  3 +# get_pkgs_inst.py - get tuple (package,installation) from popcon results file
2 4  
3 5 from operator import itemgetter
4 6 if __name__ == '__main__':
... ...
src/bin/get_program.sh
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env bash
  2 +#
  3 +# get_program.sh - get packages which have the tags 'role::program' associated
  4 +
2 5 cat /var/lib/debtags/package-tags |grep "role::program" | awk -F: '{ print $1 }'
... ...
src/bin/get_tags.sh
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env bash
  2 +#
  3 +# get_tags.py - get meaningful tags for recommendation purposes
  4 +
2 5 cat /var/lib/debtags/vocabulary | grep "Tag:" | egrep -v "culture::|devel::lang|hardware::|implemented-in::|interface::|iso15924::|made-of::|network::|protocol::|role::|scope::|secteam::|special::|uitoolkit::|x11::|TODO" | awk '{print $2}'
... ...
src/bin/pkgindex.py
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env python
2 2 """
3   - Clustering - A python script to perform clustering of popcon data.
  3 + pkgindex.py - generate a pkgs index to be used by the recommender as the
  4 + items repository, based on the pkgs filter provided by config
4 5 """
5 6 __author__ = "Tassia Camoes Araujo <tassia@gmail.com>"
6 7 __copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"
... ... @@ -38,7 +39,7 @@ if __name__ == &#39;__main__&#39;:
38 39 pkgs_list = [line.strip() for line in valid]
39 40 logging.info("Packages list length: %d" % len(pkgs_list))
40 41  
41   - # use config file or command line options
  42 + # use config file or command line options
42 43 pkgindex = data.SampleAptXapianIndex(pkgs_list,xapian.Database(cfg.axi),
43 44 cfg.axi+"-"+cfg.pkgs_filter)
44 45 end_time = datetime.datetime.now()
... ...
src/bin/popindex.py
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env python
2 2 """
3   - Clustering - A python script to perform clustering of popcon data.
  3 + popindex.py - generate a popcon index to be used by the recommender as the
  4 + users repository, based on filters provided by config
4 5 """
5 6 __author__ = "Tassia Camoes Araujo <tassia@gmail.com>"
6 7 __copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"
... ...
src/bin/rank_terms.py
  Diff comments   View file @19f1de4
1 1 #!/usr/bin/env python
  2 +#
  3 +# rank_terms.py - rank index terms by frequency
2 4  
3 5 import xapian
4 6 import os
... ...
src/data.py
  Diff comments   View file @19f1de4
... ... @@ -67,7 +67,7 @@ def print_index(index):
67 67  
68 68 class AppAptXapianIndex(xapian.WritableDatabase):
69 69 """
70   - Sample data source for packages information, mainly useful for tests.
  70 + Data source for application packages information
71 71 """
72 72 def __init__(self,axi_path,path):
73 73 xapian.WritableDatabase.__init__(self,path,
... ... @@ -93,7 +93,8 @@ class AppAptXapianIndex(xapian.WritableDatabase):
93 93  
94 94 class SampleAptXapianIndex(xapian.WritableDatabase):
95 95 """
96   - Sample data source for packages information, mainly useful for tests.
  96 + Sample data source for packages information, generated from a list of
  97 + packages.
97 98 """
98 99 def __init__(self,pkgs_list,axi,path):
99 100 xapian.WritableDatabase.__init__(self,path,
... ... @@ -106,9 +107,10 @@ class SampleAptXapianIndex(xapian.WritableDatabase):
106 107 return print_index(self)
107 108  
108 109 class PopconSubmission():
109   - def __init__(self,path,user_id=0):
  110 + def __init__(self,path,user_id=0,binary=1):
110 111 self.packages = dict()
111 112 self.path = path
  113 + self.binary = binary
112 114 self.load()
113 115 if user_id:
114 116 self.user_id = user_id
... ... @@ -142,7 +144,7 @@ class PopconSubmission():
142 144 if len(data) > 3:
143 145 exec_file = data[3]
144 146 # Binary weight
145   - if binary:
  147 + if self.binary:
146 148 self.packages[pkg] = 1
147 149 # Weights inherited from Enrico's anapop
148 150 # No executable files to track
... ... @@ -171,7 +173,7 @@ class PopconXapianIndex(xapian.WritableDatabase):
171 173 self.source_dir = os.path.expanduser(cfg.popcon_dir)
172 174 self.max_popcon = cfg.max_popcon
173 175 self.valid_pkgs = []
174   - # file format: one pkg_name per line
  176 + # file format for filter: one package name per line
175 177 with open(os.path.join(cfg.filters,cfg.pkgs_filter)) as valid_pkgs:
176 178 self.valid_pkgs = [line.strip() for line in valid_pkgs
177 179 if not line.startswith("#")]
... ...