Minor changes, mainly documentation

Tássia Camões Araújo
1 parent caf1024a
Showing 8 changed files with 25 additions and 8 deletions Show diff stats
src/bin/get_desktop.sh
src/bin/get_pkgs_inst.py
src/bin/get_program.sh
src/bin/get_tags.sh
src/bin/pkgindex.py
src/bin/popindex.py
src/bin/rank_terms.py
src/data.py
 #!/usr/bin/env bash
+#
+# get_desktop.sh - get packages which have desktop files 
+
 cd /usr/share/app-install/desktop
 sed -ne 's/X-AppInstall-Package=//p' * | sort -u
 #!/usr/bin/env python
+#
+# get_pkgs_inst.py - get tuple (package,installation) from popcon results file
 from operator import itemgetter
 if __name__ == '__main__':
 #!/usr/bin/env bash
+#
+# get_program.sh - get packages which have the tags 'role::program' associated
+
 cat /var/lib/debtags/package-tags |grep "role::program" | awk -F: '{ print $1 }'
 #!/usr/bin/env bash
+#
+# get_tags.py - get meaningful tags for recommendation purposes
+
 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}'
 #!/usr/bin/env python
 """
-    Clustering - A python script to perform clustering of popcon data.
+    pkgindex.py - generate a pkgs index to be used by the recommender as the
+                  items repository, based on the pkgs filter provided by config
 """
 __author__ = "Tassia Camoes Araujo <tassia@gmail.com>"
 __copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"
@@ -38,7 +39,7 @@ if __name__ == &#39;__main__&#39;:
         pkgs_list = [line.strip() for line in valid]
         logging.info("Packages list length: %d" % len(pkgs_list))
-    # use config file or command line options 
+    # use config file or command line options
     pkgindex = data.SampleAptXapianIndex(pkgs_list,xapian.Database(cfg.axi),
                                          cfg.axi+"-"+cfg.pkgs_filter)
     end_time = datetime.datetime.now()
 #!/usr/bin/env python
 """
-    Clustering - A python script to perform clustering of popcon data.
+    popindex.py - generate a popcon index to be used by the recommender as the
+                  users repository, based on filters provided by config
 """
 __author__ = "Tassia Camoes Araujo <tassia@gmail.com>"
 __copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"
 #!/usr/bin/env python
+#
+# rank_terms.py - rank index terms by frequency
 import xapian
 import os
@@ -67,7 +67,7 @@ def print_index(index):
 class AppAptXapianIndex(xapian.WritableDatabase):
     """
-    Sample data source for packages information, mainly useful for tests.
+    Data source for application packages information
     """
     def __init__(self,axi_path,path):
         xapian.WritableDatabase.__init__(self,path,
@@ -93,7 +93,8 @@ class AppAptXapianIndex(xapian.WritableDatabase):
 class SampleAptXapianIndex(xapian.WritableDatabase):
     """
-    Sample data source for packages information, mainly useful for tests.
+    Sample data source for packages information, generated from a list of
+    packages.
     """
     def __init__(self,pkgs_list,axi,path):
         xapian.WritableDatabase.__init__(self,path,
@@ -106,9 +107,10 @@ class SampleAptXapianIndex(xapian.WritableDatabase):
         return print_index(self)
 class PopconSubmission():
-    def __init__(self,path,user_id=0):
+    def __init__(self,path,user_id=0,binary=1):
         self.packages = dict()
         self.path = path
+        self.binary = binary
         self.load()
         if user_id:
             self.user_id = user_id
@@ -142,7 +144,7 @@ class PopconSubmission():
                         if len(data) > 3:
                             exec_file = data[3]
                             # Binary weight
-                            if binary:
+                            if self.binary:
                                 self.packages[pkg] = 1
                             # Weights inherited from Enrico's anapop
                             # No executable files to track
@@ -171,7 +173,7 @@ class PopconXapianIndex(xapian.WritableDatabase):
         self.source_dir = os.path.expanduser(cfg.popcon_dir)
         self.max_popcon = cfg.max_popcon
         self.valid_pkgs = []
-        # file format: one pkg_name per line
+        # file format for filter: one package name per line
         with open(os.path.join(cfg.filters,cfg.pkgs_filter)) as valid_pkgs:
             self.valid_pkgs = [line.strip() for line in valid_pkgs
                                if not line.startswith("#")]
1	#!/usr/bin/env bash	1	#!/usr/bin/env bash
		2	+#
		3	+# get_desktop.sh - get packages which have desktop files
		4	+
2	cd /usr/share/app-install/desktop	5	cd /usr/share/app-install/desktop
3	sed -ne 's/X-AppInstall-Package=//p' * \| sort -u	6	sed -ne 's/X-AppInstall-Package=//p' * \| sort -u
1	#!/usr/bin/env python	1	#!/usr/bin/env python
		2	+#
		3	+# get_pkgs_inst.py - get tuple (package,installation) from popcon results file
2		4
3	from operator import itemgetter	5	from operator import itemgetter
4	if __name__ == '__main__':	6	if __name__ == '__main__':
1	#!/usr/bin/env bash	1	#!/usr/bin/env bash
		2	+#
		3	+# get_program.sh - get packages which have the tags 'role::program' associated
		4	+
2	cat /var/lib/debtags/package-tags \|grep "role::program" \| awk -F: '{ print $1 }'	5	cat /var/lib/debtags/package-tags \|grep "role::program" \| awk -F: '{ print $1 }'
1	#!/usr/bin/env bash	1	#!/usr/bin/env bash
		2	+#
		3	+# get_tags.py - get meaningful tags for recommendation purposes
		4	+
2	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}'	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}'
1	#!/usr/bin/env python	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	__author__ = "Tassia Camoes Araujo <tassia@gmail.com>"	6	__author__ = "Tassia Camoes Araujo <tassia@gmail.com>"
6	__copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"	7	__copyright__ = "Copyright (C) 2011 Tassia Camoes Araujo"
1	#!/usr/bin/env python	1	#!/usr/bin/env python
		2	+#
		3	+# rank_terms.py - rank index terms by frequency
2		4
3	import xapian	5	import xapian
4	import os	6	import os