Download as
Browse Code »

Commit af80bfe9a4cd6cf70ed228b73461092d124242e6

Authored by Tássia Camões Araújo
1 parent 010147f6
Exists in master and in 1 other branch add_vagrant

Adding PopconSubmission class.

Showing 1 changed file with 39 additions and 29 deletions   Show diff stats
src/data.py
  Diff comments   View file @af80bfe
... ... @@ -74,22 +74,40 @@ class SampleAptXapianIndex(xapian.WritableDatabase):
74 74 print [term.term for term in self.get_document(doc_id).termlist()]
75 75 print "---"
76 76  
77   -#[FIXME] get pkg tags from axi and remove load_debtags_db method
78   -def load_debtags_db(db_path):
79   - """
80   - Load debtags database from the source file.
81   - """
82   - tag_filter = re.compile(r"^special::.+$|^.+::TODO$")
83   - try:
84   - db_file = open(db_path, "r")
85   - debtags_db = debtags.DB()
86   - debtags_db.read(db_file,lambda x: not tag_filter.match(x))
87   - db_file.close()
88   - return debtags_db
89   - except:
90   - logging.error("Could not load DebtagsDB from '%s'." % self.db_path)
91   - raise Error
  77 +class PopconSubmission():
  78 + def __init__(self,submission_hash):
  79 + self.hash = submission_hash
  80 + self.pkgs_list = []
92 81  
  82 + def add_pkg(self,pkg):
  83 + self.pkgs_list.append(pkg)
  84 +
  85 + def parse_submission(self,submission_path,binary=1):
  86 + """
  87 + Parse a popcon submission, generating the names of the valid packages
  88 + in the vote.
  89 + """
  90 + submission = open(submission_path)
  91 + for line in submission:
  92 + if not line.startswith("POPULARITY"):
  93 + if not line.startswith("END-POPULARITY"):
  94 + data = line[:-1].split(" ")
  95 + if len(data) > 3:
  96 + if binary:
  97 + # every installed package has the same weight
  98 + yield data[2], 1
  99 + elif data[3] == '<NOFILES>':
  100 + # No executable files to track
  101 + yield data[2], 1
  102 + elif len(data) == 4:
  103 + # Recently used packages
  104 + yield data[2], 10
  105 + elif data[4] == '<OLD>':
  106 + # Unused packages
  107 + yield data[2], 3
  108 + elif data[4] == '<RECENT-CTIME>':
  109 + # Recently installed packages
  110 + yield data[2], 8
93 111 class PopconXapianIndex(xapian.WritableDatabase,Singleton):
94 112 """
95 113 Data source for popcon submissions defined as a singleton xapian database.
... ... @@ -100,7 +118,8 @@ class PopconXapianIndex(xapian.WritableDatabase,Singleton):
100 118 """
101 119 self.path = os.path.expanduser(cfg.popcon_index)
102 120 self.popcon_dir = os.path.expanduser(cfg.popcon_dir)
103   - self.debtags_path = os.path.expanduser(cfg.tags_db)
  121 + #self.debtags_path = os.path.expanduser(cfg.tags_db)
  122 + self.axi = xapian.Database(cfg.axi)
104 123 self.load_index()
105 124  
106 125 def parse_submission(self,submission_path,binary=1):
... ... @@ -149,7 +168,6 @@ class PopconXapianIndex(xapian.WritableDatabase,Singleton):
149 168 """
150 169 if not os.path.exists(self.path):
151 170 os.makedirs(self.path)
152   - debtags_db = load_debtags_db(self.debtags_path) #[FIXME]
153 171  
154 172 try:
155 173 logging.info("Indexing popcon submissions from \'%s\'" %
... ... @@ -170,10 +188,10 @@ class PopconXapianIndex(xapian.WritableDatabase,Singleton):
170 188 logging.debug("Parsing popcon submission at \'%s\'" %
171 189 submission_path)
172 190 for pkg, freq in self.parse_submission(submission_path):
173   - doc.add_term(pkg,freq)
174   - #[FIXME] get tags from axi
175   - for tag in debtags_db.tags_of_package(pkg):
176   - doc.add_term("XT"+tag,freq)
  191 + doc.add_term("XP"+pkg,freq)
  192 + for tag in axi_search_pkg_tags(self.axi,pkg):
  193 + print tag
  194 + doc.add_term(tag,freq)
177 195 doc_id = self.add_document(doc)
178 196 logging.debug("Popcon Xapian: Indexing doc %d" % doc_id)
179 197 # python garbage collector
... ... @@ -181,14 +199,6 @@ class PopconXapianIndex(xapian.WritableDatabase,Singleton):
181 199 # flush to disk database changes
182 200 self.flush()
183 201  
184   -class PopconSubmission():
185   - def __init__(self,submission_hash):
186   - self.hash = submission_hash
187   - self.pkgs_list = []
188   -
189   - def add_pkg(self,pkg):
190   - self.pkgs_list.append(pkg)
191   -
192 202 class PopconClusteredData(Singleton):
193 203 """
194 204 Data source for popcon submissions defined as a singleton xapian database.
... ...