search.py

   1 # Tool for querying the search index that 0mirror builds when mirroring feeds.
   2 #
   3 # Copyright (C) 2011, Anders F Bjorklund
   4 # Copyright (C) 2013, Thomas Leonard
   5 # See the COPYING file for details, or visit http://0install.net.
   6 #
   7 # This version for 0mirror is based on original code for 0install:
   8 # http://thread.gmane.org/gmane.comp.file-systems.zero-install.devel/3847
   9
  10 from whoosh.index import open_dir
  11 from whoosh.qparser import MultifieldParser
  12
  13 from xml.sax.saxutils import escape, quoteattr
  14
  15 import os, sys
  16
  17 class Searcher:
  18         def __init__(self, index_dir):
  19                 index = open_dir(index_dir)
  20                 self.searcher = index.searcher()
  21
  22                 fields = ["name", "summary", "description"]
  23                 self.parser = MultifieldParser(fields, schema=index.schema)
  24
  25         def query(self, query_string, out):
  26                 """out should take unicode and encode it as necessary"""
  27                 query = self.parser.parse(unicode(query_string))
  28                 results = self.searcher.search(query)
  29                 max_score = results and max([result.score for result in results]) or 0.0
  30
  31                 out.write('<?xml version="1.0" ?>')
  32                 out.write("<results>")
  33                 for result in results:
  34                         uri = result["uri"]
  35                         name = result["name"]
  36                         summary = result.get("summary", "")
  37                         category = result.get("category", None)
  38
  39                         s = 100.0 * result.score / max_score
  40
  41                         out.write("<result uri=%s name=%s score='%d'>" % (quoteattr(uri), quoteattr(name), s))
  42                         out.write("<summary>%s</summary>" % escape(summary))
  43                         if category:
  44                                 out.write("<category>%s</category>" % escape(category))
  45                         out.write("</result>")
  46                 out.write("</results>")