Fixed whoosh index creation to work with 0.2.x. #15

[zzz-pokedex.git] / pokedex / lookup.py

diff --git a/pokedex/lookup.py b/pokedex/lookup.py
index b50871b..2b38a1a 100644 (file)
--- a/pokedex/lookup.py
+++ b/pokedex/lookup.py
@@ -1,8 +1,12 @@
 # encoding: utf8
 # encoding: utf8

+import os, os.path

 import re
 
 from sqlalchemy.sql import func
 import whoosh
 import re
 
 from sqlalchemy.sql import func
 import whoosh

+import whoosh.filedb.filestore
+import whoosh.filedb.fileindex
+import whoosh.index

 from whoosh.qparser import QueryParser
 import whoosh.spelling
 
 from whoosh.qparser import QueryParser
 import whoosh.spelling
 


@@ -13,13 +17,20 @@ import pokedex.db.tables as tables
 # retrieve something from the index
 indexed_tables = {}
 for cls in [
 # retrieve something from the index
 indexed_tables = {}
 for cls in [

+        tables.Ability,
+        tables.Item,
+        tables.Move,

         tables.Pokemon,
         tables.Pokemon,

+        tables.Type,

     ]:
     indexed_tables[cls.__tablename__] = cls
 
 # Dictionary of extra keys to file types of objects under, e.g. Pokémon can
 # also be looked up purely by number
 extra_keys = {
     ]:
     indexed_tables[cls.__tablename__] = cls
 
 # Dictionary of extra keys to file types of objects under, e.g. Pokémon can
 # also be looked up purely by number
 extra_keys = {

+    tables.Move: [
+        lambda row: u"move %d" % row.id,
+    ],

     tables.Pokemon: [
         lambda row: unicode(row.id),
     ],
     tables.Pokemon: [
         lambda row: unicode(row.id),
     ],


@@ -35,16 +46,22 @@ def get_index(session):
     if index_bits:
         return index_bits['index'], index_bits['speller']
 
     if index_bits:
         return index_bits['index'], index_bits['speller']
 

-    store = whoosh.store.RamStorage()
+    store = whoosh.filedb.filestore.RamStorage()

     schema = whoosh.fields.Schema(
         name=whoosh.fields.ID(stored=True),
         table=whoosh.fields.STORED,
         row_id=whoosh.fields.STORED,
     schema = whoosh.fields.Schema(
         name=whoosh.fields.ID(stored=True),
         table=whoosh.fields.STORED,
         row_id=whoosh.fields.STORED,

-        language_id=whoosh.fields.STORED,
+        language=whoosh.fields.STORED,
+
+        # Whoosh 0.2 explodes when using a file-stored schema with no TEXT
+        # columns.  Appease it
+        dummy=whoosh.fields.TEXT,

     )
 
     )
 

-    # Construct a straight lookup index
-    index = whoosh.index.Index(store, schema=schema, create=True)
+    index_directory = '/var/tmp/pokedex'
+    if not os.path.exists(index_directory):
+        os.mkdir(index_directory)
+    index = whoosh.index.create_in(index_directory, schema=schema)

     writer = index.writer()
 
     # Index every name in all our tables of interest
     writer = index.writer()
 
     # Index every name in all our tables of interest


@@ -69,12 +86,20 @@ def get_index(session):
 
             speller_entries.append(name)
 
 
             speller_entries.append(name)
 

-            for extra_key_func in extra_keys[cls]:
+            for extra_key_func in extra_keys.get(cls, []):

                 extra_key = extra_key_func(row)
                 writer.add_document(name=extra_key, **row_key)
 
     writer.commit()
 
                 extra_key = extra_key_func(row)
                 writer.add_document(name=extra_key, **row_key)
 
     writer.commit()
 

+    # XXX GIHWEGREHKG
+    old__schema = whoosh.spelling.SpellChecker._schema
+    def new__schema(self):
+        schema = old__schema(self)
+        schema.add('dummy', whoosh.fields.TEXT)
+        return schema
+    whoosh.spelling.SpellChecker._schema = new__schema
+

     # Construct and populate a spell-checker index.  Quicker to do it all
     # at once, as every call to add_* does a commit(), and those seem to be
     # expensive
     # Construct and populate a spell-checker index.  Quicker to do it all
     # at once, as every call to add_* does a commit(), and those seem to be
     # expensive


@@ -86,7 +111,7 @@ def get_index(session):
     # complications.
     # The below is copied from SpellChecker.add_scored_words without the check
     # for isalpha().  XXX get whoosh patched to make this unnecessary!
     # complications.
     # The below is copied from SpellChecker.add_scored_words without the check
     # for isalpha().  XXX get whoosh patched to make this unnecessary!

-    writer = whoosh.writing.IndexWriter(speller.index())
+    writer = speller.index(create=True).writer()

     for word in speller_entries:
         fields = {"word": word, "score": 1}
         for size in xrange(speller.mingram, speller.maxgram + 1):
     for word in speller_entries:
         fields = {"word": word, "score": 1}
         for size in xrange(speller.mingram, speller.maxgram + 1):


@@ -126,7 +151,7 @@ def lookup(session, name, exact_only=False):
     # Look for exact name.  A Term object does an exact match, so we don't have
     # to worry about a query parser tripping on weird characters in the input
     searcher = index.searcher()
     # Look for exact name.  A Term object does an exact match, so we don't have
     # to worry about a query parser tripping on weird characters in the input
     searcher = index.searcher()

-    query = whoosh.query.Term('name', name)
+    query = whoosh.query.Term('name', name.lower())

     results = searcher.search(query)
 
     if not exact_only:
     results = searcher.search(query)
 
     if not exact_only: