muck-solr 0.4.0
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGE_LOG +239 -0
- data/LICENSE +19 -0
- data/README.markdown +118 -0
- data/README.rdoc +107 -0
- data/Rakefile +99 -0
- data/TESTING_THE_PLUGIN +25 -0
- data/VERSION.yml +4 -0
- data/config/solr.yml +15 -0
- data/config/solr_environment.rb +32 -0
- data/lib/acts_as_solr.rb +65 -0
- data/lib/acts_as_solr/acts_methods.rb +352 -0
- data/lib/acts_as_solr/class_methods.rb +236 -0
- data/lib/acts_as_solr/common_methods.rb +89 -0
- data/lib/acts_as_solr/deprecation.rb +61 -0
- data/lib/acts_as_solr/instance_methods.rb +165 -0
- data/lib/acts_as_solr/lazy_document.rb +18 -0
- data/lib/acts_as_solr/parser_methods.rb +203 -0
- data/lib/acts_as_solr/search_results.rb +68 -0
- data/lib/acts_as_solr/solr_fixtures.rb +13 -0
- data/lib/acts_as_solr/tasks.rb +10 -0
- data/lib/acts_as_solr/tasks/database.rake +16 -0
- data/lib/acts_as_solr/tasks/solr.rake +135 -0
- data/lib/acts_as_solr/tasks/test.rake +5 -0
- data/lib/solr.rb +26 -0
- data/lib/solr/connection.rb +177 -0
- data/lib/solr/document.rb +75 -0
- data/lib/solr/exception.rb +13 -0
- data/lib/solr/field.rb +36 -0
- data/lib/solr/importer.rb +19 -0
- data/lib/solr/importer/array_mapper.rb +26 -0
- data/lib/solr/importer/delimited_file_source.rb +38 -0
- data/lib/solr/importer/hpricot_mapper.rb +27 -0
- data/lib/solr/importer/mapper.rb +51 -0
- data/lib/solr/importer/solr_source.rb +41 -0
- data/lib/solr/importer/xpath_mapper.rb +35 -0
- data/lib/solr/indexer.rb +52 -0
- data/lib/solr/request.rb +26 -0
- data/lib/solr/request/add_document.rb +58 -0
- data/lib/solr/request/base.rb +36 -0
- data/lib/solr/request/commit.rb +29 -0
- data/lib/solr/request/delete.rb +48 -0
- data/lib/solr/request/dismax.rb +46 -0
- data/lib/solr/request/index_info.rb +22 -0
- data/lib/solr/request/modify_document.rb +46 -0
- data/lib/solr/request/optimize.rb +19 -0
- data/lib/solr/request/ping.rb +36 -0
- data/lib/solr/request/select.rb +54 -0
- data/lib/solr/request/spellcheck.rb +30 -0
- data/lib/solr/request/standard.rb +402 -0
- data/lib/solr/request/update.rb +23 -0
- data/lib/solr/response.rb +27 -0
- data/lib/solr/response/add_document.rb +17 -0
- data/lib/solr/response/base.rb +42 -0
- data/lib/solr/response/commit.rb +15 -0
- data/lib/solr/response/delete.rb +13 -0
- data/lib/solr/response/dismax.rb +8 -0
- data/lib/solr/response/index_info.rb +26 -0
- data/lib/solr/response/modify_document.rb +17 -0
- data/lib/solr/response/optimize.rb +14 -0
- data/lib/solr/response/ping.rb +26 -0
- data/lib/solr/response/ruby.rb +42 -0
- data/lib/solr/response/select.rb +17 -0
- data/lib/solr/response/spellcheck.rb +20 -0
- data/lib/solr/response/standard.rb +60 -0
- data/lib/solr/response/xml.rb +39 -0
- data/lib/solr/solrtasks.rb +27 -0
- data/lib/solr/util.rb +32 -0
- data/lib/solr/xml.rb +44 -0
- data/solr/CHANGES.txt +1207 -0
- data/solr/LICENSE.txt +712 -0
- data/solr/NOTICE.txt +90 -0
- data/solr/etc/jetty.xml +205 -0
- data/solr/etc/webdefault.xml +379 -0
- data/solr/lib/easymock.jar +0 -0
- data/solr/lib/jetty-6.1.3.jar +0 -0
- data/solr/lib/jetty-util-6.1.3.jar +0 -0
- data/solr/lib/jsp-2.1/ant-1.6.5.jar +0 -0
- data/solr/lib/jsp-2.1/core-3.1.1.jar +0 -0
- data/solr/lib/jsp-2.1/jsp-2.1.jar +0 -0
- data/solr/lib/jsp-2.1/jsp-api-2.1.jar +0 -0
- data/solr/lib/servlet-api-2.4.jar +0 -0
- data/solr/lib/servlet-api-2.5-6.1.3.jar +0 -0
- data/solr/lib/xpp3-1.1.3.4.O.jar +0 -0
- data/solr/solr/README.txt +52 -0
- data/solr/solr/bin/abc +176 -0
- data/solr/solr/bin/abo +176 -0
- data/solr/solr/bin/backup +108 -0
- data/solr/solr/bin/backupcleaner +142 -0
- data/solr/solr/bin/commit +128 -0
- data/solr/solr/bin/optimize +129 -0
- data/solr/solr/bin/readercycle +129 -0
- data/solr/solr/bin/rsyncd-disable +77 -0
- data/solr/solr/bin/rsyncd-enable +76 -0
- data/solr/solr/bin/rsyncd-start +145 -0
- data/solr/solr/bin/rsyncd-stop +105 -0
- data/solr/solr/bin/scripts-util +83 -0
- data/solr/solr/bin/snapcleaner +148 -0
- data/solr/solr/bin/snapinstaller +168 -0
- data/solr/solr/bin/snappuller +248 -0
- data/solr/solr/bin/snappuller-disable +77 -0
- data/solr/solr/bin/snappuller-enable +77 -0
- data/solr/solr/bin/snapshooter +109 -0
- data/solr/solr/conf/admin-extra.html +31 -0
- data/solr/solr/conf/protwords.txt +21 -0
- data/solr/solr/conf/schema.xml +126 -0
- data/solr/solr/conf/scripts.conf +24 -0
- data/solr/solr/conf/solrconfig.xml +458 -0
- data/solr/solr/conf/stopwords.txt +57 -0
- data/solr/solr/conf/synonyms.txt +31 -0
- data/solr/solr/conf/xslt/example.xsl +132 -0
- data/solr/solr/conf/xslt/example_atom.xsl +63 -0
- data/solr/solr/conf/xslt/example_rss.xsl +62 -0
- data/solr/start.jar +0 -0
- data/solr/webapps/solr.war +0 -0
- data/test/config/solr.yml +2 -0
- data/test/db/connections/mysql/connection.rb +10 -0
- data/test/db/connections/sqlite/connection.rb +8 -0
- data/test/db/migrate/001_create_books.rb +15 -0
- data/test/db/migrate/002_create_movies.rb +12 -0
- data/test/db/migrate/003_create_categories.rb +11 -0
- data/test/db/migrate/004_create_electronics.rb +16 -0
- data/test/db/migrate/005_create_authors.rb +12 -0
- data/test/db/migrate/006_create_postings.rb +9 -0
- data/test/db/migrate/007_create_posts.rb +13 -0
- data/test/db/migrate/008_create_gadgets.rb +11 -0
- data/test/fixtures/authors.yml +9 -0
- data/test/fixtures/books.yml +13 -0
- data/test/fixtures/categories.yml +7 -0
- data/test/fixtures/db_definitions/mysql.sql +41 -0
- data/test/fixtures/electronics.yml +49 -0
- data/test/fixtures/movies.yml +9 -0
- data/test/fixtures/postings.yml +10 -0
- data/test/functional/acts_as_solr_test.rb +413 -0
- data/test/functional/association_indexing_test.rb +37 -0
- data/test/functional/faceted_search_test.rb +163 -0
- data/test/functional/multi_solr_search_test.rb +57 -0
- data/test/models/author.rb +10 -0
- data/test/models/book.rb +10 -0
- data/test/models/category.rb +8 -0
- data/test/models/electronic.rb +25 -0
- data/test/models/gadget.rb +9 -0
- data/test/models/movie.rb +17 -0
- data/test/models/novel.rb +2 -0
- data/test/models/post.rb +3 -0
- data/test/models/posting.rb +11 -0
- data/test/test_helper.rb +54 -0
- data/test/unit/acts_methods_shoulda.rb +68 -0
- data/test/unit/class_methods_shoulda.rb +85 -0
- data/test/unit/common_methods_shoulda.rb +111 -0
- data/test/unit/instance_methods_shoulda.rb +318 -0
- data/test/unit/lazy_document_shoulda.rb +34 -0
- data/test/unit/parser_instance.rb +19 -0
- data/test/unit/parser_methods_shoulda.rb +268 -0
- data/test/unit/solr_instance.rb +49 -0
- data/test/unit/test_helper.rb +24 -0
- metadata +241 -0
data/lib/solr.rb
ADDED
@@ -0,0 +1,26 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
module Solr; end
|
14
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/exception")
|
15
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/request")
|
16
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/connection")
|
17
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/response")
|
18
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/util")
|
19
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/xml")
|
20
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/importer")
|
21
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/indexer")
|
22
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/xml")
|
23
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/field")
|
24
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/request/base")
|
25
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/document")
|
26
|
+
require File.expand_path("#{File.dirname(__FILE__)}/solr/request/update")
|
@@ -0,0 +1,177 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
# TODO: add a convenience method to POST a Solr .xml file, like Solr's example post.sh
|
14
|
+
|
15
|
+
class Solr::Connection
|
16
|
+
attr_reader :url, :autocommit, :connection
|
17
|
+
|
18
|
+
# create a connection to a solr instance using the url for the solr
|
19
|
+
# application context:
|
20
|
+
#
|
21
|
+
# conn = Solr::Connection.new("http://example.com:8080/solr")
|
22
|
+
#
|
23
|
+
# if you would prefer to have all adds/updates autocommitted,
|
24
|
+
# use :autocommit => :on
|
25
|
+
#
|
26
|
+
# conn = Solr::Connection.new('http://example.com:8080/solr',
|
27
|
+
# :autocommit => :on)
|
28
|
+
|
29
|
+
def initialize(url="http://localhost:8983/solr", opts={})
|
30
|
+
@url = URI.parse(url)
|
31
|
+
unless @url.kind_of? URI::HTTP
|
32
|
+
raise "invalid http url: #{url}"
|
33
|
+
end
|
34
|
+
|
35
|
+
# TODO: Autocommit seems nice at one level, but it currently is confusing because
|
36
|
+
# only calls to Connection#add/#update/#delete, though a Connection#send(AddDocument.new(...))
|
37
|
+
# does not autocommit. Maybe #send should check for the request types that require a commit and
|
38
|
+
# commit in #send instead of the individual methods?
|
39
|
+
@autocommit = opts[:autocommit] == :on
|
40
|
+
|
41
|
+
# Not actually opening the connection yet, just setting up the persistent connection.
|
42
|
+
@connection = Net::HTTP.new(@url.host, @url.port)
|
43
|
+
|
44
|
+
@connection.read_timeout = opts[:timeout] if opts[:timeout]
|
45
|
+
end
|
46
|
+
|
47
|
+
# add a document to the index. you can pass in either a hash
|
48
|
+
#
|
49
|
+
# conn.add(:id => 123, :title => 'Tlon, Uqbar, Orbis Tertius')
|
50
|
+
#
|
51
|
+
# or a Solr::Document
|
52
|
+
#
|
53
|
+
# conn.add(Solr::Document.new(:id => 123, :title = 'On Writing')
|
54
|
+
#
|
55
|
+
# true/false will be returned to designate success/failure
|
56
|
+
|
57
|
+
def add(doc)
|
58
|
+
request = Solr::Request::AddDocument.new(doc)
|
59
|
+
response = send(request)
|
60
|
+
commit if @autocommit
|
61
|
+
return response.ok?
|
62
|
+
end
|
63
|
+
|
64
|
+
# update a document in the index (really just an alias to add)
|
65
|
+
|
66
|
+
def update(doc)
|
67
|
+
return add(doc)
|
68
|
+
end
|
69
|
+
|
70
|
+
# performs a standard query and returns a Solr::Response::Standard
|
71
|
+
#
|
72
|
+
# response = conn.query('borges')
|
73
|
+
#
|
74
|
+
# alternative you can pass in a block and iterate over hits
|
75
|
+
#
|
76
|
+
# conn.query('borges') do |hit|
|
77
|
+
# puts hit
|
78
|
+
# end
|
79
|
+
#
|
80
|
+
# options include:
|
81
|
+
#
|
82
|
+
# :sort, :default_field, :rows, :filter_queries, :debug_query,
|
83
|
+
# :explain_other, :facets, :highlighting, :mlt,
|
84
|
+
# :operator => :or / :and
|
85
|
+
# :start => defaults to 0
|
86
|
+
# :field_list => array, defaults to ["*", "score"]
|
87
|
+
|
88
|
+
def query(query, options={}, &action)
|
89
|
+
# TODO: Shouldn't this return an exception if the Solr status is not ok? (rather than true/false).
|
90
|
+
create_and_send_query(Solr::Request::Standard, options.update(:query => query), &action)
|
91
|
+
end
|
92
|
+
|
93
|
+
# performs a dismax search and returns a Solr::Response::Standard
|
94
|
+
#
|
95
|
+
# response = conn.search('borges')
|
96
|
+
#
|
97
|
+
# options are same as query, but also include:
|
98
|
+
#
|
99
|
+
# :tie_breaker, :query_fields, :minimum_match, :phrase_fields,
|
100
|
+
# :phrase_slop, :boost_query, :boost_functions
|
101
|
+
|
102
|
+
def search(query, options={}, &action)
|
103
|
+
create_and_send_query(Solr::Request::Dismax, options.update(:query => query), &action)
|
104
|
+
end
|
105
|
+
|
106
|
+
# sends a commit message to the server
|
107
|
+
def commit(options={})
|
108
|
+
response = send(Solr::Request::Commit.new(options))
|
109
|
+
return response.ok?
|
110
|
+
end
|
111
|
+
|
112
|
+
# sends an optimize message to the server
|
113
|
+
def optimize
|
114
|
+
response = send(Solr::Request::Optimize.new)
|
115
|
+
return response.ok?
|
116
|
+
end
|
117
|
+
|
118
|
+
# pings the connection and returns true/false if it is alive or not
|
119
|
+
def ping
|
120
|
+
begin
|
121
|
+
response = send(Solr::Request::Ping.new)
|
122
|
+
return response.ok?
|
123
|
+
rescue
|
124
|
+
return false
|
125
|
+
end
|
126
|
+
end
|
127
|
+
|
128
|
+
# delete a document from the index using the document id
|
129
|
+
def delete(document_id)
|
130
|
+
response = send(Solr::Request::Delete.new(:id => document_id))
|
131
|
+
commit if @autocommit
|
132
|
+
response.ok?
|
133
|
+
end
|
134
|
+
|
135
|
+
# delete using a query
|
136
|
+
def delete_by_query(query)
|
137
|
+
response = send(Solr::Request::Delete.new(:query => query))
|
138
|
+
commit if @autocommit
|
139
|
+
response.ok?
|
140
|
+
end
|
141
|
+
|
142
|
+
def info
|
143
|
+
send(Solr::Request::IndexInfo.new)
|
144
|
+
end
|
145
|
+
|
146
|
+
# send a given Solr::Request and return a RubyResponse or XmlResponse
|
147
|
+
# depending on the type of request
|
148
|
+
def send(request)
|
149
|
+
data = post(request)
|
150
|
+
Solr::Response::Base.make_response(request, data)
|
151
|
+
end
|
152
|
+
|
153
|
+
# send the http post request to solr; for convenience there are shortcuts
|
154
|
+
# to some requests: add(), query(), commit(), delete() or send()
|
155
|
+
def post(request)
|
156
|
+
response = @connection.post(@url.path + "/" + request.handler,
|
157
|
+
request.to_s,
|
158
|
+
{ "Content-Type" => request.content_type })
|
159
|
+
|
160
|
+
case response
|
161
|
+
when Net::HTTPSuccess then response.body
|
162
|
+
else
|
163
|
+
response.error!
|
164
|
+
end
|
165
|
+
|
166
|
+
end
|
167
|
+
|
168
|
+
private
|
169
|
+
|
170
|
+
def create_and_send_query(klass, options = {}, &action)
|
171
|
+
request = klass.new(options)
|
172
|
+
response = send(request)
|
173
|
+
return response unless action
|
174
|
+
response.each {|hit| action.call(hit)}
|
175
|
+
end
|
176
|
+
|
177
|
+
end
|
@@ -0,0 +1,75 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
class Solr::Document
|
14
|
+
include Enumerable
|
15
|
+
attr_accessor :boost
|
16
|
+
attr_reader :fields
|
17
|
+
|
18
|
+
# Create a new Solr::Document, optionally passing in a hash of
|
19
|
+
# key/value pairs for the fields
|
20
|
+
#
|
21
|
+
# doc = Solr::Document.new(:creator => 'Jorge Luis Borges')
|
22
|
+
def initialize(hash={})
|
23
|
+
@fields = []
|
24
|
+
self << hash
|
25
|
+
end
|
26
|
+
|
27
|
+
# Append a Solr::Field
|
28
|
+
#
|
29
|
+
# doc << Solr::Field.new(:creator => 'Jorge Luis Borges')
|
30
|
+
#
|
31
|
+
# If you are truly lazy you can simply pass in a hash:
|
32
|
+
#
|
33
|
+
# doc << {:creator => 'Jorge Luis Borges'}
|
34
|
+
def <<(fields)
|
35
|
+
case fields
|
36
|
+
when Hash
|
37
|
+
fields.each_pair do |name,value|
|
38
|
+
if value.respond_to?(:each) && !value.is_a?(String)
|
39
|
+
value.each {|v| @fields << Solr::Field.new(name => v)}
|
40
|
+
else
|
41
|
+
@fields << Solr::Field.new(name => value)
|
42
|
+
end
|
43
|
+
end
|
44
|
+
when Solr::Field
|
45
|
+
@fields << fields
|
46
|
+
else
|
47
|
+
raise "must pass in Solr::Field or Hash"
|
48
|
+
end
|
49
|
+
end
|
50
|
+
|
51
|
+
# shorthand to allow hash lookups
|
52
|
+
# doc['name']
|
53
|
+
def [](name)
|
54
|
+
field = @fields.find {|f| f.name == name.to_s}
|
55
|
+
return field.value if field
|
56
|
+
return nil
|
57
|
+
end
|
58
|
+
|
59
|
+
# shorthand to assign as a hash
|
60
|
+
def []=(name,value)
|
61
|
+
@fields << Solr::Field.new(name => value)
|
62
|
+
end
|
63
|
+
|
64
|
+
# convert the Document to a REXML::Element
|
65
|
+
def to_xml
|
66
|
+
e = Solr::XML::Element.new 'doc'
|
67
|
+
e.attributes['boost'] = @boost.to_s if @boost
|
68
|
+
@fields.each {|f| e.add_element(f.to_xml)}
|
69
|
+
return e
|
70
|
+
end
|
71
|
+
|
72
|
+
def each(*args, &blk)
|
73
|
+
fields.each(&blk)
|
74
|
+
end
|
75
|
+
end
|
@@ -0,0 +1,13 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
class Solr::Exception < Exception; end
|
data/lib/solr/field.rb
ADDED
@@ -0,0 +1,36 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
class Solr::Field
|
14
|
+
VALID_PARAMS = [:boost]
|
15
|
+
attr_accessor :name
|
16
|
+
attr_accessor :value
|
17
|
+
attr_accessor :boost
|
18
|
+
|
19
|
+
# Accepts an optional <tt>:boost</tt> parameter, used to boost the relevance of a particular field.
|
20
|
+
def initialize(params)
|
21
|
+
@boost = params[:boost]
|
22
|
+
name_key = (params.keys - VALID_PARAMS).first
|
23
|
+
@name, @value = name_key.to_s, params[name_key]
|
24
|
+
# Convert any Time values into UTC/XML schema format (which Solr requires).
|
25
|
+
@value = @value.respond_to?(:utc) ? @value.utc.xmlschema : @value.to_s
|
26
|
+
end
|
27
|
+
|
28
|
+
def to_xml
|
29
|
+
e = Solr::XML::Element.new 'field'
|
30
|
+
e.attributes['name'] = @name
|
31
|
+
e.attributes['boost'] = @boost.to_s if @boost
|
32
|
+
e.text = @value
|
33
|
+
return e
|
34
|
+
end
|
35
|
+
|
36
|
+
end
|
@@ -0,0 +1,19 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
module Solr; module Importer; end; end
|
14
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/mapper")
|
15
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/array_mapper")
|
16
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/delimited_file_source")
|
17
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/hpricot_mapper")
|
18
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/xpath_mapper")
|
19
|
+
require File.expand_path("#{File.dirname(__FILE__)}/importer/solr_source")
|
@@ -0,0 +1,26 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
|
14
|
+
|
15
|
+
class Solr::Importer::ArrayMapper < Solr::Importer::Mapper
|
16
|
+
# TODO document that initializer takes an array of Mappers [mapper1, mapper2, ... mapperN]
|
17
|
+
|
18
|
+
# TODO: make merge conflict handling configurable. as is, the last map fields win.
|
19
|
+
def map(orig_data_array)
|
20
|
+
mapped_data = {}
|
21
|
+
orig_data_array.each_with_index do |data,i|
|
22
|
+
mapped_data.merge!(@mapping[i].map(data))
|
23
|
+
end
|
24
|
+
mapped_data
|
25
|
+
end
|
26
|
+
end
|
@@ -0,0 +1,38 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
# For files with the first line containing field names
|
14
|
+
# Currently not designed for enormous files, as all lines are
|
15
|
+
# read into an array
|
16
|
+
class Solr::Importer::DelimitedFileSource
|
17
|
+
include Enumerable
|
18
|
+
|
19
|
+
def initialize(filename, splitter=/\t/)
|
20
|
+
@filename = filename
|
21
|
+
@splitter = splitter
|
22
|
+
end
|
23
|
+
|
24
|
+
def each
|
25
|
+
lines = IO.readlines(@filename)
|
26
|
+
headers = lines[0].split(@splitter).collect{|h| h.chomp}
|
27
|
+
|
28
|
+
lines[1..-1].each do |line|
|
29
|
+
data = headers.zip(line.split(@splitter).collect{|s| s.chomp})
|
30
|
+
def data.[](key)
|
31
|
+
self.assoc(key.to_s)[1]
|
32
|
+
end
|
33
|
+
|
34
|
+
yield(data)
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
end
|
@@ -0,0 +1,27 @@
|
|
1
|
+
# The ASF licenses this file to You under the Apache License, Version 2.0
|
2
|
+
# (the "License"); you may not use this file except in compliance with
|
3
|
+
# the License. You may obtain a copy of the License at
|
4
|
+
#
|
5
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
6
|
+
#
|
7
|
+
# Unless required by applicable law or agreed to in writing, software
|
8
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
9
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
10
|
+
# See the License for the specific language governing permissions and
|
11
|
+
# limitations under the License.
|
12
|
+
|
13
|
+
begin
|
14
|
+
require 'hpricot'
|
15
|
+
|
16
|
+
class Solr::Importer::HpricotMapper < Solr::Importer::Mapper
|
17
|
+
def field_data(doc, path)
|
18
|
+
doc.search(path.to_s).collect { |e| e.inner_html }
|
19
|
+
end
|
20
|
+
end
|
21
|
+
rescue LoadError => e # If we can't load hpricot
|
22
|
+
class Solr::Importer::HpricotMapper
|
23
|
+
def initialize(mapping, options={})
|
24
|
+
raise "Hpricot not installed."
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|