lib/sup/index.rb in sup-0.0.2 vs lib/sup/index.rb in sup-0.0.3
- old
+ new
@@ -1,33 +1,29 @@
## the index structure for redwood. interacts with ferret.
require 'thread'
require 'fileutils'
-require_gem 'ferret', ">= 0.10.13"
+require 'ferret'
+#require_gem 'ferret', ">= 0.10.13"
module Redwood
-class IndexError < StandardError
- attr_reader :source
-
- def initialize source, s
- super s
- @source = source
- end
-end
-
class Index
include Singleton
- attr_reader :index # debugging only
-
+ attr_reader :index
def initialize dir=BASE_DIR
@dir = dir
- @mutex = Mutex.new
@sources = {}
@sources_dirty = false
+ wsa = Ferret::Analysis::WhiteSpaceAnalyzer.new false
+ sa = Ferret::Analysis::StandardAnalyzer.new Ferret::Analysis::FULL_ENGLISH_STOP_WORDS, true
+ @analyzer = Ferret::Analysis::PerFieldAnalyzer.new wsa
+ @analyzer[:body] = sa
+ @qparser ||= Ferret::QueryParser.new :default_field => :body, :analyzer => @analyzer
+
self.class.i_am_the_instance self
end
def load
load_sources
@@ -49,22 +45,19 @@
@sources[source.id] = source
end
def source_for name; @sources.values.find { |s| s.is_source_for? name }; end
def usual_sources; @sources.values.find_all { |s| s.usual? }; end
+ def sources; @sources.values; end
def load_index dir=File.join(@dir, "ferret")
- wsa = Ferret::Analysis::WhiteSpaceAnalyzer.new false
- sa = Ferret::Analysis::StandardAnalyzer.new
- analyzer = Ferret::Analysis::PerFieldAnalyzer.new wsa
- analyzer[:body] = sa
-
if File.exists? dir
- Redwood::log "loading index"
- @index = Ferret::Index::Index.new(:path => dir, :analyzer => analyzer)
+ Redwood::log "loading index..."
+ @index = Ferret::Index::Index.new(:path => dir, :analyzer => @analyzer)
+ Redwood::log "loaded index of #{@index.size} messages"
else
- Redwood::log "creating index"
+ Redwood::log "creating index..."
field_infos = Ferret::Index::FieldInfos.new :store => :yes
field_infos.add_field :message_id
field_infos.add_field :source_id
field_infos.add_field :source_info
field_infos.add_field :date, :index => :untokenized
@@ -74,23 +67,25 @@
field_infos.add_field :from
field_infos.add_field :to
field_infos.add_field :refs
field_infos.add_field :snippet, :index => :no, :term_vector => :no
field_infos.create_index dir
- @index = Ferret::Index::Index.new(:path => dir, :analyzer => analyzer)
+ @index = Ferret::Index::Index.new(:path => dir, :analyzer => @analyzer)
end
end
## update the message by deleting and re-adding
def update_message m, source=nil, source_info=nil
docid, entry = load_entry_for_id m.id
if entry
source ||= entry[:source_id].to_i
source_info ||= entry[:source_info].to_i
end
- raise "no entry and no source info for message #{m.id}" unless source && source_info
+ ## this happens sometimes. i'm not sure why. ferret bug?
+ raise "no entry and no source info for message #{m.id}: source #{source.inspect}, info #{source_info.inspect}, entry #{entry.inspect}, query #{Ferret::Search::TermQuery.new(:message_id, m.id)}, results #{@index.search(Ferret::Search::TermQuery.new(:message_id, m.id)).inspect}" unless source && source_info
+
raise "deleting non-corresponding entry #{docid}" unless @index[docid][:message_id] == m.id
@index.delete docid
add_message m
end
@@ -107,11 +102,11 @@
## you should probably not call this on a block that doesn't break
## rather quickly because the results will probably be, as we say
## in scotland, frikkin' huuuge.
EACH_BY_DATE_NUM = 100
def each_id_by_date opts={}
- return if @index.size == 0 # otherwise ferret barfs
+ return if @index.size == 0 # otherwise ferret barfs ###TODO: remove this once my ferret patch is accepted
query = build_query opts
offset = 0
while true
results = @index.search(query, :sort => "date DESC", :limit => EACH_BY_DATE_NUM, :offset => offset)
Redwood::log "got #{results.total_hits} results for query (offset #{offset}) #{query.inspect}"
@@ -120,16 +115,19 @@
offset += EACH_BY_DATE_NUM
end
end
def num_results_for opts={}
- query = build_query opts
- x = @index.search(query).total_hits
- Redwood::log "num_results_for: have #{x} for query #{query}"
- x
+ return 0 if @index.size == 0 # otherwise ferret barfs ###TODO: remove this once my ferret patch is accepted
+ q = build_query opts
+ index.search(q).total_hits
end
+ ## yield all messages in the thread containing 'm' by repeatedly
+ ## querying the index. yields pairs of message ids and
+ ## message-building lambdas, so that building an unwanted message
+ ## can be skipped in the block if desired.
SAME_SUBJECT_DATE_LIMIT = 7
def each_message_in_thread_for m, opts={}
messages = {}
searched = {}
num_queries = 0
@@ -184,48 +182,22 @@
doc = @index[docid]
source = @sources[doc[:source_id].to_i]
#puts "building message #{doc[:message_id]} (#{source}##{doc[:source_info]})"
raise "invalid source #{doc[:source_id]}" unless source
- m =
- if source.broken?
- nil
- else
- begin
- Message.new :source => source, :source_info => doc[:source_info].to_i,
- :labels => doc[:label].split(" ").map { |s| s.intern },
- :snippet => doc[:snippet]
- rescue MessageFormatError => e
- raise IndexError.new(source, "error building message #{doc[:message_id]} at #{source}/#{doc[:source_info]}: #{e.message}")
- rescue SourceError => e
- nil
- end
- end
+ fake_header = {
+ "date" => Time.at(doc[:date].to_i),
+ "subject" => unwrap_subj(doc[:subject]),
+ "from" => doc[:from],
+ "to" => doc[:to],
+ "message-id" => doc[:message_id],
+ "references" => doc[:refs],
+ }
- unless m
- fake_header = {
- "date" => Time.at(doc[:date].to_i),
- "subject" => unwrap_subj(doc[:subject]),
- "from" => doc[:from],
- "to" => doc[:to],
- "message-id" => doc[:message_id],
- "references" => doc[:refs],
- }
-
- m = Message.new :labels => doc[:label].split(" ").map { |s| s.intern },
- :snippet => doc[:snippet], :header => fake_header,
- :body => <<EOS
-#{doc[:snippet]}...
-
-An error occurred while loading this message. It is possible that the source
-has changed, or (in the case of remote sources) is down.
-
-The error message was:
- #{source.broken_msg}
-EOS
- end
- m
+ Message.new :source => source, :source_info => doc[:source_info].to_i,
+ :labels => doc[:label].split(" ").map { |s| s.intern },
+ :snippet => doc[:snippet], :header => fake_header
end
def fresh_thread_id; @next_thread_id += 1; end
def wrap_subj subj; "__START_SUBJECT__ #{subj} __END_SUBJECT__"; end
def unwrap_subj subj; subj =~ /__START_SUBJECT__ (.*?) __END_SUBJECT__/ && $1; end
@@ -235,11 +207,11 @@
source_id =
if m.source.is_a? Integer
m.source
else
- m.source.id or raise "unregistered source #{m.source}"
+ m.source.id or raise "unregistered source #{m.source} (id #{m.source.id.inspect})"
end
to = (m.to + m.cc + m.bcc).map { |x| x.email }.join(" ")
d = {
:message_id => m.id,
@@ -303,25 +275,27 @@
contacts.keys.compact
end
protected
- ## TODO: convert this to query objects rather than strings
+ def parse_user_query_string str; @qparser.parse str; end
def build_query opts
- query = ""
- query += opts[:labels].map { |t| "+label:#{t}" }.join(" ") if opts[:labels]
- query += " +label:#{opts[:label]}" if opts[:label]
- query += " #{opts[:content]}" if opts[:content]
+ query = Ferret::Search::BooleanQuery.new
+ query.add_query opts[:qobj], :must if opts[:qobj]
+ labels = ([opts[:label]] + (opts[:labels] || [])).compact
+ labels.each { |t| query.add_query Ferret::Search::TermQuery.new("label", t.to_s), :must }
if opts[:participants]
- query += "+(" +
- opts[:participants].map { |p| "from:#{p.email} OR to:#{p.email}" }.join(" OR ") + ")"
+ q2 = Ferret::Search::BooleanQuery.new
+ opts[:participants].each do |p|
+ q2.add_query Ferret::Search::TermQuery.new("from", p.email), :should
+ q2.add_query Ferret::Search::TermQuery.new("to", p.email), :should
+ end
+ query.add_query q2, :must
end
- query += " -label:spam" unless opts[:load_spam] || opts[:labels] == :spam ||
- (opts[:labels] && opts[:labels].include?(:spam))
- query += " -label:killed" unless opts[:load_killed] || opts[:labels] == :killed ||
- (opts[:labels] && opts[:labels].include?(:killed))
+ query.add_query Ferret::Search::TermQuery.new("label", "spam"), :must_not unless opts[:load_spam] || labels.include?(:spam)
+ query.add_query Ferret::Search::TermQuery.new("label", "killed"), :must_not unless opts[:load_killed] || labels.include?(:killed)
query
end
def load_sources fn=Redwood::SOURCE_FN
@sources = Hash[*(Redwood::load_yaml_obj(fn) || []).map { |s| [s.id, s] }.flatten]
@@ -333,10 +307,10 @@
bakfn = fn + ".bak"
if File.exists? fn
File.chmod 0600, fn
FileUtils.mv fn, bakfn, :force => true unless File.exists?(bakfn) && File.size(bakfn) > File.size(fn)
end
- Redwood::save_yaml_obj @sources.values, fn
+ Redwood::save_yaml_obj @sources.values, fn
File.chmod 0600, fn
end
@sources_dirty = false
end
end