hutch-xamplr 1.0.2 → 1.0.3

data/CHANGES.txt

@@ -0,0 +1,13 @@
+
+2009-04-16 (version 1.0.2)
+
+  -- if tag starts with a number, prefix it with an 'x' (e.g. 0abc --> x0abc).
+     The problem arose if a tag like <0abc> was used in an xml file
+     used to generate xampl code. In that case, the 'attribute
+     names' associated with the tag would be based on '_abc'. The
+     change makes the attribute names be based on x0abc. The choice
+     of 'x' is arbitrary, but we can't have a leading digit, and
+     the '_' could fail to distinguish certain elements (e.g. 0abc
+     and 1abc would both be based on '_abc').
+
+

data/VERSION.yml

@@ -1,4 +1,4 @@
 --- 
-:patch: 2
+:patch: 3
 :major: 1
 :minor: 0

data/examples/random-people/people.rb

@@ -2,8 +2,8 @@ $LOAD_PATH.unshift("xampl_generated_code")
 
 require 'RandomPeople'
 
-Xampl.set_default_persister_kind(:tokyo_cabinet)
-Xampl.set_default_persister_format(:xml_format)
+#Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
 
 module RandomPeople
 

data/examples/random-people/query.rb

@@ -28,14 +28,16 @@ module RandomPeople
     found3 = []
 
     start_query_at = Time.now
-#    found1 = Xampl.transaction("random-people") do
-#      Xampl.query do | q |
-#        q.add_condition('city', :equals, 'TORONTO')
-#        q.add_condition('email', :ends_with, 'dodgit.com')
-#
-#        q.order_by('surname', :strasc)
-#      end
-#    end
+    found1 = Xampl.transaction("random-people") do
+      Xampl.query do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('email', :ends_with, 'dodgit.com')
+
+#        q.setlimit(10, 10)
+
+        q.order_by('surname', :strasc)
+      end
+    end
 
     start_find_at = Time.now
     found2 = Xampl.transaction("random-people") do
@@ -54,7 +56,7 @@ module RandomPeople
     found3 = Xampl.transaction("random-people") do
       Xampl.find_pids do | q |
         q.add_condition('city', :equals, 'TORONTO')
-#        q.add_condition('email', :ends_with, 'dodgit.com')
+        q.add_condition('email', :ends_with, 'dodgit.com')
 
 #        q.setlimit(10, 10)
 
@@ -66,7 +68,10 @@ module RandomPeople
     #  found1.each do | person_meta |
     #    puts "surname: #{ person_meta['surname']}, given: #{ person_meta['xampl'].given_name }"
     #  end
-    puts "found:: query #{ found1.size }, find_xampl: #{ found2.size }, find_pids: #{ found3.size }"
-    puts "query in: #{ start_find_at - start_query_at }, find_xampl in: #{ start_find_pids_at - start_find_at }, pids in: #{ done - start_find_pids_at}\n"
+
+    puts
+    puts "query    :: found: #{ found1.size }, in: #{ start_find_at - start_query_at }"
+    puts "find     :: found: #{ found2.size }, in: #{ start_find_pids_at - start_find_at }"
+    puts "find pids:: found: #{ found3.size }, in: #{ done - start_find_pids_at}"
   end
 end

data/examples/random-people-shared-addresses/Makefile

@@ -0,0 +1,16 @@
+
+nope:
+
+raw:
+	rm -rf repo/raw
+	time ruby rawtc.rb
+	#time ruby rawtc.rb
+	#time ruby rawtc.rb
+	#time ruby rawtc.rb
+	#time ruby rawtc.rb
+	#time ruby rawtc.rb
+
+batch:
+	rm -rf repo/random-people
+	time ruby batch-load-users.rb
+

data/examples/random-people-shared-addresses/batch-load-users.rb

@@ -0,0 +1,86 @@
+
+require 'fastercsv'
+require 'xampl_generated_code/RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  start_at = Time.now
+
+  arr_of_arrs = FasterCSV.read("random-names.csv")
+
+  parsed_at = Time.now
+
+  #GivenName,Surname,StreetAddress,City,State,ZipCode,EmailAddress,TelephoneNumber
+
+  #<people pid=''
+  #        xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            email=''
+  #            phone=''>
+  #        <address pid=''
+  #                 street-address=''
+  #                 city=''
+  #                 state=''
+  #                 postal-code=''/>
+  #    </person>
+  #</people>
+
+  base = 0
+
+  created_addresses = 0
+  shared_addresses = 0
+
+  10.times do | iter |
+    inner_start = Time.now
+    Xampl.transaction("random-people") do
+
+      base += arr_of_arrs.size
+
+      arr_of_arrs.each_with_index do | row, i |
+        person_pid = "person-#{ base + i }"
+        person = Person.new(person_pid)
+
+        person.given_name = row[0]
+        person.surname = row[1]
+        person.email = row[6]
+        person.phone = row[7]
+
+        addresses = Xampl.find_xampl do | q |
+          q.add_condition('class', :equals, Address.name)
+
+          q.add_condition('street-address', :equals, row[2])
+          q.add_condition('postal-code',    :equals, row[5])
+        end
+
+        address = addresses.first
+        if address then
+          person << address
+          shared_addresses += 1
+        else
+          address = person.new_address("address-#{ person_pid }")
+          address.street_address = row[2]
+          address.city = row[3]
+          address.state = row[4]
+          address.postal_code = row[5]
+          created_addresses += 1
+        end
+
+      end
+      puts "transaction ending..."
+    end
+    puts "iter: #{ iter } in #{ Time.now - inner_start }"
+  end
+
+  processed_at = Time.now
+
+  p "parsed in #{ parsed_at - start_at }, processed in: #{ processed_at - parsed_at }"
+  puts "   created addresses: #{ created_addresses }, shared: #{ shared_addresses }"
+
+end

data/examples/random-people-shared-addresses/find-mentions.rb

@@ -0,0 +1,50 @@
+
+require 'xampl_generated_code/RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  #<people pid=''
+  #        xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            email=''
+  #            phone=''>
+  #        <address pid=''
+  #                 street-address=''
+  #                 city=''
+  #                 state=''
+  #                 postal-code=''/>
+  #    </person>
+  #</people>
+
+  10.times do
+
+    people = Xampl.transaction("random-people") do
+      Person.find_by_query do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('email', :ends_with, 'dodgit.com')
+
+        q.order_by('surname', :strasc)
+      end
+    end
+
+    first_person = people.first
+    address = first_person.address.first
+
+    start_at = Time.now
+
+    people_at_address = Xampl.transaction("random-people") do
+      Xampl.find_mentions_of(address)
+    end
+
+    done = Time.now
+    puts "found #{ people_at_address.size } people at that address in: #{ done - start_at }, #{ people_at_address.size / (done - start_at) }/s"
+#    people_at_address.each { | person | puts person.pp_xml }
+  end
+end

data/examples/random-people-shared-addresses/find-people-by-address.rb

@@ -0,0 +1,110 @@
+
+require 'xampl_generated_code/RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  #<people pid=''
+  #        xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            email=''
+  #            phone=''>
+  #        <address pid=''
+  #                 street-address=''
+  #                 city=''
+  #                 state=''
+  #                 postal-code=''/>
+  #    </person>
+  #</people>
+
+  times = [0, 0]
+  counts = [0, 0]
+
+  10.times do | trial |
+    found1 = []
+    found2 = []
+    found3 = []
+
+    #<p:person email='Edelmira.K.Abshire@dodgit.com'
+    #          given-name='Edelmira'
+    #          phone='416-279-8804'
+    #          pid='person-29944'
+    #          surname='Abshire'
+    #          xmlns:p='http://xampl.com/people'>
+    #  <p:address city='TORONTO'
+    #             pid='person-19944'
+    #             postal_code='M1P 4W2'
+    #             state='ON'
+    #             street_address='4194 Borough Drive'/></p:person>
+
+
+    all_people = nil
+    people_inspected = 0
+    start_at = Time.now
+
+    people = Xampl.transaction("random-people") do
+      addresses = Address.find_by_query do | q |
+        q.add_condition('street-address', :equals, '4194 Borough Drive')
+        q.add_condition('postal-code',    :equals, 'M1P 4W2')
+      end
+
+      all_people = Set.new
+      people_inspected = 0
+
+      addresses.each do | address |
+        people = Person.find_by_query do | q |
+          # TODO -- want to be able to find people by the address directly, e.g. address.find_by_citation/find_by_mention & find_by_reference
+          q.add_condition('city',  :equals, address.city)
+          q.add_condition('state', :equals, address.state)
+        end
+
+        people_inspected += people.size
+
+        people.each do | person |
+          all_people << person if person.address.first == address
+        end
+      end
+    end
+
+    done_at = Time.now
+    puts "done in: #{ done_at - start_at }, #{ people_inspected } people inspected, #{ all_people.size } identified"
+    times[0] += (done_at - start_at) if 0 < trial
+    counts[0] += 1 if 0 < trial
+
+    start_at = Time.now
+
+    people = Xampl.transaction("random-people") do
+      addresses = Address.find_by_query do | q |
+        q.add_condition('street-address', :equals, '4194 Borough Drive')
+        q.add_condition('postal-code',    :equals, 'M1P 4W2')
+      end
+
+      all_people = Set.new
+      people_inspected = 0
+
+      addresses.each do | address |
+        people = Xampl.find_mentions_of(address)
+
+        people_inspected += people.size
+
+        people.each do | person |
+          all_people << person if person.address.first == address
+        end
+      end
+    end
+
+    done_at = Time.now
+    puts "done in: #{ done_at - start_at }, #{ people_inspected } people inspected, #{ all_people.size } identified"
+    times[1] += (done_at - start_at) if 0 < trial
+    counts[1] += 1 if 0 < trial
+  end
+
+  puts "0) total time: #{ times[0] }, count: #{ counts[0] }, average/s: #{ counts[0]/times[0]}"
+  puts "1) total time: #{ times[1] }, count: #{ counts[1] }, average/s: #{ counts[1]/times[1]}"
+end

data/examples/random-people-shared-addresses/optimise.rb

@@ -0,0 +1,19 @@
+$LOAD_PATH.unshift("xampl_generated_code")
+
+require 'RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  start_opt = Time.now
+  Xampl.transaction("random-people") do
+    Xampl.optimise(:indexes_only => true)
+#    Xampl.optimise
+  end
+  puts "optimised in #{ Time.now - start_opt}"
+
+end

data/examples/random-people-shared-addresses/people.rb

@@ -0,0 +1,50 @@
+
+require 'xampl_generated_code/RandomPeople'
+
+module RandomPeople
+
+  #<people pid=''
+  #        xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            email=''
+  #            phone=''>
+  #        <address pid=''
+  #                 street-address=''
+  #                 city=''
+  #                 state=''
+  #                 postal-code=''/>
+  #    </person>
+  #</people>
+
+  class Person
+
+    Xampl::TokyoCabinetPersister.add_lexical_indexs(%w{ surname city state email })
+
+    def describe_yourself
+      {
+              'surname' => self.surname,
+              'city' => self.address.first.city,
+              'state' => self.address.first.state,
+              'email' => self.email
+      }
+    end
+  end
+
+  class Address
+
+    Xampl::TokyoCabinetPersister.add_lexical_indexs(%w{ street-address postal-code city state })
+
+    def describe_yourself
+      {
+              'street-address' => self.street_address,
+              'postal-code' => self.postal_code,
+              'city' => self.city,
+              'state' => self.state
+      }
+    end
+  end
+
+end
+

data/examples/random-people-shared-addresses/query.rb

@@ -0,0 +1,78 @@
+
+require 'xampl_generated_code/RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  #<people pid=''
+  #        xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            email=''
+  #            phone=''>
+  #        <address pid=''
+  #                 street-address=''
+  #                 city=''
+  #                 state=''
+  #                 postal-code=''/>
+  #    </person>
+  #</people>
+
+  5.times do
+    found1 = []
+    found2 = []
+    found3 = []
+
+    start_query_at = Time.now
+    found1 = Xampl.transaction("random-people") do
+      Xampl.query do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('email', :ends_with, 'dodgit.com')
+
+#        q.setlimit(10, 10)
+
+        q.order_by('surname', :strasc)
+      end
+    end
+
+    start_find_at = Time.now
+    found2 = Xampl.transaction("random-people") do
+      Person.find_by_query do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('email', :ends_with, 'dodgit.com')
+
+#        q.setlimit(10, 10)
+
+        q.order_by('surname', :strasc)
+      end
+    end
+
+    start_find_pids_at = Time.now
+
+    found3 = Xampl.transaction("random-people") do
+      Xampl.find_pids do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('email', :ends_with, 'dodgit.com')
+
+#        q.setlimit(10, 10)
+
+        q.order_by('surname', :strasc)
+      end
+    end
+    done = Time.now
+
+    #  found1.each do | person_meta |
+    #    puts "surname: #{ person_meta['surname']}, given: #{ person_meta['xampl'].given_name }"
+    #  end
+
+    puts
+    puts "query    :: found: #{ found1.size }, in: #{ start_find_at - start_query_at }"
+    puts "find     :: found: #{ found2.size }, in: #{ start_find_pids_at - start_find_at }"
+    puts "find pids:: found: #{ found3.size }, in: #{ done - start_find_pids_at}"
+  end
+end

data/examples/random-people-shared-addresses/query2.rb

@@ -0,0 +1,76 @@
+$LOAD_PATH.unshift("xampl_generated_code")
+
+require 'RandomPeople'
+require 'people'
+
+Xampl.set_default_persister_kind(:tokyo_cabinet)
+#Xampl.set_default_persister_format(:xml_format)
+Xampl.set_default_persister_format(:ruby_format)
+
+module RandomPeople
+
+  #GivenName,Surname,StreetAddress,City,State,ZipCode,EmailAddress,TelephoneNumber
+  #<people pid='' xmlns="http://xampl.com/people">
+  #    <person pid=''
+  #            given-name=''
+  #            surname=''
+  #            street-address=''
+  #            city=''
+  #            state=''
+  #            postal-code=''
+  #            email=''
+  #            phone=''/>
+  #</people>
+
+  1.times do
+    found1 = []
+    found2 = []
+    found3 = []
+
+    start = Time.now
+
+    Xampl.transaction("random-people") do
+      found1 = Xampl.find_pids do | q |
+        q.add_condition('city', :equals, 'TORONTO')
+        q.add_condition('surname', :equals, 'Smith')
+      end
+
+      found2 = Xampl.find_pids do | q |
+        q.add_condition('city', :equals, 'LONDON')
+        q.add_condition('surname', :equals, 'Smith')
+      end
+
+      found3 = Xampl.find_pids do | q |
+        q.add_condition('city', :stror, 'TORONTO,LONDON')
+        q.add_condition('surname', :equals, 'Smith')
+      end
+    end
+
+    done = Time.now
+
+    puts "INDEXED found(#{ done - start }):: TORONTO #{ found1.size }, LONDON: #{ found2.size }, TORONTO,LONDON: #{ found3.size }/#{ found1.size + found2.size }"
+
+    start = Time.now
+
+    Xampl.transaction("random-people") do
+      found1 = Xampl.find_pids do | q |
+        q.add_condition('city', :equals, 'TORONTO', true, true)
+        q.add_condition('surname', :equals, 'Smith', true, true)
+      end
+
+      found2 = Xampl.find_pids do | q |
+        q.add_condition('city', :equals, 'LONDON', true, true)
+        q.add_condition('surname', :equals, 'Smith', true, true)
+      end
+
+      found3 = Xampl.find_pids do | q |
+        q.add_condition('city', :stror, 'TORONTO,LONDON', true, true)
+        q.add_condition('surname', :equals, 'Smith', true, true)
+      end
+    end
+
+    done = Time.now
+
+    puts "NOT INDEXED found(#{ done - start }):: TORONTO #{ found1.size }, LONDON: #{ found2.size }, TORONTO,LONDON: #{ found3.size }/#{ found1.size + found2.size }"
+  end
+end