documentrix 0.3.1 → 0.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c3ab97426ab9fbd4ec832a422a18d574f4c03b11f22391cbc6eded16b9ba0609
-  data.tar.gz: 991215fb26b4165a4b3562075d9742e3756218e2a24f3f32a550e232f59fb92f
+  metadata.gz: e32a72c0a1f93a96f7c3cecd185f13a0f17a2629b2dc509ec3a29fd2d7b51a41
+  data.tar.gz: 3f2c21125adf7061dcba94f1843456064245222839de0df6ddbcd8743a1aea13
 SHA512:
-  metadata.gz: d67e453e27428bcd8364349e988556af844caab5faa8a8edb3dcfba9650554b8dbfc66ad322a6f9f59d5ac64350d7006bf899a16b6c83f92909793301b6954d5
-  data.tar.gz: 49706d4984b27ee7b3d0f8ef8a9efec4a8e74d8bef78d78f15ed5c76f5bacc0faf20c36cef739aa78c1f68e5733696d16999bbd14228ae7be8e6cf9c9baf0a75
+  metadata.gz: 91dbf3ddfdeb124661ff78a4cbc1635dbb8bdf6606078aeeef77b4f9d14688d073261ad08b59f0333895c7391d83eff8a6c76467f19af4f04ca55172c1d934a5
+  data.tar.gz: d40e5a53ceeda71c7a2be37d3bfb718ffc2ddb0fddc0c6eb2346c295d06713c895493854fab5f768af075bec2595e2f23c40e893d7f7266c6f61cd95c534d733

data/CHANGES.md

@@ -1,5 +1,46 @@
 # Changes
 
+## 2026-05-22 v0.4.0
+
+### Added
+
+- Added introspection methods to `Documents`:
+    - Implemented `sources` to return an array of unique source identifiers.
+    - Implemented `each_record` to iterate over records, returning an
+      `Enumerator` if no block is provided.
+
+### Changed
+
+- Updated CI images from Alpine to Debian:
+    - Switched Ruby **4.0**, **3.4**, **3.3**, and **3.2** images to the
+      `trixie` distribution.
+    - Switched Ruby **3.1** image to the `bookworm` distribution.
+    - Updated the `dockerfile` in `.all_images.yml` to use `apt-get` instead of
+      `apk` for installing build dependencies.
+
+## 2026-05-20 v0.3.2
+
+### Performance Improvements
+
+- Optimized collections lookup by memoizing
+  `Documentrix::Documents#collections` using `@collections_cache`.
+- Added `Documentrix::Documents#invalidate_collections_cache!` to reset the
+  memoized list.
+- Integrated `invalidate_collections_cache!` into `add`, `delete`, `clear`,
+  `source_remove`, and `rename_collection` to maintain cache consistency.
+
+### Database & Cache
+
+- Standardized SQL keyword casing to uppercase for data types (e.g., `FLOAT`,
+  `TEXT`, `INTEGER`, `JSON`) within `SQLiteCache`.
+
+### Documentation & Testing
+
+- Refined documentation and return type descriptions for
+  `Documentrix::Documents#find` and `Documentrix::Documents#prefix`.
+- Updated `spec/documents_spec.rb` to ensure `invalidate_collections_cache!` is
+  correctly triggered during mutations.
+
 ## 2026-05-18 v0.3.1
 
 - Fixed scoping bugs in `clear_by_source` and `source_exist?` by implementing a

data/documentrix.gemspec

@@ -1,9 +1,9 @@
 # -*- encoding: utf-8 -*-
-# stub: documentrix 0.3.1 ruby lib
+# stub: documentrix 0.4.0 ruby lib
 
 Gem::Specification.new do |s|
   s.name = "documentrix".freeze
-  s.version = "0.3.1".freeze
+  s.version = "0.4.0".freeze
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0".freeze) if s.respond_to? :required_rubygems_version=
   s.require_paths = ["lib".freeze]

data/lib/documentrix/documents/cache/sqlite_cache.rb

@@ -443,18 +443,18 @@ class Documentrix::Documents::Cache::SQLiteCache
     @database.enable_load_extension(false)
     execute %{
       CREATE VIRTUAL TABLE IF NOT EXISTS embeddings USING vec0(
-        embedding float[#@embedding_length]
+        embedding FLOAT[#@embedding_length]
       )
     }
     execute %{
       CREATE TABLE IF NOT EXISTS records (
-        key          text NOT NULL PRIMARY KEY ON CONFLICT REPLACE,
-        text         text NOT NULL DEFAULT '',
-        embedding_id integer,
-        norm         float NOT NULL DEFAULT 0.0,
-        source       text,
-        digest       text,
-        tags         json NOT NULL DEFAULT [],
+        key          TEXT NOT NULL PRIMARY KEY ON CONFLICT REPLACE,
+        text         TEXT NOT NULL DEFAULT '',
+        embedding_id INTEGER,
+        norm         FLOAT NOT NULL DEFAULT 0.0,
+        source       TEXT,
+        digest       TEXT,
+        tags         JSON NOT NULL DEFAULT [],
         FOREIGN KEY(embedding_id) REFERENCES embeddings(id) ON DELETE CASCADE
       )
     }

data/lib/documentrix/documents.rb

@@ -162,7 +162,7 @@ class Documentrix::Documents
       infobar.progress by: batch.size
     end
     infobar.newline
-    self
+    invalidate_collections_cache!
   end
   alias << add
 
@@ -201,7 +201,9 @@ class Documentrix::Documents
   # @return [ FalseClass, TrueClass ] true if the text was removed, false
   #         otherwise.
   def delete(text)
-    @cache.delete(key(text))
+    res = @cache.delete(key(text))
+    invalidate_collections_cache! if res
+    res
   end
 
   # The size method returns the number of texts stored in the cache of this
@@ -220,7 +222,7 @@ class Documentrix::Documents
   # @return [ Documentrix::Documents ] self
   def clear(tags: nil)
     @cache.clear(tags:)
-    self
+    invalidate_collections_cache!
   end
 
   # Normalizes the source identifier to a canonical form.
@@ -318,7 +320,7 @@ class Documentrix::Documents
   def source_remove(source, digest: nil)
     source = normalize_source(source)
     @cache.clear_by_source(source, digest:, operator: '!=')
-    self
+    invalidate_collections_cache!
   end
 
   # The find method searches for strings within the cache by computing their
@@ -359,7 +361,7 @@ class Documentrix::Documents
   # @param text_count [Integer] the maximum number of records to return
   # @param opts [Hash] additional options passed to #find, such as:
   #   * :tags [Array<String>] filter results by tags
-  #   * :prompt [String] a prompt to use for the search
+  #   * :prompt [String] use for the search
   #   * :min_similarity [Numeric] minimum similarity score
   #
   # @example
@@ -379,7 +381,9 @@ class Documentrix::Documents
   #
   # @return [Array] An array of unique collection names
   def collections
-    ([ default_collection ] + @cache.collections('%s-' % class_prefix)).uniq
+    @collections_cache ||= (
+      [ default_collection ] + @cache.collections('%s-' % class_prefix)
+    ).uniq
   end
 
   # Rename the current collection, moving all keys from the old prefix to a new
@@ -395,6 +399,7 @@ class Documentrix::Documents
     new_prefix = '%s-%s-' % [ class_prefix, new_collection ]
     @cache.move_prefix(prefix, new_prefix)
     self.collection = new_collection
+    invalidate_collections_cache!
   end
 
   # The tags method returns an array of unique tags from the cache.
@@ -404,8 +409,36 @@ class Documentrix::Documents
     @cache.tags
   end
 
+  # Returns an array of all unique sources stored in the cache.
+  #
+  # @return [Array<String>] An array of unique source identifiers.
+  def sources
+    @cache.each_source.to_a
+  end
+
+ # The each_record method iterates over all records stored in the cache.
+ #
+ # @yield [record] The record being iterated over.
+ # @return [ Enumerator ] an enumerator if no block is provided.
+  def each_record(&block)
+    block or return enum_for(__method__)
+    @cache.each { |_key, record| block.(record) }
+  end
+
   private
 
+  # Resets the memoized list of collections.
+  #
+  # This is called whenever a mutation occurs that could change the set of
+  # existing collections, ensuring that the #collections method returns a
+  # fresh, accurate list on the next call.
+  #
+  # @return [ Documentrix::Documents ] self
+  def invalidate_collections_cache!
+    @collections_cache = nil
+    self
+  end
+
   # The connect_cache method initializes and returns an instance of the
   # specified cache class.
   #
@@ -482,7 +515,7 @@ class Documentrix::Documents
   # The prefix method returns a string that is used as the prefix for keys in
   # the cache of the currently configured collection.
   #
-  # @return [ String ] The prefix string
+  # @return [ String ] the prefix string
   def prefix
     '%s-%s-' % [ class_prefix, @collection ]
   end

data/lib/documentrix/version.rb

@@ -1,6 +1,6 @@
 module Documentrix
   # Documentrix version
-  VERSION         = '0.3.1'
+  VERSION         = '0.4.0'
   VERSION_ARRAY   = VERSION.split('.').map(&:to_i) # :nodoc:
   VERSION_MAJOR   = VERSION_ARRAY[0] # :nodoc:
   VERSION_MINOR   = VERSION_ARRAY[1] # :nodoc:

data/spec/documents_spec.rb

@@ -23,6 +23,8 @@ describe Documentrix::Documents do
     expect(ollama).to receive(:embed).
       with(model:, input: %w[ foo bar ], options: nil).
       and_return(double(embeddings: [ [ 0.1 ], [ 0.2 ] ]))
+    expect(documents).to receive(:invalidate_collections_cache!).
+      and_call_original
     expect(documents.add(%w[ foo bar ])).to eq documents
     expect(documents.exist?('foo')).to eq true
     expect(documents.exist?('bar')).to eq true
@@ -33,6 +35,8 @@ describe Documentrix::Documents do
     expect(ollama).to receive(:embed).
       with(model:, input: %w[ foo ], options: nil).
       and_return(double(embeddings: [ [ 0.1 ] ]))
+    expect(documents).to receive(:invalidate_collections_cache!).
+      and_call_original
     expect(documents << 'foo').to eq documents
     expect(documents.exist?('foo')).to eq true
     expect(documents.exist?('bar')).to eq false
@@ -123,6 +127,8 @@ describe Documentrix::Documents do
 
     it 'can delete texts' do
       expect(documents << 'foo').to eq documents
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       expect {
         documents.delete('foo')
       }.to change { documents.exist?('foo') }.from(true).to(false)
@@ -136,6 +142,8 @@ describe Documentrix::Documents do
 
     it 'can clear texts' do
       expect(documents << 'foo').to eq documents
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       expect {
         documents.clear
       }.to change { documents.size }.from(1).to(0)
@@ -148,9 +156,13 @@ describe Documentrix::Documents do
       expect(documents.add('foo', tags: %w[ test ])).to eq documents
       expect(documents.add('bar', tags: %w[ test2 ])).to eq documents
       expect(documents.tags.to_a).to eq %w[ test test2 ]
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       expect {
         documents.clear tags: 'test'
       }.to change { documents.size }.from(2).to(1)
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       expect {
         documents.clear tags: :test2
       }.to change { documents.size }.from(1).to(0)
@@ -166,6 +178,8 @@ describe Documentrix::Documents do
 
       expect(documents.size).to eq 3
 
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       documents.source_remove('source1')
 
       expect(documents.size).to eq 1
@@ -182,6 +196,8 @@ describe Documentrix::Documents do
       documents.collection = :foo
       documents << 'foo'
       expect(documents.collections).to eq %i[ default foo ]
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       documents.rename_collection(:bar)
       expect(documents.collection).to eq :bar
       expect(documents.collections).to eq %i[ default bar ]
@@ -195,6 +211,8 @@ describe Documentrix::Documents do
       documents.collection = :bar
       documents << 'foo'
       expect(documents.collections).to eq %i[ default foo bar ]
+      expect(documents).not_to receive(:invalidate_collections_cache!).
+        and_call_original
       expect {
         documents.rename_collection(:foo)
       }.to raise_error(ArgumentError, 'new collection foo already exists!')
@@ -208,6 +226,32 @@ describe Documentrix::Documents do
         from(:default).
         to(:new_collection)
     end
+
+    it 'returns unique sources' do
+      allow(ollama).to receive(:embed).and_return(double(embeddings: [ [ 0.1 ] ]))
+      documents.add('foo', source: 's1')
+      documents.add('bar', source: 's1')
+      documents.add('baz', source: 's2')
+      expect(documents.sources).to match_array %w[ s1 s2 ]
+    end
+
+    it 'can iterate over records' do
+      allow(ollama).to receive(:embed).and_return(double(embeddings: [ [ 0.1 ] ]))
+      documents.add('foo')
+      documents.add('bar')
+      records = []
+      documents.each_record { |r| records << r }
+      expect(records.size).to eq 2
+      expect(records.map(&:text)).to match_array %w[ foo bar ]
+    end
+
+    it 'returns an enumerator for each_record' do
+      allow(ollama).to receive(:embed).and_return(double(embeddings: [ [ 0.1 ] ]))
+      documents.add('foo')
+      documents.add('bar')
+      expect(documents.each_record).to be_a Enumerator
+      expect(documents.each_record.map(&:text)).to match_array %w[ foo bar ]
+    end
   end
 
   context 'source management' do
@@ -215,7 +259,7 @@ describe Documentrix::Documents do
       allow(documents).to receive(:compute_file_digest).and_return('d1')
       allow(documents.cache).to receive(:compute_file_digest).and_return('d1')
 
-      allow(ollama).to receive(:embed).and_return(double(embeddings: [[0.1]]))
+      allow(ollama).to receive(:embed).and_return(double(embeddings: [ [ 0.1 ] ]))
     end
 
     it 'can check if a source exists' do
@@ -243,6 +287,8 @@ describe Documentrix::Documents do
       documents.add('foo', source: 's1')
 
       expect(ollama).not_to receive(:embed)
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       documents.source_update(['foo'], source: 's1')
       expect(documents.exist?('foo')).to be true
     end
@@ -255,6 +301,8 @@ describe Documentrix::Documents do
       allow(documents.cache).to receive(:compute_file_digest).with('s1').and_return('d2')
 
       expect(ollama).to receive(:embed).once
+      expect(documents).to receive(:invalidate_collections_cache!).
+        at_least(1).and_call_original
       documents.source_update(['bar'], source: 's1')
 
       expect(documents.exist?('bar')).to be true
@@ -263,6 +311,8 @@ describe Documentrix::Documents do
 
     it 'updates the source if it is an URL' do
       expect(ollama).to receive(:embed).once
+      expect(documents).to receive(:invalidate_collections_cache!).
+        and_call_original
       documents.source_update('foo', source: 'https://www.example.com/s1')
       expect(documents.exist?('foo')).to be true
     end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: documentrix
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.4.0
 platform: ruby
 authors:
 - Florian Frank