Add a Database abstraction and helpers for various data files.

.gitignore

@@ -5,3 +5,5 @@ Gemfile.lock
 /spec/regression/**/*.db
 /spec/unit/writer/result.xlsx
 .world
+.ruby-version
+.DS_Store

CHANGELOG.md

@@ -1,3 +1,13 @@
+## 0.23.0
+
+* Add `Bmg.json` and `Bmg.yaml` factory methods, to get relations on top of
+  usual data files.
+
+* Add a `Database` abstraction, with `Database.data_folder`, `Database.sequel`
+  and `Database.xlsx` factory methods and implementations, as well as
+  `Database#to_data_folder` and `Database#to_xlsx` dump methods.
+  See README for details.
+
 ## 0.22.0 - 2024-05-17
 
 * Add the `minus` operation (also known as set difference, or EXCEPT in SQL).

README.md

@@ -15,9 +15,10 @@ further down this README.
 * [Where are base relations coming from?](#where-are-base-relations-coming-from)
   * [Memory relations](#memory-relations)
   * [Connecting to SQL databases](#connecting-to-sql-databases)
-  * [Reading files (csv, Excel, text)](#reading-files-csv-excel-text)
+  * [Reading data files](#reading-data-files-json-csv-yaml-text-xls--xlsx)
   * [Connecting to Redis databases](#connecting-to-redis-databases)
   * [Your own relations](#your-own-relations)
+* [The Database abstraction](#the-database-abstraction)
 * [List of supported operators](#supported-operators)
 * [How is this different?](#how-is-this-different)
   * [... from similar libraries](#-from-similar-libraries)
@@ -117,33 +118,36 @@ Bmg.sequel(:suppliers, sequel_db)
 #   {:array=>false})
 ```
 
-### Reading files (csv, Excel, text)
+### Reading data files (json, csv, yaml, text, xls & xlsx)
 
 Bmg provides simple adapters to read files and reach Relationland as soon as
 possible.
 
-#### CSV files
+#### JSON files
 
 ```ruby
-csv_options = { col_sep: ",", quote_char: '"' }
-r = Bmg.csv("path/to/a/file.csv", csv_options)
+r = Bmg.json("path/to/a/file.json")
 ```
 
-Options are directly transmitted to `::CSV.new`, check Ruby's standard
-library.
+The json file is expected to contain tuples of same heading.
 
-#### Excel files
+#### YAML files
 
-You will need to add [`roo`](https://github.com/roo-rb/roo) to your Gemfile to
-read `.xls` and `.xlsx` files with Bmg.
+```ruby
+r = Bmg.yaml("path/to/a/file.yaml")
+```
+
+The yaml file is expected to contain tuples of same heading.
+
+#### CSV files
 
 ```ruby
-roo_options = { skip: 1 }
-r = Bmg.excel("path/to/a/file.xls", roo_options)
+csv_options = { col_sep: ",", quote_char: '"' }
+r = Bmg.csv("path/to/a/file.csv", csv_options)
 ```
 
-Options are directly transmitted to `Roo::Spreadsheet.open`, check roo's
-documentation.
+Options are directly transmitted to `::CSV.new`, check Ruby's standard
+library.
 
 #### Text files
 
@@ -173,6 +177,19 @@ r.type.attrlist
 In this scenario, non matching lines are skipped. The `:line` attribute keeps
 being used to have at least one candidate key (so to speak).
 
+#### Excel files
+
+You will need to add [`roo`](https://github.com/roo-rb/roo) to your Gemfile to
+read `.xls` and `.xlsx` files with Bmg.
+
+```ruby
+roo_options = { skip: 1 }
+r = Bmg.excel("path/to/a/file.xls", roo_options)
+```
+
+Options are directly transmitted to `Roo::Spreadsheet.open`, check roo's
+documentation.
+
 ### Connecting to Redis databases
 
 Bmg currently requires `bmg-redis` and `redis >= 4.6` to connect
@@ -240,6 +257,52 @@ restrictions down the tree) by overriding the underscored version of operators
 Have a look at `Bmg::Algebra` for the protocol and `Bmg::Sql::Relation` for an
 example. Keep in touch with the team if you need some help.
 
+## The Database abstraction
+
+The previous section focused on obtaining *relations*. In pratice you frequently have a collection of related relations hence a *database*:
+
+* A SQL database with multiple tables
+* A list of data files, all in the same folder
+* An excel file with various sheets
+
+Bmg supports a simple Datbabase abstraction that serves those relations "by name", in a simple way. A database can also be easily dumped back to a data folder of json or csv files, or as a simple xlsx files with multiple sheets.
+
+### Connecting to a SQL Database
+
+For a SQL database, connected with Sequel:
+
+```
+db = Bmg::Database.sequel(Sequel.connect('...'))
+db.suppliers # yields a Bmg::Relation over the `suppliers` table
+```
+
+### Connecting to data files in the same folder
+
+Data files all in the same folder can be seen as a very basic form of database,
+and served as such. Bmg supports `json`, `csv` and `yaml` files in this mode:
+
+```
+db = Bmg::Database.data_folder('./my-database')
+db.suppliers # yields a Bmg::Relation over the `suppliers.(json,csv,yml)` file
+```
+
+### Dumping a Database instance
+
+As a data folder:
+
+```
+db = Bmg::Database.sequel(Sequel.connect('...'))
+db.to_data_folder('path/to/folder', 'json')
+```
+
+As an .xlsx file (any existing file will be erased, we don't support modifying
+existing files):
+
+```
+require 'bmg/xlsx'
+db.to_xlsx('path/to/file.xlsx')
+```
+
 ## Supported operators
 
 ```ruby

lib/bmg.rb

@@ -24,6 +24,16 @@ def csv(path, options = {}, type = Type::ANY)
   end
   module_function :csv
 
+  def json(path, options = {}, type = Type::ANY)
+    in_memory(path.load.map{|tuple| TupleAlgebra.symbolize_keys(tuple) })
+  end
+  module_function :json
+
+  def yaml(path, options = {}, type = Type::ANY)
+    in_memory(path.load.map{|tuple| TupleAlgebra.symbolize_keys(tuple) })
+  end
+  module_function :yaml
+
   def excel(path, options = {}, type = Type::ANY)
     Reader::Excel.new(type, path, options).spied(main_spy)
   end
@@ -57,6 +67,8 @@ def main_spy=(spy)
   require_relative 'bmg/relation/materialized'
   require_relative 'bmg/relation/proxy'
 
+  require_relative 'bmg/database'
+
   # Deprecated
   Leaf = Relation::InMemory
 end

lib/bmg/database.rb

@@ -0,0 +1,35 @@
+module Bmg
+  class Database
+
+    def self.data_folder(*args)
+      require_relative 'database/data_folder'
+      DataFolder.new(*args)
+    end
+
+    def self.sequel(*args)
+      require 'bmg/sequel'
+      require_relative 'database/sequel'
+      Sequel.new(*args)
+    end
+
+    def self.xlsx(*args)
+      require 'bmg/xlsx'
+      require_relative 'database/xlsx'
+      Xlsx.new(*args)
+    end
+
+    def to_xlsx(*args)
+      require 'bmg/xlsx'
+      Writer::Xlsx.to_xlsx(self, *args)
+    end
+
+    def to_data_folder(*args)
+      DataFolder.dump(self, *args)
+    end
+
+    def each_relation_pair
+      raise NotImplementedError
+    end
+
+  end # class Database
+end # module Bmg

lib/bmg/database/data_folder.rb

@@ -0,0 +1,67 @@
+module Bmg
+  class Database
+    class DataFolder < Database
+
+      DEFAULT_OPTIONS = {
+        data_extensions: ['json', 'csv', 'yml', 'yaml']
+      }
+
+      def initialize(folder, options = {})
+        @folder = Path(folder)
+        @options = DEFAULT_OPTIONS.merge(options)
+      end
+
+      def method_missing(name, *args, &bl)
+        return super(name, *args, &bl) unless args.empty? && bl.nil?
+        raise NotSuchRelationError(name.to_s) unless file = find_file(name)
+        read_file(file)
+      end
+
+      def each_relation_pair
+        return to_enum(:each_relation_pair) unless block_given?
+
+        @folder.glob('*') do |path|
+          next unless path.file?
+          next unless @options[:data_extensions].find {|ext|
+            path.ext == ".#{ext}" || path.ext == ext
+          }
+          yield(path.basename.rm_ext.to_sym, read_file(path))
+        end
+      end
+
+      def self.dump(database, path, ext = :json)
+        path = Path(path)
+        path.mkdir_p
+        database.each_relation_pair do |name, rel|
+          (path/"#{name}.#{ext}").write(rel.public_send(:"to_#{ext}"))
+        end
+        path
+      end
+
+    private
+
+      def read_file(file)
+        case file.ext.to_s
+        when '.json'
+          Bmg.json(file)
+        when '.csv'
+          Bmg.csv(file)
+        when '.yaml', '.yml'
+          Bmg.yaml(file)
+        else
+          raise NotSupportedError, "Unable to use #{file} as a relation"
+        end
+      end
+
+      def find_file(name)
+        exts = @options[:data_extensions]
+        exts.each do |ext|
+          target = @folder/"#{name}.#{ext}"
+          return target if target.file?
+        end
+        raise NotSuchRelationError, "#{@folder}/#{name}.#{exts.join(',')}"
+      end
+
+    end # class DataFolder
+  end # class Database
+end # module Bmg

lib/bmg/database/sequel.rb

@@ -0,0 +1,35 @@
+module Bmg
+  class Database
+    class Sequel < Database
+
+      DEFAULT_OPTIONS = {
+      }
+
+      def initialize(sequel_db, options = {})
+        @sequel_db = sequel_db
+      end
+
+      def method_missing(name, *args, &bl)
+        return super(name, *args, &bl) unless args.empty? && bl.nil?
+        raise NotSuchRelationError(name.to_s) unless @sequel_db.table_exists?(name)
+        table = @sequel_db[name]
+        rel_for(table)
+      end
+
+      def each_relation_pair
+        return to_enum(:each_relation_pair) unless block_given?
+
+        @sequel_db.tables.each do |table|
+          yield(table, rel_for(table))
+        end
+      end
+
+    protected
+
+      def rel_for(table_name)
+        Bmg.sequel(table_name, @sequel_db)
+      end
+
+    end # class Sequel
+  end # class Database
+end # module Bmg

lib/bmg/database/xlsx.rb

@@ -0,0 +1,41 @@
+module Bmg
+  class Database
+    class Xlsx < Database
+
+      DEFAULT_OPTIONS = {
+      }
+
+      def initialize(path, options = {})
+        path = Path(path) if path.is_a?(String)
+        @path = path
+        @options = options.merge(DEFAULT_OPTIONS)
+      end
+
+      def method_missing(name, *args, &bl)
+        return super(name, *args, &bl) unless args.empty? && bl.nil?
+        rel = rel_for(name)
+        raise NotSuchRelationError(name.to_s) unless rel
+        rel
+      end
+
+      def each_relation_pair
+        return to_enum(:each_relation_pair) unless block_given?
+
+        spreadsheet.sheets.each do |sheet_name|
+          yield(sheet_name.to_sym, rel_for(sheet_name))
+        end
+      end
+
+    protected
+
+      def spreadsheet
+        @spreadsheet ||= Roo::Spreadsheet.open(@path, @options)
+      end
+
+      def rel_for(sheet_name)
+        Bmg.excel(@path, { sheet: sheet_name.to_s })
+      end
+
+    end # class Sequel
+  end # class Database
+end # module Bmg

lib/bmg/error.rb

@@ -23,4 +23,7 @@ class MisuseError < Error; end
   # to backtrack to something more ruby-native.
   class NotSupportedError < Error; end
 
+  # Raised when relation (variable) is not found
+  class NotSuchRelationError < Error; end
+
 end