tsql_parser 0.1.5 → 0.1.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 81a892f94d3a26de7e1b77b16c723b67ec1ccddc8a8c6dc4251300792be75c75
-  data.tar.gz: 1a326f8c1f465444144d436d69c95180c4a6d6ff19cc90e3db93afa9e7f7484e
+  metadata.gz: d67275fb1f5b1ff478b0e8a79d99af69a2d33b53059822e882b6a0c94e68d0a1
+  data.tar.gz: 50b27357c58d8c0ca89b069ddef3a03e4565b1e354a7e322d79c9ba64440895d
 SHA512:
-  metadata.gz: 240fb09638d5005e9cce7ba057666c8fc3a5167e10071e0affd78af39a962b253914b392022ee54449f3126abbea32e70ef6a3d2c3c151607b0ac46703c9605e
-  data.tar.gz: 71dc6419e1a96499521158f1e25023afa53488f4a94fd5ff33cc46e15664533f0e6dd19a57c71fd866537c8b12ca70c64e7b72d08e0841f9aa5cbaa5c6e12667
+  metadata.gz: dd4314846a33bd5b4c0e99dd41ecb67c65494faf015fc5ec81ffcf74cd5f98c52215bc28fd1a8729e0a1694ad7a055b09f51b724b0d7f27ae945a3ed2f778d13
+  data.tar.gz: 664dce679b765bf73e229f5cafc4eb0059f540c85674d4b6d41f1e3508e7536d512c457b64075d6c0f12b70bf7d3bfabdf4887466894beceec4182f444a5b986

data/lib/parsing/config/defaults.rb

@@ -0,0 +1,56 @@
+#   __               .__
+# _/  |_  ___________|  |           ___________ _______  ______ ___________
+# \   __\/  ___/ ____/  |    ______ \____ \__  \\_  __ \/  ___// __ \_  __ \
+#  |  |  \___ < <_|  |  |__ /_____/ |  |_> > __ \|  | \/\___ \\  ___/|  | \/
+#  |__| /____  >__   |____/         |   __(____  /__|  /____  >\___  >__|
+#            \/   |__|              |__|       \/           \/     \/
+#
+# A very light-weight and opinionated T-SQL parser and formatter.
+#
+# github.com/scstauf
+#
+# path:
+#   parsing/config/defaults.rb
+# object:
+#   TSqlParser::Parsing::Defaults
+
+module TSqlParser::Parsing
+  class Defaults
+    @@default_single_char_tokens = ["(", ",", ")", "=", "+", "-", "%", "/", "*", "<", "!", ">", "'", "[", "]", ";"]
+    @@default_delimiters = [" ", "\n", "\t"]
+    @@default_tab_count = 0
+    @@default_tab = "    "
+
+    def self.set_default_tab_count(tab_count = 0)
+      @@default_tab_count = tab_count
+    end
+
+    def self.set_default_tab(tab = "    ")
+      @@default_tab = tab
+    end
+
+    def self.set_default_single_char_tokens(single_char_tokens = ["(", ",", ")", "=", "+", "-", "%", "/", "*", "<", "!", ">", "'", "[", "]", ";"])
+      @@default_single_char_tokens = single_char_tokens
+    end
+
+    def self.set_default_delimiters(delimiters = [" ", "\n", "\t"])
+      @@default_delimiters = delimiters
+    end
+
+    def self.get_default_single_char_tokens
+      @@default_single_char_tokens
+    end
+
+    def self.get_default_delimiters
+      @@default_delimiters
+    end
+
+    def self.get_default_tab_count
+      @@default_tab_count
+    end
+
+    def self.get_default_tab
+      @@default_tab
+    end
+  end
+end

data/lib/parsing/formatter.rb

@@ -15,23 +15,25 @@
 #   TSqlParser::Parsing::Formatter
 
 module TSqlParser::Parsing
+  require_relative "config/defaults"
   require_relative "parser"
   require_relative "formatters/text_formatter"
   require_relative "transformers/token_transformer"
   
   class Formatter
-    def self.format(tokens, tab_count = 0, tab = "    ")
+    def self.format(tokens, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       lines = TokenTransformer.transform(tokens)
       lines = self.cleanup_whitespace(lines)
       lines = self.insert_indentation(lines, tab_count, tab)
-      lines = self.insert_newlines(lines)
+      #lines = self.insert_newlines(lines)
       text = lines.join("\n")
-      text = TextFormatter.format_inserts(text, tab)
-      text = TextFormatter.format_updates(text, tab)
-      text = TextFormatter.format_joins(text, tab)
-      text = TextFormatter.format_wheres(text, tab)
-      text = TextFormatter.format_selects(text, tab)
-      text = TextFormatter.format_sets(text, tab)
+      text = TextFormatter.new(JOIN, text, tab).format
+      text = TextFormatter.new(INSERT, text, tab).format
+      text = TextFormatter.new(UPDATE, text, tab).format
+      text = TextFormatter.new(WHERE, text, tab).format
+      text = TextFormatter.new(SELECT, text, tab).format
+      text = TextFormatter.new(SET, text, tab).format
+      puts text
       text
     end
 
@@ -60,7 +62,7 @@ module TSqlParser::Parsing
       new_lines
     end
 
-    def self.insert_indentation(lines, tab_count = 0, tab = "    ")
+    def self.insert_indentation(lines, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       indented_lines = []
       work_lines = []
       lines.each do |line|

data/lib/parsing/formatters/format_factory.rb

@@ -0,0 +1,32 @@
+#   __               .__
+# _/  |_  ___________|  |           ___________ _______  ______ ___________
+# \   __\/  ___/ ____/  |    ______ \____ \__  \\_  __ \/  ___// __ \_  __ \
+#  |  |  \___ < <_|  |  |__ /_____/ |  |_> > __ \|  | \/\___ \\  ___/|  | \/
+#  |__| /____  >__   |____/         |   __(____  /__|  /____  >\___  >__|
+#            \/   |__|              |__|       \/           \/     \/
+#
+# A very light-weight and opinionated T-SQL parser and formatter.
+#
+# github.com/scstauf
+#
+# path:
+#   parsing/formatters/format_factory.rb
+# object:
+#   TSqlParser::Parsing::FormatFactory
+
+module TSqlParser::Parsing
+    require_relative "strategy/__formatters"
+    class FormatFactory
+        def self.get(type)
+            case type
+            when CTE then Formatters::CommonTableExpressionFormatter.new
+            when INSERT then Formatters::InsertFormatter.new
+            when JOIN then Formatters::JoinFormatter.new
+            when SELECT then Formatters::SelectFormatter.new
+            when SET then Formatters::SetFormatter.new
+            when UPDATE then Formatters::UpdateFormatter.new
+            when WHERE then Formatters::WhereFormatter.new
+            end
+        end
+    end
+end

data/lib/parsing/formatters/strategy/__defaults.rb

@@ -0,0 +1,16 @@
+#   __               .__
+# _/  |_  ___________|  |           ___________ _______  ______ ___________
+# \   __\/  ___/ ____/  |    ______ \____ \__  \\_  __ \/  ___// __ \_  __ \
+#  |  |  \___ < <_|  |  |__ /_____/ |  |_> > __ \|  | \/\___ \\  ___/|  | \/
+#  |__| /____  >__   |____/         |   __(____  /__|  /____  >\___  >__|
+#            \/   |__|              |__|       \/           \/     \/
+#
+# A very light-weight and opinionated T-SQL parser and formatter.
+#
+# github.com/scstauf
+#
+# path:
+#   parsing/formatters/strategy/__defaults.rb
+
+require_relative "../../config/defaults"
+Defaults = TSqlParser::Parsing::Defaults

data/lib/parsing/formatters/strategy/__formatters.rb

@@ -0,0 +1,31 @@
+#   __               .__
+# _/  |_  ___________|  |           ___________ _______  ______ ___________
+# \   __\/  ___/ ____/  |    ______ \____ \__  \\_  __ \/  ___// __ \_  __ \
+#  |  |  \___ < <_|  |  |__ /_____/ |  |_> > __ \|  | \/\___ \\  ___/|  | \/
+#  |__| /____  >__   |____/         |   __(____  /__|  /____  >\___  >__|
+#            \/   |__|              |__|       \/           \/     \/
+#
+# A very light-weight and opinionated T-SQL parser and formatter.
+#
+# github.com/scstauf
+#
+# path:
+#   parsing/formatters/strategy/__formatters.rb
+
+module TSqlParser::Parsing
+  require_relative "cte_formatter"
+  require_relative "set_formatter"
+  require_relative "join_formatter"
+  require_relative "insert_formatter"
+  require_relative "select_formatter"
+  require_relative "update_formatter"
+  require_relative "where_formatter"
+
+  CTE = 0
+  INSERT = 1
+  JOIN = 2
+  SELECT = 3
+  SET = 4
+  UPDATE = 5
+  WHERE = 6
+end

data/lib/parsing/formatters/strategy/base_formatter.rb

@@ -15,14 +15,24 @@
 #   TSqlParser::Parsing::Formatters::SetFormatter
 
 module TSqlParser::Parsing::Formatters
+  require_relative "__defaults"
+
   class BaseFormatter
-    def self.get_tab_count(line, tab = "    ")
-      tab_count = 0
+    def get_tab_count(line, tab = Defaults.get_default_tab)
+      tab_count = Defaults.get_default_tab_count
       while line.start_with? tab
         tab_count += 1
         line = line.sub(tab, "")
       end
       tab_count
     end
+
+    # @abstract
+    #
+    # @param [String] text
+    # @param [String] tab
+    def format(text, tab = Defaults.get_default_tab)
+      raise NotImplementedError, "#{self.class} has not implemented method '#{__method__}'"
+    end
   end
 end

data/lib/parsing/formatters/strategy/cte_formatter.rb

@@ -0,0 +1,25 @@
+#   __               .__
+# _/  |_  ___________|  |           ___________ _______  ______ ___________
+# \   __\/  ___/ ____/  |    ______ \____ \__  \\_  __ \/  ___// __ \_  __ \
+#  |  |  \___ < <_|  |  |__ /_____/ |  |_> > __ \|  | \/\___ \\  ___/|  | \/
+#  |__| /____  >__   |____/         |   __(____  /__|  /____  >\___  >__|
+#            \/   |__|              |__|       \/           \/     \/
+#
+# A very light-weight and opinionated T-SQL parser and formatter.
+#
+# github.com/scstauf
+#
+# path:
+#   parsing/formatters/cte_formatter.rb
+# object:
+#   TSqlParser::Parsing::Formatters::CommonTableExpressionFormatter
+
+module TSqlParser::Parsing::Formatters
+    require_relative "base_formatter"
+
+    class CommonTableExpressionFormatter < BaseFormatter
+      def format(text, tab = Defaults.get_default_tab)
+      end
+    end
+  end
+  

data/lib/parsing/formatters/strategy/insert_formatter.rb

@@ -18,7 +18,7 @@ module TSqlParser::Parsing::Formatters
   require_relative "base_formatter"
 
   class InsertFormatter < BaseFormatter
-    def self.format(text, tab = "    ")
+    def format(text, tab = Defaults.get_default_tab)
       formatted = []
       lines = text.split("\n")
       search = "INSERT INTO"
@@ -43,7 +43,7 @@ module TSqlParser::Parsing::Formatters
 
     private
 
-    def self.format_insert(s, tab_count = 0, tab = "    ")
+    def format_insert(s, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       return s if s.nil?
       formatted = []
       if s.include? ") VALUES ("

data/lib/parsing/formatters/strategy/join_formatter.rb

@@ -18,9 +18,11 @@ module TSqlParser::Parsing::Formatters
   require_relative "base_formatter"
 
   class JoinFormatter < BaseFormatter
-    def self.format(text, tab = "    ")
+    def format(text, tab = Defaults.get_default_tab)
       text = text.gsub(/INNER\s+JOIN/, "INNER JOIN")
                  .gsub(/LEFT\s+JOIN/, "LEFT JOIN")
+                 .gsub(/RIGHT\s+JOIN/, "RIGHT JOIN")
+                 .gsub(/CROSS\s+JOIN/, "CROSS JOIN")
       lines = text.split("\n")
       new_text = []
 

data/lib/parsing/formatters/strategy/select_formatter.rb

@@ -18,7 +18,7 @@ module TSqlParser::Parsing::Formatters
   require_relative "base_formatter"
 
   class SelectFormatter < BaseFormatter
-    def self.format(text, tab = "    ")
+    def format(text, tab = Defaults.get_default_tab)
       formatted = []
       lines = text.split("\n")
       lines.each do |line|
@@ -42,7 +42,7 @@ module TSqlParser::Parsing::Formatters
 
     private
 
-    def self.format_select(s, tab_count = 0, tab = "    ")
+    def format_select(s, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       return s if s.nil?
 
       tokens = s.split(", ")

data/lib/parsing/formatters/strategy/set_formatter.rb

@@ -15,10 +15,10 @@
 #   TSqlParser::Parsing::Formatters::SetFormatter
 
 module TSqlParser::Parsing::Formatters
-  require_relative 'base_formatter'
+  require_relative "base_formatter"
 
   class SetFormatter < BaseFormatter
-    def self.format(text, tab = "    ")
+    def format(text, tab = Defaults.get_default_tab)
       formatted = []
       lines = text.split("\n")
       wait = false
@@ -71,7 +71,7 @@ module TSqlParser::Parsing::Formatters
 
     private
 
-    def self.format_set(s, tab_count = 0, tab = "    ")
+    def format_set(s, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       return s if s.nil?
       parts = []
       builder = ""

data/lib/parsing/formatters/strategy/update_formatter.rb

@@ -18,7 +18,7 @@ module TSqlParser::Parsing::Formatters
   require_relative "base_formatter"
 
   class UpdateFormatter < BaseFormatter
-    def self.format(text, tab = "    ")
+    def format(text, tab = Defaults.get_default_tab)
       formatted = []
       lines = text.split("\n")
       lines.each do |line|
@@ -42,7 +42,7 @@ module TSqlParser::Parsing::Formatters
 
     private
 
-    def self.format_update(s, tab_count = 0, tab = "    ")
+    def format_update(s, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       return s if s.nil?
       "\n#{tab * (tab_count + 1)}#{s}"
     end

data/lib/parsing/formatters/strategy/where_formatter.rb

@@ -18,7 +18,7 @@ module TSqlParser::Parsing::Formatters
   require_relative "base_formatter"
 
   class WhereFormatter < BaseFormatter
-    def self.format(text, tab = "   ")
+    def format(text, tab = Defaults.get_default_tab)
       formatted = []
       text.split("\n").each do |line|
         first = line.strip.split(" ").first
@@ -42,7 +42,7 @@ module TSqlParser::Parsing::Formatters
 
     private
 
-    def self.format_predicate(s, tab_count = 0, tab = "    ")
+    def format_predicate(s, tab_count = Defaults.get_default_tab_count, tab = Defaults.get_default_tab)
       return s if s.nil?
       indented = []
       formatted = []

data/lib/parsing/formatters/text_formatter.rb

@@ -15,36 +15,20 @@
 #   TSqlParser::Parsing::Formatter
 
 module TSqlParser::Parsing
-  require_relative 'strategy/set_formatter'
-  require_relative 'strategy/join_formatter'
-  require_relative 'strategy/insert_formatter'
-  require_relative 'strategy/select_formatter'
-  require_relative 'strategy/update_formatter'
-  require_relative 'strategy/where_formatter'
+  require_relative "strategy/__defaults"
+  require_relative "format_factory"
 
   class TextFormatter
-    def self.format_sets(text, tab = "    ")
-      Formatters::SetFormatter.format(text, tab)
-    end
-
-    def self.format_joins(text, tab = "    ")
-      Formatters::JoinFormatter.format(text, tab)
-    end
-
-    def self.format_updates(text, tab = "    ")
-      Formatters::UpdateFormatter.format(text, tab)
-    end
-
-    def self.format_inserts(text, tab = "    ")
-      Formatters::InsertFormatter.format(text, tab)
-    end
+    attr_writer :strategy
 
-    def self.format_selects(text, tab = "    ")
-      Formatters::SelectFormatter.format(text, tab)
+    def initialize(strategy, text, tab = Defaults.get_default_tab)
+      @strategy = FormatFactory.get(strategy)
+      @text = text
+      @tab = tab
     end
 
-    def self.format_wheres(text, tab = "   ")
-      Formatters::WhereFormatter.format(text, tab)
+    def format
+      @strategy.format(@text, @tab)
     end
   end
 end

data/lib/parsing/tokenizer.rb

@@ -15,20 +15,10 @@
 #   TSqlParser::Parsing::Tokenizer
 
 module TSqlParser::Parsing
+  require_relative "config/defaults"
   require_relative "transformers/token_categorizer"
 
   class Tokenizer
-    @@default_char_delimiters = ["(", ",", ")", "=", "+", "-", "%", "/", "*", "<", "!", ">", "'", "[", "]", ";"]
-    @@default_skip_delimiters = [" ", "\n", "\t"]
-
-    def self.set_default_char_delimiters(char_delimiters = [])
-      @@default_char_delimiters = char_delimiters
-    end
-
-    def self.set_default_skip_delimiters(skip_delimiters = [])
-      @@default_skip_delimiters = skip_delimiters
-    end
-
     def self.tokenize(tsql_string)
       Tokenizer.new.basic_tokenize(tsql_string).map { |t| TokenCategorizer.categorize(t) }
     end
@@ -181,8 +171,8 @@ module TSqlParser::Parsing
       @string = false
       @string_count = 0
       @skip_count = 0
-      @char_delimiters = @@default_char_delimiters
-      @skip_delimiters = @@default_skip_delimiters
+      @char_delimiters = Defaults.get_default_single_char_tokens
+      @skip_delimiters = Defaults.get_default_delimiters
       @delimiters = ([] << @char_delimiters << @skip_delimiters).flatten.uniq
       @builder = ""
     end

data/lib/parsing/transformers/token_transformer.rb

@@ -19,59 +19,113 @@ module TSqlParser::Parsing
   require_relative "../models/flat_sql_container"
 
   class TokenTransformer
+    def initialize(tokens = [])
+      @containers = []
+      @container = nil
+      @skip_count = 0
+      @cte = ""
+      @parenthesis = 0
+      @tokens = tokens
+    end
+
     def self.transform(tokens)
-      containers = self.as_containers(tokens)
-      self.combine_containers(containers)
+      TokenTransformer.new(tokens).transform
     end
 
-    private
+    def transform
+      @tokens.each_with_index do |t, index|
+        @parenthesis += 1 if t[:open_parenthesis]
+        @parenthesis -= 1 if t[:close_parenthesis]
 
-    def self.as_containers(tokens)
-      containers = []
-      container = nil
-      skip_count = 0
-      tokens.each_with_index do |t, index|
-        if skip_count > 0
-          skip_count -= 1
-          next
-        end
+        next if self.handle_skip  
+
+        @t = t
+        @last_token = @tokens[index - 1] unless index - 1 < 0
+        @next_token = @tokens[index + 1] unless index + 1 > @tokens.size
 
-        next_token = tokens[index + 1]
+        next if self.handle_cte_end   
 
         if t[:string]
-          container.add t unless container.nil?
+          self.add_to_container
         elsif Parser.is_new_node_keyword? t[:value]
-          if not next_token.nil? and Parser.is_new_node_keyword? next_token[:value]
-            if Parser.is_new_node_composite?(t[:value], next_token[:value])
-              containers << container unless container.nil?
-              container = SqlContainer.combine(t, next_token)
-              skip_count = 1
+          if not @next_token.nil? and Parser.is_new_node_keyword? @next_token[:value]
+            if Parser.is_new_node_composite?(@t[:value], @next_token[:value])
+              self.add_container
+              @container = SqlContainer.combine(@t, @next_token)
+              @skip_count = 1
               next
             end
           end
-          containers << container unless container.nil?
-          container = SqlContainer.new(t)
+          self.add_container
+          self.new_container
         elsif t[:value] == "WITH"
-          if not next_token.nil? and not next_token[:keyword] and not next_token[:parenthesis]
-            containers << container unless container.nil?
-            container = SqlContainer.new(t)
+          if not @next_token.nil? and not @next_token[:keyword] and not @next_token[:parenthesis]
+            self.add_container
+            self.new_container
+            @cte = @next_token[:value]
+          else
+            self.add_to_container
+          end
+        elsif @t[:label]
+          self.add_container
+          self.new_container
+        else
+          self.add_to_container
+        end
+      end
+      self.add_container
+      self.get_lines_from_containers
+    end
+
+    private
+
+    def handle_skip
+      if @skip_count > 0
+        @skip_count -= 1
+        return true
+      end
+    end
+
+    def add_to_container
+      @container.add @t unless @container.nil?
+    end
+
+    def add_container
+      @containers << @container unless @container.nil?
+    end
+
+    def new_container
+      @container = SqlContainer.new(@t)
+    end
+
+    def handle_cte_end
+      if not @cte.empty? 
+        if not @last_token.nil? and @last_token[:close_parenthesis] and @parenthesis == 0
+          if @t[:keyword]
+            self.add_container
+            self.new_container
+            @cte = ""
           else
-            container.add t unless container.nil?
+            self.add_to_container
           end
-        elsif t[:label]
-          containers << container unless container.nil?
-          container = SqlContainer.new(t)
+        elsif not @last_token.nil? and not @last_token[:keyword] and @t[:value] == "AS"
+          self.add_to_container
+        elsif not @last_token.nil? and @last_token[:comma] and not @t[:keyword] and not @next_token.nil? and @next_token[:value] == "AS"
+          self.add_container
+          self.new_container
+          @cte = @t[:value]
         else
-          container.add t unless container.nil?
+          self.add_to_container
         end
+        return true
       end
-      containers << container unless container.nil?
-      FlatSqlContainer.flatten_containers(containers)
     end
 
-    def self.combine_containers(containers)
+    def get_lines_from_containers
+      @containers = FlatSqlContainer.flatten_containers(@containers)
+      
       lines = []
-      containers.each do |c|
+      @containers.each do |c|
         ct = c.get_token
 
         builder = []

data/lib/tsql_parser.rb

@@ -15,13 +15,15 @@
 #   TSqlParser
 
 module TSqlParser
+  require_relative "parsing/config/defaults"
+
   # Formats a SQL string.
   #
   # @param sql [String] the SQL string to format.
   # @param tab_count [Integer] the number of tabs to start with.
   # @param tab [String] the tab string.
   # @return [String] the formatted SQL string.
-  def self.format(sql, tab_count = 0, tab = "    ")
+  def self.format(sql, tab_count = Parsing::Defaults.get_default_tab_count, tab = Parsing::Defaults.get_default_tab)
     require_relative "parsing/formatter"
     tokens = self.parse(sql)
     Parsing::Formatter.format(tokens, tab_count, tab)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tsql_parser
 version: !ruby/object:Gem::Version
-  version: 0.1.5
+  version: 0.1.6
 platform: ruby
 authors:
 - Scott Stauffer
@@ -16,8 +16,13 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- lib/parsing/config/defaults.rb
 - lib/parsing/formatter.rb
+- lib/parsing/formatters/format_factory.rb
+- lib/parsing/formatters/strategy/__defaults.rb
+- lib/parsing/formatters/strategy/__formatters.rb
 - lib/parsing/formatters/strategy/base_formatter.rb
+- lib/parsing/formatters/strategy/cte_formatter.rb
 - lib/parsing/formatters/strategy/insert_formatter.rb
 - lib/parsing/formatters/strategy/join_formatter.rb
 - lib/parsing/formatters/strategy/select_formatter.rb
@@ -54,7 +59,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.6
+rubygems_version: 3.3.5
 signing_key: 
 specification_version: 4
 summary: A very light-weight and opinionated T-SQL parser and formatter.