jgrep 1.1.3 → 1.2.1

data/bin/jgrep

@@ -3,67 +3,87 @@
 require 'jgrep'
 require 'optparse'
 
-options = {:flat => false}
+options = {:flat => false, :start => nil}
 
-OptionParser.new do |opts|
-    opts.banner = "Usage: jgrep [options] \"expression\""
-    opts.on("-s", "--simple FIELD", "Display a single field from each of the resulting json documents") do |field|
-        options[:field] = field
-    end
-    opts.on("-f", "--flat", "Display non pretty json") do
-        options[:flat] = true
-    end
-    opts.on("-v", "--verbose", "Verbose output") do
-        JGrep::verbose_on
-    end
+begin
+    OptionParser.new do |opts|
+        opts.banner = "Usage: jgrep [options] \"expression\""
+        opts.on("-s", "--simple FIELD[s]", "Display a single field from each of the resulting json documents") do |field|
+            if (field.split(" ")).size > 1
+                options[:field] = field.split(" ")
+            else
+                options[:field] = field
+            end
+        end
 
-end.parse!
+        opts.on("-c", "--compact", "Display non pretty json") do
+            options[:flat] = true
+        end
+
+        opts.on("-f", "--flatten", "Makes output as flat as possible") do
+            JGrep::flatten_on
+        end
+
+        opts.on("-i", "--input [FILENAME]", "Specify input file to parse") do |filename|
+            options[:file] = filename
+        end
+
+        opts.on("-v", "--verbose", "Verbose output") do
+            JGrep::verbose_on
+        end
+
+        opts.on("--start [FIELD]", "Where in the data to start from") do |field|
+            options[:start] = field
+        end
+    end.parse!
+rescue OptionParser::InvalidOption => e
+    puts e.to_s.capitalize
+    exit 1
+end
 
 begin
     expression = nil
 
     #Identify the expression from command line arguments
     ARGV.each do |argument|
-        if argument =~ /<|>|=/
+        if argument =~ /<|>|=|\+|-/
             expression = argument
             ARGV.delete(argument)
         end
     end
 
-    #Load json from standard in if tty is false
+    expression = "" if expression.nil?
+
+    #Load json from standard input if tty is false
     #else find and load file from command line arugments
     unless STDIN.tty?
         json = STDIN.read
     else
-        if ARGV.size == 1
-            json = File.read(ARGV.first)
+        if options[:file]
+            json = File.read(options[:file])
         else
             raise "No json input specified"
         end
     end
 
     unless options[:field]
-        result = JGrep::jgrep((json), expression)
+        result = JGrep::jgrep((json), expression, nil, options[:start])
         unless result == []
             (options[:flat] == false) ? puts(JSON.pretty_generate(result)) : puts(result.to_json)
         end
-    else
-        if options[:field] =~ /=|<|>/
-            raise "'-s' flag requires a field to display"
-        end
 
-        result = JGrep::jgrep((json), expression)
+    else
+        if options[:field].is_a? Array
+            JGrep::validate_filters(options[:field])
+            puts(JSON.pretty_generate(JGrep::jgrep((json), expression, options[:field], options[:start])))
 
-        result.each do |r|
-            options[:field].split(".").each do |node|
-                unless r.is_a? Array
-                    r = r[node]
-                else
-                    break
-                end
-            end
-            unless r.nil?
-                (!r.is_a?(Array) && !r.is_a?(Hash)) ? puts(r) : puts(JSON.pretty_generate(r))
+        else
+            JGrep::validate_filters(options[:field])
+            result = JGrep::jgrep((json), expression, options[:field], options[:start])
+            unless result.first.is_a?(Hash) || result.flatten.first.is_a?(Hash)
+                result.map{|x| puts x unless x.nil?}
+            else
+                puts(JSON.pretty_generate(result))
             end
         end
     end

data/jgrep.gemspec

@@ -1,9 +1,9 @@
 Gem::Specification.new do |s|
     s.name = "jgrep"
-    s.version = "1.1.3"
+    s.version = "1.2.1"
 
     s.authors = ["P Loubser"]
-    s.date = %q{2011-07-27}
+    s.date = %q{2011-08-01}
     s.default_executable = "jgrep"
     s.description = "Compare a list of json documents to a simple logical language and returns matches as output"
     s.email = ["ploubser@gmail.com"]

data/lib/jgrep.rb

@@ -7,38 +7,60 @@ require 'json'
 
 module JGrep
     @verbose = false
+    @flatten = false
 
     def self.verbose_on
         @verbose = true
     end
 
+    def self.flatten_on
+        @flatten = true
+    end
+
     #Method parses json and returns documents that match the logical expression
-    def self.jgrep(json, expression)
+    def self.jgrep(json, expression, filters = nil, start = nil)
         errors = ""
         begin
-            call_stack = Parser.new(expression).execution_stack
-            result = []
+            JSON.create_id = nil
             json = JSON.parse(json)
-            json.each do |document|
-                begin
-                    if eval_statement(document, call_stack)
-                        result << document
-                    end
-                rescue Exception => e
-                    if @verbose
-                        pp document
-                        STDERR.puts "Error - #{e} \n\n"
-                    else
-                        errors = "One or more the json documents could not be parsed. Run jgrep -v for to display documents"
+            if json.is_a? Hash
+                json = [json]
+            end
+
+            json = filter_json(json, start).flatten if start
+
+            result = []
+            unless expression == ""
+                call_stack = Parser.new(expression).execution_stack
+
+                json.each do |document|
+                    begin
+                        if eval_statement(document, call_stack)
+                            result << document
+                        end
+                    rescue Exception => e
+                        if @verbose
+                            require 'pp'
+                            pp document
+                            STDERR.puts "Error - #{e} \n\n"
+                        else
+                            errors = "One or more the json documents could not be parsed. Run jgrep -v for to display documents"
+                        end
                     end
                 end
+            else
+                result = json
             end
 
             unless errors == ""
                 puts errors
             end
 
-            return result
+            unless filters
+                return result
+            else
+                filter_json(result, filters)
+            end
 
         rescue JSON::ParserError => e
             STDERR.puts "Error. Invalid JSON given"
@@ -47,6 +69,54 @@ module JGrep
 
     end
 
+    #Strips filters from json documents and returns those values as a less bloated json document
+    def self.filter_json(documents, filters)
+        result = []
+
+        if filters.is_a? Array
+            documents.each do |doc|
+                tmp_json = {}
+                filters.each do |filter|
+                    filtered_result = dig_path(doc, filter)
+                    unless (filtered_result == doc) || filtered_result.nil?
+                        tmp_json[filter] = filtered_result
+                    end
+                end
+                result << tmp_json
+            end
+
+            result = result.flatten if (result.size == 1 && @flatten == true)
+            return result
+
+        else
+            documents.each do |r|
+                filtered_result = dig_path(r, filters)
+                unless (filtered_result == r) || filtered_result.nil?
+                    result << filtered_result
+                end
+            end
+
+            result = result.flatten if (result.size == 1 && @flatten == true)
+            return result
+        end
+    end
+
+    #Validates if filters do not match any of the parser's logical tokens
+    def self.validate_filters(filters)
+        if filters.is_a? Array
+            filters.each do |filter|
+                if filter =~ /=|<|>|^and$|^or$|^!$|^not$/
+                    raise "Invalid field for -s filter : '#{filter}'"
+                end
+            end
+        else
+            if filters =~ /=|<|>|^and$|^or$|^!$|^not$/
+                raise "Invalid field for -s filter : '#{filters}'"
+            end
+        end
+        return
+    end
+
     #Correctly format values so we can do the correct type of comparison
     def self.format(kvalue, value)
         if kvalue.to_s =~ /^\d+$/ && value.to_s =~ /^\d+$/
@@ -58,6 +128,34 @@ module JGrep
         end
     end
 
+    def self.present?(document, statement)
+        statement.split(".").each do |key|
+            if document.is_a? Hash
+                if document.has_value? nil
+                    document.each do |k, v|
+                        if document[k] == nil
+                            document[k] = "null"
+                        end
+                    end
+                end
+            end
+
+            if document.is_a? Array
+                rval = false
+                document.each do |doc|
+                    rval ||= present?(doc, key)
+                end
+                return rval
+            end
+
+            document = document[key]
+            if document.nil?
+                return false
+            end
+        end
+        return true
+    end
+
     #Check if key=value is present in document
     def self.has_object?(document, statement)
 
@@ -74,7 +172,11 @@ module JGrep
         key.split(".").each_with_index do |item,i|
             tmp = tmp[item]
             if tmp.nil?
-                return false
+                if item == key.split(".").last
+                    tmp = "null"
+                else
+                    return false
+                end
             end
             result = false
             if tmp.is_a? Array
@@ -82,7 +184,7 @@ module JGrep
             end
         end
 
-        tmp, value = format(tmp, value.gsub(/"|'/, ""))
+        tmp, value = format(tmp, (value.gsub(/"|'/, "") unless value.nil?))
 
         case op
             when "="
@@ -170,6 +272,10 @@ module JGrep
                 else
                     result << has_object?(document, expression.values.first)
                 end
+            when "+"
+                result << present?(document, expression.values.first)
+            when "-"
+                result << !(present?(document, expression.values.first))
             when "and"
                 result << "&&"
             when "or"
@@ -185,4 +291,41 @@ module JGrep
 
         return eval(result.join(" "))
     end
+
+    def self.dig_path(json, path)
+
+        json = json[path.split(".").first] if json.is_a? Hash
+
+        if json.is_a?(Hash)
+            if path == path.split(".").first
+                return json
+            else
+                return dig_path(json, (path.match(/\./) ? path.split(".").drop(1).join(".") : path))
+            end
+
+        elsif json.is_a? Array
+            if path == path.split(".").first && (json.first.is_a?(Hash) && !(json.first.keys.include?(path)))
+                return json
+            else
+                tmp = []
+                json.each do |j|
+                    tmp_path = dig_path(j, (path.match(/\./) ? path.split(".").drop(1).join(".") : path))
+                    unless tmp_path.nil?
+                        tmp << tmp_path
+                    end
+                end
+                unless tmp.empty?
+                    return tmp
+                end
+            end
+
+        elsif json.nil?
+            return nil
+
+        else
+            return json
+
+        end
+
+    end
 end

data/lib/parser/parser.rb

@@ -33,7 +33,7 @@ module JGrep
                     case c_token
 
                         when "and"
-                            unless (n_token =~ /not|statement|\(/) || (scanner.token_index == scanner.arguments.size)
+                            unless (n_token =~ /not|statement|\(|\+|-/) || (scanner.token_index == scanner.arguments.size)
                                 raise "Error at column #{scanner.token_index}. \nExpected 'not', 'statement' or '('. Found '#{n_token_value}'"
                             end
 
@@ -44,7 +44,7 @@ module JGrep
                             end
 
                         when "or"
-                            unless (n_token =~ /not|statement|\(/) || (scanner.token_index == scanner.arguments.size)
+                            unless (n_token =~ /not|statement|\(|\+|-/) || (scanner.token_index == scanner.arguments.size)
                                 raise "Error at column #{scanner.token_index}. \nExpected 'not', 'statement', '('. Found '#{n_token_value}'"
                             end
 
@@ -55,7 +55,7 @@ module JGrep
                             end
 
                         when "not"
-                            unless n_token =~ /statement|\(|not/
+                            unless n_token =~ /statement|\(|not|\+|-/
                                 raise "Error at column #{scanner.token_index}. \nExpected 'statement' or '('. Found '#{n_token_value}'"
                             end
 
@@ -69,7 +69,7 @@ module JGrep
 
                             if c_token_value =~ /!=/
                                 c_token_value = c_token_value.gsub("!=", "=")
-                                @execution_stack << {"not", "not"}
+                                @execution_stack << {"not" => "not"}
                             end
 
                             unless n_token =~ /and|or|\)/
@@ -78,6 +78,21 @@ module JGrep
                                 end
                             end
 
+                        when "+"
+                            unless n_token =~ /and|or|\)/
+                                unless n_token.nil?
+                                    raise "Error at column #{scanner.token_index}. \nExpected 'and', 'or', ')'. Found '#{n_token_value}'"
+                                end
+                            end
+
+                        when "-"
+                            unless n_token =~ /and|or|\)/
+                                unless n_token.nil?
+                                    raise "Error at column #{scanner.token_index}. \nExpected 'and', 'or', ')'. Found '#{n_token_value}'"
+                                end
+                            end
+
+
                         when ")"
                             unless (n_token =~ /|and|or|not|\(/)
                                 unless n_token.nil?
@@ -87,7 +102,7 @@ module JGrep
                             parenth += 1
 
                         when "("
-                            unless n_token =~ /statement|not|\(/
+                            unless n_token =~ /statement|not|\(|\+|-/
                                 raise "Error at column #{scanner.token_index}. \nExpected 'statement', '(',  not. Found '#{n_token_value}'"
                             end
                             parenth -= 1

data/lib/parser/scanner.rb

@@ -54,6 +54,31 @@ module JGrep
                             gen_statement
                         end
 
+                    when "+"
+                        value = ""
+                        i = @token_index + 1
+
+                        begin
+                            value +=  @arguments.split("")[i]
+                            i += 1
+                        end until (i >= @arguments.size)  || (@arguments.split("")[i] =~ /\s|\)/)
+
+                        @token_index = i - 1
+                        return "+", value
+
+                    when "-"
+                        value = ""
+                        i = @token_index + 1
+
+                        begin
+                            value +=  @arguments.split("")[i]
+                            i += 1
+                        end until (i >= @arguments.size)  || (@arguments.split("")[i] =~ /\s|\)/)
+
+                        @token_index = i - 1
+                        return "-", value
+
+
                     when " "
                         return " ", " "
 

metadata

@@ -4,9 +4,9 @@ version: !ruby/object:Gem::Version
   prerelease: false
   segments: 
   - 1
+  - 2
   - 1
-  - 3
-  version: 1.1.3
+  version: 1.2.1
 platform: ruby
 authors: 
 - P Loubser
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-07-27 00:00:00 +01:00
+date: 2011-08-01 00:00:00 +01:00
 default_executable: jgrep
 dependencies: []