From 9e91f407c27cbb20c47f77de62c2f5af48c0ed01 Mon Sep 17 00:00:00 2001
From: Gwenhael Le Moine <gwenhael.le.moine@gmail.com>
Date: Sat, 30 Oct 2021 23:36:49 +0200
Subject: [PATCH] split out parser

---
 lib/parser.rb | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++
 repl.rb       | 31 +++-----------------------
 2 files changed, 65 insertions(+), 28 deletions(-)
 create mode 100644 lib/parser.rb

diff --git a/lib/parser.rb b/lib/parser.rb
new file mode 100644
index 0000000..b6e09ed
--- /dev/null
+++ b/lib/parser.rb
@@ -0,0 +1,62 @@
+# coding: utf-8
+
+class String
+  def numeric?
+    Float(self) != nil rescue false
+  end
+end
+
+def parse_input( input )
+  splitted_input = input.split(' ')
+  parsed_tree = []
+
+  regrouping = false
+  splitted_input.each do |elt|
+    next if elt.length == 1 && elt[0] == '»'
+
+    parsed_entry = { 'value' => elt }
+
+    if regrouping
+      parsed_entry = parsed_tree.pop
+
+      parsed_entry['value'] = "#{parsed_entry['value']} #{elt}".strip
+    else
+      parsed_entry['type'] = case elt[0]
+                             when '«'
+                               'PROGRAM'
+                             when '"'
+                               'STRING'
+                             when '\''
+                               'NAME'
+                             else
+                               if elt.numeric?
+                                 'NUMBER'
+                               else
+                                 'WORD' # TODO: if word isn't known then it's a NAME
+                               end
+                             end
+
+      parsed_entry['value'] = elt[1..] if %W[PROGRAM STRING NAME].include?( parsed_entry['type'] )
+    end
+
+    regrouping = ( (parsed_entry['type'] == 'NAME' && elt[-1] != '\'') ||
+                   (parsed_entry['type'] == 'STRING' && elt[-1] != '"') ||
+                   (parsed_entry['type'] == 'PROGRAM'  && elt[-1] != '»') )
+
+    parsed_entry['value'] = if ( (parsed_entry['type'] == 'NAME' && elt[-1] == '\'') ||
+                                 (parsed_entry['type'] == 'STRING'  && elt[-1] == '"') ||
+                                 (parsed_entry['type'] == 'PROGRAM' && elt[-1] == '»') )
+                              (parsed_entry['value'][..-2]).strip
+                            elsif parsed_entry['type'] == 'NUMBER'
+                              parsed_entry['value'].to_f
+                            elsif parsed_entry['type'] == 'WORD'
+                              parsed_entry['value']
+                            else
+                              parsed_entry['value']
+                            end
+
+    parsed_tree << parsed_entry
+  end
+
+  parsed_tree
+end
diff --git a/repl.rb b/repl.rb
index 8d0e99e..9318a93 100644
--- a/repl.rb
+++ b/repl.rb
@@ -3,6 +3,8 @@
 
 require 'readline'
 
+require_relative "./lib/parser"
+
 def run_REPL( stack )
   Readline.completion_proc = proc do |s|
     directory_list = Dir.glob("#{s}*")
@@ -27,33 +29,6 @@ def run_REPL( stack )
   end
 end
 
-def parse_input( input )
-  splitted_input = input.split( " " )
-  parsed_input = []
-
-  regrouping = false
-  splitted_input.each do |elt|
-    if regrouping
-      partial_elt = parsed_input.pop
-      elt = "#{partial_elt} #{elt}"
-    end
-
-    if regrouping
-      regrouping = false if (elt[0] == '\'' and elt[-1] == '\'') or (elt[0] == '"' and elt[-1] == '"') or (elt[0] == '«' and elt[-1] == '»')
-    else
-      regrouping = true if elt[0] == '\'' or elt[0] == '"' or elt[0] == '«'
-    end
-
-    # 'xx' is a name (no space allowed)
-    # "xx x x xx" is a string
-    # « xx xx xx » is a program (must have inner spaces)
-
-    parsed_input << elt
-  end
-
-  parsed_input
-end
-
 def process_input( stack, input )
   parse_input( input ).each do |elt|
     stack << elt
@@ -64,7 +39,7 @@ end
 
 def display_stack( stack )
   stack_size = stack.size
-  stack.each_with_index { |v, i| puts "#{stack_size - i}: #{v}"}
+  stack.each_with_index { |elt, i| puts "#{stack_size - i}: #{elt['value']}"}
 
   stack
 end