From bc3a2a7a96fd4bae5c69943fec3bcb842b8ba222 Mon Sep 17 00:00:00 2001
From: Gregg Kellogg <gregg@greggkellogg.net>
Date: Wed, 23 Nov 2016 12:03:38 -0800
Subject: [PATCH] Properly handle comments within middle or at end of rule

---
 .gitignore         |  1 +
 bin/ebnf           |  6 +++---
 lib/ebnf/parser.rb | 20 +++++++++++++++++---
 spec/base_spec.rb  | 12 ++++++++++++
 4 files changed, 33 insertions(+), 6 deletions(-)

diff --git a/.gitignore b/.gitignore
index 67dc548..7a48f73 100644
--- a/.gitignore
+++ b/.gitignore
@@ -17,3 +17,4 @@ tmp
 .yardoc
 _yardoc
 /Gemfile.lock
+/.byebug_history
diff --git a/bin/ebnf b/bin/ebnf
index b59982c..c64786d 100755
--- a/bin/ebnf
+++ b/bin/ebnf
@@ -17,10 +17,10 @@ options = {
   :namespace => "http://www.w3.org/ns/formats/Turtle#",
 }
 
-out = STDOUT
+input, out = nil, STDOUT
 
 OPT_ARGS = [
-  ["--dbg",             GetoptLong::NO_ARGUMENT,      "Turn on debugging output"],
+  ["--debug",           GetoptLong::NO_ARGUMENT,      "Turn on debugging output"],
   ["--bnf",             GetoptLong::NO_ARGUMENT,      "Transform EBNF to BNF"],
   ["--evaluate","-e",   GetoptLong::REQUIRED_ARGUMENT,"Evaluate argument as an EBNF document"],
   ["--ll1",             GetoptLong::REQUIRED_ARGUMENT,"Generate First/Follow rules, argument is start symbol"],
@@ -53,7 +53,7 @@ opts = GetoptLong.new(*OPT_ARGS.map {|o| o[0..-2]})
 
 opts.each do |opt, arg|
   case opt
-  when '--dbg'          then options[:debug] = true
+  when '--debug'        then options[:debug] = true
   when '--bnf'          then options[:bnf] = true
   when '--evaluate'     then input = arg
   when '--input-format' then options[:format] = arg.to_sym
diff --git a/lib/ebnf/parser.rb b/lib/ebnf/parser.rb
index 35ab720..ecb60c8 100644
--- a/lib/ebnf/parser.rb
+++ b/lib/ebnf/parser.rb
@@ -18,14 +18,20 @@ def eachRule(scanner)
           #debug("eachRule(ws)") { "[#{cur_lineno}] #{s.inspect}" }
         when s = scanner.scan(%r(/\*([^\*]|\*[^\/])*\*/)m)
           # Eat comments /* .. */
+          cur_lineno += s.count("\n")
           debug("eachRule(comment)") { "[#{cur_lineno}] #{s.inspect}" }
         when s = scanner.scan(%r(\(\*([^\*]|\*[^\)])*\*\))m)
           # Eat comments (* .. *)
+          cur_lineno += s.count("\n")
           debug("eachRule(comment)") { "[#{cur_lineno}] #{s.inspect}" }
         when s = scanner.scan(%r((#(?!x)|//).*$))
-          # Eat comments
+          # Eat comments // & #
           cur_lineno += s.count("\n")
           debug("eachRule(comment)") { "[#{cur_lineno}] #{s.inspect}" }
+        when s = scanner.scan(/\A["']/)
+          # Found a quote, scan until end of matching quote
+          s += scanner.scan_until(/#{scanner.matched}|$/)
+          r += s
         when s = scanner.scan(%r(^@terminals))
           #debug("eachRule(@terminals)") { "[#{cur_lineno}] #{s.inspect}" }
           yield(r) unless r.empty?
@@ -45,8 +51,15 @@ def eachRule(scanner)
           @lineno = cur_lineno
           r = s
         else
-          # Collect until end of line, or start of comment
-          s = scanner.scan_until(%r((?:/\*)|$)m)
+          # Collect until end of line, or start of comment or quote
+          s = scanner.scan_until(%r{(?:[/\(]\*)|#(?!x)|//|["']|$})
+          if scanner.matched.length > 0
+            # Back up scan head before ending match
+            scanner.pos = scanner.pos - scanner.matched.length
+
+            # Remove matched from end of string
+            s = s[0..-(scanner.matched.length+1)]
+          end
           cur_lineno += s.count("\n")
           #debug("eachRule(rest)") { "[#{cur_lineno}] #{s.inspect}" }
           r += s
@@ -268,6 +281,7 @@ def primary(s)
     #     ((range "^<>'{}|^`") '-\[#x00-#x20\]')
     def terminal(s)
       s = s.strip
+      #STDERR.puts s.inspect
       case m = s[0,1]
       when '"', "'" # STRING1 or STRING2
         l, s = s[1..-1].split(m.rstrip, 2)
diff --git a/spec/base_spec.rb b/spec/base_spec.rb
index 074291d..a46d490 100644
--- a/spec/base_spec.rb
+++ b/spec/base_spec.rb
@@ -34,6 +34,18 @@
         %q{((terminal STRING1 "18" (seq "\"" (star (alt CHAR (range "\t'[]()-"))) "\"")))},
       %q{[161s] WS ::= #x20 | #x9 | #xD | #xA} =>
         %q{((terminal WS "161s" (alt (hex "#x20") (hex "#x9") (hex "#xD") (hex "#xA"))))},
+      %q{[1] shexDoc ::= directive* # leading CODE} =>
+        %q{((rule shexDoc "1" (star directive)))},
+      %q{[1] shexDoc ::= directive* /* leading CODE */} =>
+        %q{((rule shexDoc "1" (star directive)))},
+      %q{[1] shexDoc ::= directive* (* leading CODE *)} =>
+        %q{((rule shexDoc "1" (star directive)))},
+      %q{[1] shexDoc ::= directive* // leading CODE} =>
+        %q{((rule shexDoc "1" (star directive)))},
+      %q{[1] shexDoc ::= /* leading CODE */ directive*} =>
+        %q{((rule shexDoc "1" (star directive)))},
+      %q{[1] shexDoc (* leading CODE *) ::= directive*} =>
+        %q{((rule shexDoc "1" (star directive)))},
     }.each do |input, expected|
       it "parses #{input.inspect}" do
         expect(parse(input).to_sxp).to produce(expected, @debug)