From cd6d5dba53efd0d5037d8a55affc7b6d52f7fc80 Mon Sep 17 00:00:00 2001
From: Rudi Grinberg <rudi.grinberg@gmail.com>
Date: Wed, 20 Jun 2018 20:00:35 +0630
Subject: [PATCH] Simplify jbuild lexer and move types around

The types should exist in the lexer module where they're used

Signed-off-by: Rudi Grinberg <rudi.grinberg@gmail.com>
---
 src/usexp/dune_lexer.mll   |  6 +++++-
 src/usexp/jbuild_lexer.mll | 25 ++++++++++++++++---------
 src/usexp/lexer0.ml        | 14 --------------
 src/usexp/lexer0.mli       |  9 ---------
 4 files changed, 21 insertions(+), 33 deletions(-)

diff --git a/src/usexp/dune_lexer.mll b/src/usexp/dune_lexer.mll
index e6e4dcf0..2c3a10a2 100644
--- a/src/usexp/dune_lexer.mll
+++ b/src/usexp/dune_lexer.mll
@@ -1,5 +1,9 @@
 {
-  open Lexer0
+open Lexer0
+
+type block_string_line_kind =
+  | With_escape_sequences
+  | Raw
 }
 
 let comment   = ';' [^ '\n' '\r']*
diff --git a/src/usexp/jbuild_lexer.mll b/src/usexp/jbuild_lexer.mll
index 4a2d546e..4cfade63 100644
--- a/src/usexp/jbuild_lexer.mll
+++ b/src/usexp/jbuild_lexer.mll
@@ -1,5 +1,14 @@
 {
   open Lexer0
+
+(* The difference between the old and new syntax is that the old
+   syntax allows backslash following by any characters other than 'n',
+   'x', ... and interpret it as it. The new syntax is stricter in
+   order to allow introducing new escape sequence in the future if
+   needed. *)
+type escape_mode =
+  | In_block_comment (* Inside #|...|# comments (old syntax) *)
+  | In_quoted_string
 }
 
 let comment   = ';' [^ '\n' '\r']*
@@ -24,7 +33,7 @@ rule token = parse
   | '"'
     { Buffer.clear escaped_buf;
       let start = Lexing.lexeme_start_p lexbuf in
-      let s = quoted_string Old_syntax lexbuf in
+      let s = quoted_string In_quoted_string lexbuf in
       lexbuf.lex_start_p <- start;
       Quoted_string s
     }
@@ -77,7 +86,7 @@ and quoted_string mode = parse
       quoted_string mode lexbuf
     }
   | eof
-    { if mode <> In_block_comment then
+    { if mode = In_block_comment then
         error lexbuf "unterminated quoted string";
       Buffer.contents escaped_buf
     }
@@ -120,14 +129,14 @@ and escape_sequence mode = parse
     }
   | (digit as c1) (digit as c2) (digit as c3)
     { let v = eval_decimal_escape c1 c2 c3 in
-      if mode <> In_block_comment && v > 255 then
+      if mode = In_quoted_string && v > 255 then
         error lexbuf "escape sequence in quoted string out of range"
           ~delta:(-1);
       Buffer.add_char escaped_buf (Char.chr v);
       Other
     }
   | digit* as s
-    { if mode <> In_block_comment then
+    { if mode = In_quoted_string then
         error lexbuf "unterminated decimal escape sequence" ~delta:(-1);
       Buffer.add_char escaped_buf '\\';
       Buffer.add_string escaped_buf s;
@@ -139,21 +148,19 @@ and escape_sequence mode = parse
       Other
     }
   | 'x' hexdigit* as s
-    { if mode <> In_block_comment then
+    { if mode = In_quoted_string then
         error lexbuf "unterminated hexadecimal escape sequence" ~delta:(-1);
       Buffer.add_char escaped_buf '\\';
       Buffer.add_string escaped_buf s;
       Other
     }
   | _ as c
-    { if mode = New_syntax then
-        error lexbuf "unknown escape sequence" ~delta:(-1);
-      Buffer.add_char escaped_buf '\\';
+    { Buffer.add_char escaped_buf '\\';
       Buffer.add_char escaped_buf c;
       Other
     }
   | eof
-    { if mode <> In_block_comment then
+    { if mode = In_quoted_string then
         error lexbuf "unterminated escape sequence" ~delta:(-1);
       Other
     }
diff --git a/src/usexp/lexer0.ml b/src/usexp/lexer0.ml
index 5ebc545a..b361acc2 100644
--- a/src/usexp/lexer0.ml
+++ b/src/usexp/lexer0.ml
@@ -31,24 +31,10 @@ let error ?(delta=0) lexbuf message =
 
 let escaped_buf = Buffer.create 256
 
-(* The difference between the old and new syntax is that the old
-   syntax allows backslash following by any characters other than 'n',
-   'x', ... and interpret it as it. The new syntax is stricter in
-   order to allow introducing new escape sequence in the future if
-   needed. *)
-type escape_mode =
-  | In_block_comment (* Inside #|...|# comments (old syntax) *)
-  | Old_syntax
-  | New_syntax
-
 type escape_sequence =
   | Newline
   | Other
 
-type block_string_line_kind =
-  | With_escape_sequences
-  | Raw
-
 let eval_decimal_char c = Char.code c - Char.code '0'
 
 let eval_decimal_escape c1 c2 c3 =
diff --git a/src/usexp/lexer0.mli b/src/usexp/lexer0.mli
index fc15c06a..b442888d 100644
--- a/src/usexp/lexer0.mli
+++ b/src/usexp/lexer0.mli
@@ -24,19 +24,10 @@ val escaped_buf : Buffer.t
 
 exception Error of Error.t
 
-type escape_mode =
-  | In_block_comment (* Inside #|...|# comments (old syntax) *)
-  | Old_syntax
-  | New_syntax
-
 type escape_sequence =
   | Newline
   | Other
 
-type block_string_line_kind =
-  | With_escape_sequences
-  | Raw
-
 val eval_decimal_char : char -> int
 
 val eval_decimal_escape : char -> char -> char -> int