Add support for block strings

Signed-off-by: Jeremie Dimino <jdimino@janestreet.com>

src/usexp/lexer.mll

@@ -56,6 +56,10 @@ type escape_sequence =
   | Other
 
 let escaped_buf = Buffer.create 256
+
+type block_string_line_kind =
+  | With_escape_sequences
+  | Raw
 }
 
 let comment   = ';' [^ '\n' '\r']*
@@ -143,7 +147,8 @@ and quoted_string_after_escaped_newline strict = parse
 
 and escape_sequence strict = parse
   | newline
-    { Newline }
+    { Lexing.new_line lexbuf;
+      Newline }
   | ['\\' '\'' '"' 'n' 't' 'b' 'r'] as c
     { let c =
         match c with
@@ -223,7 +228,7 @@ and token = parse
   | '"'
     { Buffer.clear escaped_buf;
       let start = Lexing.lexeme_start_p lexbuf in
-      let s = quoted_string true lexbuf in
+      let s = dune_quoted_string lexbuf in
       lexbuf.lex_start_p <- start;
       Quoted_string s
     }
@@ -231,3 +236,71 @@ and token = parse
     { Token.Atom (A s) }
   | eof
     { Eof }
+
+and dune_quoted_string = parse
+  | "\\|"
+    { block_string_start With_escape_sequences lexbuf }
+  | "\\>"
+    { block_string_start Raw lexbuf }
+  | ""
+    { quoted_string true lexbuf }
+
+and block_string_start kind = parse
+  | newline as s
+    { Lexing.new_line lexbuf;
+      Buffer.add_string escaped_buf s;
+      block_string_after_newline lexbuf
+    }
+  | ' '
+    { match kind with
+      | With_escape_sequences -> block_string lexbuf
+      | Raw -> raw_block_string lexbuf
+    }
+  | eof
+    { Buffer.contents escaped_buf
+    }
+  | _
+    { error lexbuf "There must be at least one space after \"\\|"
+    }
+
+and block_string = parse
+  | newline as s
+    { Lexing.new_line lexbuf;
+      Buffer.add_string escaped_buf s;
+      block_string_after_newline lexbuf
+    }
+  | '\\'
+    { match escape_sequence true lexbuf with
+      | Newline -> block_string_after_newline lexbuf
+      | Other   -> block_string               lexbuf
+    }
+  | _ as c
+    { Buffer.add_char escaped_buf c;
+      block_string lexbuf
+    }
+  | eof
+    { Buffer.contents escaped_buf
+    }
+
+and block_string_after_newline = parse
+  | blank* "\"\\|"
+    { block_string_start With_escape_sequences lexbuf }
+  | blank* "\"\\>"
+    { block_string_start Raw lexbuf }
+  | ""
+    { Buffer.contents escaped_buf
+    }
+
+and raw_block_string = parse
+  | newline as s
+    { Lexing.new_line lexbuf;
+      Buffer.add_string escaped_buf s;
+      block_string_after_newline lexbuf
+    }
+  | _ as c
+    { Buffer.add_char escaped_buf c;
+      raw_block_string lexbuf
+    }
+  | eof
+    { Buffer.contents escaped_buf
+    }

test/blackbox-tests/dune.inc

@@ -14,6 +14,14 @@
     test-cases/bad-alias-error
     (progn (run ${exe:cram.exe} -test run.t) (diff? run.t run.t.corrected))))))
 
+(alias
+ ((name block-strings)
+  (deps ((package dune) (files_recursively_in test-cases/block-strings)))
+  (action
+   (chdir
+    test-cases/block-strings
+    (progn (run ${exe:cram.exe} -test run.t) (diff? run.t run.t.corrected))))))
+
 (alias
  ((name byte-code-only)
   (deps ((package dune) (files_recursively_in test-cases/byte-code-only)))
@@ -521,6 +529,7 @@
   (deps
    ((alias aliases)
     (alias bad-alias-error)
+    (alias block-strings)
     (alias byte-code-only)
     (alias c-stubs)
     (alias configurator)
@@ -583,6 +592,7 @@
   (deps
    ((alias aliases)
     (alias bad-alias-error)
+    (alias block-strings)
     (alias byte-code-only)
     (alias c-stubs)
     (alias configurator)

test/blackbox-tests/test-cases/block-strings/dune

@@ -0,0 +1,48 @@
+; Example script taken from re2
+
+; With normal strings
+(alias
+ ((name old)
+  (action (echo "\
+ARFLAGS=rsc
+CXX=g++
+CXXFLAGS=\"-Wall -O3 -g -pthread\"
+if ! ${.ARCH_SIXTYFOUR}; then
+  CXX=\"$CXX -m32\"
+fi
+${.MAKE} -s -C libre2 clean
+${.MAKE} -s -C libre2 \\
+  ARFLAGS=\"$ARFLAGS\" \\
+  CXX=\"$CXX\" \\
+  CXXFLAGS=\"$CXXFLAGS\" \\
+  obj/libre2.a obj/so/libre2.so
+cp libre2/obj/libre2.a libre2_c_stubs.a
+cp libre2/obj/so/libre2.so dllre2_c_stubs.so
+${.MAKE} -s -C libre2 clean
+"))))
+
+; With block strings
+(alias
+ ((name new)
+  (action (echo "\> ARFLAGS=rsc
+                "\> CXX=g++
+                "\> CXXFLAGS="-Wall -O3 -g -pthread"
+                "\> if ! ${.ARCH_SIXTYFOUR}; then
+                "\>   CXX="$CXX -m32"
+                "\> fi
+                "\> ${.MAKE} -s -C libre2 clean
+                "\> ${.MAKE} -s -C libre2 \
+                "\>   ARFLAGS="$ARFLAGS" \
+                "\>   CXX="$CXX" \
+                "\>   CXXFLAGS="$CXXFLAGS" \
+                "\>   obj/libre2.a obj/so/libre2.so
+                "\> cp libre2/obj/libre2.a libre2_c_stubs.a
+                "\> cp libre2/obj/so/libre2.so dllre2_c_stubs.so
+                "\> ${.MAKE} -s -C libre2 clean
+))))
+
+(alias
+ ((name quoting-test)
+  (action (echo "\| normal: \065
+                "\> raw:    \065
+))))

test/blackbox-tests/test-cases/block-strings/run.t

@@ -0,0 +1,37 @@
+  $ dune build @old
+  ARFLAGS=rsc
+  CXX=g++
+  CXXFLAGS="-Wall -O3 -g -pthread"
+  if ! ${.ARCH_SIXTYFOUR}; then
+    CXX="$CXX -m32"
+  fi
+  ${.MAKE} -s -C libre2 clean
+  ${.MAKE} -s -C libre2 \
+    ARFLAGS="$ARFLAGS" \
+    CXX="$CXX" \
+    CXXFLAGS="$CXXFLAGS" \
+    obj/libre2$ext_lib obj/so/libre2$ext_dll
+  cp libre2/obj/libre2$ext_lib libre2_c_stubs$ext_lib
+  cp libre2/obj/so/libre2$ext_dll dllre2_c_stubs$ext_dll
+  ${.MAKE} -s -C libre2 clean
+
+  $ dune build @new
+  ARFLAGS=rsc
+  CXX=g++
+  CXXFLAGS="-Wall -O3 -g -pthread"
+  if ! ${.ARCH_SIXTYFOUR}; then
+    CXX="$CXX -m32"
+  fi
+  ${.MAKE} -s -C libre2 clean
+  ${.MAKE} -s -C libre2 \
+    ARFLAGS="$ARFLAGS" \
+    CXX="$CXX" \
+    CXXFLAGS="$CXXFLAGS" \
+    obj/libre2$ext_lib obj/so/libre2$ext_dll
+  cp libre2/obj/libre2$ext_lib libre2_c_stubs$ext_lib
+  cp libre2/obj/so/libre2$ext_dll dllre2_c_stubs$ext_dll
+  ${.MAKE} -s -C libre2 clean
+
+  $ dune build @quoting-test
+  normal: A
+  raw:    \065