From 40260a31cd197f655e6d3e0570a88d96fc1a9cac Mon Sep 17 00:00:00 2001
From: Hugo Herbelin <Hugo.Herbelin@inria.fr>
Date: Wed, 4 Oct 2017 15:37:36 +0200
Subject: Fixing BZ#5769 (variable of type "_something" was named after invalid
 "_").

---
 lib/unicode.ml  | 28 ++++++++++++++++++++++++++++
 lib/unicode.mli |  4 ++++
 2 files changed, 32 insertions(+)

(limited to 'lib')

diff --git a/lib/unicode.ml b/lib/unicode.ml
index 528d6434c..7f3743900 100644
--- a/lib/unicode.ml
+++ b/lib/unicode.ml
@@ -194,6 +194,14 @@ let is_unknown = function
   | Unknown -> true
   | Letter | IdentSep | IdentPart | Symbol -> false
 
+let is_ident_part = function
+  | IdentPart -> true
+  | Letter | IdentSep | Symbol | Unknown -> false
+
+let is_ident_sep = function
+  | IdentSep -> true
+  | Letter | IdentPart | Symbol | Unknown -> false
+
 let ident_refutation s =
   if s = ".." then None else try
     let j, n = next_utf8 s 0 in
@@ -227,6 +235,26 @@ let lowercase_first_char s =
   let j, n = next_utf8 s 0 in
   utf8_of_unicode (lowercase_unicode n)
 
+let split_at_first_letter s =
+  let n, v = next_utf8 s 0 in
+  if ((* optim *) n = 1 && s.[0] != '_') || not (is_ident_sep (classify v)) then None
+  else begin
+    let n = ref n in
+    let p = ref 0 in
+    while !n < String.length s &&
+          let n', v = next_utf8 s !n in
+          p := n';
+          (* Test if not letter *)
+          ((* optim *) n' = 1 && (s.[!n] = '_' || s.[!n] = '\''))
+          || let st = classify v in
+             is_ident_sep st || is_ident_part st
+    do n := !n + !p
+    done;
+    let s1 = String.sub s 0 !n in
+    let s2 = String.sub s !n (String.length s - !n) in
+    Some (s1,s2)
+  end
+
 (** For extraction, we need to encode unicode character into ascii ones *)
 
 let is_basic_ascii s =
diff --git a/lib/unicode.mli b/lib/unicode.mli
index a608d5f02..49b844493 100644
--- a/lib/unicode.mli
+++ b/lib/unicode.mli
@@ -30,6 +30,10 @@ val is_unknown : status -> bool
     @raise Assert_failure if the input string is empty. *)
 val lowercase_first_char : string -> string
 
+(** Split a string supposed to be an ident at the first letter;
+    as an optimization, return None if the first character is a letter *)
+val split_at_first_letter : string -> (string * string) option
+
 (** Return [true] if all UTF-8 characters in the input string are just plain
     ASCII characters. Returns [false] otherwise. *)
 val is_basic_ascii : string -> bool
-- 
cgit v1.2.3