From 526791d917f9b0804376eae02a462a3b32dd7cba Mon Sep 17 00:00:00 2001
From: Hugo Herbelin
Date: Wed, 4 Oct 2017 22:00:19 +0200
Subject: Distinguishing pseudo-letters out of the set of unicode letters.

This includes _ and insecable space which can be used in idents and
this allows more precise heuristics.
---
 lib/unicode.mli | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

(limited to 'lib/unicode.mli')

diff --git a/lib/unicode.mli b/lib/unicode.mli
index c7d7424801..a608d5f026 100644
--- a/lib/unicode.mli
+++ b/lib/unicode.mli
@@ -8,7 +8,7 @@
 
 (** Unicode utilities *)
 
-type status = Letter | IdentPart | Symbol | Unknown
+type status
 
 (** Classify a unicode char into 3 classes or unknown. *)
 val classify : int -> status
@@ -17,6 +17,15 @@ val classify : int -> status
     Return [Some (b,s)] otherwise, where [s] is an explanation and [b] is severity. *)
 val ident_refutation : string -> (bool * string) option
 
+(** Tells if a valid initial character for an identifier *)
+val is_valid_ident_initial : status -> bool
+
+(** Tells if a valid non-initial character for an identifier *)
+val is_valid_ident_trailing : status -> bool
+
+(** Tells if a character is unclassified *)
+val is_unknown : status -> bool
+
 (** First char of a string, converted to lowercase
     @raise Assert_failure if the input string is empty. *)
 val lowercase_first_char : string -> string
-- 
cgit v1.2.3


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

---
 lib/unicode.mli | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'lib/unicode.mli')

diff --git a/lib/unicode.mli b/lib/unicode.mli
index a608d5f026..49b8444937 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