From da3e1e94fa33e739e19a90e966dfbccefae42205 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Nicol=C3=A1s=20Antinori?= <nicolas_antinori@hotmail.com>
Date: Sat, 8 Feb 2025 14:05:37 -0300
Subject: [PATCH] added get_char_encoded_length function

---
 src/header/wchar/lookaheadreader.rs | 18 ++++++------------
 src/header/wchar/mod.rs             | 12 +++---------
 src/header/wchar/utf8.rs            | 15 +++++++++++++++
 3 files changed, 24 insertions(+), 21 deletions(-)

diff --git a/src/header/wchar/lookaheadreader.rs b/src/header/wchar/lookaheadreader.rs
index b1e1cfea49..516749084b 100644
--- a/src/header/wchar/lookaheadreader.rs
+++ b/src/header/wchar/lookaheadreader.rs
@@ -2,7 +2,7 @@ use super::{fseek_locked, ftell_locked, FILE, SEEK_SET};
 use crate::{
     header::{
         errno::EILSEQ,
-        wchar::{fgetwc, mbrtowc, MB_CUR_MAX},
+        wchar::{fgetwc, get_char_encoded_length, mbrtowc, MB_CUR_MAX},
         wctype::WEOF,
     },
     io::Read,
@@ -22,7 +22,6 @@ struct LookAheadBuffer {
 impl LookAheadBuffer {
     fn look_ahead(&mut self) -> Result<Option<wint_t>, i32> {
         let wchar = unsafe { *self.buf.offset(self.look_ahead) };
-        println!("-> {}", wchar as wint_t);
         if wchar == 0 {
             Ok(None)
         } else {
@@ -72,17 +71,11 @@ impl<'a> LookAheadFile<'a> {
             bytes_read += 1;
 
             if bytes_read == 1 {
-                encoded_length = if buf[0] >> 7 == 0 {
-                    1
-                } else if buf[0] >> 5 == 6 {
-                    2
-                } else if buf[0] >> 4 == 0xe {
-                    3
-                } else if buf[0] >> 3 == 0x1e {
-                    4
+                encoded_length = if let Some(el) = get_char_encoded_length(buf[0]) {
+                    el
                 } else {
                     ERRNO.set(EILSEQ);
-                    return Ok(Some(WEOF));
+                    return Ok(WEOF);
                 };
             }
 
@@ -99,9 +92,10 @@ impl<'a> LookAheadFile<'a> {
                 encoded_length,
                 ptr::null_mut(),
             );
+
+            fseek_locked(self.f, seek, SEEK_SET);
         }
 
-        unsafe { fseek_locked(self.f, seek, SEEK_SET) };
         self.look_ahead += encoded_length as i64;
 
         Ok(Some(wc as wint_t))
diff --git a/src/header/wchar/mod.rs b/src/header/wchar/mod.rs
index 7caf3dd1d2..88d690253e 100644
--- a/src/header/wchar/mod.rs
+++ b/src/header/wchar/mod.rs
@@ -10,7 +10,7 @@ use crate::{
         stdlib::{malloc, MB_CUR_MAX, MB_LEN_MAX},
         string,
         time::*,
-        wchar::lookaheadreader::LookAheadReader,
+        wchar::{lookaheadreader::LookAheadReader, utf8::get_char_encoded_length},
         wctype::*,
     },
     iter::{NulTerminated, NulTerminatedInclusive},
@@ -70,14 +70,8 @@ pub unsafe extern "C" fn fgetwc(stream: *mut FILE) -> wint_t {
         bytes_read += 1;
 
         if bytes_read == 1 {
-            encoded_length = if buf[0] >> 7 == 0 {
-                1
-            } else if buf[0] >> 5 == 6 {
-                2
-            } else if buf[0] >> 4 == 0xe {
-                3
-            } else if buf[0] >> 3 == 0x1e {
-                4
+            encoded_length = if let Some(el) = get_char_encoded_length(buf[0]) {
+                el
             } else {
                 ERRNO.set(EILSEQ);
                 return WEOF;
diff --git a/src/header/wchar/utf8.rs b/src/header/wchar/utf8.rs
index b1961c1ac8..bb127427e0 100644
--- a/src/header/wchar/utf8.rs
+++ b/src/header/wchar/utf8.rs
@@ -91,3 +91,18 @@ pub unsafe fn wcrtomb(s: *mut c_char, wc: wchar_t, ps: *mut mbstate_t) -> usize
 
     size
 }
+
+/// Gets the encoded length of a character. It is used to recognize wide characters
+pub fn get_char_encoded_length(first_byte: u8) -> Option<usize> {
+    if first_byte >> 7 == 0 {
+        Some(1)
+    } else if first_byte >> 5 == 6 {
+        Some(2)
+    } else if first_byte >> 4 == 0xe {
+        Some(3)
+    } else if first_byte >> 3 == 0x1e {
+        Some(4)
+    } else {
+        None
+    }
+}