license fixes, clippy fixes

Author: GregHanson

Makefile

@@ -1,4 +1,5 @@
-# Copyright (c) 2024-2025, Decisym, LLC
+# Copyright (c) 2025, Decisym, LLC
+# Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 lint:
 	cargo install  cargo-machete

src/bitmap_triples.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use crate::{
     common::{byte_align_bitmap, save_u32_vec},

src/builder.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use super::{bitmap_triples::BitmapTriplesBuilder, dictionary::FourSectDictBuilder};
 use crate::{rdf_reader::convert_to_nt, vocab::*};

src/common.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use std::{
     error::Error,

src/dictionary.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use hdt::{
     containers::{self, ControlType, Sequence, vbyte::encode_vbyte},
@@ -267,7 +268,7 @@ pub fn compress(set: &BTreeSet<String>, block_size: usize) -> Result<DictSectPFC
                 .unwrap_or(term.len());
 
             compressed_terms.extend_from_slice(&encode_vbyte(common_prefix_len));
-            compressed_terms.extend_from_slice(term[byte_offset..].as_bytes());
+            compressed_terms.extend_from_slice(&term.as_bytes()[byte_offset..]);
         }
 
         compressed_terms.push(0); // Null separator

src/lib.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 pub mod bitmap_triples;
 pub mod builder;

src/log_sequence.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use crate::common::save_u32_vec;
 use hdt::containers::vbyte::encode_vbyte;
@@ -39,8 +40,15 @@ impl LogSequence2 {
                     .zip(term.chars())
                     .take_while(|(a, b)| a == b)
                     .count();
+
+                let byte_offset = term
+                    .char_indices()
+                    .nth(common_prefix_len)
+                    .map(|(i, _)| i)
+                    .unwrap_or(term.len());
+
                 compressed_terms.extend_from_slice(&encode_vbyte(common_prefix_len));
-                compressed_terms.extend_from_slice(term[common_prefix_len..].as_bytes());
+                compressed_terms.extend_from_slice(&term.as_bytes()[byte_offset..]);
             };
 
             compressed_terms.push(0); // Null separator

src/main.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 //! # rdf2hdt Converter
 //!

src/mod.rs

@@ -1,4 +1,5 @@
-// Copyright (c) 2024-2025, Decisym, LLC
+// Copyright (c) 2025, Decisym, LLC
+// Licensed under the BSD 3-Clause License (see LICENSE file in the project root).
 
 use log::{debug, error, warn};
 use oxrdfio::RdfSerializer;