sacabase 2.0.0, introduce sacapart

Author: fasterthanlime

Cargo.lock

@@ -1,6 +1,7 @@
 [workspace]
 members = [
     "crates/sacabase",
+    "crates/sacapart",
     "crates/divsufsort",
     "crates/cdivsufsort",
     "crates/divsuftest",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "cdivsufsort"
-version = "1.0.0"
+version = "2.0.0"
 authors = ["Amos Wenger <[email protected]>"]
 edition = "2018"
 
@@ -15,7 +15,7 @@ license = "MIT"
 crosscheck = []
 
 [dependencies]
-sacabase = { path = "../sacabase", version = "1.0.0" }
+sacabase = { path = "../sacabase", version = "2.0.0" }
 
 [build-dependencies]
 cc = "1.0.47"

crates/cdivsufsort/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "divsufsort"
-version = "1.0.2"
+version = "2.0.0"
 authors = ["Amos Wenger <[email protected]>"]
 edition = "2018"
 
@@ -16,4 +16,4 @@ crosscheck = ["once_cell"]
 
 [dependencies]
 once_cell = { version = "1.2.0", optional = true }
-sacabase = { path = "../sacabase", version = "1.0.0" }
+sacabase = { path = "../sacabase", version = "2.0.0" }

crates/divsufsort/Cargo.toml

@@ -1,9 +1,8 @@
 use failure::Fallible;
 use size_format::SizeFormatterBinary;
-use std::{env, io::Write, process, time::Instant};
+use std::{io::Write, process, time::Instant};
 
 struct Args {
-    partitions: u32,
     free: Vec<String>,
 }
 
@@ -27,11 +26,8 @@ impl Command {
 fn main() -> Fallible<()> {
     better_panic::install();
 
-    let mut args = pico_args::Arguments::from_env();
-    let args = Args {
-        partitions: args.opt_value_from_str("--partitions")?.unwrap_or(1),
-        free: args.free()?,
-    };
+    let args = pico_args::Arguments::from_env();
+    let args = Args { free: args.free()? };
 
     if args.free.is_empty() {
         usage();

crates/divsuftest/src/main.rs

@@ -1,6 +1,6 @@
 [package]
 name = "sacabase"
-version = "1.0.0"
+version = "2.0.0"
 authors = ["Amos Wenger <[email protected]>"]
 edition = "2018"
 

crates/sacabase/Cargo.toml

@@ -2,9 +2,9 @@ use num_traits::ToPrimitive;
 use std::{cmp::min, fmt};
 
 pub struct LongestCommonSubstring<'a> {
-    text: &'a [u8],
-    start: usize,
-    len: usize,
+    pub text: &'a [u8],
+    pub start: usize,
+    pub len: usize,
 }
 
 impl<'a> fmt::Debug for LongestCommonSubstring<'a> {
@@ -18,16 +18,6 @@ impl<'a> LongestCommonSubstring<'a> {
     pub fn as_bytes(&self) -> &[u8] {
         &self.text[self.start..self.start + self.len]
     }
-
-    #[inline(always)]
-    pub fn start(&self) -> usize {
-        self.start
-    }
-
-    #[inline(always)]
-    pub fn len(&self) -> usize {
-        self.len
-    }
 }
 
 /// Returns the number of bytes `a` and `b` have in common.
@@ -52,7 +42,7 @@ pub fn longest_substring_match<'a, Index>(
     needle: &[u8],
 ) -> LongestCommonSubstring<'a>
 where
-    Index: num_traits::ToPrimitive,
+    Index: ToPrimitive,
 {
     macro_rules! sa {
         ($x: expr) => {
@@ -167,6 +157,11 @@ where
     text: &'a [u8],
 }
 
+pub trait StringIndex<'a> {
+    /// Returns the longest substring that matches `needle` in text
+    fn longest_substring_match(&self, needle: &[u8]) -> LongestCommonSubstring<'a>;
+}
+
 impl<'a, Index> SuffixArray<'a, Index>
 where
     Index: ToPrimitive,
@@ -176,19 +171,29 @@ where
         Self { sa, text }
     }
 
-    /// Returns the longest
-    pub fn longest_substring_match(&self, needle: &[u8]) -> LongestCommonSubstring<'a> {
-        longest_substring_match(self.text, &self.sa[..], needle)
-    }
-
     /// Return (text, sa), giving back ownership of `sa`
     pub fn into_parts(self) -> (&'a [u8], Vec<Index>) {
         (self.text, self.sa)
     }
 
+    /// Verifies that this suffix array is sorted.
     pub fn verify(&self) -> Result<(), NotSorted> {
         verify(self.text, &self.sa[..])
     }
+
+    /// Returns a reference to the text
+    pub fn text(&self) -> &[u8] {
+        return self.text;
+    }
+}
+
+impl<'a, Index> StringIndex<'a> for SuffixArray<'a, Index>
+where
+    Index: ToPrimitive,
+{
+    fn longest_substring_match(&self, needle: &[u8]) -> LongestCommonSubstring<'a> {
+        longest_substring_match(self.text, &self.sa[..], needle)
+    }
 }
 
 #[cfg(test)]

crates/sacabase/src/lib.rs

@@ -0,0 +1,13 @@
+[package]
+name = "sacapart"
+version = "2.0.0"
+authors = ["Amos Wenger <[email protected]>"]
+edition = "2018"
+
+[dependencies]
+sacabase = { path = "../sacabase", version = "2.0.0" }
+num-traits = "0.2.9"
+rayon = "1.2.1"
+
+[dev-dependencies]
+divsufsort = { path = "../divsufsort", version = "2.0.0" }