From 65807c40cd874514e97392e35d25b788b1a706d8 Mon Sep 17 00:00:00 2001
From: Ariel Davis <ariel.z.davis@icloud.com>
Date: Wed, 31 Jan 2024 17:15:14 -0800
Subject: [PATCH] Add str process

---
 README.md                     |   4 ++
 crates/str-process/Cargo.toml |  16 +++++
 crates/str-process/src/lib.rs | 117 ++++++++++++++++++++++++++++++++++
 3 files changed, 137 insertions(+)
 create mode 100644 crates/str-process/Cargo.toml
 create mode 100644 crates/str-process/src/lib.rs

diff --git a/README.md b/README.md
index 158879d..894f962 100644
--- a/README.md
+++ b/README.md
@@ -30,6 +30,10 @@ Turn `PascalCase` into `snake_case` and vice versa.
 
 Determine whether a sequence of patterns is exhaustive or if any of the patterns are unreachable.
 
+## `str-process`
+
+Process a string by each byte.
+
 ## `syntax-gen`
 
 Generates Rust code from an [ungrammar][].
diff --git a/crates/str-process/Cargo.toml b/crates/str-process/Cargo.toml
new file mode 100644
index 0000000..711687a
--- /dev/null
+++ b/crates/str-process/Cargo.toml
@@ -0,0 +1,16 @@
+[package]
+name = "str-process"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+publish.workspace = true
+
+[lints]
+workspace = true
+
+[lib]
+test = false
+doctest = false
+
+[dependencies]
+drop_bomb.workspace = true
diff --git a/crates/str-process/src/lib.rs b/crates/str-process/src/lib.rs
new file mode 100644
index 0000000..7ae7608
--- /dev/null
+++ b/crates/str-process/src/lib.rs
@@ -0,0 +1,117 @@
+//! Process a string by each byte.
+
+use drop_bomb::DebugDropBomb;
+
+/// The state for processing a string.
+#[derive(Debug, Default)]
+pub struct St<'a> {
+  s: &'a str,
+  idx: usize,
+}
+
+impl<'a> St<'a> {
+  /// Returns a new state for the string.
+  #[must_use]
+  pub fn new(s: &'a str) -> St<'a> {
+    St { s, idx: 0 }
+  }
+
+  /// Returns the current byte.
+  #[must_use]
+  pub fn cur(&self) -> Option<u8> {
+    self.s.as_bytes().get(self.idx).copied()
+  }
+
+  /// Returns the current byte index. Use this for marking where errors occur in the string.
+  #[must_use]
+  pub fn cur_idx(&self) -> usize {
+    self.idx
+  }
+
+  /// Advances the index by 1.
+  pub fn bump(&mut self) {
+    self.idx += 1;
+  }
+
+  /// Advances the index while `cond` holds true.
+  pub fn bump_while<F>(&mut self, mut cond: F)
+  where
+    F: FnMut(u8) -> bool,
+  {
+    while let Some(b) = self.cur() {
+      if cond(b) {
+        self.bump();
+      } else {
+        break;
+      }
+    }
+  }
+
+  /// Returns a marker that must be consumed later.
+  #[must_use]
+  pub fn mark(&self) -> Marker {
+    Marker { bomb: DebugDropBomb::new("must be passed to a `St` method"), idx: self.idx }
+  }
+
+  /// Returns a non-empty slice since the marker.
+  ///
+  /// # Panics
+  ///
+  /// If it would return an empty slice.
+  #[must_use]
+  pub fn non_empty_since(&self, m: Marker) -> &'a [u8] {
+    let start = m.idx;
+    assert!(self.did_bump_since(m));
+    &self.s.as_bytes()[start..self.idx]
+  }
+
+  /// Returns the slice since the marker.
+  ///
+  /// NOTE: allowed to return an empty slice.
+  #[must_use]
+  pub fn since(&self, mut m: Marker) -> &'a [u8] {
+    let start = m.idx;
+    m.bomb.defuse();
+    &self.s.as_bytes()[start..self.idx]
+  }
+
+  /// Returns whether the state was bumped since the marker.
+  #[must_use]
+  pub fn did_bump_since(&self, mut m: Marker) -> bool {
+    m.bomb.defuse();
+    self.idx > m.idx
+  }
+
+  /// If the next few bytes of the string are equal to prefix, advance by that much and return true.
+  /// Else return false.
+  pub fn eat_prefix(&mut self, prefix: &[u8]) -> bool {
+    let end = self.idx + prefix.len();
+    if self.s.as_bytes().get(self.idx..end).is_some_and(|bs| bs == prefix) {
+      self.idx = end;
+      true
+    } else {
+      false
+    }
+  }
+
+  /// Advances the index to the next char boundary.
+  pub fn next_str(&mut self) {
+    self.bump();
+    loop {
+      if self.s.is_char_boundary(self.idx) {
+        break;
+      }
+      match self.cur() {
+        Some(_) => self.bump(),
+        None => unreachable!("got to the end without a valid str"),
+      }
+    }
+  }
+}
+
+/// A marker for the current position.
+#[derive(Debug)]
+pub struct Marker {
+  bomb: DebugDropBomb,
+  idx: usize,
+}