From 307c246b779e089775be31699663c22e89dbd6c8 Mon Sep 17 00:00:00 2001
From: Nikita Koss <kossnikita@gmail.com>
Date: Fri, 6 Oct 2023 06:50:32 +0300
Subject: [PATCH] replace spec indices search with regex

---
 src/patch/mod.rs        | 45 +++++++++++++++++++++++++++--------------
 src/patch/peripheral.rs | 26 +++++++++++++++++++-----
 src/patch/register.rs   |  6 +++++-
 3 files changed, 56 insertions(+), 21 deletions(-)

diff --git a/src/patch/mod.rs b/src/patch/mod.rs
index 738521b9..cec4bea3 100644
--- a/src/patch/mod.rs
+++ b/src/patch/mod.rs
@@ -569,22 +569,37 @@ fn make_cpu(cmod: &Hash) -> Result<CpuBuilder> {
 }
 
 /// Find left and right indices of enumeration token in specification string
-fn spec_ind(spec: &str) -> (usize, usize) {
+///
+/// # Examples
+///
+/// ```ignore
+/// let cases = [
+///     ("RELOAD?", (6, 0)),
+///     ("TMR[1-57]_MUX", (3, 4)),
+///     ("DT[1-3]?", (2, 0)),
+///     ("GPIO[ABCDE]", (4, 0)),
+///     ("CSPT[1][7-9],CSPT[2][0-5]", (4, 0)),
+/// ];
+/// for (spec, (li, ri)) in cases {
+///     assert_eq!(spec_ind(spec), Some((li, ri)));
+/// }
+/// ```
+///
+fn spec_ind(spec: &str) -> Option<(usize, usize)> {
+    use once_cell::sync::Lazy;
+    use regex::Regex;
     let spec = spec.split(',').next().unwrap_or(spec);
-    let li = spec
-        .bytes()
-        .position(|b| b == b'*')
-        .or_else(|| spec.bytes().position(|b| b == b'?'))
-        .or_else(|| spec.bytes().position(|b| b == b'['))
-        .unwrap();
-    let ri = spec
-        .bytes()
-        .rev()
-        .position(|b| b == b'*')
-        .or_else(|| spec.bytes().rev().position(|b| b == b'?'))
-        .or_else(|| spec.bytes().rev().position(|b| b == b']'))
-        .unwrap();
-    (li, ri)
+    static RE: Lazy<Regex> = Lazy::new(|| {
+        Regex::new(r"^\w*((?:[\?*]|\[\d+(?:-\d+)?\]|\[[a-zA-Z]+(?:-[a-zA-Z]+)?\])+)\w*$").unwrap()
+    });
+    let Some(caps) = RE.captures(spec) else {
+        return None;
+    };
+    let spec = caps.get(0).unwrap();
+    let token = caps.get(1).unwrap();
+    let li = token.start();
+    let ri = spec.end() - token.end();
+    Some((li, ri))
 }
 
 fn check_offsets(offsets: &[u32], dim_increment: u32) -> bool {
diff --git a/src/patch/peripheral.rs b/src/patch/peripheral.rs
index c4d70314..3edb9cd5 100644
--- a/src/patch/peripheral.rs
+++ b/src/patch/peripheral.rs
@@ -1135,7 +1135,11 @@ fn collect_in_array(
         return Err(anyhow!("{path}: registers {rspec} not found"));
     }
     registers.sort_by_key(|r| r.address_offset);
-    let (li, ri) = spec_ind(rspec);
+    let Some((li, ri)) = spec_ind(rspec) else {
+        return Err(anyhow!(
+            "`{rspec}` contains no tokens or contains more than one token"
+        ));
+    };
     let dim = registers.len();
     let dim_index = if rmod.contains_key(&"_start_from_zero".to_yaml()) {
         (0..dim).map(|v| v.to_string()).collect::<Vec<_>>()
@@ -1241,10 +1245,18 @@ fn collect_in_cluster(
                 .iter()
                 .map(|r| {
                     let match_rspec = matchsubspec(&r.name, rspec).unwrap();
-                    let (li, ri) = spec_ind(match_rspec);
-                    r.name[li..r.name.len() - ri].to_string()
+                    let Some((li, ri)) = spec_ind(match_rspec) else {
+                        return Err(anyhow!(
+                            "`{match_rspec}` contains no tokens or contains more than one token"
+                        ));
+                    };
+                    Ok(r.name[li..r.name.len() - ri].to_string())
                 })
-                .collect::<Vec<_>>();
+                .collect::<Result<Vec<_>, _>>();
+            let new_dim_index = match new_dim_index {
+                Ok(v) => v,
+                Err(e) => return Err(e),
+            };
             if first {
                 dim = registers.len();
                 dim_index = new_dim_index;
@@ -1316,7 +1328,11 @@ fn collect_in_cluster(
             reg.name = if let Some(name) = rmod.get_str("name")? {
                 name.into()
             } else {
-                let (li, ri) = spec_ind(&rspec);
+                let Some((li, ri)) = spec_ind(&rspec) else {
+                    return Err(anyhow!(
+                        "`{rspec}` contains no tokens or contains more than one token"
+                    ));
+                };
                 format!("{}{}", &rspec[..li], &rspec[rspec.len() - ri..])
             };
             if let Some(desc) = rmod.get_str("description")? {
diff --git a/src/patch/register.rs b/src/patch/register.rs
index fc40abc9..03c7fea3 100644
--- a/src/patch/register.rs
+++ b/src/patch/register.rs
@@ -373,7 +373,11 @@ impl RegisterExt for Register {
                 return Err(anyhow!("{}: fields {fspec} not found", self.name));
             }
             fields.sort_by_key(|f| f.bit_range.offset);
-            let (li, ri) = spec_ind(fspec);
+            let Some((li, ri)) = spec_ind(fspec) else {
+                return Err(anyhow!(
+                    "`{fspec}` contains no tokens or contains more than one token"
+                ));
+            };
             let dim = fields.len();
             let dim_index = if fmod.contains_key(&"_start_from_zero".to_yaml()) {
                 (0..dim).map(|v| v.to_string()).collect::<Vec<_>>()