Make regex dependency optional

`regex` was used only in four trivial cases that could be implemented
more simply, either naively or using memchr, without losing performance.
As such the dependency needlessly increases build time, size of binary
and attack surface.

This change makes `regex` optional and defaults to `naive`/`memchr`
implementations. This *improves* performance a bit. The dependency
could've been removed entirely but was kept in case regression is
discovered on another platform and to make comparing the performance
easier. It can be removed in the future if the code is proven to be
reliable.

Author: Kixunil

Cargo.toml

@@ -31,7 +31,13 @@ lazy_static = "^1.4"
 libc = { version = "^0.2", optional = true }
 parking_lot = "^0.11"
 protobuf = { version = "^2.0", optional = true }
-regex = "^1.3"
+# DO NOT RELY ON THIS FEATURE TO STAY AVAILABLE!
+# It doesn't change the API.
+# Intended for testing/debugging only.
+# It can affect the performance.
+# Report any interesting findings, especially if the performance IMPROVES with `regex` turned ON.
+regex = { version = "^1.3", optional = true }
+memchr = "^2.3"
 reqwest = { version = "^0.11", features = ["blocking"], optional = true }
 thiserror = "^1.0"
 

src/desc.rs

@@ -5,32 +5,71 @@ use std::collections::{BTreeSet, HashMap};
 use std::hash::Hasher;
 
 use fnv::FnvHasher;
-use regex::Regex;
 
 use crate::errors::{Error, Result};
 use crate::metrics::SEPARATOR_BYTE;
 use crate::proto::LabelPair;
 
-// Details of required format are at
-// https://prometheus.io/docs/concepts/data_model/#metric-names-and-labels
-fn is_valid_metric_name(name: &str) -> bool {
-    lazy_static! {
-        static ref VALIDATOR: Regex =
-            Regex::new("^[a-zA-Z_:][a-zA-Z0-9_:]*$").expect("Regex to be valid.");
+#[cfg(not(feature = "regex"))]
+mod validation {
+    fn matches_charset_without_colon(c: char) -> bool {
+        c.is_ascii_alphabetic() || c == '_'
     }
 
-    VALIDATOR.is_match(name)
+    fn matches_charset_with_colon(c: char) -> bool {
+        matches_charset_without_colon(c) || c == ':'
+    }
+
+    // check if there's at least one char
+    // the charset_validator returns true on zeroth char
+    // the charset_validator returns true on all remaining chars or they are digits if it returned
+    // false
+    // Equivalent to regex ^[?][?0-9]*$ where ? denotes char set as validated by charset_validator
+    fn is_valid_ident<F: FnMut(char) -> bool>(input: &str, mut charset_validator: F) -> bool {
+        let mut chars = input.chars();
+        let zeroth = chars.next();
+        zeroth
+            .and_then(|zeroth| if charset_validator(zeroth) { Some(chars.all(|c| charset_validator(c) || c.is_digit(10))) } else { None })
+            .unwrap_or(false)
+    }
+
+    // Details of required format are at
+    // https://prometheus.io/docs/concepts/data_model/#metric-names-and-labels
+    pub(super) fn is_valid_metric_name(name: &str) -> bool {
+        is_valid_ident(name, matches_charset_with_colon)
+    }
+
+    pub(super) fn is_valid_label_name(name: &str) -> bool {
+        is_valid_ident(name, matches_charset_without_colon)
+    }
 }
 
-fn is_valid_label_name(name: &str) -> bool {
-    lazy_static! {
-        static ref VALIDATOR: Regex =
-            Regex::new("^[a-zA-Z_][a-zA-Z0-9_]*$").expect("Regex to be valid.");
+#[cfg(feature = "regex")]
+mod validation {
+    use regex::Regex;
+
+    pub(super) fn is_valid_metric_name(name: &str) -> bool {
+        lazy_static! {
+            static ref VALIDATOR: Regex =
+                Regex::new("^[a-zA-Z_:][a-zA-Z0-9_:]*$").expect("Regex to be valid.");
+        }
+
+        VALIDATOR.is_match(name)
     }
 
-    VALIDATOR.is_match(name)
+    pub(super) fn is_valid_label_name(name: &str) -> bool {
+        lazy_static! {
+            static ref VALIDATOR:
+                Regex = Regex::new("^[a-zA-Z_][a-zA-Z0-9_]*$").expect("Regex to be valid.");
+        }
+
+        VALIDATOR.is_match(name)
+
+    }
 }
 
+use validation::*;
+
 /// The descriptor used by every Prometheus [`Metric`](crate::core::Metric). It is essentially
 /// the immutable meta-data of a metric. The normal metric implementations
 /// included in this package manage their [`Desc`] under the hood.

src/encoder/text.rs

@@ -1,6 +1,5 @@
 // Copyright 2019 TiKV Project Authors. Licensed under Apache-2.0.
 
-use regex::{Match, Regex};
 use std::borrow::Cow;
 use std::io::Write;
 
@@ -216,26 +215,42 @@ fn label_pairs_to_text(
     Ok(())
 }
 
-/// `escape_string` replaces `\` by `\\`, new line character by `\n`, and `"` by `\"` if
-/// `include_double_quote` is true.
-///
-/// Implementation adapted from
-/// https://lise-henry.github.io/articles/optimising_strings.html
-fn escape_string(v: &str, include_double_quote: bool) -> Cow<'_, str> {
+#[cfg(feature = "regex")]
+fn find_first_occurence(v: &str, include_double_quote: bool) -> Option<usize> {
+    use regex::{Match, Regex};
+
     // Regex compilation is expensive. Use `lazy_static` to compile the regexes
     // once per process lifetime and not once per function invocation.
     lazy_static! {
         static ref ESCAPER: Regex = Regex::new("(\\\\|\n)").expect("Regex to be valid.");
         static ref QUOTED_ESCAPER: Regex = Regex::new("(\\\\|\n|\")").expect("Regex to be valid.");
     }
 
-    let first_occurence = if include_double_quote {
+    if include_double_quote {
         QUOTED_ESCAPER.find(v)
     } else {
         ESCAPER.find(v)
     }
     .as_ref()
-    .map(Match::start);
+    .map(Match::start)
+}
+
+#[cfg(not(feature = "regex"))]
+fn find_first_occurence(v: &str, include_double_quote: bool) -> Option<usize> {
+    if include_double_quote {
+        memchr::memchr3(b'\\', b'\n', b'\"', v.as_bytes())
+    } else {
+        memchr::memchr2(b'\\', b'\n', v.as_bytes())
+    }
+}
+
+/// `escape_string` replaces `\` by `\\`, new line character by `\n`, and `"` by `\"` if
+/// `include_double_quote` is true.
+///
+/// Implementation adapted from
+/// https://lise-henry.github.io/articles/optimising_strings.html
+fn escape_string(v: &str, include_double_quote: bool) -> Cow<'_, str> {
+    let first_occurence = find_first_occurence(v, include_double_quote);
 
     if let Some(first) = first_occurence {
         let mut escaped = String::with_capacity(v.len() * 2);