From 0b32e80146065d0d8783d769aec21d674c32d551 Mon Sep 17 00:00:00 2001
From: hgrsd <daniel@hgrsd.nl>
Date: Tue, 2 Jul 2024 19:26:29 +0100
Subject: [PATCH] perf: string inference runs fewer expensive tests

---
 Cargo.lock          |  2 +-
 Cargo.toml          |  2 +-
 src/infer_string.rs | 54 ++++++++++++++++++++++++++++-----------------
 3 files changed, 36 insertions(+), 22 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 6026fed..bc19874 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -177,7 +177,7 @@ dependencies = [
 
 [[package]]
 name = "drivel"
-version = "0.2.1"
+version = "0.2.2"
 dependencies = [
  "chrono",
  "clap",
diff --git a/Cargo.toml b/Cargo.toml
index 9ebc9bc..cf5460a 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -3,7 +3,7 @@ name = "drivel"
 description = "Infer a schema from JSON input, and generate synthetic data based on the inferred schema."
 license = "MIT"
 authors = ["Daniël Hogers <daniel@hgrsd.nl>"]
-version = "0.2.1"
+version = "0.2.2"
 edition = "2021"
 repository = "https://github.com/hgrsd/drivel"
 
diff --git a/src/infer_string.rs b/src/infer_string.rs
index 65e26a0..e75d538 100644
--- a/src/infer_string.rs
+++ b/src/infer_string.rs
@@ -12,26 +12,40 @@ lazy_static! {
 }
 
 pub(crate) fn infer_string_type(s: &str) -> StringType {
-    if ISO_DATE_REGEX.is_match(s) {
-        StringType::IsoDate
-    } else if chrono::DateTime::parse_from_rfc2822(s).is_ok() {
-        StringType::DateTimeISO8601
-    } else if chrono::DateTime::parse_from_rfc3339(s).is_ok() {
-        StringType::DateTimeISO8601
-    } else if UUIDREGEX.is_match(s) {
-        StringType::UUID
-    } else if EMAIL_REGEX.is_match(s) {
-        StringType::Email
-    } else if url::Url::parse(s).is_ok() {
-        StringType::Url
-    } else if HOSTNAME_REGEX.is_match(s) {
-        StringType::Hostname
-    } else {
-        StringType::Unknown {
-            strings_seen: vec![s.to_owned()],
-            chars_seen: s.chars().collect(),
-            min_length: Some(s.len()),
-            max_length: Some(s.len()),
+    if s.len() == 36 && UUIDREGEX.is_match(s) {
+        return StringType::UUID;
+    }
+
+    if s.contains('@') && EMAIL_REGEX.is_match(s) {
+        return StringType::Email;
+    }
+
+    if s.contains('.') {
+        if url::Url::parse(s).is_ok() {
+            return StringType::Url;
+        }
+        if HOSTNAME_REGEX.is_match(s) {
+            return StringType::Hostname;
+        }
+    }
+
+    if s.chars().take(1).all(|char| char.is_numeric()) {
+        if ISO_DATE_REGEX.is_match(s) {
+            return StringType::IsoDate;
+        }
+        if chrono::DateTime::parse_from_rfc3339(s).is_ok() {
+            return StringType::DateTimeISO8601;
         }
     }
+
+    if chrono::DateTime::parse_from_rfc2822(s).is_ok() {
+        return StringType::DateTimeISO8601;
+    }
+
+    return StringType::Unknown {
+        strings_seen: vec![s.to_owned()],
+        chars_seen: s.chars().collect(),
+        min_length: Some(s.len()),
+        max_length: Some(s.len()),
+    };
 }