Builtin string functions - downcase, len, match(regex), replace(regex…

…), trim, upcase. New dependency regex.

Cargo.toml

@@ -16,6 +16,7 @@ name = "evalexpr"
 path = "src/lib.rs"
 
 [dependencies]
+regex = { version = "1" }
 serde = { version = "1", optional = true}
 serde_derive = { version = "1", optional = true}
 

README.md

@@ -214,6 +214,12 @@ This crate offers a set of builtin functions.
 |------------|-----------------|-------------|
 | min | >= 1 | Returns the minimum of the arguments |
 | max | >= 1 | Returns the maximum of the arguments |
+| downcase | 1 | Returns lower-case version of string |
+| len | 1 | Return the character length of string argument |
+| match | 2 | Returns true if first string argument matches regex in second |
+| replace | 3 | Returns string with matches replaced by third argument |
+| trim | 1 | Strips whitespace from start and end of string |
+| upcase | 1 | Returns upper-case version of string |
 
 The `min` and `max` functions can deal with a mixture of integer and floating point arguments.
 They return the result as the type it was passed into the function.

src/error/display.rs

@@ -84,6 +84,7 @@ impl fmt::Display for EvalexprError {
             ModulationError { dividend, divisor } => {
                 write!(f, "Error modulating {} % {}", dividend, divisor)
             },
+            InvalidRegex { regex, message } => write!(f, "Regular expression {} is invalid: {}", regex, message),
             ContextNotManipulable => write!(f, "Cannot manipulate context"),
             IllegalEscapeSequence(string) => write!(f, "Illegal escape sequence: {}", string),
             CustomMessage(message) => write!(f, "Error: {}", message),

src/error/mod.rs

@@ -167,6 +167,14 @@ pub enum EvalexprError {
         divisor: Value,
     },
 
+    /// This regular expression could not be parsed
+    InvalidRegex {
+        /// The invalid regular expression
+        regex: String,
+        /// Failure message from the regex engine
+        message: String,
+    },
+
     /// A modification was attempted on a `Context` that does not allow modifications.
     ContextNotManipulable,
 
@@ -279,6 +287,11 @@ impl EvalexprError {
     pub(crate) fn modulation_error(dividend: Value, divisor: Value) -> Self {
         EvalexprError::ModulationError { dividend, divisor }
     }
+
+    /// Constructs `EvalexprError::InvalidRegex(regex)`
+    pub fn invalid_regex(regex: String, message: String) -> Self {
+        EvalexprError::InvalidRegex{ regex, message }
+    }
 }
 
 /// Returns `Ok(())` if the actual and expected parameters are equal, and `Err(Error::WrongOperatorArgumentAmount)` otherwise.

src/function/builtin.rs

@@ -1,8 +1,25 @@
+use regex::Regex;
+
+use crate::error::*;
 use value::{FloatType, IntType};
 use EvalexprError;
 use Function;
 use Value;
 
+fn regex_with_local_errors(re_str: &str) -> Result<Regex, EvalexprError> {
+    match Regex::new(re_str) {
+        Ok(re) => Ok(re),
+        Err(regex::Error::Syntax(message)) =>
+            Err(EvalexprError::invalid_regex(re_str.to_string(), message)),
+        Err(regex::Error::CompiledTooBig(max_size)) =>
+            Err(EvalexprError::invalid_regex(
+                re_str.to_string(),
+                format!("Regex exceeded max size {}", max_size))
+            ),
+        Err(err) => Err(EvalexprError::CustomMessage(err.to_string())),
+    }
+}
+
 pub fn builtin_function(identifier: &str) -> Option<Function> {
     match identifier {
         "min" => Some(Function::new(
@@ -53,6 +70,60 @@ pub fn builtin_function(identifier: &str) -> Option<Function> {
                 }
             }),
         )),
+
+        // string functions
+
+        "downcase" => Some(Function::new(
+            Some(1),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                Ok(Value::from(subject.to_lowercase()))
+            }),
+        )),
+        "len" => Some(Function::new(
+            Some(1),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                Ok(Value::from(subject.len() as i64))
+            }),
+        )),
+        "match" => Some(Function::new(
+            Some(2),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                let re_str = expect_string(&arguments[1])?;
+                match regex_with_local_errors(re_str) {
+                    Ok(re) => Ok(Value::Boolean(re.is_match(subject))),
+                    Err(err) => Err(err)
+                }
+            }),
+        )),
+        "replace" => Some(Function::new(
+            Some(3),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                let re_str = expect_string(&arguments[1])?;
+                let repl = expect_string(&arguments[2])?;
+                match regex_with_local_errors(re_str) {
+                    Ok(re) => Ok(Value::String(re.replace_all(subject, repl).to_string())),
+                    Err(err) => Err(err),
+                }
+            }),
+        )),
+        "trim" => Some(Function::new(
+            Some(1),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                Ok(Value::from(subject.trim()))
+            }),
+        )),
+        "upcase" => Some(Function::new(
+            Some(1),
+            Box::new(|arguments| {
+                let subject = expect_string(&arguments[0])?;
+                Ok(Value::from(subject.to_uppercase()))
+            }),
+        )),
         _ => None,
     }
 }

src/lib.rs

@@ -347,6 +347,7 @@
 
 #![warn(missing_docs)]
 
+extern crate regex;
 #[cfg(test)]
 extern crate ron;
 #[cfg(feature = "serde_support")]

tests/integration.rs

@@ -275,15 +275,57 @@ fn test_n_ary_functions() {
         Ok(Value::Int(3))
     );
     assert_eq!(eval_with_context("count 5", &context), Ok(Value::Int(1)));
+}
 
+#[test]
+fn test_builtin_functions() {
     assert_eq!(
-        eval_with_context("min(4.0, 3)", &context),
+        eval("min(4.0, 3)"),
         Ok(Value::Int(3))
     );
     assert_eq!(
-        eval_with_context("max(4.0, 3)", &context),
+        eval("max(4.0, 3)"),
         Ok(Value::Float(4.0))
     );
+    assert_eq!(
+        eval("downcase(\"FOOBAR\")"),
+        Ok(Value::from("foobar"))
+    );
+    assert_eq!(
+        eval("len(\"foobar\")"),
+        Ok(Value::Int(6))
+    );
+    assert_eq!(
+        eval("match(\"foobar\", \"[ob]{3}\")"),
+        Ok(Value::Boolean(true))
+    );
+    assert_eq!(
+        eval("match(\"gazonk\", \"[ob]{3}\")"),
+        Ok(Value::Boolean(false))
+    );
+    match eval("match(\"foo\", \"[\")") {
+        Err(EvalexprError::InvalidRegex{ regex, message }) => {
+            assert_eq!(regex, "[");
+            assert!(message.contains("unclosed character class"));
+        },
+        v => panic!(v),
+    };
+    assert_eq!(
+        eval("replace(\"foobar\", \".*?(o+)\", \"b$1\")"),
+        Ok(Value::String("boobar".to_owned()))
+    );
+    assert_eq!(
+        eval("replace(\"foobar\", \".*?(i+)\", \"b$1\")"),
+        Ok(Value::String("foobar".to_owned()))
+    );
+    assert_eq!(
+        eval("trim(\"  foo  bar \")"),
+        Ok(Value::from("foo  bar"))
+    );
+    assert_eq!(
+        eval("upcase(\"foobar\")"),
+        Ok(Value::from("FOOBAR"))
+    );
 }
 
 #[test]