From f87170a5e3bbc726a947c5f4993a6da53481e7f5 Mon Sep 17 00:00:00 2001
From: LinZhihao-723 <zh.lin@mail.utoronto.ca>
Date: Fri, 13 Dec 2024 15:17:53 -0500
Subject: [PATCH] Add unit tests for schema parser

---
 examples/schema.yaml               |  2 +-
 src/error_handling/error.rs        |  1 +
 src/nfa/nfa.rs                     |  8 ++---
 src/parser/mod.rs                  |  2 +-
 src/parser/schema_parser/parser.rs | 58 +++++++++++++++++++++++++++---
 5 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/examples/schema.yaml b/examples/schema.yaml
index ddf42c1..bbfc5a7 100644
--- a/examples/schema.yaml
+++ b/examples/schema.yaml
@@ -6,7 +6,7 @@ timestamp:
   # E.g. 2015-01-31 15:50:45
   - '\d{4}\-\d{2}\-\d{2} \d{2}:\d{2}:\d{2}'
 
-delimiters: '\t\r\n:,!;%'
+delimiters: "\t\r\n:,!;%"
 
 variables:
   int: '\-{0,1}\d+'
diff --git a/src/error_handling/error.rs b/src/error_handling/error.rs
index 337f51a..738b1a3 100644
--- a/src/error_handling/error.rs
+++ b/src/error_handling/error.rs
@@ -4,6 +4,7 @@ use regex_syntax::ast;
 pub enum Error {
     RegexParsingError(ast::Error),
     YamlParsingError(serde_yaml::Error),
+    IOError(std::io::Error),
     UnsupportedAstNodeType(&'static str),
     NoneASCIICharacters,
     NegationNotSupported(&'static str),
diff --git a/src/nfa/nfa.rs b/src/nfa/nfa.rs
index 8e6ac6a..056c4c7 100644
--- a/src/nfa/nfa.rs
+++ b/src/nfa/nfa.rs
@@ -36,14 +36,10 @@ pub struct Transition {
 impl Debug for Transition {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         if 0 == self.symbol_onehot_encoding {
-            return write!(
-                f,
-                "{:?} -> {:?}, symbol: {}",
-                self.from, self.to, "epsilon"
-            );
+            return write!(f, "{:?} -> {:?}, symbol: {}", self.from, self.to, "epsilon");
         }
 
-        let mut char_vec : Vec<char> = Vec::new();
+        let mut char_vec: Vec<char> = Vec::new();
         for i in 0..128u8 {
             let mask = 1u128 << i;
             if mask & self.symbol_onehot_encoding == mask {
diff --git a/src/parser/mod.rs b/src/parser/mod.rs
index c6f6966..844c775 100644
--- a/src/parser/mod.rs
+++ b/src/parser/mod.rs
@@ -1,2 +1,2 @@
 pub(crate) mod regex_parser;
-mod schema_parser;
+pub(crate) mod schema_parser;
diff --git a/src/parser/schema_parser/parser.rs b/src/parser/schema_parser/parser.rs
index 7c4bc3f..252e879 100644
--- a/src/parser/schema_parser/parser.rs
+++ b/src/parser/schema_parser/parser.rs
@@ -1,11 +1,12 @@
 use crate::error_handling::Error::{
-    InvalidSchema, MissingSchemaKey, NoneASCIICharacters, YamlParsingError,
+    IOError, InvalidSchema, MissingSchemaKey, NoneASCIICharacters, YamlParsingError,
 };
 use crate::error_handling::Result;
 use crate::parser::regex_parser::parser::RegexParser;
 use regex_syntax::ast::Ast;
 use serde_yaml::Value;
 use std::collections::{HashMap, HashSet};
+use std::io::Read;
 
 pub struct TimestampSchema {
     regex: String,
@@ -78,15 +79,18 @@ impl ParsedSchema {
         &self.schemas
     }
 
-    pub fn has_delimiter(&self, delimiter: u8) -> bool {
-        self.delimiters.contains(&delimiter)
+    pub fn has_delimiter(&self, delimiter: char) -> bool {
+        if false == delimiter.is_ascii() {
+            return false;
+        }
+        self.delimiters.contains(&(delimiter as u8))
     }
 }
 
 impl ParsedSchema {
     const TIMESTAMP_KEY: &'static str = "timestamp";
     const VAR_KEY: &'static str = "variables";
-    const DELIMITER_EKY: &'static str = "delimiter";
+    const DELIMITER_EKY: &'static str = "delimiters";
 
     pub fn parse_from_str(yaml_content: &str) -> Result<ParsedSchema> {
         match Self::load_kv_pairs_from_yaml_content(yaml_content) {
@@ -95,6 +99,19 @@ impl ParsedSchema {
         }
     }
 
+    pub fn parse_from_file(yaml_file_path: &str) -> Result<ParsedSchema> {
+        match std::fs::File::open(yaml_file_path) {
+            Ok(mut file) => {
+                let mut contents = String::new();
+                if let Err(e) = file.read_to_string(&mut contents) {
+                    return Err(IOError(e));
+                }
+                Self::parse_from_str(contents.as_str())
+            }
+            Err(e) => Err(IOError(e)),
+        }
+    }
+
     fn get_key_value<'a>(
         kv_map: &'a HashMap<String, Value>,
         key: &'static str,
@@ -162,3 +179,36 @@ impl ParsedSchema {
         }))
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_read_example_schema_file() -> Result<()> {
+        let project_root = env!("CARGO_MANIFEST_DIR");
+        let schema_path = std::path::Path::new(project_root)
+            .join("examples")
+            .join("schema.yaml");
+        let parsed_schema = ParsedSchema::parse_from_file(schema_path.to_str().unwrap())?;
+
+        assert_eq!(parsed_schema.get_schemas().len(), 7);
+        for (schema_id, schema) in parsed_schema.get_schemas().iter().enumerate() {
+            match schema {
+                Schema::Timestamp(schema) => {
+                    assert!(schema_id < 3)
+                }
+                Schema::Var(schema) => {
+                    assert!(schema_id >= 3)
+                }
+            }
+        }
+
+        let delimiters: Vec<char> = vec!['\t', '\n', '\r', ':', ',', '!', ';', '%'];
+        for delimiter in delimiters {
+            assert!(parsed_schema.has_delimiter(delimiter));
+        }
+
+        Ok(())
+    }
+}