Allow backslash as last char in attributes

src/parser/sections/tests/components.rs

@@ -21,7 +21,7 @@
 use super::super::components::*;
 
 use crate::parser::KeywordParam;
-use crate::types::{Layer, Mirror, RectangleRef, TextPar, XYRef};
+use crate::types::{Attribute, Layer, Mirror, RectangleRef, TextPar, XYRef};
 
 #[test]
 fn test_example_components() {
@@ -226,3 +226,95 @@ fn test_component_with_improperly_formatted_strings() {
         },]
     );
 }
+
+#[test]
+fn test_component_with_multiple_attributes() {
+    let params = vec![
+        KeywordParam {
+            keyword: "COMPONENT",
+            parameter: "Foo",
+        },
+        KeywordParam {
+            keyword: "PLACE",
+            parameter: "0.0 1.0",
+        },
+        KeywordParam {
+            keyword: "LAYER",
+            parameter: "TOP",
+        },
+        KeywordParam {
+            keyword: "ROTATION",
+            parameter: "0",
+        },
+        KeywordParam {
+            keyword: "SHAPE",
+            parameter: "Foo 0 0",
+        },
+        KeywordParam {
+            keyword: "DEVICE",
+            parameter: "Device Foo",
+        },
+        KeywordParam {
+            keyword: "ATTRIBUTE",
+            parameter: r#"COMPONENT_1 "OperatingTemp" "-40°C - +85°C""#,
+        },
+        KeywordParam {
+            keyword: "ATTRIBUTE",
+            parameter: r#"COMPONENT_2 "Technology" "SMD""#,
+        },
+        KeywordParam {
+            keyword: "ATTRIBUTE",
+            parameter: r#"COMPONENT_3 "Manufacturer" "Foobar Tech.""#,
+        },
+        KeywordParam {
+            keyword: "ATTRIBUTE",
+            parameter: r#"COMPONENT_4 "ComponentLinkUrl" "\\foobar.tech\unc\path\which\ends\with\a\backslash\""#,
+        },
+    ];
+
+    let components = parse_components(&params).unwrap();
+
+    assert_eq!(
+        components,
+        vec![Component {
+            name: "Foo".to_string(),
+            device: "Device Foo".to_string(),
+            place: XYRef {
+                x: 0.0,
+                y: 1.0
+            },
+            layer: Layer::Top,
+            rotation: 0.0,
+            shape: Shape {
+                name: "Foo".to_string(),
+                mirror: Mirror::Not,
+                flip: false
+            },
+            subcomponents: vec![],
+            texts: vec![],
+            sheet: None,
+            attributes: vec![
+                Attribute {
+                    category: "COMPONENT_1".to_string(),
+                    name: "OperatingTemp".to_string(),
+                    data: "-40°C - +85°C".to_string()
+                },
+                Attribute {
+                    category: "COMPONENT_2".to_string(),
+                    name: "Technology".to_string(),
+                    data: "SMD".to_string()
+                },
+                Attribute {
+                    category: "COMPONENT_3".to_string(),
+                    name: "Manufacturer".to_string(),
+                    data: "Foobar Tech.".to_string()
+                },
+                Attribute {
+                    category: "COMPONENT_4".to_string(),
+                    name: "ComponentLinkUrl".to_string(),
+                    data: r#"\\foobar.tech\unc\path\which\ends\with\a\backslash\"#.to_string()
+                },
+            ],
+        },]
+    );
+}

src/parser/types/string.rs

@@ -20,7 +20,7 @@
 
 use nom::branch::alt;
 use nom::bytes::complete::{take_while, take_while1};
-use nom::character::complete::char;
+use nom::character::complete::{char, satisfy};
 use nom::combinator::{map, not, peek, value};
 use nom::multi::many0;
 use nom::sequence::{delimited, preceded};
@@ -33,7 +33,8 @@ enum QuotedStringFragment<'a> {
 }
 
 fn is_valid_char(c: char) -> bool {
-    matches!(c, ' '..='~')
+    // Accept all Unicode except control characters
+    !c.is_control()
 }
 
 fn is_valid_char_but_not_backslash_or_quote(c: char) -> bool {
@@ -46,8 +47,17 @@ fn is_valid_char_but_not_space(c: char) -> bool {
 
 fn backslash_sequence(s: &str) -> IResult<&str, char> {
     alt((
-        // A backslash before a quote mark becomes a quote mark
-        preceded(char('\\'), value('"', char('"'))),
+        // A backslash before a quote mark becomes a quote mark, only if not at the end and followed by a printable character
+        preceded(
+            char('\\'),
+            preceded(
+                peek((
+                    char('"'),
+                    satisfy(is_valid_char),
+                )),
+                value('"', char('"')),
+            ),
+        ),
         // Backslash before any other character is not an escape sequence--it's just a literal backslash
         value('\\', char('\\')),
     ))
@@ -112,6 +122,8 @@ mod tests {
         assert_eq!(backslash_sequence(r#"\;"#), Ok((";", '\\')));
         assert_eq!(backslash_sequence(r#"\\;"#), Ok((r#"\;"#, '\\')));
         assert_eq!(backslash_sequence(r#"\";"#), Ok((";", '"')));
+        assert_eq!(backslash_sequence(r#"\""#), Ok(("\"", '\\')));
+        assert_eq!(backslash_sequence(r#"\"#), Ok(("", '\\')));
     }
 
     #[test]
@@ -184,8 +196,12 @@ mod tests {
         assert!(unquoted_string("\"qoY@M;").is_err());
         assert!(unquoted_string(r#""A""#).is_err());
 
-        // No non-ASCII characters
-        assert_eq!(unquoted_string("V3\"'😀"), Ok(("😀", "V3\"'")));
-        assert_eq!(unquoted_string("A'😀A4%"), Ok(("😀A4%", "A'")));
+        // Unicode characters are allowed
+        assert_eq!(unquoted_string("V3\"'😀"), Ok(("", "V3\"'😀")));
+        assert_eq!(unquoted_string("A'😀A4%"), Ok(("", "A'😀A4%")));
+
+        // Control characters are not allowed
+        assert_eq!(unquoted_string("abc\u{0007}def"), Ok(("\u{0007}def", "abc")));
+        assert_eq!(unquoted_string("\u{0001}abc"), Ok(("\u{0001}abc", "")));
     }
 }

src/parser/types/tests/attrib_ref.rs

@@ -60,3 +60,18 @@ fn test_ok() {
         ))
     );
 }
+
+#[test]
+fn test_attrib_ref_quoted() {
+    assert_eq!(
+        attrib_ref(r#"COMPONENT "OperatingTemp" "-40°C - +85°C""#),
+        Ok((
+            "",
+            Attribute {
+                category: "COMPONENT".to_string(),
+                name: "OperatingTemp".to_string(),
+                data: "-40°C - +85°C".to_string()
+            }
+        ))
+    );
+}