Add color And \\ulnone

Cargo.toml

@@ -5,7 +5,7 @@ repository = "https://github.com/d0rianb/rtf-parser"
 version = "0.2.1"
 edition = "2021"
 license = "MIT"
-keywords = ["rtf", "rich" ,"text", "format", "parser"]
+keywords = ["rtf", "rich", "text", "format", "parser"]
 categories = ["parsing", "parser-implementations"]
 exclude = ["*.rtf", ".idea"]
 
@@ -15,4 +15,7 @@ opt-level = 3
 
 [dependencies]
 derivative = "2.2.0"
+serde = { version = "1.0", optional = true, features = ["derive"] }
 
+[features]
+serde_support = ["serde"]

src/document.rs

@@ -1,6 +1,17 @@
 use crate::header::RtfHeader;
 use crate::parser::StyleBlock;
 
+#[cfg(feature="serde_support")]
+use serde::{Deserialize, Serialize};
+
+#[cfg(feature="serde_support")]
+#[derive(Debug, Default, Clone, PartialEq, Deserialize, Serialize)]
+pub struct RtfDocument {
+    pub header: RtfHeader,
+    pub body: Vec<StyleBlock>,
+}
+
+#[cfg(not(feature="serde_support"))]
 #[derive(Debug, Default, Clone, PartialEq)]
 pub struct RtfDocument {
     pub header: RtfHeader,

src/header.rs

@@ -1,37 +1,95 @@
 use std::collections::HashMap;
 
+#[cfg(feature="serde_support")]
+use serde::{Deserialize, Serialize};
+
 use crate::paragraph::Paragraph;
 use crate::parser::Painter;
 use crate::tokens::{ControlWord, Token};
 
+pub type ColorRef = u16;
+pub type ColorTable = HashMap<ColorRef, Color>;
+
 pub type FontRef = u16;
 pub type FontTable = HashMap<FontRef, Font>;
 
 pub type StyleRef = u16;
 pub type StyleSheet = HashMap<StyleRef, Style>;
 
 /// Style for the StyleSheet
+#[cfg(feature="serde_support")]
+#[derive(Hash, Default, Debug, Clone, PartialEq, Deserialize, Serialize)]
+pub struct Style {
+    painter: Painter,
+    paragraph: Paragraph,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Hash, Default, Debug, Clone, PartialEq)]
 pub struct Style {
     painter: Painter,
     paragraph: Paragraph,
 }
 
 /// Information about the document, including references to fonts & styles
+#[cfg(feature="serde_support")]
+#[derive(Default, Debug, Clone, PartialEq, Deserialize, Serialize)]
+pub struct RtfHeader {
+    pub character_set: CharacterSet,
+    pub font_table: FontTable,
+    pub color_table: ColorTable,
+    pub stylesheet: StyleSheet,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Default, Debug, Clone, PartialEq)]
 pub struct RtfHeader {
     pub character_set: CharacterSet,
     pub font_table: FontTable,
+    pub color_table: ColorTable,
     pub stylesheet: StyleSheet,
 }
 
+#[cfg(feature="serde_support")]
+#[derive(Hash, Default, Clone, Debug, PartialEq, Deserialize, Serialize)]
+pub struct Font {
+    pub name: String,
+    pub character_set: u8,
+    pub font_family: FontFamily,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Hash, Default, Clone, Debug, PartialEq)]
 pub struct Font {
     pub name: String,
     pub character_set: u8,
     pub font_family: FontFamily,
 }
 
+#[cfg(feature="serde_support")]
+#[derive(Hash, Default, Clone, Debug, PartialEq, Deserialize, Serialize)]
+pub struct Color {
+    pub red: u16,
+    pub green: u16,
+    pub blue: u16,
+}
+#[cfg(not(feature="serde_support"))]
+#[derive(Hash, Default, Clone, Debug, PartialEq)]
+pub struct Color {
+    pub red: u16,
+    pub green: u16,
+    pub blue: u16,
+}
+
+#[cfg(feature="serde_support")]
+#[allow(dead_code)]
+#[derive(Debug, PartialEq, Default, Clone, Deserialize, Serialize)]
+pub enum CharacterSet {
+    #[default]
+    Ansi,
+    Mac,
+    Pc,
+    Pca,
+    Ansicpg(u16),
+}
+#[cfg(not(feature="serde_support"))]
 #[allow(dead_code)]
 #[derive(Debug, PartialEq, Default, Clone)]
 pub enum CharacterSet {
@@ -53,6 +111,21 @@ impl CharacterSet {
     }
 }
 
+#[cfg(feature="serde_support")]
+#[allow(dead_code)]
+#[derive(Debug, PartialEq, Hash, Clone, Default, Deserialize, Serialize)]
+pub enum FontFamily {
+    #[default]
+    Nil,
+    Roman,
+    Swiss,
+    Modern,
+    Script,
+    Decor,
+    Tech,
+    Bidi,
+}
+#[cfg(not(feature="serde_support"))]
 #[allow(dead_code)]
 #[derive(Debug, PartialEq, Hash, Clone, Default)]
 pub enum FontFamily {

src/lexer.rs

@@ -121,6 +121,12 @@ impl Lexer {
                     let control_word = ControlWord::from(ident)?;
                     let mut ret = vec![Token::ControlSymbol(control_word)];
                     recursive_tokenize!(tail, ret);
+
+                    // \u1234 \u1234 is ok, but \u1234  \u1234 is lost a space, \u1234   \u1234 lost two spaces, and so on
+                    if control_word.0 == ControlWord::Unicode && tail.len() > 0 {
+                        ret.push(Token::PlainText(tail));
+                    }
+
                     return Ok(ret);
                 }
                 '*' => Ok(vec![Token::IgnorableDestination]),
@@ -147,8 +153,9 @@ impl Lexer {
 
 #[cfg(test)]
 pub(crate) mod tests {
+    use crate::header::Color;
     use crate::lexer::Lexer;
-    use crate::tokens::ControlWord::{Ansi, Bold, FontNumber, FontSize, FontTable, Italic, Par, Pard, Rtf, Underline, Unknown};
+    use crate::tokens::ControlWord::{Ansi, Bold, FontNumber, ColorNumber, FontSize, FontTable, Italic, Par, Pard, Rtf, Underline, Unknown};
     use crate::tokens::Property::*;
     use crate::tokens::Token::*;
 
@@ -202,7 +209,7 @@ if (a == b) \{\
             vec![
                 ControlSymbol((FontNumber, Value(0))),
                 ControlSymbol((FontSize, Value(24))),
-                ControlSymbol((Unknown("\\cf"), Value(0))),
+                ControlSymbol((ColorNumber, Value(0))),
                 PlainText("test de code "),
                 CRLF,
                 PlainText("if (a == b) "),
@@ -266,7 +273,7 @@ if (a == b) \{\
                 OpeningBracket,
                 ControlSymbol((Unknown("\\partightenfactor"), Value(0))),
                 ControlSymbol((FontSize, Value(24))),
-                ControlSymbol((Unknown("\\cf"), Value(0))),
+                ControlSymbol((ColorNumber, Value(0))),
                 PlainText("Font size 12,"),
                 ControlSymbol((FontNumber, Value(0))),
                 ControlSymbol((Bold, None)),

src/paragraph.rs

@@ -1,7 +1,18 @@
 // Define the paragraph related structs and enums
 
+#[cfg(feature="serde_support")]
+use serde::{Deserialize, Serialize};
 use crate::tokens::ControlWord;
 
+#[cfg(feature="serde_support")]
+#[derive(Debug, Default, Clone, PartialEq, Hash, Deserialize, Serialize)]
+pub struct Paragraph {
+    pub alignment: Alignment,
+    pub spacing: Spacing,
+    pub indent: Indentation,
+    pub tab_width: i32,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Debug, Default, Clone, PartialEq, Hash)]
 pub struct Paragraph {
     pub alignment: Alignment,
@@ -11,6 +22,16 @@ pub struct Paragraph {
 }
 
 /// Alignement of a paragraph (left, right, center, justify)
+#[cfg(feature="serde_support")]
+#[derive(Debug, Default, Clone, Copy, PartialEq, Hash, Deserialize, Serialize)]
+pub enum Alignment {
+    #[default]
+    LeftAligned,    // \ql
+    RightAligned,   // \qr
+    Center,         // \qc
+    Justify,        // \qj
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Debug, Default, Clone, Copy, PartialEq, Hash)]
 pub enum Alignment {
     #[default]
@@ -33,6 +54,15 @@ impl From<&ControlWord<'_>> for Alignment {
 }
 
 /// The vertical margin before / after a block of text
+#[cfg(feature="serde_support")]
+#[derive(Debug, Default, Clone, PartialEq, Hash, Deserialize, Serialize)]
+pub struct Spacing {
+    pub before: i32,
+    pub after: i32,
+    pub between_line: SpaceBetweenLine,
+    pub line_multiplier: i32,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Debug, Default, Clone, PartialEq, Hash)]
 pub struct Spacing {
     pub before: i32,
@@ -41,6 +71,15 @@ pub struct Spacing {
     pub line_multiplier: i32,
 }
 
+#[cfg(feature="serde_support")]
+#[derive(Default, Debug, Clone, PartialEq, Hash, Deserialize, Serialize)]
+pub enum SpaceBetweenLine {
+    Value(i32),
+    #[default]
+    Auto,
+    Invalid,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Default, Debug, Clone, PartialEq, Hash)]
 pub enum SpaceBetweenLine {
     Value(i32),
@@ -64,6 +103,14 @@ impl From<i32> for SpaceBetweenLine {
 }
 
 // This struct can not be an enum because left-indent and right-ident can both be defined at the same time
+#[cfg(feature="serde_support")]
+#[derive(Default, Debug, Clone, PartialEq, Hash, Deserialize, Serialize)]
+pub struct Indentation {
+    pub left: i32,
+    pub right: i32,
+    pub first_line: i32,
+}
+#[cfg(not(feature="serde_support"))]
 #[derive(Default, Debug, Clone, PartialEq, Hash)]
 pub struct Indentation {
     pub left: i32,