Add some instr text (#385)

* feat: Add tc instr

* feat: Add pageref

docx-core/src/documents/elements/instr_pageref.rs

@@ -0,0 +1,67 @@
+use serde::Serialize;
+
+use crate::documents::*;
+
+// https://c-rex.net/projects/samples/ooxml/e1/Part4/OOXML_P4_DOCX_PAGEREFPAGEREF_topic_ID0EHXK1.html
+#[derive(Serialize, Debug, Clone, PartialEq, Default)]
+pub struct InstrPAGEREF {
+    pub page_ref: String,
+    pub hyperlink: bool,
+    pub relative_position: bool,
+}
+
+impl InstrPAGEREF {
+    pub fn new(r: impl Into<String>) -> Self {
+        Self {
+            page_ref: r.into(),
+            ..Default::default()
+        }
+    }
+
+    pub fn hyperlink(mut self) -> Self {
+        self.hyperlink = true;
+        self
+    }
+
+    pub fn relative_position(mut self) -> Self {
+        self.relative_position = true;
+        self
+    }
+}
+
+impl BuildXML for InstrPAGEREF {
+    fn build(&self) -> Vec<u8> {
+        let mut instr = format!("PAGEREF {}", self.page_ref);
+
+        if self.relative_position {
+            instr = format!("{} \\p", instr);
+        }
+
+        if self.hyperlink {
+            instr = format!("{} \\h", instr);
+        }
+
+        instr.into()
+    }
+}
+
+impl std::str::FromStr for InstrPAGEREF {
+    type Err = ();
+
+    fn from_str(instr: &str) -> Result<Self, Self::Err> {
+        let mut s = instr.split(' ');
+        let text = s.next();
+        let mut page_ref = InstrPAGEREF::new(text.unwrap_or_default());
+        loop {
+            if let Some(i) = s.next() {
+                match i {
+                    "\\h" => page_ref = page_ref.hyperlink(),
+                    "\\p" => page_ref = page_ref.relative_position(),
+                    _ => {}
+                }
+            } else {
+                return Ok(page_ref);
+            }
+        }
+    }
+}

docx-core/src/documents/elements/instr_tc.rs

@@ -0,0 +1,99 @@
+use serde::Serialize;
+
+use crate::documents::*;
+
+// https://c-rex.net/projects/samples/ooxml/e1/Part4/OOXML_P4_DOCX_TCTC_topic_ID0EU2N1.html
+#[derive(Serialize, Debug, Clone, PartialEq, Default)]
+pub struct InstrTC {
+    pub text: String,
+    // \n Omits the page number for the entry.
+    pub omits_page_number: bool,
+    pub level: Option<usize>,
+    // \f
+    pub item_type_identifier: Option<String>,
+}
+
+impl InstrTC {
+    pub fn new(text: impl Into<String>) -> Self {
+        Self {
+            text: text.into(),
+            ..Default::default()
+        }
+    }
+
+    pub fn omits_page_number(mut self) -> Self {
+        self.omits_page_number = true;
+        self
+    }
+
+    pub fn level(mut self, level: usize) -> Self {
+        self.level = Some(level);
+        self
+    }
+
+    pub fn item_type_identifier(mut self, t: impl Into<String>) -> Self {
+        self.item_type_identifier = Some(t.into());
+        self
+    }
+}
+
+impl BuildXML for InstrTC {
+    fn build(&self) -> Vec<u8> {
+        let mut instr = format!("TC {}", self.text);
+
+        if let Some(ref t) = self.item_type_identifier {
+            instr = format!("{} \\f {}", instr, t);
+        }
+
+        if let Some(level) = self.level {
+            instr = format!("{} \\l {}", instr, level);
+        }
+
+        if self.omits_page_number {
+            instr = format!("{} \\n", instr);
+        }
+
+        instr.into()
+    }
+}
+
+fn parse_level(i: &str) -> Option<usize> {
+    let r = i.replace("&quot;", "").replace("\"", "");
+    if let Ok(l) = usize::from_str(&r) {
+        return Some(l);
+    }
+    None
+}
+
+impl std::str::FromStr for InstrTC {
+    type Err = ();
+
+    fn from_str(instr: &str) -> Result<Self, Self::Err> {
+        let mut s = instr.split(' ');
+        let text = s.next();
+        let mut tc = InstrTC::new(text.unwrap_or_default());
+        loop {
+            if let Some(i) = s.next() {
+                match i {
+                    "\\f" => {
+                        if let Some(r) = s.next() {
+                            let r = r.replace("&quot;", "").replace("\"", "");
+                            tc = tc.item_type_identifier(r);
+                        }
+                    }
+                    "\\l" => {
+                        if let Some(r) = s.next() {
+                            if let Some(l) = parse_level(r) {
+                                tc = tc.level(l);
+                            }
+                        }
+                    }
+                    "\\n" => tc = tc.omits_page_number(),
+                    _ => {}
+                }
+            } else {
+                return Ok(tc);
+            }
+        }
+    }
+}

docx-core/src/documents/elements/instr_text.rs

@@ -7,6 +7,8 @@ use crate::xml_builder::*;
 #[derive(Debug, Clone, PartialEq)]
 pub enum InstrText {
     TOC(InstrToC),
+    TC(InstrTC),
+    PAGEREF(InstrPAGEREF),
     Unsupported(String),
 }
 
@@ -14,6 +16,8 @@ impl BuildXML for Box<InstrText> {
     fn build(&self) -> Vec<u8> {
         let instr = match self.as_ref() {
             InstrText::TOC(toc) => toc.build(),
+            InstrText::TC(tc) => tc.build(),
+            InstrText::PAGEREF(page_ref) => page_ref.build(),
             InstrText::Unsupported(s) => s.as_bytes().to_vec(),
         };
         XMLBuilder::new()
@@ -36,6 +40,18 @@ impl Serialize for InstrText {
                 t.serialize_field("data", s)?;
                 t.end()
             }
+            InstrText::TC(ref s) => {
+                let mut t = serializer.serialize_struct("TC", 2)?;
+                t.serialize_field("type", "tc")?;
+                t.serialize_field("data", s)?;
+                t.end()
+            }
+            InstrText::PAGEREF(ref s) => {
+                let mut t = serializer.serialize_struct("PAGEREF", 2)?;
+                t.serialize_field("type", "pageref")?;
+                t.serialize_field("data", s)?;
+                t.end()
+            }
             InstrText::Unsupported(ref s) => {
                 let mut t = serializer.serialize_struct("Unsupported", 2)?;
                 t.serialize_field("type", "unsupported")?;
@@ -62,4 +78,16 @@ mod tests {
             r#"<w:instrText>TOC \o &quot;1-3&quot;</w:instrText>"#
         );
     }
+
+    #[test]
+    fn test_pageref_instr() {
+        let b = Box::new(InstrText::PAGEREF(
+            InstrPAGEREF::new("_Toc90425847").hyperlink(),
+        ))
+        .build();
+        assert_eq!(
+            str::from_utf8(&b).unwrap(),
+            r#"<w:instrText>PAGEREF _Toc90425847 \h</w:instrText>"#
+        );
+    }
 }

docx-core/src/documents/elements/mod.rs

@@ -18,7 +18,6 @@ mod default_tab_stop;
 mod delete;
 mod delete_text;
 mod div;
-mod instr_toc;
 mod doc_defaults;
 mod doc_grid;
 mod doc_id;
@@ -33,7 +32,10 @@ mod highlight;
 mod indent;
 mod indent_level;
 mod insert;
+mod instr_pageref;
+mod instr_tc;
 mod instr_text;
+mod instr_toc;
 mod italic;
 mod italic_cs;
 mod justification;
@@ -136,7 +138,10 @@ pub use highlight::*;
 pub use indent::*;
 pub use indent_level::*;
 pub use insert::*;
+pub use instr_pageref::*;
+pub use instr_tc::*;
 pub use instr_text::*;
+pub use instr_toc::*;
 pub use italic::*;
 pub use italic_cs::*;
 pub use justification::*;
@@ -204,4 +209,3 @@ pub use wp_anchor::*;
 pub use wps_shape::*;
 pub use wps_text_box::*;
 pub use zoom::*;
-pub use instr_toc::*;

docx-core/src/reader/instr_text.rs

@@ -29,10 +29,19 @@ impl ElementReader for InstrText {
                                 return Err(ReaderError::XMLReadError);
                             } else {
                                 if instr.starts_with("TOC") {
-                                    for i in instr.split(' ') {
-                                        dbg!(i);
+                                    if let Ok(instr) = InstrToC::from_str(instr) {
+                                        return Ok(InstrText::TOC(instr));
+                                    }
+                                }
+                                if instr.starts_with("TC") {
+                                    if let Ok(instr) = InstrTC::from_str(instr) {
+                                        return Ok(InstrText::TC(instr));
+                                    }
+                                }
+                                if instr.starts_with("PAGEREF") {
+                                    if let Ok(instr) = InstrPAGEREF::from_str(instr) {
+                                        return Ok(InstrText::PAGEREF(instr));
                                     }
-                                    return Ok(InstrText::TOC(InstrToC::new()));
                                 }
                                 return Ok(InstrText::Unsupported(instr.to_string()));
                             }

docx-wasm/test/__snapshots__/index.test.js.snap

@@ -23126,15 +23126,19 @@ Object {
                         "caption_label_including_numbers": null,
                         "entry_and_page_number_separator": null,
                         "entry_bookmark_name": null,
-                        "hide_tab_and_page_numbers_in_webview": false,
-                        "hyperlink": false,
+                        "heading_styles_range": Array [
+                          1,
+                          3,
+                        ],
+                        "hide_tab_and_page_numbers_in_webview": true,
+                        "hyperlink": true,
                         "preserve_new_line": false,
                         "preserve_tab": false,
                         "seq_field_identifier_for_prefix": null,
                         "sequence_and_page_numbers_separator": null,
                         "styles_with_levels": Array [],
                         "tc_field_identifier": null,
-                        "use_applied_paragraph_line_level": false,
+                        "use_applied_paragraph_line_level": true,
                       },
                       "type": "toc",
                     },
@@ -23351,8 +23355,12 @@ Object {
                 "children": Array [
                   Object {
                     "data": Object {
-                      "data": "PAGEREF _Toc89816785 \\\\h",
-                      "type": "unsupported",
+                      "data": Object {
+                        "hyperlink": true,
+                        "page_ref": "PAGEREF",
+                        "relative_position": false,
+                      },
+                      "type": "pageref",
                     },
                     "type": "instrText",
                   },
@@ -23695,8 +23703,12 @@ Object {
                 "children": Array [
                   Object {
                     "data": Object {
-                      "data": "PAGEREF _Toc89816786 \\\\h",
-                      "type": "unsupported",
+                      "data": Object {
+                        "hyperlink": true,
+                        "page_ref": "PAGEREF",
+                        "relative_position": false,
+                      },
+                      "type": "pageref",
                     },
                     "type": "instrText",
                   },
@@ -24039,8 +24051,12 @@ Object {
                 "children": Array [
                   Object {
                     "data": Object {
-                      "data": "PAGEREF _Toc89816787 \\\\h",
-                      "type": "unsupported",
+                      "data": Object {
+                        "hyperlink": true,
+                        "page_ref": "PAGEREF",
+                        "relative_position": false,
+                      },
+                      "type": "pageref",
                     },
                     "type": "instrText",
                   },
@@ -24383,8 +24399,12 @@ Object {
                 "children": Array [
                   Object {
                     "data": Object {
-                      "data": "PAGEREF _Toc89816788 \\\\h",
-                      "type": "unsupported",
+                      "data": Object {
+                        "hyperlink": true,
+                        "page_ref": "PAGEREF",
+                        "relative_position": false,
+                      },
+                      "type": "pageref",
                     },
                     "type": "instrText",
                   },