Read hyperlink instr (#518)

* fix: read hyperlink instr

* fix

* fix

* fix

* rc7

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

@@ -4,7 +4,9 @@ use crate::documents::*;
 use crate::types::*;
 use crate::xml_builder::*;
 
-#[derive(Serialize, Debug, Clone, PartialEq)]
+#[derive(Serialize, Debug, Clone, PartialEq, ts_rs::TS)]
+#[ts(export)]
+#[serde(rename_all = "camelCase")]
 pub struct FieldChar {
     pub field_char_type: FieldCharType,
     pub dirty: bool,

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

@@ -0,0 +1,68 @@
+use serde::Serialize;
+
+// https://c-rex.net/projects/samples/ooxml/e1/Part4/OOXML_P4_DOCX_HYPERLINKHYPERLINK_topic_ID0EFYG1.html
+#[derive(Serialize, Debug, Clone, PartialEq, Default, ts_rs::TS)]
+#[ts(export)]
+#[serde(rename_all = "camelCase")]
+pub struct InstrHyperlink {
+    pub target: String,
+    // \l
+    pub anchor: bool,
+}
+
+impl InstrHyperlink {
+    pub fn new(target: impl Into<String>) -> Self {
+        Self {
+            target: target.into(),
+            ..Default::default()
+        }
+    }
+}
+
+// impl BuildXML for instrHyperlink {
+//     fn build(&self) -> Vec<u8> {
+// TODO:
+//     }
+// }
+
+impl std::str::FromStr for InstrHyperlink {
+    type Err = ();
+
+    fn from_str(instr: &str) -> Result<Self, Self::Err> {
+        let mut s = instr.split(' ');
+        let mut target = "".to_string();
+        let mut anchor = false;
+        loop {
+            if let Some(i) = s.next() {
+                match i {
+                    "\\l" => {
+                        anchor = true;
+                    }
+                    "\\m" => {
+                        // TODO:
+                    }
+                    "\\n" => {
+                        // TODO:
+                    }
+                    "\\o" => {
+                        // TODO: Support later
+                        let _ = s.next();
+                    }
+                    "\\t" => {
+                        // TODO: Support later
+                        let _ = s.next();
+                    }
+                    _ => {
+                        target = i.replace("&quot;", "").replace("\"", "").to_string();
+                    }
+                }
+            } else {
+                // FIXME: For now, return error if target is not found
+                if target.is_empty() {
+                    return Err(());
+                }
+                return Ok(Self { target, anchor });
+            }
+        }
+    }
+}

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

@@ -4,6 +4,7 @@ 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)]
+#[serde(rename_all = "camelCase")]
 pub struct InstrPAGEREF {
     pub page_ref: String,
     pub hyperlink: bool,

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

@@ -4,6 +4,7 @@ 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)]
+#[serde(rename_all = "camelCase")]
 pub struct InstrTC {
     pub text: String,
     // \n Omits the page number for the entry.

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

@@ -9,6 +9,7 @@ pub enum InstrText {
     TOC(InstrToC),
     TC(InstrTC),
     PAGEREF(InstrPAGEREF),
+    HYPERLINK(InstrHyperlink),
     Unsupported(String),
 }
 
@@ -18,6 +19,7 @@ impl BuildXML for Box<InstrText> {
             InstrText::TOC(toc) => toc.build(),
             InstrText::TC(tc) => tc.build(),
             InstrText::PAGEREF(page_ref) => page_ref.build(),
+            InstrText::HYPERLINK(_link) => todo!(),
             InstrText::Unsupported(s) => s.as_bytes().to_vec(),
         };
         XMLBuilder::new()
@@ -52,6 +54,12 @@ impl Serialize for InstrText {
                 t.serialize_field("data", s)?;
                 t.end()
             }
+            InstrText::HYPERLINK(ref s) => {
+                let mut t = serializer.serialize_struct("HYPERLINK", 2)?;
+                t.serialize_field("type", "hyperlink")?;
+                t.serialize_field("data", s)?;
+                t.end()
+            }
             InstrText::Unsupported(ref s) => {
                 let mut t = serializer.serialize_struct("Unsupported", 2)?;
                 t.serialize_field("type", "unsupported")?;

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

@@ -2,7 +2,8 @@ use serde::Serialize;
 
 use crate::documents::*;
 
-#[derive(Serialize, Debug, Clone, PartialEq, Default)]
+#[derive(Serialize, Debug, Clone, PartialEq, Default, ts_rs::TS)]
+#[ts(export)]
 pub struct StyleWithLevel(pub (String, usize));
 
 impl StyleWithLevel {
@@ -11,7 +12,9 @@ impl StyleWithLevel {
     }
 }
 // https://c-rex.net/projects/samples/ooxml/e1/Part4/OOXML_P4_DOCX_TOCTOC_topic_ID0ELZO1.html
-#[derive(Serialize, Debug, Clone, PartialEq, Default)]
+#[derive(Serialize, Debug, Clone, PartialEq, Default, ts_rs::TS)]
+#[ts(export)]
+#[serde(rename_all = "camelCase")]
 pub struct InstrToC {
     // \o If no heading range is specified, all heading levels used in the document are listed.
     #[serde(skip_serializing_if = "Option::is_none")]

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

@@ -34,6 +34,7 @@ mod hyperlink;
 mod indent;
 mod indent_level;
 mod insert;
+mod instr_hyperlink;
 mod instr_pageref;
 mod instr_tc;
 mod instr_text;
@@ -148,6 +149,7 @@ pub use hyperlink::*;
 pub use indent::*;
 pub use indent_level::*;
 pub use insert::*;
+pub use instr_hyperlink::*;
 pub use instr_pageref::*;
 pub use instr_tc::*;
 pub use instr_text::*;

docx-core/src/reader/instr_text.rs

@@ -38,6 +38,11 @@ impl ElementReader for InstrText {
                                         return Ok(InstrText::TC(instr));
                                     }
                                 }
+                                if instr.starts_with("HYPERLINK") {
+                                    if let Ok(instr) = InstrHyperlink::from_str(instr) {
+                                        return Ok(InstrText::HYPERLINK(instr));
+                                    }
+                                }
                                 if instr.starts_with("PAGEREF") {
                                     if let Ok(instr) = InstrPAGEREF::from_str(instr) {
                                         return Ok(InstrText::PAGEREF(instr));

docx-core/src/types/field_char_type.rs

@@ -10,7 +10,9 @@ use wasm_bindgen::prelude::*;
 use super::errors;
 
 #[wasm_bindgen]
-#[derive(Copy, Clone, Debug, Deserialize, Serialize, PartialEq)]
+#[derive(Copy, Clone, Debug, Deserialize, Serialize, PartialEq, ts_rs::TS)]
+#[ts(export)]
+#[serde(rename_all = "camelCase")]
 pub enum FieldCharType {
     Begin,
     Separate,

docx-wasm/js/json/bindings/FieldChar.ts

@@ -0,0 +1,3 @@
+import type { FieldCharType } from "./FieldCharType";
+
+export interface FieldChar { fieldCharType: FieldCharType, dirty: boolean, }

docx-wasm/js/json/bindings/FieldCharType.ts

@@ -0,0 +1,2 @@
+
+export type FieldCharType = "begin" | "separate" | "end" | "unsupported";

docx-wasm/js/json/bindings/InstrHyperlink.ts

@@ -0,0 +1,2 @@
+
+export interface InstrHyperlink { target: string, anchor: boolean, }

docx-wasm/js/json/bindings/InstrToC.ts

@@ -0,0 +1,3 @@
+import type { StyleWithLevel } from "./StyleWithLevel";
+
+export interface InstrToC { headingStylesRange?: [number, number], tcFieldLevelRange?: [number, number], omitPageNumbersLevelRange?: [number, number], entryBookmarkName?: string, stylesWithLevels: Array<StyleWithLevel>, entryAndPageNumberSeparator?: string, sequenceAndPageNumbersSeparator?: string, captionLabel: string | null, captionLabelIncludingNumbers?: string, seqFieldIdentifierForPrefix?: string, tcFieldIdentifier?: string, hyperlink: boolean, preserveTab: boolean, preserveNewLine: boolean, useAppliedParagraphLineLevel: boolean, hideTabAndPageNumbersInWebview: boolean, }

docx-wasm/js/json/bindings/StyleWithLevel.ts

@@ -0,0 +1,2 @@
+
+export type StyleWithLevel = [string, number];

docx-wasm/js/json/run.ts

@@ -4,6 +4,9 @@ import { CommentRangeStartJSON, CommentRangeEndJSON } from "..";
 import { BorderType } from "../border";
 import { InsertJSON, DeleteJSON } from "./paragraph";
 import { VertAlignType } from "../run";
+import { FieldChar } from "./bindings/FieldChar";
+import { InstrHyperlink } from "./bindings/InstrHyperlink";
+import { InstrToC } from "./bindings/InstrToC";
 
 export type TextBorderJSON = {
   borderType: BorderType;
@@ -53,7 +56,9 @@ export type RunChildJSON =
   | DrawingJSON
   | ShapeJSON
   | CommentRangeStartJSON
-  | CommentRangeEndJSON;
+  | CommentRangeEndJSON
+  | FieldCharJSON
+  | InstrTextJSON;
 
 export type TextJSON = {
   type: "text";
@@ -89,3 +94,21 @@ export type RunJSON = {
     children: RunChildJSON[];
   };
 };
+
+export type FieldCharJSON = {
+  type: "fieldChar";
+  data: FieldChar;
+};
+
+export type InstrTextJSON = {
+  type: "instrText";
+  data:
+    | {
+        type: "hyperlink";
+        data: InstrHyperlink;
+      }
+    | {
+        type: "toc";
+        data: InstrToC;
+      };
+};

docx-wasm/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docx-wasm",
-  "version": "0.0.276-rc4",
+  "version": "0.0.276-rc7",
   "main": "dist/node/index.js",
   "browser": "dist/web/index.js",
   "author": "bokuweb <bokuweb12@gmail.com>",

docx-wasm/test/index.test.js

@@ -159,6 +159,12 @@ describe("reader", () => {
     const json = w.readDocx(buffer);
     expect(json).toMatchSnapshot();
   });
+
+  test("should read hyperlink instr", () => {
+    const buffer = readFileSync("../fixtures/instr_links/instr_links.docx");
+    const json = w.readDocx(buffer);
+    expect(json).toMatchSnapshot();
+  });
 });
 
 describe("writer", () => {