Merge pull request #173 from developer0hye/fix/pptx-nowrap-regression

fix: restore pptx mixed-script no-wrap titles

Author: developer0hye

crates/office2pdf/src/render/typst_gen_fixed_page_textbox_tests.rs

@@ -1039,6 +1039,113 @@ fn test_fixed_page_text_box_no_wrap_keeps_latin_text_extractable() {
     );
 }
 
+#[test]
+fn test_fixed_page_text_box_no_wrap_keeps_mixed_script_titles_unbroken() {
+    let doc = make_doc(vec![make_fixed_page(
+        960.0,
+        540.0,
+        vec![FixedElement {
+            x: 100.0,
+            y: 120.0,
+            width: 320.0,
+            height: 40.0,
+            kind: FixedElementKind::TextBox(crate::ir::TextBoxData {
+                content: vec![Block::Paragraph(Paragraph {
+                    style: ParagraphStyle {
+                        alignment: Some(Alignment::Center),
+                        ..ParagraphStyle::default()
+                    },
+                    runs: vec![Run {
+                        text: "III. 기술부문".to_string(),
+                        style: TextStyle {
+                            font_size: Some(28.0),
+                            ..TextStyle::default()
+                        },
+                        href: None,
+                        footnote: None,
+                    }],
+                })],
+                padding: Insets::default(),
+                vertical_align: crate::ir::TextBoxVerticalAlign::Top,
+                fill: None,
+                opacity: None,
+                stroke: None,
+                shape_kind: None,
+                no_wrap: true,
+                auto_fit: false,
+            }),
+        }],
+    )]);
+    let output = generate_typst(&doc).unwrap();
+    assert!(
+        output.source.contains("I\u{2060}I\u{2060}I\u{2060}.")
+            && output
+                .source
+                .contains("\u{00A0}\u{2060}기\u{2060}술\u{2060}부\u{2060}문"),
+        "Expected mixed-script no-wrap title to keep the full heading unbreakable, got:\n{}",
+        output.source,
+    );
+}
+
+#[test]
+fn test_fixed_page_text_box_no_wrap_preserves_mixed_script_titles_across_runs() {
+    let doc = make_doc(vec![make_fixed_page(
+        960.0,
+        540.0,
+        vec![FixedElement {
+            x: 100.0,
+            y: 120.0,
+            width: 320.0,
+            height: 40.0,
+            kind: FixedElementKind::TextBox(crate::ir::TextBoxData {
+                content: vec![Block::Paragraph(Paragraph {
+                    style: ParagraphStyle {
+                        alignment: Some(Alignment::Center),
+                        ..ParagraphStyle::default()
+                    },
+                    runs: vec![
+                        Run {
+                            text: "III.".to_string(),
+                            style: TextStyle {
+                                font_size: Some(28.0),
+                                ..TextStyle::default()
+                            },
+                            href: None,
+                            footnote: None,
+                        },
+                        Run {
+                            text: " 기술부문".to_string(),
+                            style: TextStyle {
+                                font_size: Some(40.0),
+                                ..TextStyle::default()
+                            },
+                            href: None,
+                            footnote: None,
+                        },
+                    ],
+                })],
+                padding: Insets::default(),
+                vertical_align: crate::ir::TextBoxVerticalAlign::Top,
+                fill: None,
+                opacity: None,
+                stroke: None,
+                shape_kind: None,
+                no_wrap: true,
+                auto_fit: false,
+            }),
+        }],
+    )]);
+    let output = generate_typst(&doc).unwrap();
+    assert!(
+        output.source.contains("I\u{2060}I\u{2060}I\u{2060}.")
+            && output
+                .source
+                .contains("\u{00A0}\u{2060}기\u{2060}술\u{2060}부\u{2060}문"),
+        "Expected mixed-script no-wrap title to stay unbroken across runs, got:\n{}",
+        output.source,
+    );
+}
+
 #[test]
 fn test_fixed_page_text_box_auto_fit_short_text_uses_scale_to_fit() {
     let doc = make_doc(vec![make_fixed_page(

crates/office2pdf/src/render/typst_gen_text.rs

@@ -133,12 +133,23 @@ pub(super) fn generate_runs_with_tabs_no_wrap(
     runs: &[Run],
     tab_stops: Option<&[TabStop]>,
 ) {
+    let preserve_cjk_no_wrap: bool = runs
+        .iter()
+        .filter(|run| run.footnote.is_none())
+        .any(|run| run.text.chars().any(is_cjk_like));
+    let mut no_wrap_state: NoWrapState = NoWrapState::default();
     let transformed_runs: Vec<Run> = runs
         .iter()
         .map(|run| {
             let mut transformed_run: Run = run.clone();
             if transformed_run.footnote.is_none() {
-                transformed_run.text = no_wrap_text(&transformed_run.text);
+                transformed_run.text = no_wrap_text(
+                    &transformed_run.text,
+                    preserve_cjk_no_wrap,
+                    &mut no_wrap_state,
+                );
+            } else {
+                no_wrap_state = NoWrapState::default();
             }
             transformed_run
         })
@@ -147,6 +158,12 @@ pub(super) fn generate_runs_with_tabs_no_wrap(
     generate_runs_with_tabs(out, &transformed_runs, tab_stops);
 }
 
+#[derive(Clone, Copy, Default)]
+struct NoWrapState {
+    previous_visible_char: Option<char>,
+    previous_non_breaking_space: bool,
+}
+
 /// Emits Typst variable bindings for a non-first tab segment: measurement,
 /// decimal anchor (if applicable), default remainder, advance, fill, and
 /// the accumulated prefix content variable.
@@ -208,29 +225,44 @@ pub(super) fn generate_runs(out: &mut String, runs: &[Run]) {
     }
 }
 
-fn no_wrap_text(text: &str) -> String {
+fn no_wrap_text(text: &str, preserve_cjk_no_wrap: bool, state: &mut NoWrapState) -> String {
+    if !preserve_cjk_no_wrap {
+        return text.to_string();
+    }
+
     let mut out: String = String::new();
-    let mut previous_visible_char: Option<char> = None;
 
     for ch in text.chars() {
         if matches!(ch, '\t' | PPTX_SOFT_LINE_BREAK_CHAR) {
             out.push(ch);
-            previous_visible_char = None;
+            *state = NoWrapState::default();
+            continue;
+        }
+
+        if ch == ' ' {
+            out.push('\u{00A0}');
+            state.previous_visible_char = None;
+            state.previous_non_breaking_space = true;
             continue;
         }
 
-        if previous_visible_char.is_some_and(|prev| needs_cjk_no_wrap_joiner(prev, ch)) {
+        if state.previous_non_breaking_space
+            || state
+                .previous_visible_char
+                .is_some_and(|prev| needs_no_wrap_joiner(prev, ch))
+        {
             out.push('\u{2060}');
         }
         out.push(ch);
-        previous_visible_char = (!ch.is_whitespace()).then_some(ch);
+        state.previous_visible_char = Some(ch);
+        state.previous_non_breaking_space = false;
     }
 
     out
 }
 
-fn needs_cjk_no_wrap_joiner(previous: char, current: char) -> bool {
-    is_cjk_like(previous) && is_cjk_like(current)
+fn needs_no_wrap_joiner(previous: char, current: char) -> bool {
+    !previous.is_whitespace() && !current.is_whitespace()
 }
 
 fn is_cjk_like(ch: char) -> bool {