Implement multiparagraph values.

diff --git a/rust/pspp/src/spv/read/html.rs b/rust/pspp/src/spv/read/html.rs
index 4cc4b12caf29cdf2ebf44a60ec46143cbadd5a56..7f4b61c55bb0e10828fcbef524d575015436cc9e 100644 (file)
--- a/rust/pspp/src/spv/read/html.rs
+++ b/rust/pspp/src/spv/read/html.rs
@@ -438,6 +438,41 @@ impl Markup {
             Some(results)
         }
     }
+
+    fn append(&mut self, other: &mut Self) {
+        let mut a = take(self).into_seq();
+        let mut b = take(other).into_seq();
+        if let Some(Markup::Text(head)) = a.last_mut()
+            && let Some(Markup::Text(tail)) = b.first()
+        {
+            head.push_str(&tail);
+            a.extend(b.drain(1..));
+        } else {
+            a.append(&mut b);
+        }
+        *self = Self::from_seq(a);
+    }
+
+    fn into_seq(self) -> Vec<Markup> {
+        if self.is_empty() {
+            Vec::new()
+        } else if let Markup::Seq(markups) = self {
+            markups
+        } else {
+            vec![self]
+        }
+    }
+
+    fn from_seq(seq: Vec<Self>) -> Self {
+        if seq.is_empty() {
+            Self::default()
+        } else {
+            match <[Self; 1]>::try_from(seq) {
+                Ok([singleton]) => singleton,
+                Err(multiple) => Self::Seq(multiple),
+            }
+        }
+    }
 }
 
 /// A block of styled text.
@@ -547,10 +582,16 @@ impl Document {
         }
     }
 
-    /// Returns the document converted to a [Value].  If the document contains
-    /// more than one [Block], only the first one appears in the [Value].
+    /// Returns the document converted to a [Value].  If the document has
+    /// multiple [Block]s, then they are concatenated with new-lines in between.
     pub fn into_value(self) -> Value {
-        self.0.into_iter().next().unwrap_or_default().into_value()
+        let mut iter = self.0.into_iter();
+        let mut block = iter.next().unwrap_or_default();
+        for mut additional in iter {
+            block.markup.append(&mut Markup::Text(String::from("\n")));
+            block.markup.append(&mut additional.markup);
+        }
+        block.into_value()
     }
 
     /// Returns the document converted to XHTML, except that the result will not
@@ -1098,6 +1139,38 @@ NPAR TEST
         );
     }
 
+    /// From the corpus, anonymized.
+    ///
+    /// This tests treatment of multiple paragraphs in a context where we
+    /// usually expect only one.
+    #[test]
+    fn multiparagraph_value() {
+        let text = r##"<xml>&lt;html>
+  
+
+  &lt;head>
+
+  &lt;/head>
+  &lt;body>
+    &lt;p style="margin-top: 0">
+      H0:There is no association between X and Y
+    &lt;/p>
+    &lt;p style="margin-top: 0">
+      H1:There is association between X and Y
+    &lt;/p>
+  &lt;/body>
+&lt;/html>
+</xml>"##;
+        let content = quick_xml::de::from_str::<String>(text).unwrap();
+        let html = Document::from_html(&content);
+        let s = html.into_value().display(()).to_string();
+        assert_eq!(
+            s,
+            "H0:There is no association between X and Y
+H1:There is association between X and Y"
+        );
+    }
+
     /// Checks that the `escape-html` feature is enabled in [quick_xml], since
     /// we need that to resolve `&nbsp;` and other HTML entities.
     #[test]