fix(HTML): Fixes to list and table encoding for compatability with ad…

…dress resolving algorithm
stencila · Oct 11, 2021 · a57d4b0 · a57d4b0
1 parent b8ffeda
commit a57d4b0
Show file tree

Hide file tree

Showing 10 changed files with 352 additions and 161 deletions.
diff --git a/fixtures/fragments/html/table.html b/fixtures/fragments/html/table.html
@@ -0,0 +1,32 @@
+<p>Example based on how a Pandas dat a frame is encoded to HTML in a Jupyter notebook:</p>
+
+<table>
+  <thead>
+    <tr>
+      <th></th>
+      <th>a</th>
+      <th>b</th>
+      <th>c</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <th>0</th>
+      <td>1</td>
+      <td>2</td>
+      <td>3</td>
+    </tr>
+    <tr>
+      <th>1</th>
+      <td>4</td>
+      <td>5</td>
+      <td>6</td>
+    </tr>
+    <tr>
+      <th>2</th>
+      <td>7</td>
+      <td>8</td>
+      <td>9</td>
+    </tr>
+  </tbody>
+</table>
diff --git a/rust/src/methods/decode/html.rs b/rust/src/methods/decode/html.rs
@@ -9,8 +9,8 @@ use stencila_schema::{
     Delete, Emphasis, Heading, ImageObjectSimple, InlineContent, Link, List, ListItem,
     ListItemContent, ListOrder, Node, NontextualAnnotation, NumberValidator, Paragraph, Parameter,
     Quote, QuoteBlock, StringValidator, Strong, Subscript, Superscript, TableCell,
-    TableCellContent, TableRow, TableRowRowType, TableSimple, ThematicBreak, ValidatorTypes,
-    VideoObjectSimple,
+    TableCellCellType, TableCellContent, TableRow, TableRowRowType, TableSimple, ThematicBreak,
+    ValidatorTypes, VideoObjectSimple,
 };
 
 /// Decode a HTML document to a `Node`
@@ -535,6 +535,12 @@ fn decode_table_cells(node: &NodeRef, context: &Context) -> Vec<TableCell> {
         .filter_map(|child| {
             if let Some(element) = child.as_element() {
                 if matches!(element.name.local, local_name!("td") | local_name!("th")) {
+                    let cell_type = if matches!(element.name.local, local_name!("th")) {
+                        Some(TableCellCellType::Header)
+                    } else {
+                        None
+                    };
+
                     let blocks = decode_blocks(&child, context);
                     let content = if blocks.len() > 1 {
                         Some(TableCellContent::VecBlockContent(blocks))
@@ -548,7 +554,9 @@ fn decode_table_cells(node: &NodeRef, context: &Context) -> Vec<TableCell> {
                             None
                         }
                     };
+
                     return Some(TableCell {
+                        cell_type,
                         content,
                         ..Default::default()
                     });

diff --git a/rust/src/methods/decode/snapshots/html_articles@simple.html.snap b/rust/src/methods/decode/snapshots/html_articles@simple.html.snap
@@ -57,12 +57,14 @@ input_file: fixtures/articles/simple.html
           "cells": [
             {
               "type": "TableCell",
+              "cellType": "Header",
               "content": [
                 "Group"
               ]
             },
             {
               "type": "TableCell",
+              "cellType": "Header",
               "content": [
                 "Value"
               ]

diff --git a/rust/src/methods/decode/snapshots/html_fragments@table.html.snap b/rust/src/methods/decode/snapshots/html_fragments@table.html.snap
@@ -0,0 +1,140 @@
+---
+source: rust/src/methods/decode/html.rs
+expression: "decode_fragment(&content, false)"
+input_file: fixtures/fragments/html/table.html
+
+---
+[
+  {
+    "type": "Paragraph",
+    "content": [
+      "Example based on how a Pandas dat a frame is encoded to HTML in a Jupyter notebook:"
+    ]
+  },
+  {
+    "type": "Table",
+    "rows": [
+      {
+        "type": "TableRow",
+        "cells": [
+          {
+            "type": "TableCell",
+            "cellType": "Header"
+          },
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "a"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "b"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "c"
+            ]
+          }
+        ],
+        "rowType": "Header"
+      },
+      {
+        "type": "TableRow",
+        "cells": [
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "0"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "1"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "2"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "3"
+            ]
+          }
+        ]
+      },
+      {
+        "type": "TableRow",
+        "cells": [
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "1"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "4"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "5"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "6"
+            ]
+          }
+        ]
+      },
+      {
+        "type": "TableRow",
+        "cells": [
+          {
+            "type": "TableCell",
+            "cellType": "Header",
+            "content": [
+              "2"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "7"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "8"
+            ]
+          },
+          {
+            "type": "TableCell",
+            "content": [
+              "9"
+            ]
+          }
+        ]
+      }
+    ]
+  }
+]
diff --git a/rust/src/methods/decode/snapshots/ipynb_articles@kitchen-sink.ipynb.snap b/rust/src/methods/decode/snapshots/ipynb_articles@kitchen-sink.ipynb.snap
@@ -583,22 +583,26 @@ input_file: fixtures/articles/kitchen-sink.ipynb
               "type": "TableRow",
               "cells": [
                 {
-                  "type": "TableCell"
+                  "type": "TableCell",
+                  "cellType": "Header"
                 },
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "a"
                   ]
                 },
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "b"
                   ]
                 },
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "c"
                   ]
@@ -611,6 +615,7 @@ input_file: fixtures/articles/kitchen-sink.ipynb
               "cells": [
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "0"
                   ]
@@ -640,6 +645,7 @@ input_file: fixtures/articles/kitchen-sink.ipynb
               "cells": [
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "1"
                   ]
@@ -669,6 +675,7 @@ input_file: fixtures/articles/kitchen-sink.ipynb
               "cells": [
                 {
                   "type": "TableCell",
+                  "cellType": "Header",
                   "content": [
                     "2"
                   ]