Sort mark rendering by total frequency in the block.

klette · klette · commit 14bddb8a7bfe · 2021-04-14T23:17:45.000+02:00
Ensures that we prioritize starting with the "broadest" mark capturing the most content.

I suspect there are lurking bugs here on huge blocks, but this makes the test suite pass for now
diff --git a/sanity_html/dataclasses.py b/sanity_html/dataclasses.py
@@ -45,6 +45,7 @@ class Block:
     children: list[dict] = field(default_factory=list)
     markDefs: list[dict] = field(default_factory=list)
     marker_definitions: dict[str, Type[MarkerDefinition]] = field(init=False)
+    marker_frequencies: dict[str, int] = field(init=False)
 
     def __post_init__(self) -> None:
         """
@@ -54,6 +55,17 @@ def __post_init__(self) -> None:
         we can directly look up both annotation marks or decorator marks.
         """
         self.marker_definitions = get_marker_definitions(self.markDefs)
+        self.marker_frequencies = self._compute_marker_frequencies()
+
+    def _compute_marker_frequencies(self) -> dict[str, int]:
+        counts: dict[str, int] = {}
+        for child in self.children:
+            for mark in child.get('marks', []):
+                if mark in counts:
+                    counts[mark] += 1
+                else:
+                    counts[mark] = 0
+        return counts
 
     def get_node_siblings(self, node: Union[dict, Span]) -> Tuple[Optional[dict], Optional[dict]]:
         """Return the sibling nodes (prev, next) to the given node."""
diff --git a/sanity_html/renderer.py b/sanity_html/renderer.py
@@ -102,15 +102,16 @@ def _render_span(self, span: Span, block: Block) -> str:
         prev_marks = prev_node.get('marks', []) if prev_node else []
         next_marks = next_node.get('marks', []) if next_node else []
 
-        for mark in span.marks:
+        sorted_marks = sorted(span.marks, key=lambda x: -block.marker_frequencies[x])
+        for mark in sorted_marks:
             if mark in prev_marks:
                 continue
             marker_callable = block.marker_definitions[mark]()
             result += marker_callable.render_prefix(span, mark, block)
 
         result += html.escape(span.text)
 
-        for mark in reversed(span.marks):
+        for mark in reversed(sorted_marks):
             if mark in next_marks:
                 continue
 
diff --git a/tests/fixtures/upstream/009-messy-link-text.json b/tests/fixtures/upstream/009-messy-link-text.json
@@ -1 +1,37 @@
-{"input":{"_type":"block","children":[{"_key":"a1ph4","_type":"span","marks":["zomgLink"],"text":"Sanity"},{"_key":"b374","_type":"span","marks":[],"text":" can be used to power almost any "},{"_key":"ch4r1i3","_type":"span","marks":["zomgLink","strong","em"],"text":"app"},{"_key":"d3174","_type":"span","marks":["em","zomgLink"],"text":" or website"},{"_key":"ech0","_type":"span","marks":[],"text":"."}],"markDefs":[{"_key":"zomgLink","_type":"link","href":"https://sanity.io/"}],"style":"blockquote"},"output":"<blockquote><a href=\"https://sanity.io/\">Sanity</a> can be used to power almost any <a href=\"https://sanity.io/\"><em><strong>app</strong> or website</em></a>.</blockquote>"}
+{
+  "input": {
+    "_type": "block",
+    "children": [
+      {
+        "_key": "a1ph4",
+        "_type": "span",
+        "marks": ["zomgLink"],
+        "text": "Sanity"
+      },
+      {
+        "_key": "b374",
+        "_type": "span",
+        "marks": [],
+        "text": " can be used to power almost any "
+      },
+      {
+        "_key": "ch4r1i3",
+        "_type": "span",
+        "marks": ["zomgLink", "strong", "em"],
+        "text": "app"
+      },
+      {
+        "_key": "d3174",
+        "_type": "span",
+        "marks": ["em", "zomgLink"],
+        "text": " or website"
+      },
+      { "_key": "ech0", "_type": "span", "marks": [], "text": "." }
+    ],
+    "markDefs": [
+      { "_key": "zomgLink", "_type": "link", "href": "https://sanity.io/" }
+    ],
+    "style": "blockquote"
+  },
+  "output": "<blockquote><a href=\"https://sanity.io/\">Sanity</a> can be used to power almost any <a href=\"https://sanity.io/\"><em><strong>app</strong> or website</em></a>.</blockquote>"
+}
