fix: keep the text of unchanged Recalbox elements

Every <bios> element the model held was rendered again, so attribute order, alignment, a line break inside md5 and the case of a digest changed in 222 elements that stated the same thing. The diff deleted 451 lines for 136 corrections. An element whose attributes say what Recalbox wrote now keeps Recalbox's own text: 175 lines, the corrections.

scripts/exporter/recalbox_exporter.py

@@ -11,6 +11,8 @@ when the platform stated it explicitly.
 
 from __future__ import annotations
 
+import html
+import re
 import sys
 from pathlib import Path
 from xml.etree.ElementTree import Comment, ParseError
@@ -33,6 +35,32 @@ SCHEMA_URL = (
 )
 
 
+_ATTRIBUTE = re.compile(r'([A-Za-z]+)\s*=\s*"([^"]*)"')
+_SYSTEM_BLOCK = re.compile(r"<system\b.*?</system>", re.S)
+_BIOS_ELEMENT = re.compile(r"[ \t]*<bios\b.*?/>", re.S)
+
+
+def _stated(text: str) -> dict[str, str]:
+    """What an element says, without its layout: attribute order, spacing,
+    line breaks inside a value and the case of a digest are not content."""
+    attrs = {k: " ".join(html.unescape(v).split()) for k, v in _ATTRIBUTE.findall(text)}
+    if "md5" in attrs:
+        attrs["md5"] = ",".join(h.strip().lower() for h in attrs["md5"].split(","))
+    return attrs
+
+
+def _original_elements(original: str) -> list[dict[str, str]]:
+    """Each system block's <bios> elements as written, by path, in order."""
+    blocks = []
+    for block in _SYSTEM_BLOCK.finditer(original):
+        elements: dict[str, str] = {}
+        for match in _BIOS_ELEMENT.finditer(block.group(0)):
+            raw = match.group(0)
+            elements.setdefault(_stated(raw).get("path", ""), raw)
+        blocks.append(elements)
+    return blocks
+
+
 class Exporter(BaseExporter):
     """Write Recalbox's es_bios.xml, corrected."""
 
@@ -179,17 +207,26 @@ class Exporter(BaseExporter):
             lines.extend(body)
             lines.append("  </system>")
 
-        # Each block of theirs corrects the entries it holds.
+        # Each block of theirs corrects the entries it holds. An element that
+        # states what Recalbox wrote keeps Recalbox's text: re-rendered, 222
+        # unchanged elements made 180 lines of diff the report never named.
+        written = _original_elements(original) if original else []
         blocks: list[tuple[str, str, list[str]]] = []
-        for platform, fullname, children in theirs:
+        for index, (platform, fullname, children) in enumerate(theirs):
             elements = ours.get(platform, (None, {}))[1]
+            raw = written[index] if index < len(written) else {}
             body: list[str] = []
             for child in children:
                 if isinstance(child, str):
                     body.append(f"    {child}")
                     continue
-                mine = elements.pop(child.get("path", ""), None)
-                body.append(mine[1] if mine else element(child))
+                path = child.get("path", "")
+                mine = elements.pop(path, None)
+                kept = raw.get(" ".join(path.split()))
+                if kept is not None and (mine is None or _stated(mine[1]) == _stated(kept)):
+                    body.append(kept)
+                else:
+                    body.append(mine[1] if mine else element(child))
             blocks.append((platform, fullname, body))
 
         # An addition joins the block named like its file, else the first