From d8005e720a86433fd83b3c94f7ca7d3ea033e921 Mon Sep 17 00:00:00 2001
From: "codeflash-ai[bot]"
<148906541+codeflash-ai[bot]@users.noreply.github.com>
Date: Tue, 2 Dec 2025 05:29:04 +0000
Subject: [PATCH] Optimize _wrap_repr
The optimized version achieves an **81% speedup** by eliminating the expensive `"".join()` operation on a 27-element list and replacing it with direct string concatenation using f-string formatting.
**Key optimizations applied:**
1. **Eliminated list construction and join**: The original code created a 27-element list of string literals and then called `"".join()` on it. The optimized version uses Python's automatic string literal concatenation within parentheses, which happens at compile time rather than runtime.
2. **Reduced string operations**: Instead of 27 separate string objects being joined at runtime, the optimized version creates one large string template with only 2 variable interpolations (`height` and `r`).
3. **Minor conditional optimization**: Changed `end is False` to `not end` for slightly better performance.
**Why this leads to speedup:**
- `"".join()` on many small strings requires Python to iterate through the list, calculate total length, allocate a new string buffer, and copy each element
- String literal concatenation in parentheses is optimized by Python's compiler into a single string constant
- The line profiler shows the join operation took 16.6% of total time in the original, which is completely eliminated
**Impact on workloads:**
Based on the function reference, `_wrap_repr` is called in a loop within `summarize_children()` for each child node in a data tree structure. This means the 81% speedup will be multiplied across potentially many child nodes, making tree rendering significantly faster. The test results show consistent 50-100% speedups across all test cases, with larger inputs still benefiting substantially (54-68% faster for large-scale tests).
The optimization is particularly effective for this HTML templating use case where most of the output is static markup with only a few dynamic variables.
---
xarray/datatree_/datatree/formatting_html.py | 66 ++++++++++----------
1 file changed, 34 insertions(+), 32 deletions(-)
diff --git a/xarray/datatree_/datatree/formatting_html.py b/xarray/datatree_/datatree/formatting_html.py
index 547b567a396..d947c6d4efc 100644
--- a/xarray/datatree_/datatree/formatting_html.py
+++ b/xarray/datatree_/datatree/formatting_html.py
@@ -19,9 +19,11 @@ def summarize_children(children: Mapping[str, Any]) -> str:
lines_callback = lambda n, c, end: _wrap_repr(node_repr(n, c), end=end)
children_html = "".join(
- lines_callback(n, c, end=False) # Long lines
- if i < N_CHILDREN
- else lines_callback(n, c, end=True) # Short lines
+ (
+ lines_callback(n, c, end=False) # Long lines
+ if i < N_CHILDREN
+ else lines_callback(n, c, end=True)
+ ) # Short lines
for i, (n, c) in enumerate(children.items())
)
@@ -98,35 +100,35 @@ def _wrap_repr(r: str, end: bool = False) -> str:
"""
# height of line
end = bool(end)
- height = "100%" if end is False else "1.2em"
- return "".join(
- [
- "",
- "
",
- "
",
- "
",
- "
",
- "
",
- "
"
+ "
"
+ "
"
+ "
"
+ "
"
+ "
"
+ "