Disallow single-line implicit concatenated strings

[MichaReiser]

Summary

I mainly put this up because I'm curious about the ecosystem changes. A proper implementation has to take the result of FormatImplicitConcatenatedFlat::new into account instead of using the simplified approximation that forces the parent to expand when any string part is a raw or triple quoted string.

The goal of this change is to fix the formatter and ISC001 incompatibility by preventing that the formatter ever introduces a single-line implicitly concatenated string.

Related to https://github.com/astral-sh/ruff/issues/8272

Test Plan

[github-actions[bot]]

ruff-ecosystem results

Formatter (stable)

✅ ecosystem check detected no format changes.

Formatter (preview)

ℹ️ ecosystem check detected format changes. (+18 -6 lines in 4 files in 3 projects; 51 projects unchanged)

bokeh/bokeh (+11 -3 lines across 2 files)

ruff format --preview

examples/styling/mathtext/latex_bessel.py~L21

     width=700,
     height=500,
     title=(
-        r"Bessel functions of the first kind: $$J_\alpha(x) = \sum_{m=0}^{\infty}" r"\frac{(-1)^m}{m!\:\Gamma(m+\alpha+1)} \left(\frac{x}{2}\right)^{2m+\alpha}$$"
+        r"Bessel functions of the first kind: $$J_\alpha(x) = \sum_{m=0}^{\infty}"
+        r"\frac{(-1)^m}{m!\:\Gamma(m+\alpha+1)} \left(\frac{x}{2}\right)^{2m+\alpha}$$"
     ),
 )
 p.x_range.range_padding = 0

src/bokeh/core/property/color.py~L126

             Regex(r"^#[0-9a-fA-F]{4}$"),
             Regex(r"^#[0-9a-fA-F]{6}$"),
             Regex(r"^#[0-9a-fA-F]{8}$"),
-            Regex(r"^rgba\(((25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*," r"\s*?){2}(25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*," r"\s*([01]\.?\d*?)\)"),
-            Regex(r"^rgb\(((25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*," r"\s*?){2}(25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*?\)"),
+            Regex(
+                r"^rgba\(((25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*,"
+                r"\s*?){2}(25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*,"
+                r"\s*([01]\.?\d*?)\)"
+            ),
+            Regex(
+                r"^rgb\(((25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*,"
+                r"\s*?){2}(25[0-5]|2[0-4]\d|1\d{1,2}|\d\d?)\s*?\)"
+            ),
             Tuple(Byte, Byte, Byte),
             Tuple(Byte, Byte, Byte, Percent),
             RGB,

pandas-dev/pandas (+4 -2 lines across 1 file)

ruff format --preview

pandas/tests/tools/test_to_datetime.py~L1346

         assert res is NaT
 
         msg = "|".join([
-            r'^time data "a" doesn\'t match format "%H:%M:%S". ' f"{PARSING_ERR_MSG}$",
+            r'^time data "a" doesn\'t match format "%H:%M:%S". '
+            f"{PARSING_ERR_MSG}$",
             r'^Given date string "a" not likely a datetime$',
             r'^unconverted data remains when parsing with format "%H:%M:%S": "9". '
             f"{PARSING_ERR_MSG}$",

pandas/tests/tools/test_to_datetime.py~L1396

 
         msg = "|".join([
             r'^Given date string "a" not likely a datetime$',
-            r'^time data "a" doesn\'t match format "%H:%M:%S". ' f"{PARSING_ERR_MSG}$",
+            r'^time data "a" doesn\'t match format "%H:%M:%S". '
+            f"{PARSING_ERR_MSG}$",
             r'^unconverted data remains when parsing with format "%H:%M:%S": "9". '
             f"{PARSING_ERR_MSG}$",
             r"^second must be in 0..59: 00:01:99$",

rotki/rotki (+3 -1 lines across 1 file)

ruff format --preview

rotkehlchen/exchanges/coinbase.py~L68

 LEGACY_RE: re.Pattern = re.compile(r"^[\w]+$")
 NEW_RE: re.Pattern = re.compile(r"^organizations/[\w-]+/apiKeys/[\w-]+$")
 PRIVATE_KEY_RE: re.Pattern = re.compile(
-    r"^-----BEGIN EC PRIVATE KEY-----\n" r"[\w+/=\n]+" r"-----END EC PRIVATE KEY-----\n?$",
+    r"^-----BEGIN EC PRIVATE KEY-----\n"
+    r"[\w+/=\n]+"
+    r"-----END EC PRIVATE KEY-----\n?$",
     re.MULTILINE,
 )
 

[MichaReiser]

I think this is a smart change. We should document it in the Black deviations though.

Thanks for the feedback. I plan to document the deviations as part of promoting the new style guide.