Fix update_github_links.py not to produce double parens (#4558)

This happened if you copy-paste title from github-generated commit into
changelog.

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: denik

CHANGELOG.md

@@ -9,22 +9,22 @@
 ## Release v0.289.0 (2026-02-19)
 
 ### CLI
-* Make auth profiles respect DATABRICKS_CLI_PATH env var (([#4467](https://github.com/databricks/cli/pull/4467)))
-* Fix arrow key navigation in prompts on Windows (([#4501](https://github.com/databricks/cli/pull/4501)))
+* Make auth profiles respect DATABRICKS_CLI_PATH env var ([#4467](https://github.com/databricks/cli/pull/4467))
+* Fix arrow key navigation in prompts on Windows ([#4501](https://github.com/databricks/cli/pull/4501))
 
 ### Bundles
 * Log artifact build output in debug mode ([#4208](https://github.com/databricks/cli/pull/4208))
 * Fix bundle init not working in Azure Government ([#4286](https://github.com/databricks/cli/pull/4286))
-* Allow single and double quotes in environment dependencies (([#4511](https://github.com/databricks/cli/pull/4511)))
-* Use purge option when deleting alerts (([#4505](https://github.com/databricks/cli/pull/4505)))
+* Allow single and double quotes in environment dependencies ([#4511](https://github.com/databricks/cli/pull/4511))
+* Use purge option when deleting alerts ([#4505](https://github.com/databricks/cli/pull/4505))
 * engine/direct: Replace server_side_default with more precise backend_default rule in bundle plan ([#4490](https://github.com/databricks/cli/pull/4490))
-* engine/direct: Extend pipelines recreate_on_changes configuration (([#4499](https://github.com/databricks/cli/pull/4499)))
+* engine/direct: Extend pipelines recreate_on_changes configuration ([#4499](https://github.com/databricks/cli/pull/4499))
 * engine/direct: Added support for UC external locations (direct only) ([#4484](https://github.com/databricks/cli/pull/4484))
 
 ### Dependency updates
-* Upgrade Go SDK to v0.106.0 (([#4486](https://github.com/databricks/cli/pull/4486)))
-* Upgrade Terraform provider to v1.106.0 (([#4542](https://github.com/databricks/cli/pull/4542)))
-* Upgrade Go to 1.25.7 (([#4547](https://github.com/databricks/cli/pull/4547)))
+* Upgrade Go SDK to v0.106.0 ([#4486](https://github.com/databricks/cli/pull/4486))
+* Upgrade Terraform provider to v1.106.0 ([#4542](https://github.com/databricks/cli/pull/4542))
+* Upgrade Go to 1.25.7 ([#4547](https://github.com/databricks/cli/pull/4547))
 
 
 ## Release v0.288.0 (2026-02-12)
@@ -49,7 +49,7 @@
 * Add support for Lakebase resources (postgres_projects, postgres_branches, postgres_endpoints) ([#4423](https://github.com/databricks/cli/pull/4423))
 * Add missing values to SchemaGrantPrivilege enum ([#4380](https://github.com/databricks/cli/pull/4380))
 * Added support for UC catalogs (only in direct mode) ([#4342](https://github.com/databricks/cli/pull/4342))
-* engine/direct: Fix updates to fields being ignored for database_catalogs, synced_database_tables (([#4388](https://github.com/databricks/cli/pull/4388)))
+* engine/direct: Fix updates to fields being ignored for database_catalogs, synced_database_tables ([#4388](https://github.com/databricks/cli/pull/4388))
 * engine/direct: Fix migration for database_instances, database_catalog, synced_database_tables ([#4424](https://github.com/databricks/cli/pull/4424))
 * Show warning message if valueFrom is used instead of value_from in app config section ([$4429](https://github.com/databricks/cli/pull/4429))
 
@@ -77,9 +77,9 @@
 * Pass additional Azure DevOps `SYSTEM_*` environment variables to Terraform for OIDC authentication: `SYSTEM_COLLECTIONURI`, `SYSTEM_DEFINITIONID`, `SYSTEM_HOSTTYPE`, `SYSTEM_JOBID`, `SYSTEM_TEAMPROJECT` ([#4318](https://github.com/databricks/cli/pull/4318))
 * Add support for valueFrom property (similar to app.yaml) inside Apps config field in bundle configuration ([#4297](https://github.com/databricks/cli/pull/4297))
 * engine/direct: Support bind & unbind. ([#4279](https://github.com/databricks/cli/pull/4279))
-* engine/direct: Ignore changes between nulls and empty slices/maps (([#4313](https://github.com/databricks/cli/pull/4313)))
-* engine/direct: Ignore changes between nulls and empty structs (([#4338](https://github.com/databricks/cli/pull/4338)))
-* On terraform `bundle plan -o json` will no longer include plan_version key, it's intended for direct engine only (([#4314](https://github.com/databricks/cli/pull/4314)))
+* engine/direct: Ignore changes between nulls and empty slices/maps ([#4313](https://github.com/databricks/cli/pull/4313))
+* engine/direct: Ignore changes between nulls and empty structs ([#4338](https://github.com/databricks/cli/pull/4338))
+* On terraform `bundle plan -o json` will no longer include plan_version key, it's intended for direct engine only ([#4314](https://github.com/databricks/cli/pull/4314))
 
 
 ## Release v0.283.0 (2026-01-14)
@@ -191,7 +191,7 @@ To disable this, set the environment variable DATABRICKS_CACHE_ENABLED to false.
 ### Bundles
 * Add `default-minimal` template for users who want a clean slate without sample code ([#3885](https://github.com/databricks/cli/pull/3885))
 * Add validation that served_models and served_entities are not used at the same time. Add client side translation logic. ([#3880](https://github.com/databricks/cli/pull/3880))
-* Filter out provided permissions, to select a single permissions level per principal (([#3904](https://github.com/databricks/cli/pull/3904))).
+* Filter out provided permissions, to select a single permissions level per principal ([#3904](https://github.com/databricks/cli/pull/3904)).
 
 
 ## Release v0.276.0 (2025-11-05)

tools/update_github_links.py

@@ -17,24 +17,38 @@
 
 DEFAULT_FILES = ("NEXT_CHANGELOG.md", "CHANGELOG.md")
 
-# Regex that matches an *already converted* link, e.g.:
-#   ([#1234](https://github.com/databricks/cli/pull/1234))
-# The groups capture the PR number in the text and in the URL respectively so
-# they can be compared for consistency.
+# Canonical form: ([#1234](https://github.com/databricks/cli/pull/1234))
 CONVERTED_LINK_RE = re.compile(
     r"\(\[#(?P<num_text>\d+)\]\("  # ([#1234](
     r"https://github\.com/databricks/cli/pull/(?P<num_url>\d+)"  # …/pull/1234
     r"\)\)"  # ))
 )
 
-# Regex that matches a *raw* reference, `#1234`, that is **not** already inside
-# a converted link.  The negative look-behind ensures the # is not preceded by
-# a literal '[' which would indicate an already converted link.
-RAW_REF_RE = re.compile(r"(?<!\[)#(?P<num>\d+)\b")
+# Double-paren form produced by a previous incorrect run:
+#   (([#1234](https://github.com/databricks/cli/pull/1234)))
+DOUBLE_PAREN_LINK_RE = re.compile(
+    r"\(\(\[#(?P<num>\d+)\]\("
+    r"https://github\.com/databricks/cli/pull/\d+"
+    r"\)\)\)"
+)
+
+# Raw reference already wrapped in parens: (#1234)
+PAREN_RAW_REF_RE = re.compile(r"\(#(?P<num>\d+)\)")
+
+# Bare raw reference not already part of a converted link or paren-wrapped ref.
+# Negative look-behinds: '[' means it's inside a converted link; '(' means
+# it will be handled by PAREN_RAW_REF_RE above.
+RAW_REF_RE = re.compile(r"(?<!\[)(?<!\()#(?P<num>\d+)\b")
 
 
 def find_mismatched_links(text):
-    """Return texts of mismatching converted links."""
+    """Return texts of mismatching converted links.
+
+    >>> find_mismatched_links("([#1234](https://github.com/databricks/cli/pull/1234))")
+    []
+    >>> find_mismatched_links("([#1234](https://github.com/databricks/cli/pull/9999))")
+    ['Converted link numbers differ: text #1234 vs URL #9999 — …([#1234](https://github.com/databricks/cli/pull/9999))…']
+    """
     mismatches = []
     for m in CONVERTED_LINK_RE.finditer(text):
         num_text, num_url = m.group("num_text"), m.group("num_url")
@@ -45,13 +59,48 @@ def find_mismatched_links(text):
 
 
 def convert_raw_references(text):
-    """Convert raw `#1234` references to markdown links."""
+    """Convert raw `#1234` references to markdown links.
+
+    Already-converted single-paren links are left unchanged:
+
+    >>> convert_raw_references("([#1234](https://github.com/databricks/cli/pull/1234))")
+    '([#1234](https://github.com/databricks/cli/pull/1234))'
+
+    Double-paren links from a previous incorrect run are collapsed to single-paren:
+
+    >>> convert_raw_references("(([#1234](https://github.com/databricks/cli/pull/1234)))")
+    '([#1234](https://github.com/databricks/cli/pull/1234))'
+
+    A raw reference with surrounding parens becomes a single-paren link (not double):
 
-    def _repl(match):
-        num = match.group("num")
+    >>> convert_raw_references("(#3456)")
+    '([#3456](https://github.com/databricks/cli/pull/3456))'
+
+    A bare raw reference gets wrapped in a single-paren link:
+
+    >>> convert_raw_references("#3456")
+    '([#3456](https://github.com/databricks/cli/pull/3456))'
+
+    Idempotent: running twice produces the same result:
+
+    >>> t = "(#3456) and #7890"
+    >>> convert_raw_references(convert_raw_references(t)) == convert_raw_references(t)
+    True
+    """
+
+    def _make_link(num):
         return f"([#{num}](https://github.com/databricks/cli/pull/{num}))"
 
-    return RAW_REF_RE.sub(_repl, text)
+    # Fix existing double-paren links produced by a previous incorrect run.
+    text = DOUBLE_PAREN_LINK_RE.sub(lambda m: _make_link(m.group("num")), text)
+
+    # Convert (#1234) — parens already present, replace the whole token.
+    text = PAREN_RAW_REF_RE.sub(lambda m: _make_link(m.group("num")), text)
+
+    # Convert bare #1234 — not preceded by [ (converted) or ( (paren-wrapped).
+    text = RAW_REF_RE.sub(lambda m: _make_link(m.group("num")), text)
+
+    return text
 
 
 def process_file(path):