feat: improve case conversion

.github/workflows/ci.yml

@@ -2,6 +2,7 @@ name: CI
 
 on:
   pull_request:
+    types: [opened, synchronize, reopened, labeled]
   push:
     branches: [main]
 

db2sql/domain/policy/identifier.py

@@ -4,12 +4,22 @@
 
 import re
 
-_CAMEL_RE = re.compile(r"(?<!^)(?=[A-Z])")
+# Split before a capitalized word so acronyms stay glued: "HTTPServer" → "HTTP_Server".
+_CAMEL_BOUNDARY_RE = re.compile(r"(.)([A-Z][a-z]+)")
+# Split between lowercase/digit and uppercase to catch trailing acronyms: "userID" → "user_ID".
+_LOWER_UPPER_RE = re.compile(r"([a-z0-9])([A-Z])")
 
 
 def to_snake_case(name: str) -> str:
-    """Convert CamelCase to snake_case (idempotent on already-snake_case input)."""
-    return _CAMEL_RE.sub("_", name).lower()
+    """Convert CamelCase / PascalCase to snake_case.
+
+    Keeps acronym runs intact (``HTTPServer`` → ``http_server``,
+    ``UserID`` → ``user_id``) and leaves all-caps identifiers as a single
+    word (``MYTABLE`` → ``mytable``). Idempotent on snake_case input.
+    """
+    name = _CAMEL_BOUNDARY_RE.sub(r"\1_\2", name)
+    name = _LOWER_UPPER_RE.sub(r"\1_\2", name)
+    return name.lower()
 
 
 def normalize_identifier(name: str, preserve_case: bool) -> str:

docs/cli.rst

@@ -193,7 +193,19 @@ Output
    column names) are kept exactly as they appear in the source database.
 
    When disabled (the default), identifiers are converted to ``snake_case``
-   so they work without quoting in PostgreSQL.
+   so they work without quoting in PostgreSQL.  The conversion keeps
+   acronym runs glued together and collapses all-caps identifiers to a
+   single word:
+
+   ===================== =====================
+   Source identifier     Converted
+   ===================== =====================
+   ``CamelCase``         ``camel_case``
+   ``HTTPServer``        ``http_server``
+   ``UserID``            ``user_id``
+   ``MYTABLE``           ``mytable``
+   ``customer_ID``       ``customer_id``
+   ===================== =====================
 
 .. option:: --transaction / --no-transaction
 

docs/configuration.rst

@@ -269,7 +269,11 @@ dump
    * - ``preserve_case``
      - ``false``
      - Preserve identifier case as-is.  When ``false``, identifiers are
-       converted to ``snake_case``.
+       converted to ``snake_case`` — acronym runs stay glued
+       (``HTTPServer`` → ``http_server``, ``UserID`` → ``user_id``) and
+       all-caps names collapse to a single word
+       (``MYTABLE`` → ``mytable``).  See :option:`--preserve-case` for
+       the full table of examples.
    * - ``limit_records``
      - ``-1``
      - Maximum rows per table.  ``-1`` means no limit.
@@ -642,7 +646,8 @@ Oracle connections are configured almost entirely through
      preserve_case: false
      default_data_format: copy
      mapping_schemas:
-       HR: hr           # rewrite Oracle's upper-case owner to snake_case
+       HR: human_resources  # optional rename; snake_case normalization
+                            # alone would already produce "hr"
 
 Oracle → MSSQL
 ~~~~~~~~~~~~~~

tests/unit/domain/policy/test_identifier.py

@@ -11,17 +11,27 @@
     "name, expected",
     [
         ("CamelCase", "camel_case"),
-        ("HTTPServer", "h_t_t_p_server"),
+        ("HTTPServer", "http_server"),
         ("already_snake", "already_snake"),
         ("lower", "lower"),
         ("Single", "single"),
-        ("UserID", "user_i_d"),
+        ("UserID", "user_id"),
+        ("XMLParser", "xml_parser"),
+        ("getHTTPResponseCode", "get_http_response_code"),
+        ("MYTABLE", "mytable"),
+        ("customer_ID", "customer_id"),
+        ("", ""),
     ],
 )
 def test_to_snake_case(name: str, expected: str) -> None:
     assert to_snake_case(name) == expected
 
 
+def test_to_snake_case_is_idempotent() -> None:
+    once = to_snake_case("HTTPServerName")
+    assert to_snake_case(once) == once
+
+
 def test_normalize_identifier_preserve_case_returns_input() -> None:
     assert normalize_identifier("UserName", preserve_case=True) == "UserName"