Merge pull request #1574 from dbcli/RW/refactor-is-inside-quotes

Refactor `is_inside_quotes()`

Author: rolandwalker

changelog.md

@@ -10,6 +10,7 @@ Bug Fixes
 Internal
 --------
 * Tune Codex reviews.
+* Refactor `is_inside_quotes()` detection.
 
 
 1.54.0 (2026/02/16)

mycli/packages/completion_engine.py

@@ -1,6 +1,6 @@
 import functools
 import re
-from typing import Any
+from typing import Any, Literal
 
 import sqlparse
 from sqlparse.sql import Comparison, Identifier, Token, Where
@@ -22,7 +22,7 @@ def _enum_value_suggestion(text_before_cursor: str, full_text: str) -> dict[str,
     match = _ENUM_VALUE_RE.search(text_before_cursor)
     if not match:
         return None
-    if _is_inside_quotes(text_before_cursor, match.start("lhs")):
+    if is_inside_quotes(text_before_cursor, match.start("lhs")):
         return None
 
     lhs = match.group("lhs")
@@ -43,25 +43,82 @@ def _is_where_or_having(token: Token | None) -> bool:
     return bool(token and token.value and token.value.lower() in ("where", "having"))
 
 
+def _find_doubled_backticks(text: str) -> list[int]:
+    length = len(text)
+    doubled_backticks: list[int] = []
+    backtick = '`'
+
+    for index in range(0, length):
+        ch = text[index]
+        if ch != backtick:
+            index += 1
+            continue
+        if index + 1 < length and text[index + 1] == backtick:
+            doubled_backticks.append(index)
+            doubled_backticks.append(index + 1)
+            index += 2
+            continue
+        index += 1
+
+    return doubled_backticks
+
+
 @functools.lru_cache(maxsize=128)
-def _is_inside_quotes(text: str, pos: int) -> bool:
+def is_inside_quotes(text: str, pos: int) -> Literal[False, 'single', 'double', 'backtick']:
     in_single = False
     in_double = False
+    in_backticks = False
     escaped = False
-
-    for ch in text[:pos]:
-        if escaped:
+    doubled_backtick_positions = []
+    single_quote = "'"
+    double_quote = '"'
+    backtick = '`'
+    backslash = '\\'
+
+    # scanning the string twice seems to be needed to handle doubled backticks
+    if backtick in text:
+        doubled_backtick_positions = _find_doubled_backticks(text)
+
+    length = len(text)
+    if pos < 0:
+        pos = length + pos
+        pos = max(pos, 0)
+    pos = min(length, pos)
+
+    # optimization
+    up_to_pos = text[:pos]
+    if backtick not in up_to_pos and single_quote not in up_to_pos and double_quote not in up_to_pos:
+        return False
+
+    for index in range(0, pos):
+        ch = text[index]
+        if index in doubled_backtick_positions:
+            index += 1
+            continue
+        if escaped and (in_double or in_single):
             escaped = False
+            index += 1
             continue
-        if ch == "\\":
+        if ch == backslash and (in_double or in_single):
             escaped = True
+            index += 1
             continue
-        if ch == "'" and not in_double:
+        if ch == backtick and not in_double and not in_single:
+            in_backticks = not in_backticks
+        elif ch == single_quote and not in_double and not in_backticks:
             in_single = not in_single
-        elif ch == '"' and not in_single:
+        elif ch == double_quote and not in_single and not in_backticks:
             in_double = not in_double
-
-    return in_single or in_double
+        index += 1
+
+    if in_single:
+        return 'single'
+    elif in_double:
+        return 'double'
+    elif in_backticks:
+        return 'backtick'
+    else:
+        return False
 
 
 def suggest_type(full_text: str, text_before_cursor: str) -> list[dict[str, Any]]:
@@ -197,7 +254,7 @@ def suggest_based_on_last_token(
         # less efficient, but handles all cases
         # in fact, this is quite slow, but not as slow as offering completions!
         # faster would be to peek inside the Pygments lexer run by prompt_toolkit -- how?
-        if _is_inside_quotes(text_before_cursor, -1):
+        if is_inside_quotes(text_before_cursor, -1) in ['single', 'double']:
             return []
 
     if isinstance(token, str):
@@ -383,7 +440,7 @@ def suggest_based_on_last_token(
         # "CREATE DATABASE <newdb> WITH TEMPLATE <db>"
         return [{"type": "database"}]
 
-    elif _is_inside_quotes(text_before_cursor, -1):
+    elif is_inside_quotes(text_before_cursor, -1) in ['single', 'double']:
         return []
 
     elif token_v.endswith(",") or is_operand(token_v) or token_v in ["=", "and", "or"]:

test/test_completion_engine.py

@@ -3,7 +3,11 @@
 import pytest
 
 from mycli.packages import special
-from mycli.packages.completion_engine import suggest_type
+from mycli.packages.completion_engine import (
+    _find_doubled_backticks,
+    is_inside_quotes,
+    suggest_type,
+)
 
 
 def sorted_dicts(dicts):
@@ -628,3 +632,81 @@ def test_quoted_where():
     text = "'where i=';"
     suggestions = suggest_type(text, text)
     assert suggestions == [{"type": "keyword"}]
+
+
+def test_find_doubled_backticks_none():
+    text = 'select `ab`'
+    assert _find_doubled_backticks(text) == []
+
+
+def test_find_doubled_backticks_some():
+    text = 'select `a``b`'
+    assert _find_doubled_backticks(text) == [9, 10]
+
+
+def test_inside_quotes_01():
+    text = "select '"
+    assert is_inside_quotes(text, len(text)) == 'single'
+
+
+def test_inside_quotes_02():
+    text = "select '\\'"
+    assert is_inside_quotes(text, len(text)) == 'single'
+
+
+def test_inside_quotes_03():
+    text = "select '`"
+    assert is_inside_quotes(text, len(text)) == 'single'
+
+
+def test_inside_quotes_04():
+    text = 'select "'
+    assert is_inside_quotes(text, len(text)) == 'double'
+
+
+def test_inside_quotes_05():
+    text = 'select "\\"\''
+    assert is_inside_quotes(text, len(text)) == 'double'
+
+
+def test_inside_quotes_06():
+    text = 'select ""'
+    assert is_inside_quotes(text, len(text)) is False
+
+
+@pytest.mark.parametrize(
+    ["text", "position", "expected"],
+    [
+        ("select `'",      len("select `'"),  'backtick'),
+        ("select `' ",     len("select `' "), 'backtick'),
+        ("select `'",      -1,  'backtick'),
+        ("select `'",      -2,  False),
+        ('select `ab` ',   -1,  False),
+        ('select `ab` ',   -2,  'backtick'),
+        ('select `a``b` ', -1,  False),
+        ('select `a``b` ', -2,  'backtick'),
+        ('select `a``b` ', -3,  'backtick'),
+        ('select `a``b` ', -4,  'backtick'),
+        ('select `a``b` ', -5,  'backtick'),
+        ('select `a``b` ', -6,  'backtick'),
+        ('select `a``b` ', -7,  False),
+    ]
+)  # fmt: skip
+def test_inside_quotes_backtick_01(text, position, expected):
+    assert is_inside_quotes(text, position) == expected
+
+
+def test_inside_quotes_backtick_02():
+    """Empty backtick pairs are treated as a doubled (escaped) backtick.
+    This is okay because it is invalid SQL, and we don't have to complete on it.
+    """
+    text = 'select ``'
+    assert is_inside_quotes(text, -1) is False
+
+
+def test_inside_quotes_backtick_03():
+    """Empty backtick pairs are treated as a doubled (escaped) backtick.
+    This is okay because it is invalid SQL, and we don't have to complete on it.
+    """
+    text = 'select ``'
+    assert is_inside_quotes(text, -2) is False