Add true fuzzy-match completions with rapidfuzz

changelog.md

@@ -7,6 +7,7 @@ Features
 * Default to standards-compliant `utf8mb4` character set.
 * Stream input from STDIN to consume less memory, adding `--noninteractive` and `--format=` CLI arguments.
 * Remove suggested quoting on completions for identifiers with uppercase.
+* Add true fuzzy-match completions with rapidfuzz.
 
 
 Bug Fixes

mycli/sqlcompleter.py

@@ -8,6 +8,7 @@
 from prompt_toolkit.completion import CompleteEvent, Completer, Completion
 from prompt_toolkit.completion.base import Document
 from pygments.lexers._mysql_builtins import MYSQL_DATATYPES, MYSQL_FUNCTIONS, MYSQL_KEYWORDS
+import rapidfuzz
 
 from mycli.packages.completion_engine import suggest_type
 from mycli.packages.filepaths import complete_path, parse_path, suggest_path
@@ -996,6 +997,25 @@ def find_matches(
                     completions.append(item)
                     continue
 
+            if len(text) >= 4:
+                rapidfuzz_matches = rapidfuzz.process.extract(
+                    text,
+                    collection,
+                    scorer=rapidfuzz.fuzz.WRatio,
+                    # todo: maybe make our own processor which only does case-folding
+                    # because underscores are valuable info
+                    processor=rapidfuzz.utils.default_process,
+                    limit=20,
+                    score_cutoff=75,
+                )
+                for elt in rapidfuzz_matches:
+                    item, _score, _type = elt
+                    if len(item) < len(text) / 1.5:
+                        continue
+                    if item in completions:
+                        continue
+                    completions.append(item)
+
         else:
             match_end_limit = len(text) if start_only else None
             for item in collection:

pyproject.toml

@@ -21,6 +21,7 @@ dependencies = [
     "pyperclip >= 1.8.1",
     "pycryptodomex",
     "pyfzf >= 0.3.1",
+    "rapidfuzz ~= 3.14.3",
 ]
 
 [build-system]

test/test_smart_completion_public_schema_only.py

@@ -422,6 +422,18 @@ def test_table_names_inter_partial(completer, complete_event):
     result = list(completer.get_completions(Document(text=text, cursor_position=position), complete_event))
     assert result == [
         Completion(text="time_zone_leap_second", start_position=-9),
+        Completion(text='time_zone_name', start_position=-9),
+        Completion(text='time_zone_transition', start_position=-9),
+        Completion(text='time_zone_transition_type', start_position=-9),
+    ]
+
+
+def test_table_names_fuzzy(completer, complete_event):
+    text = "SELECT * FROM tim_leap"
+    position = len("SELECT * FROM tim_leap")
+    result = list(completer.get_completions(Document(text=text, cursor_position=position), complete_event))
+    assert result == [
+        Completion(text="time_zone_leap_second", start_position=-8),
     ]