Merge pull request #2148 from mabel-dev/#2146

#2146

opteryx/__version__.py

@@ -1,4 +1,4 @@
-__build__ = 913
+__build__ = 915
 
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

opteryx/managers/expression/ops.py

@@ -180,71 +180,24 @@ def _inner_filter_operations(arr, operator, value):
         return list_ops.cython_allop_neq(arr[0], value)
 
     if operator == "AnyOpILike":
-        patterns = value[0]
+        from opteryx.utils.sql import regex_match_any
 
-        combined_regex_pattern = r"|".join(sql_like_to_regex(p) for p in patterns if p)
-        combined_regex = re.compile(combined_regex_pattern, re.IGNORECASE)
-
-        out = numpy.zeros(arr.size, dtype=bool)
-        for i, row in enumerate(arr):
-            if row is None:
-                out[i] = None
-                continue
-            if row.size == 0:
-                continue
-            out[i] = any(combined_regex.search(elem) for elem in row)
-
-        return out
+        return regex_match_any(arr, value[0], flags=re.IGNORECASE)
 
     if operator == "AnyOpLike":
-        patterns = value[0]
+        from opteryx.utils.sql import regex_match_any
 
-        combined_regex_pattern = r"|".join(sql_like_to_regex(p) for p in patterns if p)
-        combined_regex = re.compile(combined_regex_pattern)
+        return regex_match_any(arr, value[0])
 
-        out = numpy.zeros(arr.size, dtype=bool)
-        for i, row in enumerate(arr):
-            if row is None:
-                out[i] = None
-                continue
-            if row.size == 0:
-                continue
-            out[i] = any(combined_regex.search(elem) for elem in row)
-
-        return out
     if operator == "AnyOpNotLike":
-        patterns = value[0]
-
-        combined_regex_pattern = r"|".join(sql_like_to_regex(p) for p in patterns if p)
-        combined_regex = re.compile(combined_regex_pattern)
+        from opteryx.utils.sql import regex_match_any
 
-        out = numpy.zeros(arr.size, dtype=bool)
-        for i, row in enumerate(arr):
-            if row is None:
-                out[i] = None
-                continue
-            if row.size == 0:
-                continue
-            out[i] = any(combined_regex.search(elem) for elem in row)
-
-        return numpy.invert(out)
+        return regex_match_any(arr, value[0], invert=True)
 
     if operator == "AnyOpNotILike":
-        patterns = value[0]
-
-        combined_regex_pattern = r"|".join(sql_like_to_regex(p) for p in patterns if p)
-        combined_regex = re.compile(combined_regex_pattern, re.IGNORECASE)
-
-        out = numpy.zeros(arr.size, dtype=bool)
-        for i, row in enumerate(arr):
-            if row is None:
-                out[i] = None
-                continue
-            if row.size == 0:
-                continue
-            out[i] = any(combined_regex.search(elem) for elem in row)
+        from opteryx.utils.sql import regex_match_any
 
-        return numpy.invert(out)
+        return regex_match_any(arr, value[0], flags=re.IGNORECASE, invert=True)
 
     if operator == "AtQuestion":
         import simdjson

opteryx/utils/sql.py

@@ -1,6 +1,8 @@
 import re
 from typing import List
 
+import numpy
+
 ESCAPE_SPECIAL_CHARS = re.compile(r"([.^$*+?{}[\]|()\\])")
 
 
@@ -114,3 +116,84 @@ def split_sql_statements(sql: str) -> List[str]:
         statements.append("".join(buffer).strip())
 
     return [s for s in statements if s != ""]
+
+
+def regex_match_any(
+    arr: numpy.ndarray,
+    patterns: List[str],
+    flags: int = re.NOFLAG,
+    invert: bool = False,
+) -> numpy.ndarray:
+    """
+    Evaluates whether each row in `arr` matches ANY of the given LIKE patterns.
+    Patterns are converted to regexes, combined, and compiled once.
+
+    Parameters:
+        arr: numpy.ndarray
+            1D array of rows. Each element can be:
+                - None
+                - A single string/bytes
+                - A list/tuple/array of strings/bytes
+              (all non-None elements are assumed to be the same structure).
+        patterns: List[str]
+            A list of SQL LIKE patterns. These get combined into a single regex.
+        flags: int, optional
+            Flags to pass to `re.compile()`, e.g. re.IGNORECASE for ILIKE.
+
+    Returns:
+        numpy.ndarray:
+            A 1D object array with True, False, or None,
+            indicating whether each row did (or did not) match the patterns.
+    """
+    # 1) Combine the LIKE patterns into a single compiled regex
+    #    (Empty patterns list => empty string => matches nothing)
+    combined_pattern_str = r"|".join(sql_like_to_regex(p) for p in patterns if p)
+    # If there are no valid patterns, we build a "never match" pattern
+    if not combined_pattern_str:
+        combined_pattern_str = r"(?!x)"  # Negative lookahead to never match
+
+    combined_regex = re.compile(combined_pattern_str, flags=flags)
+
+    # 2) Create the output array (dtype=object so we can store None/bool)
+    out = numpy.empty(arr.size, dtype=object)
+
+    # 3) Determine if the array consists of single strings or lists-of-strings
+    first_non_none = None
+    for x in arr:
+        if x is not None:
+            first_non_none = x
+            break
+
+    # If the entire array is None, just return all None
+    if first_non_none is None:
+        out[:] = None
+        return out
+
+    single_string_mode = isinstance(first_non_none, (str, bytes))
+
+    # 4) Main loop
+    if single_string_mode:
+        # Single-string mode
+        for i, row in enumerate(arr):
+            if row is None:
+                out[i] = None
+            else:
+                # Match or not?
+                is_match = combined_regex.search(row) is not None
+                out[i] = (not is_match) if invert else is_match
+    else:
+        # Lists-of-strings mode
+        for i, row in enumerate(arr):
+            if row is None:
+                out[i] = None
+            else:
+                # row is assumed to be an iterable of strings/bytes
+                if row.size == 0:
+                    # Probably a numpy array with zero length
+                    is_match = False
+                else:
+                    # If anything in the row matches, it's True
+                    is_match = any(combined_regex.search(elem) for elem in row)
+                out[i] = (not is_match) if invert else is_match
+
+    return out

tests/sql_battery/test_shapes_and_errors_battery.py

@@ -1857,29 +1857,60 @@
         ("SELECT name, missions FROM $astronauts WHERE missions ILIKE ANY ('%Apoll%')", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('%Apoll%', 'mission')", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions ILIKE ANY ('%Apoll%', 'mission')", 34, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY '%apoll%'", 357, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY '%apoll%'", 323, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%apoll%')", 357, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%apoll%')", 323, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%')", 323, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%Apoll%')", 323, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%', 'mission')", 323, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%Apoll%', 'mission')", 323, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY '%apoll%'", 334, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY '%apoll%'", 300, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%apoll%')", 334, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%apoll%')", 300, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%')", 300, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%Apoll%')", 300, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%', 'mission')", 300, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT ILIKE ANY ('%Apoll%', 'mission')", 300, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('Apoll%', 'Gemini%', 'Mercury%')", 37, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('Apoll%', 'Gemini%', 'Mercury%')", 320, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('Apoll%', 'Gemini%', 'Mercury%')", 297, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ()", 0, 2, SqlError),
         ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ()", 0, 2, SqlError),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('%Apoll%', null)", 34, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%', null)", 323, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('%Apoll%', null)", 300, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('%aPoll%')", 0, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions ILIKE ANY ('%aPoll%')", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('Apollo 11')", 3, 2, None),
-        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('Apollo 11')", 354, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE missions NOT LIKE ANY ('Apollo 11')", 331, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('Apollo_%')", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('Apo__o%')", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('%Apoll%', 123)", 34, 2, None),
         ("SELECT name, missions FROM $astronauts WHERE missions LIKE ANY ('%pattern1%', '%pattern2%', '%pattern3%', '%pattern4%', '%pattern5%', '%pattern6%', '%pattern7%', '%pattern8%', '%pattern9%', '%pattern10%', '%pattern11%', '%pattern12%', '%pattern13%', '%pattern14%', '%pattern15%', '%pattern16%', '%pattern17%', '%pattern18%', '%pattern19%', '%pattern20%', '%pattern21%', '%pattern22%', '%pattern23%', '%pattern24%', '%pattern25%', '%pattern26%', '%pattern27%', '%pattern28%', '%pattern29%', '%pattern30%', '%pattern31%', '%pattern32%', '%pattern33%', '%pattern34%', '%pattern35%', '%pattern36%', '%pattern37%', '%pattern38%', '%pattern39%', '%pattern40%', '%pattern41%', '%pattern42%', '%pattern43%', '%pattern44%', '%pattern45%', '%pattern46%', '%pattern47%', '%pattern48%', '%pattern49%', '%pattern50%');", 0, 2, None),
 
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY '%armstrong%'", 0, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name ILIKE ANY '%armstrong%'", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%arms%')", 0, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name ILIKE ANY ('%arms%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Armstrong%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name ILIKE ANY ('%Arms%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Armstrong%', 'mission')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name ILIKE ANY ('%Armstrong%', 'mission')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY '%armstrong%'", 357, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT ILIKE ANY '%armstrong%'", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('%armstrong%')", 357, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT ILIKE ANY ('%armstrong%')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('%Armstrong%')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT ILIKE ANY ('%Armstrong%')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('%Armstrong%', 'mission')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT ILIKE ANY ('%Armstrong%', 'mission')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Armstrong%', '%Aldrin%', '%Collins%')", 4, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('%Armstrong%', '%Aldrin%', '%Collins%')", 353, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ()", 0, 2, SqlError),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ()", 0, 2, SqlError),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Armstrong%', null)", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('%Armstrong%', null)", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%aRmstrong%')", 0, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name ILIKE ANY ('%aRmstrong%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('Neil A. Armstrong')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name NOT LIKE ANY ('Neil A. Armstrong')", 356, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%__Armstrong%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Arm__rong%')", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%Armstrong%', 123)", 1, 2, None),
+        ("SELECT name, missions FROM $astronauts WHERE name LIKE ANY ('%pattern1%', '%pattern2%', '%pattern3%', '%pattern4%', '%pattern5%', '%pattern6%', '%pattern7%', '%pattern8%', '%pattern9%', '%pattern10%', '%pattern11%', '%pattern12%', '%pattern13%', '%pattern14%', '%pattern15%', '%pattern16%', '%pattern17%', '%pattern18%', '%pattern19%', '%pattern20%', '%pattern21%', '%pattern22%', '%pattern23%', '%pattern24%', '%pattern25%', '%pattern26%', '%pattern27%', '%pattern28%', '%pattern29%', '%pattern30%', '%pattern31%', '%pattern32%', '%pattern33%', '%pattern34%', '%pattern35%', '%pattern36%', '%pattern37%', '%pattern38%', '%pattern39%', '%pattern40%', '%pattern41%', '%pattern42%', '%pattern43%', '%pattern44%', '%pattern45%', '%pattern46%', '%pattern47%', '%pattern48%', '%pattern49%', '%pattern50%');", 0, 2, None),
+
         # ****************************************************************************************
 
         # These are queries which have been found to return the wrong result or not run correctly

tests/storage/test_cache_memcached.py

@@ -6,7 +6,6 @@
 
 import os
 import sys
-import pytest
 
 os.environ["OPTERYX_DEBUG"] = "1"
 
@@ -62,6 +61,7 @@ def test_memcached_cache():
     assert stats.get("remote_cache_hits", 0) >= stats["blobs_read"], str(stats)
     assert stats.get("cache_misses", 0) == 0, stats
 
+@skip_if(is_arm() or is_windows() or is_mac())
 def test_memcache_stand_alone():
     os.environ["OPTERYX_DEBUG"] = "1"
     from opteryx.managers.cache import MemcachedCache
@@ -110,6 +110,7 @@ def threaded_cache_operations(cache: MemcachedCache, payloads: list):
     for thread in threads:
         thread.join()
 
+@skip_if(is_arm() or is_windows() or is_mac())
 def test_memcache_threaded():
     os.environ["OPTERYX_DEBUG"] = "1"
 
@@ -128,7 +129,7 @@ def test_memcache_threaded():
         if result:
             assert result == load, f"Post-thread check failed: {result} != {load}"
 
-
+@skip_if(is_arm() or is_windows() or is_mac())
 def test_skip_on_error():
     from opteryx.managers.cache import MemcachedCache
     cache = MemcachedCache()