dcs-sdk 1.4.9__py3-none-any.whl → 1.5.1__py3-none-any.whl

data_diff/abcs/database_types.py

@@ -14,15 +14,35 @@
 
 import decimal
 from abc import ABC, abstractmethod
-from datetime import datetime
+from datetime import date, datetime
 from typing import Collection, List, Optional, Tuple, Type, TypeVar, Union
 
 import attrs
 
-from data_diff.utils import ArithAlphanumeric, ArithUnicodeString, ArithUUID, Unknown
+from data_diff.utils import (
+    ArithAlphanumeric,
+    ArithDate,
+    ArithDateTime,
+    ArithTimestamp,
+    ArithTimestampTZ,
+    ArithUnicodeString,
+    ArithUUID,
+    Unknown,
+)
 
 DbPath = Tuple[str, ...]
-DbKey = Union[int, str, bytes, ArithUUID, ArithAlphanumeric, ArithUnicodeString]
+DbKey = Union[
+    int,
+    str,
+    bytes,
+    ArithUUID,
+    ArithAlphanumeric,
+    ArithUnicodeString,
+    ArithDateTime,
+    ArithDate,
+    ArithTimestamp,
+    ArithTimestampTZ,
+]
 DbTime = datetime
 
 N = TypeVar("N")
@@ -150,23 +170,66 @@ class TemporalType(PrecisionType):
 
 
 @attrs.define(frozen=True)
-class Timestamp(TemporalType):
-    pass
+class IKey(ABC):
+    "Interface for ColType, for using a column as a key in table."
+
+    @property
+    @abstractmethod
+    def python_type(self) -> type:
+        "Return the equivalent Python type of the key"
+
+    def make_value(self, value):
+        if isinstance(value, self.python_type):
+            return value
+        return self.python_type(value)
 
 
 @attrs.define(frozen=True)
-class TimestampTZ(TemporalType):
-    pass
+class Timestamp(TemporalType, IKey):
+    @property
+    def python_type(self) -> type:
+        return ArithTimestamp
+
+    def make_value(self, value):
+        if isinstance(value, ArithTimestamp):
+            return value
+        return ArithTimestamp(value)
 
 
 @attrs.define(frozen=True)
-class Datetime(TemporalType):
-    pass
+class TimestampTZ(TemporalType, IKey):
+    @property
+    def python_type(self) -> type:
+        return ArithTimestampTZ
+
+    def make_value(self, value):
+        if isinstance(value, ArithTimestampTZ):
+            return value
+        return ArithTimestampTZ(value)
 
 
 @attrs.define(frozen=True)
-class Date(TemporalType):
-    pass
+class Datetime(TemporalType, IKey):
+    @property
+    def python_type(self) -> type:
+        return ArithDateTime
+
+    def make_value(self, value):
+        if isinstance(value, ArithDateTime):
+            return value
+        return ArithDateTime(value)
+
+
+@attrs.define(frozen=True)
+class Date(TemporalType, IKey):
+    @property
+    def python_type(self) -> type:
+        return ArithDate
+
+    def make_value(self, value):
+        if isinstance(value, ArithDate):
+            return value
+        return ArithDate(value)
 
 
 @attrs.define(frozen=True)
@@ -190,21 +253,6 @@ class Float(FractionalType):
     python_type = float
 
 
-@attrs.define(frozen=True)
-class IKey(ABC):
-    "Interface for ColType, for using a column as a key in table."
-
-    @property
-    @abstractmethod
-    def python_type(self) -> type:
-        "Return the equivalent Python type of the key"
-
-    def make_value(self, value):
-        if isinstance(value, self.python_type):
-            return value
-        return self.python_type(value)
-
-
 @attrs.define(frozen=True)
 class Decimal(FractionalType, IKey):  # Snowflake may use Decimal as a key
     @property
@@ -284,6 +332,11 @@ class String_VaryingAlphanum(String_Alphanum):
     pass
 
 
+@attrs.define(frozen=True)
+class String_VaryingUnicode(ColType_Unicode, StringType):
+    pass
+
+
 @attrs.define(frozen=True)
 class String_FixedAlphanum(String_Alphanum):
     length: int

data_diff/databases/base.py

@@ -55,6 +55,7 @@ from data_diff.abcs.compiler import AbstractCompiler, Compilable
 from data_diff.abcs.database_types import (
     JSON,
     ArithAlphanumeric,
+    ArithUnicodeString,
     Array,
     Boolean,
     ColType,
@@ -69,6 +70,7 @@ from data_diff.abcs.database_types import (
     String_Alphanum,
     String_UUID,
     String_VaryingAlphanum,
+    String_VaryingUnicode,
     Struct,
     TemporalType,
     Text,
@@ -115,7 +117,11 @@ from data_diff.queries.extras import (
 )
 from data_diff.schema import RawColumnInfo
 from data_diff.utils import (
+    ArithDate,
+    ArithDateTime,
     ArithString,
+    ArithTimestamp,
+    ArithTimestampTZ,
     ArithUUID,
     SybaseDriverTypes,
     is_uuid,
@@ -311,6 +317,12 @@ class BaseDialect(abc.ABC):
                 return f"CAST('{elem.uuid}' AS UUID)"
             s = f"'{elem.uuid}'"
             return s.upper() if elem.uppercase else s.lower() if elem.lowercase else s
+        elif isinstance(elem, (ArithDateTime, ArithTimestamp, ArithTimestampTZ)):
+            return self.timestamp_value(elem._dt)
+        elif isinstance(elem, ArithDate):
+            from datetime import time
+
+            return self.timestamp_value(datetime.combine(elem._date, time.min))
         elif isinstance(elem, ArithString):
             return f"'{elem}'"
         assert False, elem
@@ -756,6 +768,16 @@ class BaseDialect(abc.ABC):
             return v.code
         elif isinstance(v, ArithAlphanumeric):
             return f"'{v._str}'"
+        elif isinstance(v, ArithUnicodeString):
+            return f"'{v._str}'"
+        elif isinstance(v, ArithDate):
+            return f"'{str(v)}'"
+        elif isinstance(v, ArithTimestamp):
+            return f"'{str(v)}'"
+        elif isinstance(v, ArithTimestampTZ):
+            return f"'{str(v)}'"
+        elif isinstance(v, ArithDateTime):
+            return self.timestamp_value(v._dt)
         return repr(v)
 
     def constant_values(self, rows) -> str:
@@ -1239,9 +1261,16 @@ class Database(abc.ABC):
                         logger.debug(
                             f"Mixed Alphanum/Non-Alphanum values detected in column {'.'.join(table_path)}.{col_name}. It cannot be used as a key."
                         )
+                        # Fallback to Unicode string type
+                        assert col_name in col_dict
+                        col_dict[col_name] = String_VaryingUnicode(collation=col_dict[col_name].collation)
                     else:
                         assert col_name in col_dict
                         col_dict[col_name] = String_VaryingAlphanum(collation=col_dict[col_name].collation)
+                else:
+                    # All samples failed alphanum test, fallback to Unicode string
+                    assert col_name in col_dict
+                    col_dict[col_name] = String_VaryingUnicode(collation=col_dict[col_name].collation)
 
         return col_dict
 

data_diff/databases/sybase.py

@@ -22,10 +22,12 @@ from loguru import logger
 from data_diff.abcs.database_types import (
     JSON,
     Boolean,
+    ColType,
     ColType_UUID,
     Date,
     Datetime,
     DbPath,
+    DbTime,
     Decimal,
     Float,
     FractionalType,
@@ -49,6 +51,7 @@ from data_diff.databases.base import (
     ThreadedDatabase,
     import_helper,
 )
+from data_diff.schema import RawColumnInfo
 
 
 @import_helper("sybase")
@@ -196,9 +199,25 @@ class Dialect(BaseDialect):
         return f"VALUES {values}"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:
+        varchar_type = (
+            "VARCHAR"
+            if (self.sybase_driver_type.is_iq or self.query_config_for_free_tds["freetds_query_chosen"])
+            else "NVARCHAR"
+        )
+
+        # Handle Date type - return YYYY-MM-DD format
+        if isinstance(coltype, Date):
+            return (
+                f"CASE WHEN {value} IS NULL THEN NULL "
+                f"ELSE "
+                f"CAST(DATEPART(YEAR, {value}) AS CHAR(4)) + '-' + "
+                f"RIGHT('0' + CAST(DATEPART(MONTH, {value}) AS VARCHAR(2)), 2) + '-' + "
+                f"RIGHT('0' + CAST(DATEPART(DAY, {value}) AS VARCHAR(2)), 2) "
+                f"END"
+            )
         if isinstance(coltype, Datetime):
             if coltype.precision == 4:
-                return f"CAST({value} AS VARCHAR(100))"
+                return f"CAST({value} AS {varchar_type}(100))"
             if coltype.precision > 0:
                 return (
                     f"CASE WHEN {value} IS NULL THEN NULL "
@@ -227,6 +246,58 @@ class Dialect(BaseDialect):
             return f"CAST({value} AS VARCHAR(100))"
         return f"CAST({value} AS NVARCHAR(100))"
 
+    def timestamp_value(self, t: DbTime) -> str:
+        """Provide SQL for the given timestamp value - match normalize_timestamp precision"""
+        # Use consistent formatting that matches what normalize_timestamp produces
+        # This ensures exact equality comparisons work correctly
+        formatted = t.strftime("%Y-%m-%d %H:%M:%S")
+        if t.microsecond > 0:
+            # Always use 3-digit milliseconds to match normalize_timestamp output
+            # which uses DATEPART(MILLISECOND, value) giving 3 digits
+            milliseconds = t.microsecond // 1000
+            formatted += f".{milliseconds:03d}"
+        return f"'{formatted}'"
+
+    def timestamp_equality_condition(self, column: str, timestamp_value: str) -> str:
+        """Generate a timestamp equality condition that handles precision mismatches"""
+        # For Sybase, we need to handle the case where stored values have microsecond precision
+        # but our query values only have millisecond precision
+
+        # Extract the timestamp without quotes
+        clean_value = timestamp_value.strip("'")
+
+        # If the value has fractional seconds, create a range query
+        if "." in clean_value:
+            # For a value like '2020-01-01 00:02:33.951'
+            # We want to match anything from .951000 to .951999 microseconds
+            base_value = clean_value
+            next_ms_value = self._increment_millisecond(clean_value)
+
+            return f"({column} >= '{base_value}' AND {column} < '{next_ms_value}')"
+        else:
+            # No fractional seconds, use exact match
+            return f"{column} = '{clean_value}'"
+
+    def _increment_millisecond(self, timestamp_str: str) -> str:
+        """Increment the millisecond part of a timestamp string"""
+        from datetime import datetime, timedelta
+
+        try:
+            # Parse the timestamp
+            if "." in timestamp_str:
+                dt = datetime.strptime(timestamp_str, "%Y-%m-%d %H:%M:%S.%f")
+            else:
+                dt = datetime.strptime(timestamp_str, "%Y-%m-%d %H:%M:%S")
+
+            # Add 1 millisecond
+            dt_incremented = dt + timedelta(milliseconds=1)
+
+            # Format back to string with millisecond precision
+            return dt_incremented.strftime("%Y-%m-%d %H:%M:%S.%f")[:-3]
+        except ValueError:
+            # Fallback to original value if parsing fails
+            return timestamp_str
+
     def normalize_number(self, value: str, coltype: FractionalType) -> str:
         return self.to_string(f"CAST({value} AS DECIMAL(38, {coltype.precision}))")
 
@@ -339,6 +410,20 @@ class Dialect(BaseDialect):
             return f"CONVERT(VARCHAR({ch_len}), {value})"
         return f"CONVERT(NVARCHAR({ch_len}), {value})"
 
+    def parse_type(self, table_path: DbPath, info: RawColumnInfo) -> ColType:
+        """Override base parse_type to handle datetime columns that should be treated as dates"""
+
+        # Check if this is a datetime column that should be treated as a date
+        if info.data_type == "datetime":
+            # Sybase IQ stores DATE columns as datetime with precision=4
+            # and DATETIME columns as datetime with precision=8
+            if info.datetime_precision == 4:
+                return Date(
+                    precision=info.datetime_precision,
+                    rounds=self.ROUNDS_ON_PREC_LOSS,
+                )
+        return super().parse_type(table_path, info)
+
     def parse_table_name(self, name: str) -> DbPath:
         "Parse the given table name into a DbPath"
         self.TABLE_NAMES.append(name.split(".")[-1])
@@ -389,7 +474,7 @@ class Sybase(ThreadedDatabase):
         username = self._args.get("user", None)
         password = self._args.get("password", None)
         driver = self._args.get("driver", None)
-        max_query_timeout = 60 * 60  # 3600 seconds
+        max_query_timeout = 60 * 5  # 300 seconds
 
         if self.dialect.sybase_driver_type.is_freetds:
             conn_dict = {
@@ -583,7 +668,13 @@ class Sybase(ThreadedDatabase):
                     self.dialect.query_config_for_free_tds["ase_query_chosen"] = True
                     return ase_query
                 else:
+                    max_temp_space_usage_query = "SET TEMPORARY OPTION MAX_TEMP_SPACE_PER_CONNECTION = 5120"
+                    if self._query_cursor(self._conn.cursor(), max_temp_space_usage_query, test_query=True):
+                        logger.info("Max temporary space usage set successfully.")
+                    else:
+                        logger.warning("Failed to set max temporary space usage, continuing with default settings.")
                     logger.info("Sybase IQ Detected")
+
                     self.dialect.query_config_for_free_tds["freetds_query_chosen"] = True
                     return iq_query
             except Exception as e:

data_diff/hashdiff_tables.py

@@ -446,6 +446,10 @@ class HashDiffer(TableDiffer):
         segment_index=None,
         segment_count=None,
     ):
+        # Check if level exceeds maximum allowed recursion depth
+        if level > 15:
+            raise RecursionError(f"Maximum recursion level exceeded: {level} > 15")
+
         # Initialize diff tracker if not already done
         self._initialize_diff_tracker(table1, table2)
 
@@ -547,6 +551,10 @@ class HashDiffer(TableDiffer):
         level=0,
         max_rows=None,
     ):
+        # Check if level exceeds maximum allowed recursion depth
+        if level > 15:
+            raise RecursionError(f"Maximum recursion level exceeded: {level} > 15")
+
         assert table1.is_bounded and table2.is_bounded
 
         # Initialize diff tracker if not already done

data_diff/table_segment.py

@@ -44,7 +44,11 @@ from data_diff.queries.extras import (
 )
 from data_diff.schema import RawColumnInfo, Schema, create_schema
 from data_diff.utils import (
+    ArithDate,
+    ArithDateTime,
     ArithString,
+    ArithTimestamp,
+    ArithTimestampTZ,
     ArithUnicodeString,
     Vector,
     safezip,
@@ -62,10 +66,14 @@ def split_key_space(min_key: DbKey, max_key: DbKey, count: int) -> List[DbKey]:
     if max_key - min_key <= count:
         count = 1
 
-    if isinstance(min_key, ArithString) or isinstance(min_key, ArithUnicodeString):
+    # Handle arithmetic string types (including temporal types)
+    if isinstance(
+        min_key, (ArithString, ArithUnicodeString, ArithDateTime, ArithDate, ArithTimestamp, ArithTimestampTZ)
+    ):
         assert type(min_key) is type(max_key)
         checkpoints = min_key.range(max_key, count)
     else:
+        # Handle numeric types
         if isinstance(min_key, Decimal):
             min_key = float(min_key)
         if isinstance(max_key, Decimal):
@@ -351,7 +359,15 @@ class TableSegment:
                         continue
                     mk_v = schema.make_value(val)
                     constant_val = self.database.dialect._constant_value(mk_v)
-                    where_expr = f"{quoted} = {constant_val}"
+
+                    # Special handling for Sybase timestamp equality to handle precision mismatches
+                    if hasattr(self.database.dialect, "timestamp_equality_condition") and hasattr(
+                        mk_v, "_dt"
+                    ):  # Check if it's a datetime-like object
+                        where_expr = self.database.dialect.timestamp_equality_condition(quoted, constant_val)
+                    else:
+                        where_expr = f"{quoted} = {constant_val}"
+
                     and_exprs.append(Code(where_expr))
                 if and_exprs:
                     key_exprs.append(and_(*and_exprs))

data_diff/utils.py

@@ -21,7 +21,7 @@ import string
 import threading
 from abc import abstractmethod
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import date, datetime, time
 from typing import (
     Any,
     Dict,
@@ -179,11 +179,230 @@ def _any_to_uuid(v: Union[str, int, UUID, "ArithUUID"]) -> UUID:
         raise ValueError(f"Cannot convert a value to UUID: {v!r}")
 
 
+def _any_to_datetime(v: Union[str, datetime, date, "ArithDateTime"]) -> datetime:
+    """Convert various types to datetime object."""
+    if isinstance(v, ArithDateTime):
+        return v._dt
+    elif isinstance(v, datetime):
+        return v
+    elif isinstance(v, date):
+        return datetime.combine(v, time.min)
+    elif isinstance(v, str):
+        # Try specific formats first to preserve original precision
+        try:
+            # Handle format: YYYY-MM-DD HH:MM:SS.mmm (3-digit milliseconds)
+            return datetime.strptime(v, "%Y-%m-%d %H:%M:%S.%f")
+        except ValueError:
+            try:
+                # Handle format: YYYY-MM-DD HH:MM:SS
+                return datetime.strptime(v, "%Y-%m-%d %H:%M:%S")
+            except ValueError:
+                try:
+                    # Handle format: YYYY-MM-DD
+                    return datetime.strptime(v, "%Y-%m-%d")
+                except ValueError:
+                    # Last resort: try ISO format parsing
+                    try:
+                        return datetime.fromisoformat(v.replace("Z", "+00:00"))
+                    except ValueError:
+                        raise ValueError(f"Cannot parse datetime string: {v!r}")
+    else:
+        raise ValueError(f"Cannot convert value to datetime: {v!r}")
+
+
+def _any_to_date(v: Union[str, datetime, date, "ArithDate"]) -> date:
+    """Convert various types to date object."""
+    if isinstance(v, ArithDate):
+        return v._date
+    elif isinstance(v, date):
+        return v
+    elif isinstance(v, datetime):
+        return v.date()
+    elif isinstance(v, str):
+        try:
+            return datetime.fromisoformat(v.replace("Z", "+00:00")).date()
+        except ValueError:
+            try:
+                return datetime.strptime(v, "%Y-%m-%d").date()
+            except ValueError:
+                raise ValueError(f"Cannot parse date string: {v!r}")
+    else:
+        raise ValueError(f"Cannot convert value to date: {v!r}")
+
+
+@attrs.define(frozen=True, eq=False, order=False)
+class ArithDateTime(ArithString):
+    """A datetime that supports basic arithmetic and range operations for database diffing."""
+
+    _dt: datetime = attrs.field(converter=_any_to_datetime)
+
+    def range(self, other: "ArithDateTime", count: int) -> List[Self]:
+        """Generate evenly spaced datetime checkpoints between self and other."""
+        assert isinstance(other, ArithDateTime)
+
+        start_ts = self._dt.timestamp()
+        end_ts = other._dt.timestamp()
+
+        checkpoints = split_space(start_ts, end_ts, count)
+        return [self.new(datetime.fromtimestamp(ts)) for ts in checkpoints]
+
+    def __int__(self) -> int:
+        """Convert to timestamp for arithmetic operations."""
+        return int(self._dt.timestamp())
+
+    def __add__(self, other: Union[int, float]) -> Self:
+        """Add seconds to the datetime."""
+        if isinstance(other, (int, float)):
+            new_ts = self._dt.timestamp() + other
+            return self.new(datetime.fromtimestamp(new_ts))
+        return NotImplemented
+
+    def __sub__(self, other: Union["ArithDateTime", int, float]):
+        """Subtract seconds or another datetime."""
+        if isinstance(other, (int, float)):
+            new_ts = self._dt.timestamp() - other
+            return self.new(datetime.fromtimestamp(new_ts))
+        elif isinstance(other, ArithDateTime):
+            return self._dt.timestamp() - other._dt.timestamp()
+        return NotImplemented
+
+    def __eq__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt == other._dt
+        return NotImplemented
+
+    def __ne__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt != other._dt
+        return NotImplemented
+
+    def __gt__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt > other._dt
+        return NotImplemented
+
+    def __lt__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt < other._dt
+        return NotImplemented
+
+    def __ge__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt >= other._dt
+        return NotImplemented
+
+    def __le__(self, other: object) -> bool:
+        if isinstance(other, ArithDateTime):
+            return self._dt <= other._dt
+        return NotImplemented
+
+    def __str__(self) -> str:
+        """Return ISO format string."""
+        return self._dt.isoformat()
+
+    def __repr__(self) -> str:
+        return f"ArithDateTime({self._dt!r})"
+
+
+@attrs.define(frozen=True, eq=False, order=False)
+class ArithDate(ArithString):
+    """A date that supports basic arithmetic and range operations for database diffing."""
+
+    _date: date = attrs.field(converter=_any_to_date)
+
+    def range(self, other: "ArithDate", count: int) -> List[Self]:
+        """Generate evenly spaced date checkpoints between self and other."""
+        assert isinstance(other, ArithDate)
+
+        start_ordinal = self._date.toordinal()
+        end_ordinal = other._date.toordinal()
+
+        checkpoints = split_space(start_ordinal, end_ordinal, count)
+        return [self.new(date.fromordinal(int(ordinal))) for ordinal in checkpoints]
+
+    def __int__(self) -> int:
+        """Convert to ordinal for arithmetic operations."""
+        return self._date.toordinal()
+
+    def __add__(self, other: int) -> Self:
+        """Add days to the date."""
+        if isinstance(other, int):
+            new_ordinal = self._date.toordinal() + other
+            return self.new(date.fromordinal(new_ordinal))
+        return NotImplemented
+
+    def __sub__(self, other: Union["ArithDate", int]):
+        """Subtract days or another date."""
+        if isinstance(other, int):
+            new_ordinal = self._date.toordinal() - other
+            return self.new(date.fromordinal(new_ordinal))
+        elif isinstance(other, ArithDate):
+            return self._date.toordinal() - other._date.toordinal()
+        return NotImplemented
+
+    def __eq__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date == other._date
+        return NotImplemented
+
+    def __ne__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date != other._date
+        return NotImplemented
+
+    def __gt__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date > other._date
+        return NotImplemented
+
+    def __lt__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date < other._date
+        return NotImplemented
+
+    def __ge__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date >= other._date
+        return NotImplemented
+
+    def __le__(self, other: object) -> bool:
+        if isinstance(other, ArithDate):
+            return self._date <= other._date
+        return NotImplemented
+
+    def __str__(self) -> str:
+        """Return ISO format date string."""
+        return self._date.isoformat()
+
+    def __repr__(self) -> str:
+        return f"ArithDate({self._date!r})"
+
+
+@attrs.define(frozen=True, eq=False, order=False)
+class ArithTimestamp(ArithDateTime):
+    """A timestamp that inherits from ArithDateTime but with explicit timestamp semantics."""
+
+    def __repr__(self) -> str:
+        return f"ArithTimestamp({self._dt!r})"
+
+
+@attrs.define(frozen=True, eq=False, order=False)
+class ArithTimestampTZ(ArithDateTime):
+    """A timezone-aware timestamp that extends ArithDateTime."""
+
+    def __repr__(self) -> str:
+        return f"ArithTimestampTZ({self._dt!r})"
+
+    def __str__(self) -> str:
+        """Return ISO format string with timezone info."""
+        return self._dt.isoformat()
+
+
 @attrs.define(frozen=True, eq=False, order=False)
 class ArithUnicodeString(ArithString):
     """A Unicode string for arbitrary text keys, supporting lexicographical ordering and checkpoint generation across databases."""
 
-    string: str = attrs.field(converter=str)
+    _str: str = attrs.field(converter=str)
 
     @staticmethod
     def split_space(start: int, end: int, count: int) -> List[int]:
@@ -197,10 +416,10 @@ class ArithUnicodeString(ArithString):
 
     def _str_to_int(self) -> int:
         """Convert string to an integer for interpolation, handling empty strings and Unicode."""
-        if not self.string:
+        if not self._str:
             return 0  # Handle empty string
         result = 0
-        for char in self.string:
+        for char in self._str:
             result = result * 256 + ord(char)
         return result
 
@@ -220,9 +439,9 @@ class ArithUnicodeString(ArithString):
         assert isinstance(other, ArithUnicodeString), "Other must be an ArithUnicodeString"
 
         # Handle edge case: same or empty strings
-        if self.string == other.string or count <= 0:
+        if self._str == other._str or count <= 0:
             return []
-        if not self.string or not other.string:
+        if not self._str or not other._str:
             return [self.new("a") for _ in range(count)] if count > 0 else []
 
         # Ensure min_key < max_key
@@ -233,17 +452,25 @@ class ArithUnicodeString(ArithString):
         start_int = min_key._str_to_int()
         end_int = max_key._str_to_int()
 
+        # If the range is too small, return empty list
+        if end_int - start_int <= count:
+            return []
+
         # Generate checkpoints
         checkpoints_int = self.split_space(start_int, end_int, count)
 
         # Convert back to strings and create instances
-        checkpoints = [self.new(self._int_to_str(i)) for i in checkpoints_int]
-
-        # Filter to ensure min_key < x < max_key
-        filtered_checkpoints = [cp for cp in checkpoints if min_key < cp < max_key]
-
-        # Fallback to ensure non-empty list for close keys
-        return filtered_checkpoints or checkpoints[:count] or [self.new(self._int_to_str(start_int + 1))]
+        checkpoints = []
+        for i in checkpoints_int:
+            # Ensure checkpoint is valid and within bounds
+            if start_int < i < end_int:
+                checkpoint_str = self._int_to_str(i)
+                checkpoint = self.new(checkpoint_str)
+                # Double-check the string comparison bounds
+                if min_key < checkpoint < max_key:
+                    checkpoints.append(checkpoint)
+
+        return checkpoints
 
     def __int__(self) -> int:
         """Convert to integer representation for arithmetic."""
@@ -268,46 +495,46 @@ class ArithUnicodeString(ArithString):
     def __eq__(self, other: object) -> bool:
         """Check equality with another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string == other.string
+            return self._str == other._str
         return NotImplemented
 
     def __ne__(self, other: object) -> bool:
         """Check inequality with another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string != other.string
+            return self._str != other._str
         return NotImplemented
 
     def __gt__(self, other: object) -> bool:
         """Check if greater than another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string > other.string
+            return self._str > other._str
         return NotImplemented
 
     def __lt__(self, other: object) -> bool:
         """Check if less than another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string < other.string
+            return self._str < other._str
         return NotImplemented
 
     def __ge__(self, other: object) -> bool:
         """Check if greater than or equal to another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string >= other.string
+            return self._str >= other._str
         return NotImplemented
 
     def __le__(self, other: object) -> bool:
         """Check if less than or equal to another ArithUnicodeString."""
         if isinstance(other, ArithUnicodeString):
-            return self.string <= other.string
+            return self._str <= other._str
         return NotImplemented
 
     def __str__(self) -> str:
         """Return the string representation, escaped for SQL."""
-        return self.string.replace("'", "''")
+        return self._str.replace("'", "''")
 
     def __repr__(self) -> str:
         """Return a detailed representation."""
-        return f"ArithUnicodeString(string={self.string!r})"
+        return f"ArithUnicodeString(string={self._str!r})"
 
 
 @attrs.define(frozen=True, eq=False, order=False)

dcs_sdk/__version__.py

@@ -12,4 +12,4 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 
-__version__ = "1.4.9"
+__version__ = "1.5.1"

dcs_sdk/sdk/data_diff/data_differ.py

@@ -628,9 +628,9 @@ class DBTableDiffer:
                         key = []
                         for val in tup:
                             if isinstance(val, str) and val.isdigit():
-                                key.append(int(val))
+                                key.append((0, int(val)))
                             else:
-                                key.append(val)
+                                key.append((1, str(val)))
                         return tuple(key)
 
                     return sorted(keys, key=sort_key)

{dcs_sdk-1.4.9.dist-info → dcs_sdk-1.5.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dcs-sdk
-Version: 1.4.9
+Version: 1.5.1
 Summary: SDK for DataChecks
 Author: Waterdip Labs
 Author-email: hello@waterdip.ai
@@ -60,7 +60,7 @@ Requires-Dist: vertica-python (>=1.4.0) ; extra == "vertica" or extra == "all-db
 Description-Content-Type: text/markdown
 
 <h1 align="center">
-  DCS SDK v1.4.9
+  DCS SDK v1.5.1
 </h1>
 
 > SDK for DataChecks

{dcs_sdk-1.4.9.dist-info → dcs_sdk-1.5.1.dist-info}/RECORD

@@ -2,11 +2,11 @@ data_diff/__init__.py,sha256=TGNvifF-dIbitACA6PS4sORtHEcTUukcrZz2NEIsL9U,10336
 data_diff/__main__.py,sha256=UvFvBKU74202bfRcIO_Wk-SU8WmnNuDK_1YVJpueMlc,16969
 data_diff/abcs/__init__.py,sha256=RkfhRKLXEForLCs4rZkTf0qc_b0TokSggSAcKI4yfZg,610
 data_diff/abcs/compiler.py,sha256=RuGhGlLTQuCzOJfYxa4gjcADsyvbZ9yZPuDuY6XH8Rk,785
-data_diff/abcs/database_types.py,sha256=dHE6K6UtqFwX6LTjOqJu1OOb_XJBXODeMcZQ0cdsRzk,9892
+data_diff/abcs/database_types.py,sha256=OTx6GSUQnuQYkRN7yN0Di8mF-Wt7rTtyzjAQPenHzzw,11054
 data_diff/config.py,sha256=uRcoVVhPjVZqgQNwr18v6sPq06cGXDLemTUyitU57zA,4998
 data_diff/databases/__init__.py,sha256=NrBm1Paj7jkHZ_hQCD-4-Q1eeDdh3v9_bz1DkPDOv9g,1680
 data_diff/databases/_connect.py,sha256=nGsmtzDSKN8CK8zMkdcGZz0iExzkJDYw-PGebIkmQgc,11151
-data_diff/databases/base.py,sha256=Fd_OzWOL97YiTXE_oYsBthXVVLvJGbU5Y4JgR77Sra0,51904
+data_diff/databases/base.py,sha256=3UPNVmG6XLVPMLmA3cGFZH-B2JIRGW--hiotkJEIh4k,53196
 data_diff/databases/bigquery.py,sha256=PDwSkmWRW26gUl2SMOyIsiYtgrqghptYqG8_SaaiXb4,14709
 data_diff/databases/clickhouse.py,sha256=5DsW8UpyYsWI8I3AlPUvHWYdWdWpqzRsvsVwgqEyaLw,7554
 data_diff/databases/databricks.py,sha256=6cdwfAspg1GIgWFlQByvFcW_Hz1mnJeI9_2kZhOP8b4,9334
@@ -18,13 +18,13 @@ data_diff/databases/postgresql.py,sha256=QrvYDsMSECQiPfsl85UQxY9ivxTRXnEwjAfRjd-
 data_diff/databases/presto.py,sha256=1Z8iDV2pc35Bu7DuUerFuFLbrwgSHSkBYmJ72JlZSZ8,7116
 data_diff/databases/redshift.py,sha256=-gFWs3NCcevO4s6c4zV3_LYihK24fUd5BADTKahubjw,8122
 data_diff/databases/snowflake.py,sha256=7G6fvVJXOtTvXmSfWCxTslF4WohoscQoiqcmJIN684A,7910
-data_diff/databases/sybase.py,sha256=2SwTbbNGPw324wiyFtGnuuvJUG2ppgHbYtx4R_fwmxw,27480
+data_diff/databases/sybase.py,sha256=ALbOshD-AUctGUSbnjvomvOr767laShusZLQ2CbdwBk,31755
 data_diff/databases/trino.py,sha256=VIN3gMJvT4oSYuXCJ1QnngVL2gjjEYMFw87QTHgjs8c,2328
 data_diff/databases/vertica.py,sha256=2dSDZp6qOEvUVPldI5Tgn7Sm3dCpC3vNXJL3qb3FDvQ,5529
 data_diff/diff_tables.py,sha256=Ey88gUr9Wh8UVsgRlBCY3CACIYfHL52PxJSrd821aqg,20060
 data_diff/errors.py,sha256=4Yru8yOwyuDuBlTABnGCvJMSpe6-rbLJpNnVHeTTyHU,745
 data_diff/format.py,sha256=QFDjdZaBVf_N-jfKiX4ppOUdpXTPZXmv1j0pc1RiOoc,10245
-data_diff/hashdiff_tables.py,sha256=K-JoMimAwACB3mqQLIdv5P5joYWyRMSZKcbDJWz5dlk,27993
+data_diff/hashdiff_tables.py,sha256=nbE3mfILWUYT9Z38P0lKskD_DrNOV4G_752EfmwGBxg,28339
 data_diff/info_tree.py,sha256=yHtFSoXuu6oBafLYOYQjUSKlB-DnAAd08U9HOEAdTPI,2799
 data_diff/joindiff_tables.py,sha256=fyrEYjyh2BX1vGibwVZLYM1V6JJTOY-uGXY-KInvMkw,17612
 data_diff/lexicographic_space.py,sha256=bBoCbbH1Mla9jNOq1b5RuwjAxSVU7gWkra673tPBwXQ,8305
@@ -37,20 +37,20 @@ data_diff/queries/base.py,sha256=pT-iaII7Nlu-w-Cuq9fhoNKX7-GSxkQ3Fk8K-tMkk60,964
 data_diff/queries/extras.py,sha256=aUm-ifj3BMlz4o4bbuHtmnvHZuptYAKGS5yWTHmNpvc,1270
 data_diff/query_utils.py,sha256=R7ZfRwcvv9Zf4zWXNln4tr_OxLmDI7CPmmCahYfHxlo,2101
 data_diff/schema.py,sha256=QoYSSB3k-svLXz680uRgsI4qjii8BFKOOQvheqtgEbs,2413
-data_diff/table_segment.py,sha256=-zYdppv8lW8ZfDRzZvmARm1fYxGHhAYPurR0r0GmWh8,21350
+data_diff/table_segment.py,sha256=3UYy7FkNq4FjxPbmE7KH7OJ15ZRhH7Db5N_HSOnQ-dA,22014
 data_diff/thread_utils.py,sha256=_692ERjnWfHKaZsLdg7CNfkKiRd66y7_kpgDwzntp44,3831
-data_diff/utils.py,sha256=IHKaFfyizOiA1By0sfnTra7CJvgEDwr5KPGjev0kx0A,23328
+data_diff/utils.py,sha256=8vbVbJvBSqtRfrzvQsz79qIAF2bHn1dnufijstDxddU,31345
 data_diff/version.py,sha256=Wk0ovyBlLEF2UaWLWEcVBLFElREtIxi7TU1hD3CuTFI,634
 dcs_sdk/__init__.py,sha256=RkfhRKLXEForLCs4rZkTf0qc_b0TokSggSAcKI4yfZg,610
 dcs_sdk/__main__.py,sha256=Qn8stIaQGrdLjHQ-H7xO0T-brtq5RWZoWU9QvqoarV8,683
-dcs_sdk/__version__.py,sha256=lS23jYuOvSiv7SJ1XHD4d1O93eassDRAcs2EuasDZ-0,633
+dcs_sdk/__version__.py,sha256=pM5-cAegZHkaKKxyTKGtuMAzXU_ZrNtyizfbN_4nido,633
 dcs_sdk/cli/__init__.py,sha256=RkfhRKLXEForLCs4rZkTf0qc_b0TokSggSAcKI4yfZg,610
 dcs_sdk/cli/cli.py,sha256=LyrRk972OL9pTqrvBeXWBu5rUDAN17lQ1g8FdSRW_8M,4299
 dcs_sdk/sdk/__init__.py,sha256=skrZcgWWJBL6NXTUERywJ3qRJRemgpDXyW7lPg1FJk8,2107
 dcs_sdk/sdk/config/__init__.py,sha256=RkfhRKLXEForLCs4rZkTf0qc_b0TokSggSAcKI4yfZg,610
 dcs_sdk/sdk/config/config_loader.py,sha256=Sg0gvleGrOcDaX41amgw6Uj6c_s4Zel6gQalIp9CltI,21416
 dcs_sdk/sdk/data_diff/__init__.py,sha256=RkfhRKLXEForLCs4rZkTf0qc_b0TokSggSAcKI4yfZg,610
-dcs_sdk/sdk/data_diff/data_differ.py,sha256=8WBDz-q-Li5OyJcG_a6JRnDPCJBvIsMki-DrD4EgDzQ,35575
+dcs_sdk/sdk/data_diff/data_differ.py,sha256=RH9C9OthahU8eeJzmUgQe24UZGGuBu7X2a_DrhTCa-U,35590
 dcs_sdk/sdk/rules/__init__.py,sha256=_BkKcE_jfdDQI_ECdOamJaefMKEXrKpYjPpnBQXl_Xs,657
 dcs_sdk/sdk/rules/rules_mappping.py,sha256=fxakVkf7B2cVkYSO946LTim_HmMsl6lBDBqZjTTsSPI,1292
 dcs_sdk/sdk/rules/rules_repository.py,sha256=x0Rli-wdnHAmXm5526go_qC3P-eFRt-4L7fs4hNqC-g,7564
@@ -65,7 +65,7 @@ dcs_sdk/sdk/utils/similarity_score/levenshtein_distance_provider.py,sha256=puAWP
 dcs_sdk/sdk/utils/table.py,sha256=mXArkEYmxf7mz_DJvyHVHIN9RyxaF708I70n52JrptA,18268
 dcs_sdk/sdk/utils/themes.py,sha256=Meo2Yldv4uyPpEqI7qdA28Aa6sxtwUU1dLKKm4QavjM,1403
 dcs_sdk/sdk/utils/utils.py,sha256=vF2zAvgt__Y8limicWTEWRyn41SBVJN81ZCTBRy6hQg,11907
-dcs_sdk-1.4.9.dist-info/METADATA,sha256=JUBldmhqJC8gu9G2eftTjkcf8rgWsPgsPQkk5tlkHWw,6221
-dcs_sdk-1.4.9.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
-dcs_sdk-1.4.9.dist-info/entry_points.txt,sha256=zQtrZL7YuaKtt6WPwihCTV1BRXnqBkaY6zUGdYJbBSg,49
-dcs_sdk-1.4.9.dist-info/RECORD,,
+dcs_sdk-1.5.1.dist-info/METADATA,sha256=9uoe5HtGCzBAOVWSNNHpV79337iosk6r0sS3QrK33OU,6221
+dcs_sdk-1.5.1.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
+dcs_sdk-1.5.1.dist-info/entry_points.txt,sha256=zQtrZL7YuaKtt6WPwihCTV1BRXnqBkaY6zUGdYJbBSg,49
+dcs_sdk-1.5.1.dist-info/RECORD,,