feat: rename to_md5 to md5_as_hex

Author: pik94

data_diff/databases/base.py

@@ -403,7 +403,7 @@ def render_checksum(self, c: Compiler, elem: Checksum) -> str:
     def render_concat(self, c: Compiler, elem: Concat) -> str:
         if self._prevent_overflow_when_concat:
             items = [
-                f"{self.compile(c, Code(self.to_md5(self.to_string(self.compile(c, expr)))))}" for expr in elem.exprs
+                f"{self.compile(c, Code(self.md5_as_hex(self.to_string(self.compile(c, expr)))))}" for expr in elem.exprs
             ]
 
         # We coalesce because on some DBs (e.g. MySQL) concat('a', NULL) is NULL
@@ -787,7 +787,7 @@ def md5_as_int(self, s: str) -> str:
         "Provide SQL for computing md5 and returning an int"
 
     @abstractmethod
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         """Method to calculate MD5"""
 
     @abstractmethod

data_diff/databases/bigquery.py

@@ -134,7 +134,7 @@ def parse_table_name(self, name: str) -> DbPath:
     def md5_as_int(self, s: str) -> str:
         return f"cast(cast( ('0x' || substr(TO_HEX(md5({s})), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS})) as int64) as numeric) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/clickhouse.py

@@ -105,7 +105,7 @@ def md5_as_int(self, s: str) -> str:
             f"reinterpretAsUInt128(reverse(unhex(lowerUTF8(substr(hex(MD5({s})), {substr_idx}))))) - {CHECKSUM_OFFSET}"
         )
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"hex(MD5({s}))"
 
     def normalize_number(self, value: str, coltype: FractionalType) -> str:

data_diff/databases/databricks.py

@@ -82,7 +82,7 @@ def parse_table_name(self, name: str) -> DbPath:
     def md5_as_int(self, s: str) -> str:
         return f"cast(conv(substr(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 16, 10) as decimal(38, 0)) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/duckdb.py

@@ -100,7 +100,7 @@ def current_timestamp(self) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"('0x' || SUBSTRING(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS},{CHECKSUM_HEXDIGITS}))::BIGINT - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/mssql.py

@@ -151,7 +151,7 @@ def normalize_number(self, value: str, coltype: NumericType) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"convert(bigint, convert(varbinary, '0x' + RIGHT(CONVERT(NVARCHAR(32), HashBytes('MD5', {s}), 2), {CHECKSUM_HEXDIGITS}), 1)) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"HashBytes('MD5', {s})"
 
 

data_diff/databases/mysql.py

@@ -101,7 +101,7 @@ def set_timezone_to_utc(self) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"conv(substring(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 16, 10) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/oracle.py

@@ -137,7 +137,7 @@ def md5_as_int(self, s: str) -> str:
         # TODO: Find a way to use UTL_RAW.CAST_TO_BINARY_INTEGER ?
         return f"to_number(substr(standard_hash({s}, 'MD5'), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 'xxxxxxxxxxxxxxx') - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"standard_hash({s}, 'MD5'"
 
     def normalize_uuid(self, value: str, coltype: ColType_UUID) -> str:

data_diff/databases/postgresql.py

@@ -98,7 +98,7 @@ def type_repr(self, t) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"('x' || substring(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}))::bit({_CHECKSUM_BITSIZE})::bigint - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/presto.py

@@ -128,7 +128,7 @@ def current_timestamp(self) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"cast(from_base(substr(to_hex(md5(to_utf8({s}))), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 16) as decimal(38, 0)) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"to_hex(md5(to_utf8({s})))"
 
     def normalize_uuid(self, value: str, coltype: ColType_UUID) -> str:

data_diff/databases/redshift.py

@@ -48,7 +48,7 @@ def type_repr(self, t) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"strtol(substring(md5({s}), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 16)::decimal(38) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/snowflake.py

@@ -76,7 +76,7 @@ def type_repr(self, t) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"BITAND(md5_number_lower64({s}), {CHECKSUM_MASK}) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"md5_number_lower64({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

data_diff/databases/vertica.py

@@ -110,7 +110,7 @@ def current_timestamp(self) -> str:
     def md5_as_int(self, s: str) -> str:
         return f"CAST(HEX_TO_INTEGER(SUBSTRING(MD5({s}), {1 + MD5_HEXDIGITS - CHECKSUM_HEXDIGITS})) AS NUMERIC(38, 0)) - {CHECKSUM_OFFSET}"
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         return f"MD5({s})"
 
     def normalize_timestamp(self, value: str, coltype: TemporalType) -> str:

tests/test_query.py

@@ -76,7 +76,7 @@ def optimizer_hints(self, s: str):
     def md5_as_int(self, s: str) -> str:
         raise NotImplementedError
 
-    def to_md5(self, s: str) -> str:
+    def md5_as_hex(self, s: str) -> str:
         raise NotImplementedError
 
     def normalize_number(self, value: str, coltype: FractionalType) -> str: