gh-132983: Split _zstd_set_c_parameters (GH-133921)

(cherry picked from commit 11f7a93)

Co-authored-by: Adam Turner <[email protected]>

Author: AA-Turner

Lib/compression/zstd/_zstdfile.py

@@ -1,7 +1,6 @@
 import io
 from os import PathLike
-from _zstd import (ZstdCompressor, ZstdDecompressor, ZstdError,
-                   ZSTD_DStreamOutSize)
+from _zstd import ZstdCompressor, ZstdDecompressor, ZSTD_DStreamOutSize
 from compression._common import _streams
 
 __all__ = ('ZstdFile', 'open')

Lib/test/test_zstd.py

@@ -65,6 +65,10 @@
 
 SUPPORT_MULTITHREADING = False
 
+C_INT_MIN = -(2**31)
+C_INT_MAX = (2**31) - 1
+
+
 def setUpModule():
     global SUPPORT_MULTITHREADING
     SUPPORT_MULTITHREADING = CompressionParameter.nb_workers.bounds() != (0, 0)
@@ -196,14 +200,21 @@ def test_simple_compress_bad_args(self):
         self.assertRaises(TypeError, ZstdCompressor, zstd_dict=b"abcd1234")
         self.assertRaises(TypeError, ZstdCompressor, zstd_dict={1: 2, 3: 4})
 
-        with self.assertRaises(ValueError):
-            ZstdCompressor(2**31)
-        with self.assertRaises(ValueError):
-            ZstdCompressor(options={2**31: 100})
+        # valid range for compression level is [-(1<<17), 22]
+        msg = r'illegal compression level {}; the valid range is \[-?\d+, -?\d+\]'
+        with self.assertRaisesRegex(ValueError, msg.format(C_INT_MAX)):
+            ZstdCompressor(C_INT_MAX)
+        with self.assertRaisesRegex(ValueError, msg.format(C_INT_MIN)):
+            ZstdCompressor(C_INT_MIN)
+        msg = r'illegal compression level; the valid range is \[-?\d+, -?\d+\]'
+        with self.assertRaisesRegex(ValueError, msg):
+            ZstdCompressor(level=-(2**1000))
+        with self.assertRaisesRegex(ValueError, msg):
+            ZstdCompressor(level=2**1000)
 
-        with self.assertRaises(ZstdError):
+        with self.assertRaises(ValueError):
             ZstdCompressor(options={CompressionParameter.window_log: 100})
-        with self.assertRaises(ZstdError):
+        with self.assertRaises(ValueError):
             ZstdCompressor(options={3333: 100})
 
         # Method bad arguments
@@ -254,18 +265,32 @@ def test_compress_parameters(self):
              }
         ZstdCompressor(options=d)
 
-        # larger than signed int, ValueError
         d1 = d.copy()
-        d1[CompressionParameter.ldm_bucket_size_log] = 2**31
-        self.assertRaises(ValueError, ZstdCompressor, options=d1)
+        # larger than signed int
+        d1[CompressionParameter.ldm_bucket_size_log] = C_INT_MAX
+        with self.assertRaises(ValueError):
+            ZstdCompressor(options=d1)
+        # smaller than signed int
+        d1[CompressionParameter.ldm_bucket_size_log] = C_INT_MIN
+        with self.assertRaises(ValueError):
+            ZstdCompressor(options=d1)
 
-        # clamp compressionLevel
+        # out of bounds compression level
         level_min, level_max = CompressionParameter.compression_level.bounds()
-        compress(b'', level_max+1)
-        compress(b'', level_min-1)
-
-        compress(b'', options={CompressionParameter.compression_level:level_max+1})
-        compress(b'', options={CompressionParameter.compression_level:level_min-1})
+        with self.assertRaises(ValueError):
+            compress(b'', level_max+1)
+        with self.assertRaises(ValueError):
+            compress(b'', level_min-1)
+        with self.assertRaises(ValueError):
+            compress(b'', 2**1000)
+        with self.assertRaises(ValueError):
+            compress(b'', -(2**1000))
+        with self.assertRaises(ValueError):
+            compress(b'', options={
+                CompressionParameter.compression_level: level_max+1})
+        with self.assertRaises(ValueError):
+            compress(b'', options={
+                CompressionParameter.compression_level: level_min-1})
 
         # zstd lib doesn't support MT compression
         if not SUPPORT_MULTITHREADING:
@@ -278,19 +303,19 @@ def test_compress_parameters(self):
 
         # out of bounds error msg
         option = {CompressionParameter.window_log:100}
-        with self.assertRaisesRegex(ZstdError,
-                (r'Error when setting zstd compression parameter "window_log", '
-                 r'it should \d+ <= value <= \d+, provided value is 100\. '
-                 r'\((?:32|64)-bit build\)')):
+        with self.assertRaisesRegex(
+            ValueError,
+            "compression parameter 'window_log' received an illegal value 100; "
+            r'the valid range is \[-?\d+, -?\d+\]',
+        ):
             compress(b'', options=option)
 
     def test_unknown_compression_parameter(self):
         KEY = 100001234
         option = {CompressionParameter.compression_level: 10,
                   KEY: 200000000}
-        pattern = (r'Invalid zstd compression parameter.*?'
-                   fr'"unknown parameter \(key {KEY}\)"')
-        with self.assertRaisesRegex(ZstdError, pattern):
+        pattern = rf"invalid compression parameter 'unknown parameter \(key {KEY}\)'"
+        with self.assertRaisesRegex(ValueError, pattern):
             ZstdCompressor(options=option)
 
     @unittest.skipIf(not SUPPORT_MULTITHREADING,
@@ -385,12 +410,22 @@ def test_simple_decompress_bad_args(self):
         self.assertRaises(TypeError, ZstdDecompressor, options=b'abc')
 
         with self.assertRaises(ValueError):
-            ZstdDecompressor(options={2**31 : 100})
+            ZstdDecompressor(options={C_INT_MAX: 100})
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(options={C_INT_MIN: 100})
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(options={0: C_INT_MAX})
+        with self.assertRaises(OverflowError):
+            ZstdDecompressor(options={2**1000: 100})
+        with self.assertRaises(OverflowError):
+            ZstdDecompressor(options={-(2**1000): 100})
+        with self.assertRaises(OverflowError):
+            ZstdDecompressor(options={0: -(2**1000)})
 
-        with self.assertRaises(ZstdError):
-            ZstdDecompressor(options={DecompressionParameter.window_log_max:100})
-        with self.assertRaises(ZstdError):
-            ZstdDecompressor(options={3333 : 100})
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(options={DecompressionParameter.window_log_max: 100})
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(options={3333: 100})
 
         empty = compress(b'')
         lzd = ZstdDecompressor()
@@ -403,26 +438,52 @@ def test_decompress_parameters(self):
         d = {DecompressionParameter.window_log_max : 15}
         ZstdDecompressor(options=d)
 
-        # larger than signed int, ValueError
         d1 = d.copy()
-        d1[DecompressionParameter.window_log_max] = 2**31
-        self.assertRaises(ValueError, ZstdDecompressor, None, d1)
+        # larger than signed int
+        d1[DecompressionParameter.window_log_max] = 2**1000
+        with self.assertRaises(OverflowError):
+            ZstdDecompressor(None, d1)
+        # smaller than signed int
+        d1[DecompressionParameter.window_log_max] = -(2**1000)
+        with self.assertRaises(OverflowError):
+            ZstdDecompressor(None, d1)
+
+        d1[DecompressionParameter.window_log_max] = C_INT_MAX
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(None, d1)
+        d1[DecompressionParameter.window_log_max] = C_INT_MIN
+        with self.assertRaises(ValueError):
+            ZstdDecompressor(None, d1)
 
         # out of bounds error msg
         options = {DecompressionParameter.window_log_max:100}
-        with self.assertRaisesRegex(ZstdError,
-                (r'Error when setting zstd decompression parameter "window_log_max", '
-                 r'it should \d+ <= value <= \d+, provided value is 100\. '
-                 r'\((?:32|64)-bit build\)')):
+        with self.assertRaisesRegex(
+            ValueError,
+            "decompression parameter 'window_log_max' received an illegal value 100; "
+            r'the valid range is \[-?\d+, -?\d+\]',
+        ):
+            decompress(b'', options=options)
+
+        # out of bounds deecompression parameter
+        options[DecompressionParameter.window_log_max] = C_INT_MAX
+        with self.assertRaises(ValueError):
+            decompress(b'', options=options)
+        options[DecompressionParameter.window_log_max] = C_INT_MIN
+        with self.assertRaises(ValueError):
+            decompress(b'', options=options)
+        options[DecompressionParameter.window_log_max] = 2**1000
+        with self.assertRaises(OverflowError):
+            decompress(b'', options=options)
+        options[DecompressionParameter.window_log_max] = -(2**1000)
+        with self.assertRaises(OverflowError):
             decompress(b'', options=options)
 
     def test_unknown_decompression_parameter(self):
         KEY = 100001234
         options = {DecompressionParameter.window_log_max: DecompressionParameter.window_log_max.bounds()[1],
                   KEY: 200000000}
-        pattern = (r'Invalid zstd decompression parameter.*?'
-                   fr'"unknown parameter \(key {KEY}\)"')
-        with self.assertRaisesRegex(ZstdError, pattern):
+        pattern = rf"invalid decompression parameter 'unknown parameter \(key {KEY}\)'"
+        with self.assertRaisesRegex(ValueError, pattern):
             ZstdDecompressor(options=options)
 
     def test_decompress_epilogue_flags(self):
@@ -1425,11 +1486,11 @@ def test_init_bad_mode(self):
             ZstdFile(io.BytesIO(COMPRESSED_100_PLUS_32KB), "rw")
 
         with self.assertRaisesRegex(TypeError,
-                                    r"NOT be a CompressionParameter"):
+                                    r"not be a CompressionParameter"):
             ZstdFile(io.BytesIO(), 'rb',
                      options={CompressionParameter.compression_level:5})
         with self.assertRaisesRegex(TypeError,
-                                    r"NOT be a DecompressionParameter"):
+                                    r"not be a DecompressionParameter"):
             ZstdFile(io.BytesIO(), 'wb',
                      options={DecompressionParameter.window_log_max:21})
 
@@ -1440,19 +1501,19 @@ def test_init_bad_check(self):
         with self.assertRaises(TypeError):
             ZstdFile(io.BytesIO(), "w", level='asd')
         # CHECK_UNKNOWN and anything above CHECK_ID_MAX should be invalid.
-        with self.assertRaises(ZstdError):
+        with self.assertRaises(ValueError):
             ZstdFile(io.BytesIO(), "w", options={999:9999})
-        with self.assertRaises(ZstdError):
+        with self.assertRaises(ValueError):
             ZstdFile(io.BytesIO(), "w", options={CompressionParameter.window_log:99})
 
         with self.assertRaises(TypeError):
             ZstdFile(io.BytesIO(COMPRESSED_100_PLUS_32KB), "r", options=33)
 
-        with self.assertRaises(ValueError):
+        with self.assertRaises(OverflowError):
             ZstdFile(io.BytesIO(COMPRESSED_100_PLUS_32KB),
                              options={DecompressionParameter.window_log_max:2**31})
 
-        with self.assertRaises(ZstdError):
+        with self.assertRaises(ValueError):
             ZstdFile(io.BytesIO(COMPRESSED_100_PLUS_32KB),
                              options={444:333})
 
@@ -1468,7 +1529,7 @@ def test_init_close_fp(self):
             tmp_f.write(DAT_130K_C)
             filename = tmp_f.name
 
-        with self.assertRaises(ValueError):
+        with self.assertRaises(TypeError):
             ZstdFile(filename, options={'a':'b'})
 
         # for PyPy

Modules/_zstd/_zstdmodule.c

@@ -103,16 +103,13 @@ static const ParameterInfo dp_list[] = {
 };
 
 void
-set_parameter_error(const _zstd_state* const state, int is_compress,
-                    int key_v, int value_v)
+set_parameter_error(int is_compress, int key_v, int value_v)
 {
     ParameterInfo const *list;
     int list_size;
-    char const *name;
     char *type;
     ZSTD_bounds bounds;
-    int i;
-    char pos_msg[128];
+    char pos_msg[64];
 
     if (is_compress) {
         list = cp_list;
@@ -126,8 +123,8 @@ set_parameter_error(const _zstd_state* const state, int is_compress,
     }
 
     /* Find parameter's name */
-    name = NULL;
-    for (i = 0; i < list_size; i++) {
+    char const *name = NULL;
+    for (int i = 0; i < list_size; i++) {
         if (key_v == (list+i)->parameter) {
             name = (list+i)->parameter_name;
             break;
@@ -149,20 +146,16 @@ set_parameter_error(const _zstd_state* const state, int is_compress,
         bounds = ZSTD_dParam_getBounds(key_v);
     }
     if (ZSTD_isError(bounds.error)) {
-        PyErr_Format(state->ZstdError,
-                     "Invalid zstd %s parameter \"%s\".",
+        PyErr_Format(PyExc_ValueError, "invalid %s parameter '%s'",
                      type, name);
         return;
     }
 
     /* Error message */
-    PyErr_Format(state->ZstdError,
-                 "Error when setting zstd %s parameter \"%s\", it "
-                 "should %d <= value <= %d, provided value is %d. "
-                 "(%d-bit build)",
-                 type, name,
-                 bounds.lowerBound, bounds.upperBound, value_v,
-                 8*(int)sizeof(Py_ssize_t));
+    PyErr_Format(PyExc_ValueError,
+        "%s parameter '%s' received an illegal value %d; "
+        "the valid range is [%d, %d]",
+        type, name, value_v, bounds.lowerBound, bounds.upperBound);
 }
 
 static inline _zstd_state*

Modules/_zstd/_zstdmodule.h

@@ -49,7 +49,6 @@ set_zstd_error(const _zstd_state* const state,
                const error_type type, size_t zstd_ret);
 
 extern void
-set_parameter_error(const _zstd_state* const state, int is_compress,
-                    int key_v, int value_v);
+set_parameter_error(int is_compress, int key_v, int value_v);
 
 #endif  // !ZSTD_MODULE_H