cpython: 2c514c382a2a (original) (raw)

--- a/Objects/unicodeobject.c +++ b/Objects/unicodeobject.c @@ -5496,33 +5496,39 @@ PyUnicode_DecodeUTF16Stateful(const char int kind = PyUnicode_KIND(unicode); void *data = PyUnicode_DATA(unicode); while (_q < aligned_end) {

           unsigned long block = * (unsigned long *) _q;[](#l1.7)

           unsigned short *pblock = (unsigned short*)&block;[](#l1.8)

```
           union {[](#l1.9)
```

               unsigned long as_long;[](#l1.10)

               unsigned short units[sizeof(long) / sizeof(short)];[](#l1.11)

               unsigned char bytes[sizeof(long)];[](#l1.12)

           } block, block_copy;[](#l1.13)
           Py_UCS4 maxch;[](#l1.14)

           block.as_long = *(unsigned long *) _q;[](#l1.16)
           if (native_ordering) {[](#l1.17)
               /* Can use buffer directly */[](#l1.18)

               if (block & FAST_CHAR_MASK)[](#l1.19)

               if (block.as_long & FAST_CHAR_MASK)[](#l1.20)
                   break;[](#l1.21)
           }[](#l1.22)
           else {[](#l1.23)
               /* Need to byte-swap */[](#l1.24)

               unsigned char *_p = (unsigned char*)pblock;[](#l1.25)

               if (block & SWAPPED_FAST_CHAR_MASK)[](#l1.26)

               block_copy = block;[](#l1.27)

               if (block.as_long & SWAPPED_FAST_CHAR_MASK)[](#l1.29)
                   break;[](#l1.30)

```
               _p[0] = _q[1];[](#l1.31)
```
```
               _p[1] = _q[0];[](#l1.32)
```
```
               _p[2] = _q[3];[](#l1.33)
```
```
               _p[3] = _q[2];[](#l1.34)
```

               block.bytes[0] = block_copy.bytes[1];[](#l1.35)

               block.bytes[1] = block_copy.bytes[0];[](#l1.36)

               block.bytes[2] = block_copy.bytes[3];[](#l1.37)

               block.bytes[3] = block_copy.bytes[2];[](#l1.38)

#if (SIZEOF_LONG == 8)

```
               _p[4] = _q[5];[](#l1.40)
```
```
               _p[5] = _q[4];[](#l1.41)
```
```
               _p[6] = _q[7];[](#l1.42)
```
```
               _p[7] = _q[6];[](#l1.43)
```

-#endif

```
           }[](#l1.45)
```

           maxch = Py_MAX(pblock[0], pblock[1]);[](#l1.46)

               block.bytes[4] = block_copy.bytes[5];[](#l1.47)

               block.bytes[5] = block_copy.bytes[4];[](#l1.48)

               block.bytes[6] = block_copy.bytes[7];[](#l1.49)

               block.bytes[7] = block_copy.bytes[6];[](#l1.50)

+#endif

```
           }[](#l1.52)
```

           maxch = Py_MAX(block.units[0], block.units[1]);[](#l1.53)

#if SIZEOF_LONG == 8

           maxch = Py_MAX(maxch, Py_MAX(pblock[2], pblock[3]));[](#l1.55)

           maxch = Py_MAX(maxch, Py_MAX(block.units[2], block.units[3]));[](#l1.56)

#endif if (maxch > PyUnicode_MAX_CHAR_VALUE(unicode)) { if (unicode_widen(&unicode, maxch) < 0) @@ -5530,11 +5536,11 @@ PyUnicode_DecodeUTF16Stateful(const char kind = PyUnicode_KIND(unicode); data = PyUnicode_DATA(unicode); }

           PyUnicode_WRITE(kind, data, outpos++, pblock[0]);[](#l1.64)

           PyUnicode_WRITE(kind, data, outpos++, pblock[1]);[](#l1.65)

           PyUnicode_WRITE(kind, data, outpos++, block.units[0]);[](#l1.66)

           PyUnicode_WRITE(kind, data, outpos++, block.units[1]);[](#l1.67)

#if SIZEOF_LONG == 8

           PyUnicode_WRITE(kind, data, outpos++, pblock[2]);[](#l1.69)

           PyUnicode_WRITE(kind, data, outpos++, pblock[3]);[](#l1.70)

           PyUnicode_WRITE(kind, data, outpos++, block.units[2]);[](#l1.71)

           PyUnicode_WRITE(kind, data, outpos++, block.units[3]);[](#l1.72)

#endif _q += SIZEOF_LONG; }