[buffer] Templatize UTF-* functions

This commit is contained in:
Behdad Esfahbod 2014-07-16 14:49:55 -04:00
parent e634fed428
commit 625dbf141a
2 changed files with 172 additions and 157 deletions

View File

@ -1298,6 +1298,8 @@ hb_buffer_add_utf (hb_buffer_t *buffer,
unsigned int item_offset,
int item_length)
{
typedef hb_utf_t<T> utf_t;
assert (buffer->content_type == HB_BUFFER_CONTENT_TYPE_UNICODE ||
(!buffer->len && buffer->content_type == HB_BUFFER_CONTENT_TYPE_INVALID));
@ -1305,7 +1307,7 @@ hb_buffer_add_utf (hb_buffer_t *buffer,
return;
if (text_length == -1)
text_length = hb_utf_strlen (text);
text_length = utf_t::strlen (text);
if (item_length == -1)
item_length = text_length - item_offset;
@ -1328,7 +1330,7 @@ hb_buffer_add_utf (hb_buffer_t *buffer,
while (start < prev && buffer->context_len[0] < buffer->CONTEXT_LENGTH)
{
hb_codepoint_t u;
prev = hb_utf_prev (prev, start, &u);
prev = utf_t::prev (prev, start, &u);
buffer->context[0][buffer->context_len[0]++] = u;
}
}
@ -1339,7 +1341,7 @@ hb_buffer_add_utf (hb_buffer_t *buffer,
{
hb_codepoint_t u;
const T *old_next = next;
next = hb_utf_next (next, end, &u);
next = utf_t::next (next, end, &u);
buffer->add (u, old_next - (const T *) text);
}
@ -1349,7 +1351,7 @@ hb_buffer_add_utf (hb_buffer_t *buffer,
while (next < end && buffer->context_len[1] < buffer->CONTEXT_LENGTH)
{
hb_codepoint_t u;
next = hb_utf_next (next, end, &u);
next = utf_t::next (next, end, &u);
buffer->context[1][buffer->context_len[1]++] = u;
}

View File

@ -29,202 +29,215 @@
#include "hb-private.hh"
template <typename T, bool validate=true> struct hb_utf_t;
/* UTF-8 */
static inline const uint8_t *
hb_utf_next (const uint8_t *text,
const uint8_t *end,
hb_codepoint_t *unicode)
template <>
struct hb_utf_t<uint8_t, true>
{
/* Written to only accept well-formed sequences.
* Based on ideas from ICU's U8_NEXT.
* Generates a -1 for each ill-formed byte. */
hb_codepoint_t c = *text++;
if (c > 0x7Fu)
static inline const uint8_t *
next (const uint8_t *text,
const uint8_t *end,
hb_codepoint_t *unicode)
{
if (hb_in_range (c, 0xC2u, 0xDFu)) /* Two-byte */
/* Written to only accept well-formed sequences.
* Based on ideas from ICU's U8_NEXT.
* Generates a -1 for each ill-formed byte. */
hb_codepoint_t c = *text++;
if (c > 0x7Fu)
{
unsigned int t1;
if (likely (text < end &&
(t1 = text[0] - 0x80u) <= 0x3Fu))
if (hb_in_range (c, 0xC2u, 0xDFu)) /* Two-byte */
{
c = ((c&0x1Fu)<<6) | t1;
text++;
}
else
goto error;
}
else if (hb_in_range (c, 0xE0u, 0xEFu)) /* Three-byte */
{
unsigned int t1, t2;
if (likely (1 < end - text &&
(t1 = text[0] - 0x80u) <= 0x3Fu &&
(t2 = text[1] - 0x80u) <= 0x3Fu))
{
c = ((c&0xFu)<<12) | (t1<<6) | t2;
if (unlikely (c < 0x0800u || hb_in_range (c, 0xD800u, 0xDFFFu)))
unsigned int t1;
if (likely (text < end &&
(t1 = text[0] - 0x80u) <= 0x3Fu))
{
c = ((c&0x1Fu)<<6) | t1;
text++;
}
else
goto error;
text += 2;
}
else
goto error;
}
else if (hb_in_range (c, 0xF0u, 0xF4u)) /* Four-byte */
{
unsigned int t1, t2, t3;
if (likely (2 < end - text &&
(t1 = text[0] - 0x80u) <= 0x3Fu &&
(t2 = text[1] - 0x80u) <= 0x3Fu &&
(t3 = text[2] - 0x80u) <= 0x3Fu))
else if (hb_in_range (c, 0xE0u, 0xEFu)) /* Three-byte */
{
c = ((c&0x7u)<<18) | (t1<<12) | (t2<<6) | t3;
if (unlikely (!hb_in_range (c, 0x10000u, 0x10FFFFu)))
unsigned int t1, t2;
if (likely (1 < end - text &&
(t1 = text[0] - 0x80u) <= 0x3Fu &&
(t2 = text[1] - 0x80u) <= 0x3Fu))
{
c = ((c&0xFu)<<12) | (t1<<6) | t2;
if (unlikely (c < 0x0800u || hb_in_range (c, 0xD800u, 0xDFFFu)))
goto error;
text += 2;
}
else
goto error;
}
else if (hb_in_range (c, 0xF0u, 0xF4u)) /* Four-byte */
{
unsigned int t1, t2, t3;
if (likely (2 < end - text &&
(t1 = text[0] - 0x80u) <= 0x3Fu &&
(t2 = text[1] - 0x80u) <= 0x3Fu &&
(t3 = text[2] - 0x80u) <= 0x3Fu))
{
c = ((c&0x7u)<<18) | (t1<<12) | (t2<<6) | t3;
if (unlikely (!hb_in_range (c, 0x10000u, 0x10FFFFu)))
goto error;
text += 3;
}
else
goto error;
text += 3;
}
else
goto error;
}
else
goto error;
}
*unicode = c;
return text;
error:
*unicode = -1;
return text;
}
static inline const uint8_t *
hb_utf_prev (const uint8_t *text,
const uint8_t *start,
hb_codepoint_t *unicode)
{
const uint8_t *end = text--;
while (start < text && (*text & 0xc0) == 0x80 && end - text < 4)
text--;
if (likely (hb_utf_next (text, end, unicode) == end))
*unicode = c;
return text;
*unicode = -1;
return end - 1;
}
error:
*unicode = -1;
return text;
}
static inline const uint8_t *
prev (const uint8_t *text,
const uint8_t *start,
hb_codepoint_t *unicode)
{
const uint8_t *end = text--;
while (start < text && (*text & 0xc0) == 0x80 && end - text < 4)
text--;
static inline unsigned int
hb_utf_strlen (const uint8_t *text)
{
return strlen ((const char *) text);
}
if (likely (next (text, end, unicode) == end))
return text;
*unicode = -1;
return end - 1;
}
static inline unsigned int
strlen (const uint8_t *text)
{
return ::strlen ((const char *) text);
}
};
/* UTF-16 */
static inline const uint16_t *
hb_utf_next (const uint16_t *text,
const uint16_t *end,
hb_codepoint_t *unicode)
template <>
struct hb_utf_t<uint16_t, true>
{
hb_codepoint_t c = *text++;
if (likely (!hb_in_range (c, 0xD800u, 0xDFFFu)))
static inline const uint16_t *
next (const uint16_t *text,
const uint16_t *end,
hb_codepoint_t *unicode)
{
*unicode = c;
return text;
}
hb_codepoint_t c = *text++;
if (likely (hb_in_range (c, 0xD800u, 0xDBFFu)))
{
/* High-surrogate in c */
hb_codepoint_t l;
if (text < end && ((l = *text), likely (hb_in_range (l, 0xDC00u, 0xDFFFu))))
if (likely (!hb_in_range (c, 0xD800u, 0xDFFFu)))
{
/* Low-surrogate in l */
*unicode = (c << 10) + l - ((0xD800u << 10) - 0x10000u + 0xDC00u);
text++;
return text;
*unicode = c;
return text;
}
if (likely (hb_in_range (c, 0xD800u, 0xDBFFu)))
{
/* High-surrogate in c */
hb_codepoint_t l;
if (text < end && ((l = *text), likely (hb_in_range (l, 0xDC00u, 0xDFFFu))))
{
/* Low-surrogate in l */
*unicode = (c << 10) + l - ((0xD800u << 10) - 0x10000u + 0xDC00u);
text++;
return text;
}
}
/* Lonely / out-of-order surrogate. */
*unicode = -1;
return text;
}
/* Lonely / out-of-order surrogate. */
*unicode = -1;
return text;
}
static inline const uint16_t *
hb_utf_prev (const uint16_t *text,
const uint16_t *start,
hb_codepoint_t *unicode)
{
const uint16_t *end = text--;
hb_codepoint_t c = *text;
if (likely (!hb_in_range (c, 0xD800u, 0xDFFFu)))
static inline const uint16_t *
prev (const uint16_t *text,
const uint16_t *start,
hb_codepoint_t *unicode)
{
*unicode = c;
return text;
const uint16_t *end = text--;
hb_codepoint_t c = *text;
if (likely (!hb_in_range (c, 0xD800u, 0xDFFFu)))
{
*unicode = c;
return text;
}
if (likely (start < text && hb_in_range (c, 0xDC00u, 0xDFFFu)))
text--;
if (likely (next (text, end, unicode) == end))
return text;
*unicode = -1;
return end - 1;
}
if (likely (start < text && hb_in_range (c, 0xDC00u, 0xDFFFu)))
text--;
if (likely (hb_utf_next (text, end, unicode) == end))
return text;
*unicode = -1;
return end - 1;
}
static inline unsigned int
hb_utf_strlen (const uint16_t *text)
{
unsigned int l = 0;
while (*text++) l++;
return l;
}
static inline unsigned int
strlen (const uint16_t *text)
{
unsigned int l = 0;
while (*text++) l++;
return l;
}
};
/* UTF-32 */
static inline const uint32_t *
hb_utf_next (const uint32_t *text,
const uint32_t *end HB_UNUSED,
hb_codepoint_t *unicode)
template <bool validate>
struct hb_utf_t<uint32_t, validate>
{
hb_codepoint_t c = *text++;
if (unlikely (c > 0x10FFFFu || hb_in_range (c, 0xD800u, 0xDFFFu)))
goto error;
*unicode = c;
return text;
static inline const uint32_t *
next (const uint32_t *text,
const uint32_t *end HB_UNUSED,
hb_codepoint_t *unicode)
{
hb_codepoint_t c = *text++;
if (validate && unlikely (c > 0x10FFFFu || hb_in_range (c, 0xD800u, 0xDFFFu)))
goto error;
*unicode = c;
return text;
error:
*unicode = -1;
return text;
}
error:
*unicode = -1;
return text;
}
static inline const uint32_t *
hb_utf_prev (const uint32_t *text,
const uint32_t *start HB_UNUSED,
hb_codepoint_t *unicode)
{
hb_utf_next (text - 1, text, unicode);
return text - 1;
}
static inline const uint32_t *
prev (const uint32_t *text,
const uint32_t *start HB_UNUSED,
hb_codepoint_t *unicode)
{
next (text - 1, text, unicode);
return text - 1;
}
static inline unsigned int
hb_utf_strlen (const uint32_t *text)
{
unsigned int l = 0;
while (*text++) l++;
return l;
}
static inline unsigned int
strlen (const uint32_t *text)
{
unsigned int l = 0;
while (*text++) l++;
return l;
}
};
#endif /* HB_UTF_PRIVATE_HH */