/*
* __wt_huffman_confchk --
* Verify Huffman configuration.
*/
int
__wt_huffman_confchk(WT_SESSION_IMPL *session, WT_CONFIG_ITEM *v)
{
if (v->len == 0)
return (0);
/* Standard Huffman encodings, no work to be done. */
if (WT_STRING_MATCH("english", v->str, v->len))
return (0);
if (WT_STRING_MATCH("none", v->str, v->len))
return (0);
return (__huffman_confchk_file(session, v, NULL, NULL));
}
/*
* __wt_huffman_read --
* Read a Huffman table from a file.
*/
static int
__wt_huffman_read(WT_SESSION_IMPL *session, WT_CONFIG_ITEM *ip,
struct __wt_huffman_table **tablep, u_int *entriesp, u_int *numbytesp)
{
struct __wt_huffman_table *table, *tp;
FILE *fp;
WT_DECL_RET;
int64_t symbol, frequency;
u_int entries, lineno;
int is_utf8;
*tablep = NULL;
*entriesp = *numbytesp = 0;
fp = NULL;
table = NULL;
/*
* Try and open the backing file.
*/
WT_RET(__huffman_confchk_file(session, ip, &is_utf8, &fp));
/*
* UTF-8 table is 256 bytes, with a range of 0-255.
* UTF-16 is 128KB (2 * 65536) bytes, with a range of 0-65535.
*/
if (is_utf8) {
entries = UINT8_MAX;
*numbytesp = 1;
WT_ERR(__wt_calloc_def(session, entries, &table));
} else {
entries = UINT16_MAX;
*numbytesp = 2;
WT_ERR(__wt_calloc_def(session, entries, &table));
}
for (tp = table, lineno = 1; (ret =
fscanf(fp, "%" SCNi64 " %" SCNi64, &symbol, &frequency)) != EOF;
++tp, ++lineno) {
if (lineno > entries)
WT_ERR_MSG(session, EINVAL,
"Huffman table file %.*s is corrupted, "
"more than %" PRIu32 " entries",
(int)ip->len, ip->str, entries);
if (ret != 2)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman table file %.*s is corrupted: "
"expected two unsigned integral values",
lineno, (int)ip->len, ip->str);
if (symbol < 0 || symbol > entries)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman file %.*s is corrupted; "
"symbol %" PRId64 " not in range, maximum "
"value is %u",
lineno, (int)ip->len, ip->str, symbol, entries);
if (frequency < 0 || frequency > UINT32_MAX)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman file %.*s is corrupted; "
"frequency %" PRId64 " not in range, maximum "
"value is %" PRIu32,
lineno, (int)ip->len, ip->str, frequency,
(uint32_t)UINT32_MAX);
tp->symbol = (uint32_t)symbol;
tp->frequency = (uint32_t)frequency;
}
ret = ferror(fp) ? WT_ERROR : 0;
*entriesp = lineno - 1;
*tablep = table;
if (0) {
err: __wt_free(session, table);
}
(void)__wt_fclose(&fp, WT_FHANDLE_READ);
return (ret);
}
/*
* __wt_huffman_read --
* Read a Huffman table from a file.
*/
static int
__wt_huffman_read(WT_SESSION_IMPL *session, WT_CONFIG_ITEM *ip,
struct __wt_huffman_table **tablep, u_int *entriesp, u_int *numbytesp)
{
struct __wt_huffman_table *table, *tp;
WT_DECL_ITEM(tmp);
WT_DECL_RET;
WT_FSTREAM *fs;
int64_t symbol, frequency;
u_int entries, lineno;
int n;
bool is_utf8;
*tablep = NULL;
*entriesp = *numbytesp = 0;
fs = NULL;
table = NULL;
/*
* Try and open the backing file.
*/
WT_RET(__huffman_confchk_file(session, ip, &is_utf8, &fs));
/*
* UTF-8 table is 256 bytes, with a range of 0-255.
* UTF-16 is 128KB (2 * 65536) bytes, with a range of 0-65535.
*/
if (is_utf8) {
entries = UINT8_MAX;
*numbytesp = 1;
WT_ERR(__wt_calloc_def(session, entries, &table));
} else {
entries = UINT16_MAX;
*numbytesp = 2;
WT_ERR(__wt_calloc_def(session, entries, &table));
}
WT_ERR(__wt_scr_alloc(session, 0, &tmp));
for (tp = table, lineno = 1;; ++tp, ++lineno) {
WT_ERR(__wt_getline(session, fs, tmp));
if (tmp->size == 0)
break;
n = sscanf(
tmp->data, "%" SCNi64 " %" SCNi64, &symbol, &frequency);
/*
* Entries is 0-based, that is, there are (entries +1) possible
* values that can be configured. The line number is 1-based, so
* adjust the test for too many entries, and report (entries +1)
* in the error as the maximum possible number of entries.
*/
if (lineno > entries + 1)
WT_ERR_MSG(session, EINVAL,
"Huffman table file %.*s is corrupted, "
"more than %" PRIu32 " entries",
(int)ip->len, ip->str, entries + 1);
if (n != 2)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman table file %.*s is corrupted: "
"expected two unsigned integral values",
lineno, (int)ip->len, ip->str);
if (symbol < 0 || symbol > entries)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman file %.*s is corrupted; "
"symbol %" PRId64 " not in range, maximum "
"value is %u",
lineno, (int)ip->len, ip->str, symbol, entries);
if (frequency < 0 || frequency > UINT32_MAX)
WT_ERR_MSG(session, EINVAL,
"line %u of Huffman file %.*s is corrupted; "
"frequency %" PRId64 " not in range, maximum "
"value is %" PRIu32,
lineno, (int)ip->len, ip->str, frequency,
(uint32_t)UINT32_MAX);
tp->symbol = (uint32_t)symbol;
tp->frequency = (uint32_t)frequency;
}
*entriesp = lineno - 1;
*tablep = table;
if (0) {
err: __wt_free(session, table);
}
(void)__wt_fclose(session, &fs);
__wt_scr_free(session, &tmp);
return (ret);
}
