/* This file is included into "Fuzzy.xs". The reason for having it as
a separate file from "Fuzzy.xs" is so that this file can easily be
edited in Emacs C mode without the C mode causing problems when
editing "Fuzzy.xs". */
/* Get memory via Perl. */
#define get_memory(value, number, what) { \
Newxz (value, number, what); \
if (! value) { \
croak ("%s:%d: " \
"Could not allocate memory for %d %s", \
__FILE__, __LINE__, number, #what); \
} \
text_fuzzy->n_mallocs++; \
}
typedef enum {
tfp_ok,
tfp_unicode_failure,
tfp_text_fuzzy_error,
}
tfp_status_t;
#define TFPCALL(x) { \
tfp_status_t status = x; \
if (status != tfp_ok) { \
return status; \
} \
}
/* Send a bad return value from one of the C routines in
"text-fuzzy.c" back to the user via Perl's error handlers. The
parameters "file_name" and "line_number" are the name of the C file
and the line number where the error occurred in the C file. These
are discarded by this error handler. */
int perl_error_handler (const char * file_name, int line_number,
const char * format, ...)
{
va_list a;
// warn ("%s:%d: ", file_name, line_number);
va_start (a, format);
vcroak (format, & a);
va_end (a);
return 0;
}
#define SMALL 0x1000
/* Decide how many ints to allocate for "text_fuzzy->b.unicode". It
has to be bigger than "minimum", the actual length of the
string. Also, we don't want to keep reallocating it, so make it
large enough for most of the cases (SMALL). */
static void fake_length (text_fuzzy_t * text_fuzzy, int minimum)
{
int r = SMALL;
again:
if (minimum < r) {
text_fuzzy->b_unicode_length = r;
return;
}
r *= 2;
if (r > STRING_MAX_CHARS) {
/* Stupid value. */
croak ("String length %d longer than maximum allowed for, %d.\n",
minimum, STRING_MAX_CHARS);
}
goto again;
}
/* Allocate the memory for b. */
static void
allocate_b_unicode (text_fuzzy_t * text_fuzzy, int b_length)
{
if (! text_fuzzy->b.unicode) {
/* We have not allocated any memory yet. */
fake_length (text_fuzzy, b_length);
get_memory (text_fuzzy->b.unicode,
text_fuzzy->b_unicode_length, int);
}
else if (b_length > text_fuzzy->b_unicode_length) {
/* "b" is bigger than what we allowed for. */
fake_length (text_fuzzy, b_length);
/* "Renew" is "realloc" for Perl. See "perldoc perlapi". */
Renew (text_fuzzy->b.unicode, text_fuzzy->b_unicode_length, int);
}
}
/* Given a Perl string in "text" which is marked as being Unicode
characters, use Perl's Unicode handlers to turn it into a string of
integers. */
static tfp_status_t
sv_to_int_ptr (SV * text, text_fuzzy_string_t * tfs)
{
#if PERL_REVISION == 5 && PERL_VERSION < 16
int i;
const U8 * utf;
STRLEN curlen;
STRLEN length;
const unsigned char * stuff;
stuff = (const unsigned char *) SvPV (text, length);
utf = stuff;
curlen = length;
for (i = 0; i < tfs->ulength; i++) {
STRLEN len;
/* The documentation for "utf8n_to_uvuni" can be found in
"perldoc perlapi". There is an online version here:
"http://perldoc.perl.org/perlapi.html#Unicode-Support". */
tfs->unicode[i] = utf8n_to_uvuni (utf, curlen, & len, 0);
curlen -= len;
utf += len;
}
return tfp_ok;
#else /* Perl version/revision */
int i;
const U8 * utf;
const U8 * send;
STRLEN length;
utf = (const U8 *) SvPV (text, length);
send = utf + length;
for (i = 0; i < tfs->ulength; i++) {
STRLEN len;
tfs->unicode[i] = (int) utf8_to_uvchr_buf (utf, send, & len);
if (len == -1 || tfs->unicode[i] == 0) {
return tfp_unicode_failure;
}
utf += len;
}
return tfp_ok;
#endif /* Perl version/revision */
}
/* Convert a Perl SV into the text_fuzzy_t structure. */
static tfp_status_t
sv_to_text_fuzzy (SV * text, text_fuzzy_t ** text_fuzzy_ptr)
{
STRLEN length;
const unsigned char * stuff;
text_fuzzy_t * text_fuzzy;
int i;
int is_utf8;
char * copy;
/* Allocate memory for "text_fuzzy". */
get_memory (text_fuzzy, 1, text_fuzzy_t);
text_fuzzy->max_distance = NO_MAX_DISTANCE;
/* Copy the string in "text" into "text_fuzzy". */
stuff = (const unsigned char *) SvPV (text, length);
get_memory (copy, length + 1, char);
for (i = 0; i < (int) length; i++) {
copy[i] = stuff[i];
}
copy[length] = '\0';
text_fuzzy->text.length = length;
text_fuzzy->text.text = copy;
is_utf8 = SvUTF8 (text);
if (is_utf8) {
/* Put the Unicode version of the string into
"text_fuzzy->text". */
text_fuzzy->unicode = 1;
text_fuzzy->text.ulength = sv_len_utf8 (text);
get_memory (text_fuzzy->text.unicode, text_fuzzy->text.ulength, int);
TFPCALL(sv_to_int_ptr(text, & text_fuzzy->text));
if (text_fuzzy->text.ulength > 0) {
/* Generate the Unicode alphabet. */
TEXT_FUZZY (generate_ualphabet (text_fuzzy));
}
}
else {
if (text_fuzzy->text.length > 0) {
TEXT_FUZZY (generate_alphabet (text_fuzzy));
}
}
TEXT_FUZZY (allocate_edits (text_fuzzy));
* text_fuzzy_ptr = text_fuzzy;
return tfp_ok;
}
static tfp_status_t
sv_to_text_fuzzy_string (SV * word, text_fuzzy_t * text_fuzzy)
{
STRLEN length;
char * nonu;
text_fuzzy->b.text = SvPV (word, length);
text_fuzzy->b.allocated = 0;
text_fuzzy->b.length = length;
if (SvUTF8 (word) || text_fuzzy->unicode) {
/* Make a Unicode version of b. */
text_fuzzy->b.ulength = sv_len_utf8 (word);
allocate_b_unicode (text_fuzzy, text_fuzzy->b.ulength);
sv_to_int_ptr (word, & text_fuzzy->b);
if (! text_fuzzy->unicode) {
/* Make a non-Unicode version of b. */
int i;
text_fuzzy->b.length = text_fuzzy->b.ulength;
text_fuzzy->b.allocated = 1;
get_memory (nonu, text_fuzzy->b.length + 1, char);
for (i = 0; i < text_fuzzy->b.ulength; i++) {
int c;
c = text_fuzzy->b.unicode[i];
if (c <= 0x80) {
nonu[i] = c;
}
else {
/* Put a non-matching character in there. */
nonu[i] = text_fuzzy->invalid_char;
}
}
text_fuzzy->b.text = nonu;
}
}
return tfp_ok;
}
static void
free_text (text_fuzzy_t * text_fuzzy)
{
if (text_fuzzy->b.allocated) {
Safefree (text_fuzzy->b.text);
text_fuzzy->n_mallocs--;
text_fuzzy->b.text = 0;
text_fuzzy->b.allocated = 0;
}
}
/* The following definitions relate to the macros "FAIL" and
"FAIL_MSG" in "text-fuzzy.c.in". */
#undef TEXT_FUZZY_USER_ERROR
#define TEXT_FUZZY_USER_ERROR -1
static int
text_fuzzy_sv_distance (text_fuzzy_t * text_fuzzy, SV * word)
{
sv_to_text_fuzzy_string (word, text_fuzzy);
TEXT_FUZZY (compare_single (text_fuzzy));
free_text (text_fuzzy);
if (text_fuzzy->found) {
return text_fuzzy->distance;
}
else {
return text_fuzzy->max_distance + 1;
}
}
static int
text_fuzzy_av_distance (text_fuzzy_t * text_fuzzy, AV * words, AV * wantarray)
{
int i;
int n_words;
int nearest;
if (wantarray) {
text_fuzzy->wantarray = 1;
}
TEXT_FUZZY (begin_scanning (text_fuzzy));
nearest = -1;
n_words = av_len (words) + 1;
/* Check for empty array. */
if (n_words == 0) {
return -1;
}
for (i = 0; i < n_words; i++) {
SV * word;
word = * av_fetch (words, i, 0);
sv_to_text_fuzzy_string (word, text_fuzzy);
text_fuzzy->offset = i;
TEXT_FUZZY (compare_single (text_fuzzy));
free_text (text_fuzzy);
if (text_fuzzy->found) {
nearest = i;
if (! text_fuzzy->wantarray && text_fuzzy->distance == 0) {
/* Stop the search if there is an exact
match. Note that "no_exact" is checked in
"compare_single", so we don't need to check it
here. */
break;
}
}
}
text_fuzzy->distance = text_fuzzy->max_distance;
/* Set the maximum distance back to the user's value. */
TEXT_FUZZY (end_scanning (text_fuzzy));
/* If the user wants an array of values, we go through the linked
list and collect them into "wantarray". Because we went through
the list of words from the top to the bottom, gathering
whatever was the minimum value at that point in the progress,
our list may contain false hits which must be discarded. */
if (text_fuzzy->wantarray) {
int n_candidates;
int * candidates;
int i;
TEXT_FUZZY (get_candidates (text_fuzzy, & n_candidates,
& candidates));
if (n_candidates > 0) {
for (i = 0; i < n_candidates; i++) {
SV * offset;
offset = newSViv (candidates[i]);
av_push (wantarray, offset);
}
TEXT_FUZZY (free_candidates (text_fuzzy, candidates));
}
}
return nearest;
}
/* Free the memory allocated to "text_fuzzy" and check that there has
not been a memory leak. */
static int text_fuzzy_free (text_fuzzy_t * text_fuzzy)
{
if (text_fuzzy->b.unicode) {
Safefree (text_fuzzy->b.unicode);
text_fuzzy->n_mallocs--;
}
/* See the comments in "text-fuzzy.c.in" about why this is
necessary. */
TEXT_FUZZY (free_memory (text_fuzzy));
if (text_fuzzy->unicode) {
Safefree (text_fuzzy->text.unicode);
text_fuzzy->n_mallocs--;
}
Safefree (text_fuzzy->text.text);
text_fuzzy->n_mallocs--;
if (text_fuzzy->n_mallocs != 1) {
warn ("memory leak: n_mallocs %d != 1", text_fuzzy->n_mallocs);
}
Safefree (text_fuzzy);
return 0;
}