mirrors/kitty
1
0
Fork 0
mirror of https://github.com/kovidgoyal/kitty synced 2026-07-07 17:43:53 +02:00
Code Issues Packages Projects Releases Wiki Activity

Unicode input: When searching by name search for prefix matches as well as whole word matches

Browse Source 
So now hori matches both "hori" and "horizontal". Switched to a
prefix-trie internally.
This commit is contained in:
Kovid Goyal
2018-04-24 07:45:20 +05:30
parent 9d67198ff9
commit 0b99bb534f
9 changed files with 32513 additions and 30796 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
kittens/unicode_input/main.py
View File

@@ -56,7 +56,7 @@ def name(cp):
@lru_cache(maxsize=256)
def codepoints_matching_search(parts):
ans = []
if parts and parts[0]:
if parts and parts[0] and len(parts[0]) > 1:
codepoints = points_for_word(parts[0])
for word in parts[1:]:
pts = points_for_word(word)

63099
kittens/unicode_input/names.h generated
View File

File diff suppressed because one or more lines are too long

59
kittens/unicode_input/unicode_names.c
View File

@@ -9,32 +9,61 @@
static PyObject*
all_words(PYNOARG) {
PyObject *ans = PyTuple_New(arraysz(idx_to_word));
PyObject *ans = PyTuple_New(arraysz(all_words_map));
if (!ans) return NULL;
for (size_t i = 0; i < arraysz(idx_to_word); i++) {
PyObject *w = PyUnicode_FromString(idx_to_word[i]);
for (size_t i = 0; i < arraysz(all_words_map); i++) {
PyObject *w = PyUnicode_FromString(all_words_map[i]);
if (w == NULL) { Py_DECREF(ans); return NULL; }
PyTuple_SET_ITEM(ans, i, w);
}
return ans;
}
static inline void
add_matches(const word_trie *wt, char_type *codepoints, size_t *pos, const size_t sz) {
size_t num = mark_groups[wt->match_offset];
for (size_t i = wt->match_offset + 1; i < wt->match_offset + 1 + num && *pos < sz; i++, (*pos)++) {
codepoints[*pos] = mark_to_cp[mark_groups[i]];
}
}
static void
process_trie_node(const word_trie *wt, char_type *codepoints, size_t *pos, const size_t sz) {
if (wt->match_offset) add_matches(wt, codepoints, pos, sz);
size_t num_children = children_array[wt->children_offset];
if (!num_children) return;
for (size_t c = wt->children_offset + 1; c < wt->children_offset + 1 + num_children; c++) {
if (*pos > sz) return;
uint32_t x = children_array[c];
process_trie_node(&all_trie_nodes[x >> 8], codepoints, pos, sz);
}
}
static inline PyObject*
codepoints_for_word(const char *word, size_t len) {
PyObject *ans = PyFrozenSet_New(NULL); if (ans == NULL) return NULL;
const unsigned short *words = words_for_first_letter[(unsigned)*word];
if (words == NULL) return ans;
for (unsigned short i = 1; i <= words[0]; i++) {
unsigned short word_idx = words[i];
const char *w = idx_to_word[word_idx];
if (strncmp(word, w, len) == 0 && strlen(w) == len) {
const char_type* codepoints = codepoints_for_word_idx[word_idx];
for (char_type i = 1; i <= codepoints[0]; i++) {
PyObject *t = PyLong_FromUnsignedLong(codepoints[i]); if (t == NULL) { Py_DECREF(ans); return NULL; }
int ret = PySet_Add(ans, t); Py_DECREF(t); if (ret != 0) { Py_DECREF(ans); return NULL; }
const word_trie *wt = all_trie_nodes;
for (size_t i = 0; i < len; i++) {
unsigned char ch = word[i];
size_t num_children = children_array[wt->children_offset];
if (!num_children) return PyFrozenSet_New(NULL);
bool found = false;
for (size_t c = wt->children_offset + 1; c < wt->children_offset + 1 + num_children; c++) {
uint32_t x = children_array[c];
if ((x & 0xff) == ch) {
found = true;
wt = &all_trie_nodes[x >> 8];
break;
}
break;
}
if (!found) return PyFrozenSet_New(NULL);
}
static char_type codepoints[1024];
size_t cpos = 0;
process_trie_node(wt, codepoints, &cpos, arraysz(codepoints));
PyObject *ans = PyFrozenSet_New(NULL); if (ans == NULL) return NULL;
for (size_t i = 0; i < cpos; i++) {
PyObject *t = PyLong_FromUnsignedLong(codepoints[i]); if (t == NULL) { Py_DECREF(ans); return NULL; }
int ret = PySet_Add(ans, t); Py_DECREF(t); if (ret != 0) { Py_DECREF(ans); return NULL; }
}
return ans;
}