Skip to content

Commit ceb069a

Browse files
leti367leti367
committed
use malloc
1 parent 4e1359b commit ceb069a

File tree

3 files changed

+94
-34
lines changed

3 files changed

+94
-34
lines changed

‎wasm/blingfire.js‎

Lines changed: 1 addition & 1 deletion
Some generated files are not rendered by default. Learn more about customizing how changed files appear on GitHub.

‎wasm/blingfire.wasm‎

100644100755
-5.47 KB
Binary file not shown.

‎wasm/blingfire_wrapper.js‎

Lines changed: 93 additions & 33 deletions
Original file line numberDiff line numberDiff line change
@@ -12,15 +12,30 @@ export function TextToWords(s) {
1212

1313
var len = Module["lengthBytesUTF8"](s);
1414

15-
var inUtf8 = Module["stackAlloc"](len + 1); // if we don't do +1 this library won't copy the last character
15+
var inUtf8 = Module["_malloc"](len + 1); // if we don't do +1 this library won't copy the last character
1616
Module["stringToUTF8"](s, inUtf8, len + 1); // since it always also needs a space for a 0-char
1717

1818
var MaxOutLength = (len << 1) + 1; // worst case every character is a token
19-
var outUtf8 = Module["stackAlloc"](MaxOutLength);
19+
var outUtf8 = Module["_malloc"](MaxOutLength);
2020

21-
var actualLen = Module["_TextToWords"](inUtf8, len, outUtf8, MaxOutLength);
22-
if(0 > actualLen || actualLen > MaxOutLength) {
23-
return null;
21+
try
22+
{
23+
var actualLen = Module["_TextToWords"](inUtf8, len, outUtf8, MaxOutLength);
24+
if(0 > actualLen || actualLen > MaxOutLength) {
25+
return null;
26+
}
27+
}
28+
finally
29+
{
30+
if (inUtf8 != 0)
31+
{
32+
Module["_free"](inUtf8);
33+
}
34+
35+
if (outUtf8 != 0)
36+
{
37+
Module["_free"](outUtf8);
38+
}
2439
}
2540

2641
return Module["UTF8ToString"](outUtf8);
@@ -31,15 +46,30 @@ export function TextToSentences(s) {
3146

3247
var len = Module["lengthBytesUTF8"](s);
3348

34-
var inUtf8 = Module["stackAlloc"](len + 1); // if we don't do +1 this library won't copy the last character
49+
var inUtf8 = Module["_malloc"](len + 1); // if we don't do +1 this library won't copy the last character
3550
Module["stringToUTF8"](s, inUtf8, len + 1); // since it always also needs a space for a 0-char
3651

3752
var MaxOutLength = (len << 1) + 1; // worst case every character is a token
38-
var outUtf8 = Module["stackAlloc"](MaxOutLength);
53+
var outUtf8 = Module["_malloc"](MaxOutLength);
3954

40-
var actualLen = Module["_TextToSentences"](inUtf8, len, outUtf8, MaxOutLength);
41-
if(0 > actualLen || actualLen > MaxOutLength) {
42-
return null;
55+
try
56+
{
57+
var actualLen = Module["_TextToSentences"](inUtf8, len, outUtf8, MaxOutLength);
58+
if(0 > actualLen || actualLen > MaxOutLength) {
59+
return null;
60+
}
61+
}
62+
finally
63+
{
64+
if (inUtf8 != 0)
65+
{
66+
Module["_free"](inUtf8);
67+
}
68+
69+
if (outUtf8 != 0)
70+
{
71+
Module["_free"](outUtf8);
72+
}
4373
}
4474

4575
return Module["UTF8ToString"](outUtf8);
@@ -108,33 +138,48 @@ export function TextToIds(handle, s, max_len, unk = 0) {
108138

109139
// convert input JS string to UTF-8
110140
var len = Module["lengthBytesUTF8"](s);
111-
var inUtf8 = Module["stackAlloc"](len + 1); // if we don't do +1 this library won't copy the last character
141+
var inUtf8 = Module["_malloc"](len + 1); // if we don't do +1 this library won't copy the last character
112142
Module["stringToUTF8"](s, inUtf8, len + 1); // since it always also needs a space for a 0-char
113143

114144
// allocate space for ids on stack (it is faster and we don't need to delete it manualy)
115145
var MaxOutLength = max_len;
116-
var IdsOut = Module["stackAlloc"](MaxOutLength * 4); // sizeof(int)
146+
var IdsOut = Module["_malloc"](MaxOutLength * 4); // sizeof(int)
117147

118-
// get the IDS from BlingFire
119-
var actualLen = Module["_TextToIds"](h, inUtf8, len, IdsOut, MaxOutLength, unk);
120-
if(0 >= actualLen) {
121-
return null;
122-
}
148+
try
149+
{
150+
// get the IDS from BlingFire
151+
var actualLen = Module["_TextToIds"](h, inUtf8, len, IdsOut, MaxOutLength, unk);
152+
if(0 >= actualLen) {
153+
return null;
154+
}
123155

124-
// get the smallest between actualLen and MaxOutLength
125-
var actualLenOrMax = actualLen < MaxOutLength ? actualLen : MaxOutLength;
156+
// get the smallest between actualLen and MaxOutLength
157+
var actualLenOrMax = actualLen < MaxOutLength ? actualLen : MaxOutLength;
126158

127-
// read bytes
128-
var tmp = Module["HEAPU8"].subarray(IdsOut, IdsOut + (actualLenOrMax * 4));
159+
// read bytes
160+
var tmp = Module["HEAPU8"].subarray(IdsOut, IdsOut + (actualLenOrMax * 4));
129161

130-
// allocate JS array
131-
var ids = new Int32Array(actualLenOrMax);
162+
// allocate JS array
163+
var ids = new Int32Array(actualLenOrMax);
132164

133-
// decode bytes into int array (I could not find how to make a cast here)
134-
var i = 0;
135-
var j = 0;
136-
for(; i < actualLenOrMax; i++, j+=4) {
137-
ids[i] = tmp[j] + (tmp[j + 1] * 256) + (tmp[j + 2] * 65536) + (tmp[j + 3] * 16777216);
165+
// decode bytes into int array (I could not find how to make a cast here)
166+
var i = 0;
167+
var j = 0;
168+
for(; i < actualLenOrMax; i++, j+=4) {
169+
ids[i] = tmp[j] + (tmp[j + 1] * 256) + (tmp[j + 2] * 65536) + (tmp[j + 3] * 16777216);
170+
}
171+
}
172+
finally
173+
{
174+
if (inUtf8 != 0)
175+
{
176+
Module["_free"](inUtf8);
177+
}
178+
179+
if (IdsOut != 0)
180+
{
181+
Module["_free"](IdsOut);
182+
}
138183
}
139184

140185
return ids;
@@ -153,15 +198,30 @@ export function WordHyphenation(handle, s, hyp = 0x2D) {
153198

154199
var len = Module["lengthBytesUTF8"](s);
155200

156-
var inUtf8 = Module["stackAlloc"](len + 1); // if we don't do +1 this library won't copy the last character
201+
var inUtf8 = Module["_malloc"](len + 1); // if we don't do +1 this library won't copy the last character
157202
Module["stringToUTF8"](s, inUtf8, len + 1); // since it always also needs a space for a 0-char
158203

159204
var MaxOutLength = (len << 1) + 1; // worst case hyphen after every character
160-
var outUtf8 = Module["stackAlloc"](MaxOutLength);
205+
var outUtf8 = Module["_malloc"](MaxOutLength);
161206

162-
var actualLen = Module["_WordHyphenationWithModel"](inUtf8, len, outUtf8, MaxOutLength, h, hyp);
163-
if(0 > actualLen || actualLen > MaxOutLength) {
164-
return null;
207+
try
208+
{
209+
var actualLen = Module["_WordHyphenationWithModel"](inUtf8, len, outUtf8, MaxOutLength, h, hyp);
210+
if(0 > actualLen || actualLen > MaxOutLength) {
211+
return null;
212+
}
213+
}
214+
finally
215+
{
216+
if (inUtf8 != 0)
217+
{
218+
Module["_free"](inUtf8);
219+
}
220+
221+
if (outUtf8 != 0)
222+
{
223+
Module["_free"](outUtf8);
224+
}
165225
}
166226

167227
return Module["UTF8ToString"](outUtf8);

0 commit comments

Comments
 (0)