projects
/
SXSI
/
TextCollection.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Implementation of LZ-index for extracting text
[SXSI/TextCollection.git]
/
TextCollectionBuilder.cpp
diff --git
a/TextCollectionBuilder.cpp
b/TextCollectionBuilder.cpp
index
1d49f39
..
7228f72
100644
(file)
--- a/
TextCollectionBuilder.cpp
+++ b/
TextCollectionBuilder.cpp
@@
-43,6
+43,9
@@
TextCollectionBuilder::TextCollectionBuilder(unsigned samplerate, ulong estimate
p_->numberOfSamples = 0;
// Current params: 8 bytes, no samples, buffer size n/10 bytes.
p_->numberOfSamples = 0;
// Current params: 8 bytes, no samples, buffer size n/10 bytes.
+ // Buffer size is always at least 15MB:
+ if (estimatedInputLength < TEXTCOLLECTION_DEFAULT_INPUT_LENGTH)
+ estimatedInputLength = TEXTCOLLECTION_DEFAULT_INPUT_LENGTH;
p_->sa = new CSA::RLCSABuilder(8, 0, estimatedInputLength/10);
assert(p_->sa->isOk());
p_->sa = new CSA::RLCSABuilder(8, 0, estimatedInputLength/10);
assert(p_->sa->isOk());