more int fixups
--HG-- branch : WINDOWSPORT
This commit is contained in:
@@ -223,7 +223,7 @@ inline bool TextSplit::emitterm(bool isspan, string &w, int pos,
|
||||
{
|
||||
LOGDEB2(("TextSplit::emitterm: [%s] pos %d\n", w.c_str(), pos));
|
||||
|
||||
size_t l = w.length();
|
||||
int l = int(w.length());
|
||||
|
||||
#ifdef TEXTSPLIT_STATS
|
||||
// Update word length statistics. Do this before we filter out
|
||||
@@ -232,7 +232,7 @@ inline bool TextSplit::emitterm(bool isspan, string &w, int pos,
|
||||
m_stats.newsamp(m_wordChars);
|
||||
#endif
|
||||
|
||||
if (l > 0 && l < (unsigned)m_maxWordLength) {
|
||||
if (l > 0 && l < m_maxWordLength) {
|
||||
// 1 byte word: we index single ascii letters and digits, but
|
||||
// nothing else. We might want to turn this into a test for a
|
||||
// single utf8 character instead ?
|
||||
|
||||
@@ -184,7 +184,7 @@ private:
|
||||
// Current span. Might be jf.dockes@wanadoo.f
|
||||
std::string m_span;
|
||||
|
||||
std::vector <std::pair<unsigned int, unsigned int> > m_words_in_span;
|
||||
std::vector <std::pair<int, int> > m_words_in_span;
|
||||
|
||||
// Current word: no punctuation at all in there. Byte offset
|
||||
// relative to the current span and byte length
|
||||
|
||||
Reference in New Issue
Block a user