Make dehyphenate (co-worker->coworker) optional

src/common/rclconfig.cpp

@@ -321,8 +321,9 @@ bool RclConfig::updateMainConfig()
     m_mdrstate.init(m_conf);
 
     setKeyDir(cstr_null);
-    bool nocjk = false;
-    if (getConfParam("nocjk", &nocjk) && nocjk == true) {
+
+    bool bvalue = false;
+    if (getConfParam("nocjk", &bvalue) && bvalue == true) {
 	TextSplit::cjkProcessing(false);
     } else {
 	int ngramlen;
@@ -333,14 +334,18 @@ bool RclConfig::updateMainConfig()
 	}
     }
 
-    bool nonum = false;
-    if (getConfParam("nonumbers", &nonum) && nonum == true) {
+    bvalue = false;
+    if (getConfParam("nonumbers", &bvalue) && bvalue == true) {
 	TextSplit::noNumbers();
     }
 
-    bool fnmpathname = true;
-    if (getConfParam("skippedPathsFnmPathname", &fnmpathname)
-	&& fnmpathname == false) {
+    bvalue = false;
+    if (getConfParam("dehyphenate", &bvalue)) {
+	TextSplit::deHyphenate(bvalue);
+    }
+
+    bvalue = true;
+    if (getConfParam("skippedPathsFnmPathname", &bvalue) && bvalue == false) {
 	FsTreeWalker::setNoFnmPathname();
     }
 

src/common/textsplit.cpp

@@ -212,6 +212,7 @@ bool TextSplit::isCJK(int c)
 bool          TextSplit::o_processCJK = true;
 unsigned int  TextSplit::o_CJKNgramLen = 2;
 bool          TextSplit::o_noNumbers = false;
+bool          TextSplit::o_deHyphenate = false;
 
 // Final term checkpoint: do some checking (the kind which is simpler
 // to do here than in the main loop), then send term to our client.
@@ -309,7 +310,8 @@ bool TextSplit::words_from_span(int bp)
     // Byte position of the span start
     int spboffs = bp - m_span.size();
 
-    if (spanwords == 2 && m_span[m_words_in_span[0].second] == '-') {
+    if (o_deHyphenate && spanwords == 2 && 
+	m_span[m_words_in_span[0].second] == '-') {
 	unsigned int s0 = m_words_in_span[0].first;
 	unsigned int l0 = m_words_in_span[0].second - m_words_in_span[0].first;
 	unsigned int s1 = m_words_in_span[1].first;

src/common/textsplit.h

@@ -52,6 +52,13 @@ public:
 	o_noNumbers = true;
     }
 
+    // Given [co-worker] as input, do we also generate [coworker] ?
+    // Set by rclconfig
+    static bool o_deHyphenate;
+    static void deHyphenate(bool on) {
+	o_deHyphenate = on;
+    }
+
     enum Flags {
         // Default: will return spans and words (a_b, a, b)
         TXTS_NONE = 0,