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;
+    bvalue = false;
-    if (getConfParam("nonumbers", &nonum) && nonum == true) {
+    if (getConfParam("nonumbers", &bvalue) && bvalue == true) {
 	TextSplit::noNumbers();
     }
 
-    bool fnmpathname = true;
+    bvalue = false;
-    if (getConfParam("skippedPathsFnmPathname", &fnmpathname)
+    if (getConfParam("dehyphenate", &bvalue)) {
-	&& fnmpathname == false) {
+	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, 

src/doc/user/usermanual.xml

@@ -5591,6 +5591,15 @@ skippedPaths = ~/somedir/*.txt
             </listitem>
           </varlistentry>
 
+          <varlistentry><term><varname>dehyphenate</varname></term>
+            <listitem><para>Determines if, given an input of
+            <literal>co-worker</literal>, we add a term for
+            <literal>coworker</literal>. This possibility is new in version
+            1.22, and on by default. Setting the variable to off allows
+            restoring the previous behaviour.</para>
+            </listitem>
+          </varlistentry>
+
           <varlistentry><term><varname>nocjk</varname></term>
             <listitem><para>If this set to true, specific east asian
             (Chinese Korean Japanese) characters/word splitting is

src/sampleconf/recoll.conf.in

@@ -134,6 +134,13 @@ indexstemminglanguages = english
 # are not performed by unac, but I cant imagine someone typing the composed
 # forms in a search.
 unac_except_trans = ßss œoe Œoe æae Æae ffff fifi flfl
+# Turn off the indexing of numbers: may reduce the index size if you have
+# no use for them 
+# nonumbers = 0
+
+# Turn off indexing "coworker" for an input of "co-worker" (in addition to
+# co, worker, "co worker". Default is on as of version 1.22
+# dehyphenate = 1
 
 # Maximum expansion count for a single term (ie: when using wildcards).
 # We used to not limit this at all (except for filenames where the limit