add capability to remember page breaks generated by, e.g. pdftotext, and use them to start an external viewer on a match page

src/common/textsplit.cpp

@@ -86,7 +86,7 @@ public:
 	for (i = 0; i  < strlen(wild); i++)
 	    charclasses[int(wild[i])] = WILD;
 
-	char special[] = ".@+-,#'_\n\r";
+	char special[] = ".@+-,#'_\n\r\f";
 	for (i = 0; i  < strlen(special); i++)
 	    charclasses[int(special[i])] = special[i];
 
@@ -316,6 +316,7 @@ bool TextSplit::text_to_words(const string &in)
     m_inNumber = false;
     m_wordStart = m_wordLen = m_prevpos = m_prevlen = m_wordpos = m_spanpos = 0;
     int curspanglue = 0;
+    bool pagepending = false;
 
     // Running count of non-alphanum chars. Reset when we see one;
     int nonalnumcnt = 0;
@@ -369,6 +370,10 @@ bool TextSplit::text_to_words(const string &in)
 		    return false;
 		m_inNumber = false;
 	    }
+	    if (pagepending) {
+		pagepending = false;
+		newpage(m_wordpos);
+	    }
 	    break;
 
 	case WILD:
@@ -521,7 +526,10 @@ bool TextSplit::text_to_words(const string &in)
 		goto SPACE;
 	    }
 	    break;
-
+	case '\f':
+	    pagepending = true;
+	    goto SPACE;
+	    break;
 #ifdef RCL_SPLIT_CAMELCASE
             // Camelcase handling. 
             // If we get uppercase ascii after lowercase ascii, emit word.

src/common/textsplit.h

@@ -20,10 +20,8 @@
 #include <string>
 #include <vector>
 
-#ifndef NO_NAMESPACES
 using std::string;
 using std::vector;
-#endif
 
 class Utf8Iter;
 
@@ -78,6 +76,12 @@ public:
 			  int bte   // byte offset of first char after term
 			  ) = 0; 
 
+    /** Called when we encounter formfeed \f 0x0c. Override to use the event.
+     * Mostly or exclusively used with pdftoxx output. Other filters mostly 
+     * just don't know about pages. */
+    virtual void newpage(int /*pos*/)
+    {
+    }
 
     // Static utility functions: