switch to use external REPP. Delete boost escape function from utilit…

…y, since it

is included in REPP.


git-svn-id: https://pet.opendfki.de/repos/pet/main@929 4200e16c-5112-0410-ac55-d7fb557a720a

cheap/Makefile.am

@@ -9,7 +9,7 @@ endif
 include $(top_srcdir)/common/Makefile.common
 
 # Add cheap-specific CPPFLAGS/CXXFLAGS/LDFLAGS/LIBS as determined by configure
-CPPFLAGS += -I$(top_srcdir)/common -I$(top_srcdir)/fspp @CHEAPCPPFLAGS@
+CPPFLAGS += -I$(top_srcdir)/common -I$(top_srcdir)/fspp -I$(top_srcdir)/repp @CHEAPCPPFLAGS@
 CXXFLAGS += @CHEAPCXXFLAGS@
 LDFLAGS += @CHEAPLDFLAGS@
 LIBS += @CHEAPLIBS@
@@ -82,7 +82,9 @@ cheaplibsources = \
 	vpm.h vpm.cpp \
 	pcfg.h pcfg.cpp \
 	yy-tokenizer.cpp yy-tokenizer.h \
-	repp.cpp repp.h \
+	$(top_srcdir)/repp/repp.cpp $(top_srcdir)/repp/repp.h \
+	$(top_srcdir)/repp/tdl_options.cpp $(top_srcdir)/repp/tdl_options.h \
+	$(top_srcdir)/repp/repp_from_pet.cpp $(top_srcdir)/repp/repp_from_pet.h \
 	tagger.cpp tagger.h \
 	$(top_srcdir)/common/bitcode.cpp \
 	$(top_srcdir)/common/chunk-alloc.cpp \

cheap/cheap.cpp

@@ -33,6 +33,7 @@
 #include "yy-tokenizer.h"
 #ifdef HAVE_BOOST_REGEX_ICU_HPP
 #include "repp.h"
+#include "repp_from_pet.h"
 #endif
 #include "tagger.h"
 #include "lingo-tokenizer.h"
@@ -647,7 +648,15 @@ bool load_grammar(string initial_name) {
    case TOKENIZER_REPP:
 #ifdef HAVE_BOOST_REGEX_ICU_HPP
     {
-      tok = new tReppTokenizer();
+      string name = get_opt_string("opt_repp");
+      settings *reppsettings;
+      if(!name.empty()) {
+        reppsettings = new settings(name, cheap_settings->base(), "reading");
+        if(!reppsettings->valid())
+          throw tError("Unable to read REPP configuration '" + name + "'.");
+        cheap_settings->install(reppsettings);
+      }
+      tok = createReppTokenizer(cheap_settings);
     }
     break;
 #else

common/utility.cpp

@@ -509,41 +509,3 @@ socket_readline(int socket, char *string, int length, bool ffp) {
 
 } /* socket_readline() */
 
-#ifndef FLOP
-#ifdef HAVE_BOOST_REGEX_ICU_HPP
-/* escape literal braces in regex before giving it to Boost, since Boost::Regex
- * is not completely pcre compatible
- */
-string boostescape(string esc) 
-{
-  boost::smatch res;
-  boost::u32regex quantifier 
-    = boost::make_u32regex("^([{](?:[0-9]+(?:,[0-9]*)?)[}])");
-  int len = esc.length();
-  for (int x=0; x < len; ++x) {
-    if (esc.at(x) == '{') {
-      if (x > 0 && esc.at(x-1) == '\\') 
-        continue; //already escaped, keep going
-      if (x > 1 && esc.at(x-2) == '\\' && isalpha(esc.at(x-1))) {
-        //probably the start of a backslash seq. find and skip past the 
-        //closing brace. If not a backslash seq, ambiguous - let boost complain
-        while (x < len && esc.at(x) != '}')
-          x++;
-      } else if (boost::u32regex_search(esc.substr(x), res, quantifier)) {
-        //quantifier, skip past closing brace
-        x += string(res[0]).length()-1;
-      } else {
-        esc.replace(x,1, "\\{", 2);
-        x++;
-        len++;
-      }
-    } else if (esc.at(x) == '}') {
-      esc.replace(x,1, "\\}", 2);
-      x++;
-      len++;
-    }
-  }
-  return esc;
-}
-#endif
-#endif

common/utility.h

@@ -196,11 +196,4 @@ int socket_readline(int, char *, int, bool = false);
 void print_borland_heap(FILE *f);
 #endif
 
-#ifndef FLOP
-#ifdef HAVE_BOOST_REGEX_ICU_HPP
-/* work around pcre incompatibility in Boost::Regex */
-std::string boostescape(std::string esc);
-#endif
-#endif
-
 #endif