sen component added

configure

@@ -8,7 +8,7 @@ CUR_DIR=$(pwd)
 SRC_DIR="${CUR_DIR}/src"
 
 # list of components to compile
-CMPLIST="compiledic cor dgp fla gph grp gue kon kor kot lem mar rm12 rs12 sen-nl ser tags tok.c tok.l tok.pl unfla"
+CMPLIST="compiledic cor dgp fla gph grp gue kon kor kot lem mar rm12 rs12 sen sen-nl ser tags tok.c tok.l tok.pl unfla"
 COMP=
 
 

src/sen/Makefile

@@ -0,0 +1,32 @@
+include ../../config.mak
+
+ifeq ($(BUILD_STATIC), yes)
+  LDFLAGS += -static
+endif
+
+LDFLAGS +=
+CFLAGS += -O2
+
+sen: lex.yy.c
+	$(CC) $(CFLAGS) -o sen lex.yy.c -lfl $(LDFLAGS)
+
+lex.yy.c: sen.l
+	$(FLEX) sen.l
+
+.PHONY: install
+install:
+ifdef BIN_DIR
+	install -m 0755 sen $(BIN_DIR)
+endif
+
+.PHONY: uninstall
+uninstall:
+ifdef BIN_DIR
+	rm $(BIN_DIR)/sen
+endif
+
+clean: clean.flex
+	rm sen || true
+
+clean.flex:
+	rm lex.yy.c || true

src/sen/sen.l

@@ -0,0 +1,97 @@
+%{
+
+  void print_EOS(void);
+  void set_position(void);
+
+  int pos=0;
+  int len=0;
+
+%}
+
+ul      [A-Z¡ÆÊ£ÑÓ¦¯¬]
+ll      [a-z±æê³ñ󶿼]
+l       ul|ll
+n       [0-9]+
+s       [ \t]+
+
+
+ab1     (mgr|in¿|prof|hab|doc|dyr|kier|zast)
+ab2     (ul|pl|al)
+ab3	(str|ryc|rys)
+ab4	(np|tzw)
+
+abrv    ({ab1}|{ab2}|{ab3})
+
+SEG     .*\n
+N       {n}{s}{n}{s}N{s}.*\n
+S       {n}{s}{n}{s}S{s}.*\n
+P       {n}{s}{n}{s}P{s}.*\n
+W       {n}{s}{n}{s}W{s}.*\n
+UL      {n}{s}{n}{s}W{s}{ul}.*\n
+Cap     {n}{s}{n}{s}W{s}{ul}{ll}*.*\n
+POINT   {n}{s}{n}{s}P{s}\.({s}.*)?\n
+QMARK   {n}{s}{n}{s}P{s}\?({s}.*)?\n
+EXCL    {n}{s}{n}{s}P{s}\!({s}.*)?\n
+DASH    {n}{s}{n}{s}P{s}\-({s}.*)?\n
+POINTS  {POINT}+
+
+ABRV    {n}{s}{n}{s}W{s}{abrv}({s}.*)?\n
+
+EOS     {POINT}|{POINTS}|{QMARK}|{EXCL}
+
+
+%%
+
+
+{N}({POINT}{N})+          { ECHO; set_position(); }
+({UL}{POINT}{S}?)+{Cap}   { ECHO; set_position(); }
+{ABRV}{POINT}             { ECHO; set_position(); }
+
+
+{P}/{S}{DASH}             { ECHO; set_position(); print_EOS(); }
+{EOS}/{S}({Cap}|{P}|{N})  { ECHO; set_position(); print_EOS(); }
+
+.*                        { ECHO; set_position(); }
+
+<<EOF>>                   { printf("%04d 00 EOS *\n",pos+len); exit(1); }
+
+%%
+
+int main()
+{
+  printf("0000 00 BOS *\n");
+  yylex();
+  return 0;
+}
+
+int yywrap()
+{
+  return 1;
+}
+
+void set_position()
+{
+  char *lastseg;
+  char *tmp;
+
+  yytext[yyleng-1] = '\0';
+
+  tmp = strrchr(yytext, '\n');
+  if( tmp )
+  {
+    lastseg = tmp + 1;
+  }
+  else
+  {
+    lastseg = yytext;
+  }
+
+  sscanf(lastseg, "%d %d", &pos, &len);
+  yytext[yyleng-1] = '\n';
+
+}
+
+void print_EOS()
+{
+  printf("%04d 00 EOS *\n%04d 00 BOS *\n", pos+len, pos+len);
+}

src/ser/ser

@@ -11,7 +11,7 @@ use File::Temp;
 use File::HomeDir;
 
 my $LIB_DIR="/usr/local/lib/utt";
-my $systemconfigfile='/usr/local/etc/utt/ser.conf';
+my $systemconfigfile='/etc/utt/ser.conf';
 my $userconfigfile=home()."/.utt/ser.conf";
 
 Getopt::Long::Configure('no_ignore_case_always');