Changeset 243d027

Timestamp:

01/18/13 18:46:38 (12 years ago)

Author:

Tomasz Obrebski <to@…>

Branches:

master

Children:

e0cd003

Parents:

18e1952

git-author:

Tomasz Obrebski <to@…> (01/18/13 18:46:38)

git-committer:

Tomasz Obrebski <to@…> (01/18/13 18:46:38)

Message:

tok accepts options and prints help message now

Files:

• Makefile (modified) (1 diff)
• src/tok.l/Makefile (modified) (1 diff)
• src/tok.l/tok.l (modified) (1 diff)
• src/tok.l/tok_cmdline.ggo (modified) (1 diff)

Makefile

@@ -100 +100 @@
 install_share: share
         cd $(CUR_DIR)/share && make install; cd $(CUR_DIR)
-        echo grammar compilation...
+        #grammar compilation...
         dgc
 

src/tok.l/Makefile

@@ -9 +9 @@
 
 tok: lex.yy.c
-        $(CC) $(CFLAGS) -lfl -o tok lex.yy.c $(LDFLAGS)
+        $(CC) $(CFLAGS) -o tok lex.yy.c tok_cmdline.c -lfl $(LDFLAGS)
 
 lex.yy.c: tok_cmdline.h tok_cmdline.c
         $(FLEX) tok.l
 
-tok_cmdline.h tok_cmdline.c:
+tok_cmdline.h tok_cmdline.c: tok_cmdline.ggo
         $(GENGETOPT) -i tok_cmdline.ggo --conf-parser --file=tok_cmdline
 

src/tok.l/tok.l

@@ -56 +56 @@
 %%
 
-/*int main(int argc, char** argv)
+int main(int argc, char** argv)
 {
-  if (cmdline_parser(argc, argv, &args) != 0) return 1;
+  if (cmdline_parser(argc, argv, &args) != 0) exit(1);
   setlocale(LC_CTYPE,"");
+
   setlocale(LC_COLLATE,"");
   yylex();
   return 0;
-}*/
+}
 
 int yywrap()

src/tok.l/tok_cmdline.ggo

@@ -1 +1 @@
 package "tok"
 version "0.1"
+usage   "tok [OPTIONS]"
+purpose "tok transforms raw text into UTT format."
 
-option "interactive"            i       "Interactive mode." flag off
+description "OPTIONS"
+
+option "interactive"            i       "Interactive mode (no output buffering)." flag off
+
+text "
+DESCRIPTION
+
+tok reads from standard input, identifies tokens on the basis of their orthographic form and writes a sequence of segments in UTT format to
+the standard output. The type of the token is printed as the type field.
+
+OUTPUT FORMAT
+
+UTT-file with four fields: start, length, type, and form. In the type field five types of tokens are distinguished:
+
+  W (word) - continuous sequence of letters
+  N (number) - continuous sequence of digits
+  S (space) - continuous sequence of space characters
+  P (punctuation) - single printable character other than W, N, S
+  B (unprintable character) - single unprintable character
+
+USAGE EXAMPLE
+
+      tok
+"