etisserant@0: /*
etisserant@0:  * (c) 2003 Mario de Sousa
etisserant@0:  *
etisserant@0:  * Offered to the public under the terms of the GNU General Public License
etisserant@0:  * as published by the Free Software Foundation; either version 2 of the
etisserant@0:  * License, or (at your option) any later version.
etisserant@0:  *
etisserant@0:  * This program is distributed in the hope that it will be useful, but
etisserant@0:  * WITHOUT ANY WARRANTY; without even the implied warranty of
etisserant@0:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
etisserant@0:  * Public License for more details.
etisserant@0:  *
etisserant@0:  * This code is made available on the understanding that it will not be
etisserant@0:  * used in safety-critical situations without a full and competent review.
etisserant@0:  */
etisserant@0: 
etisserant@0: /*
etisserant@0:  * An IEC 61131-3 IL and ST compiler.
etisserant@0:  *
etisserant@0:  * Based on the
etisserant@0:  * FINAL DRAFT - IEC 61131-3, 2nd Ed. (2001-12-10)
etisserant@0:  *
etisserant@0:  */
etisserant@0: 
etisserant@0: /*
etisserant@0:  * Stage 1
etisserant@0:  * =======
etisserant@0:  *
etisserant@0:  * This file contains the lexical tokens definitions, from which
etisserant@0:  * the flex utility will generate a lexical parser function.
etisserant@0:  */
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /*****************************/
etisserant@0: /* Lexical Parser Options... */
etisserant@0: /*****************************/
etisserant@0: 
etisserant@0: /* The lexical analyser will never work in interactive mode,
etisserant@0:  * i.e., it will only process programs saved to files, and never
etisserant@0:  * programs being written inter-actively by the user.
etisserant@0:  * This option saves the resulting parser from calling the
etisserant@0:  * isatty() function, that seems to be generating some compile
etisserant@0:  * errors under some (older?) versions of flex.
etisserant@0:  */
etisserant@0: %option never-interactive
etisserant@0: 
etisserant@0: /* Have the lexical analyser use a 'char *yytext' instead of an
etisserant@0:  * array of char 'char yytext[??]' to store the lexical token.
etisserant@0:  */
etisserant@0: %pointer
etisserant@0: 
etisserant@0: 
etisserant@0: /* Have the lexical analyser ignore the case of letters.
etisserant@0:  * This will occur for all the tokens and keywords, but
etisserant@0:  * the resulting text handed up to the syntax parser
etisserant@0:  * will not be changed, and keep the original case
etisserant@0:  * of the letters in the input file.
etisserant@0:  */
etisserant@0: %option case-insensitive
etisserant@0: 
etisserant@0: /* Have the generated lexical analyser keep track of the
etisserant@0:  * line number it is currently analysing.
etisserant@0:  * This is used to pass up to the syntax parser
etisserant@0:  * the number of the line on which the current
etisserant@0:  * token was found. It will enable the syntax parser
etisserant@0:  * to generate more informatve error messages...
etisserant@0:  */
etisserant@0: %option yylineno
etisserant@0: 
etisserant@0: /* required for the use of the yy_pop_state() and
etisserant@0:  * yy_push_state() functions
etisserant@0:  */
etisserant@0: %option stack
etisserant@0: 
etisserant@0: /* The '%option stack' also requests the inclusion of 
etisserant@0:  * the yy_top_state(), however this function is not
etisserant@0:  * currently being used. This means that the compiler
etisserant@0:  * is complaining about the existance of this function.
etisserant@0:  * The following option removes the yy_top_state()
etisserant@0:  * function from the resulting c code, so the compiler 
etisserant@0:  * no longer complains.
etisserant@0:  */
etisserant@0: %option noyy_top_state
etisserant@0: 
etisserant@0: /**************************************************/
etisserant@0: /* External Variable and Function declarations... */
etisserant@0: /**************************************************/
etisserant@0: 
etisserant@0: 
etisserant@0: %{
etisserant@0: /* Define TEST_MAIN to include a main() function.
etisserant@0:  * Useful for testing the parser generated by flex.
etisserant@0:  */
etisserant@0: /*
etisserant@0: #define TEST_MAIN
etisserant@0: */
etisserant@0: /* If lexical parser is compiled by itself, we need to define the following
etisserant@0:  * constant to some string. Under normal circumstances LIBDIRECTORY is set
etisserant@0:  * in the syntax parser header file...
etisserant@0:  */
etisserant@0: #ifdef TEST_MAIN
etisserant@0: #define LIBDIRECTORY "just_testing"
etisserant@0: #endif
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /* Required for strdup() */
etisserant@0: #include <string.h>
etisserant@0: 
etisserant@0: /* Required only for the declaration of abstract syntax classes
etisserant@0:  * (class symbol_c; class token_c; class list_c;)
etisserant@0:  * These will not be used in flex, but the token type union defined
etisserant@0:  * in iec.hh contains pointers to these classes, so we must include
etisserant@0:  * it here.
etisserant@0:  */
etisserant@0: #include "../absyntax/absyntax.hh"
etisserant@0: 
mario@15: 
etisserant@0: /* generated by bison.
etisserant@0:  * Contains the definition of the token constants, and the
etisserant@0:  * token value type YYSTYPE (in our case, a 'const char *')
etisserant@0:  */
etisserant@0: #include "iec.y.hh"
mario@15: #include "stage1_2_priv.hh"
mario@15: 
etisserant@0: 
etisserant@0: /* Variable defined by the bison parser,
etisserant@0:  * where the value of the tokens will be stored
etisserant@0:  */
etisserant@0: extern YYSTYPE yylval;
etisserant@0: 
etisserant@0: /* The name of the file currently being parsed...
etisserant@0:  * This variable is declared and read from the code generated by bison!
etisserant@0:  * Note that flex accesses and updates this global variable
etisserant@0:  * apropriately whenever it comes across an (*#include <filename> *)
etisserant@0:  * directive...
etisserant@0:  */
mario@15: /*
mario@15:  NOTE: already defined in iec.y.hh 
etisserant@0: extern const char *current_filename;
mario@15: */
mario@15: 
etisserant@0: 
etisserant@0: /* We will not be using unput() in our flex code... */
etisserant@0: #define YY_NO_UNPUT
etisserant@0: 
etisserant@0: /* Variable defined by the bison parser.
etisserant@0:  * It must be initialised with the location
etisserant@0:  * of the token being parsed.
etisserant@0:  * This is only needed if we want to keep
etisserant@0:  * track of the locations, in order to give
etisserant@0:  * more meaningful error messages!
etisserant@0:  */
etisserant@0: extern YYLTYPE yylloc;
etisserant@0: 
etisserant@0: /* Macro that is executed for every action.
etisserant@0:  * We use it to pass the location of the token
etisserant@0:  * back to the bison parser...
etisserant@0:  */
etisserant@0: #define YY_USER_ACTION { 					\
etisserant@0: 	yylloc.first_line = yylloc.last_line = yylineno;	\
etisserant@0: 	yylloc.first_column = yylloc.last_column = 0;		\
etisserant@0: 	}
etisserant@0: 
etisserant@0: 
etisserant@0: /* Since this lexical parser we defined only works in ASCII based
etisserant@0:  * systems, we might as well make sure it is being compiled on
etisserant@0:  * one...
etisserant@0:  * Lets check a few random characters...
etisserant@0:  */
etisserant@0: #if (('a' != 0x61) || ('A' != 0x41) || ('z' != 0x7A) || ('Z' != 0x5A) || \
etisserant@0:      ('0' != 0x30) || ('9' != 0x39) || ('(' != 0x28) || ('[' != 0x5B))
etisserant@0: #error This lexical analyser is not portable to a non ASCII based system.
etisserant@0: #endif
etisserant@0: 
etisserant@0: 
etisserant@0: /* Function only called from within flex, but defined
etisserant@0:  * in iec.y!
lbessard@3:  * We declare it here...
etisserant@0:  *
etisserant@0:  * Search for a symbol in either of the two symbol tables
etisserant@0:  * and return the token id of the first symbol found.
etisserant@0:  * Searches first in the variables, and only if not found
etisserant@0:  * does it continue searching in the library elements
etisserant@0:  */
etisserant@0: //token_id_t get_identifier_token(const char *identifier_str);
etisserant@0: int get_identifier_token(const char *identifier_str);
etisserant@0: %}
etisserant@0: 
etisserant@0: 
etisserant@0: /***************************************************/
etisserant@0: /* Forward Declaration of functions defined later. */
etisserant@0: /***************************************************/
etisserant@0: 
etisserant@0: %{
etisserant@0: /* return all the text in the current token back to the input stream. */
etisserant@0: void unput_text(unsigned int n);
etisserant@0: %}
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /****************************/
etisserant@0: /* Lexical Parser States... */
etisserant@0: /****************************/
etisserant@0: 
etisserant@0: /* NOTE: Our psrser can parse st or il code, intermixed
etisserant@0:  *       within the same file.
etisserant@0:  *       With IL we come across the issue of the EOL (end of line) token.
etisserant@0:  *       ST, and the declaration parts of IL do not use this token!
etisserant@0:  *       If the lexical analyser were to issue this token during ST
etisserant@0:  *       language parsing, or during the declaration of data types,
etisserant@0:  *       function headers, etc. in IL, the syntax parser would crash.
etisserant@0:  *
etisserant@0:  *       We can solve this issue using one of three methods:
etisserant@0:  *        (1) Augment all the syntax that does not accept the EOL
etisserant@0:  *            token to simply ignore it. This makes the syntax
etisserant@0:  *            definition (in iec.y) very cluttered!
etisserant@0:  *        (2) Let the lexical parser figure out which language
etisserant@0:  *            it is parsing, and decide whether or not to issue
etisserant@0:  *            the EOL token. This requires the lexical parser
etisserant@0:  *            to have knowledge of the syntax!, making for a poor
etisserant@0:  *            overall organisation of the code. It would also make it
etisserant@0:  *            very difficult to understand the lexical parser as it
etisserant@0:  *            would use several states, and a state machine to transition
etisserant@0:  *            between the states. The state transitions would be
etisserant@0:  *            intermingled with the lexical parser defintion!
etisserant@0:  *        (3) Use a mixture of (1) and (2). The lexical analyser
etisserant@0:  *            merely distinguishes between function headers and function
etisserant@0:  *            bodies, but no longer makes a distinction between il and
etisserant@0:  *            st language bodies. When parsing a body, it will return
etisserant@0:  *            the EOL token. In other states '\n' will be ignored as
etisserant@0:  *            whitespace.
etisserant@0:  *            The ST language syntax has been augmented in the syntax
etisserant@0:  *            parser configuration to ignore any EOL tokens that it may
etisserant@0:  *            come across!
etisserant@0:  *            This option has both drawbacks of option (1) and (2), but
etisserant@0:  *            much less intensely.
etisserant@0:  *            The syntax that gets cluttered is limited to the ST statements
etisserant@0:  *            (which is rather limited, compared to the function headers and
etisserant@0:  *            data type declarations, etc...), while the state machine in
etisserant@0:  *            the lexical parser becomes very simple. All state transitions
etisserant@0:  *            can be handled within the lexical parser by itself, and can be
etisserant@0:  *            easily identified. Thus knowledge of the syntax required by
etisserant@0:  *            the lexical parser is very limited!
etisserant@0:  *
etisserant@0:  * Amazingly enough, I (Mario) got to implement option (3)
etisserant@0:  * at first, requiring two basic states, decl and body.
etisserant@0:  * The lexical parser will enter the body state when
etisserant@0:  * it is parsing the body of a function/program/function block. The
etisserant@0:  * state transition is done when we find a VAR_END that is not followed
etisserant@0:  * by a VAR! This is the syntax knowledge that gets included in the
etisserant@0:  * lexical analyser with this option!
etisserant@0:  * Unfortunately, getting the st syntax parser to ignore EOL anywhere
etisserant@0:  * where they might appear leads to conflicts. This is due to the fact
etisserant@0:  * that the syntax parser uses the single look-ahead token to remove
etisserant@0:  * possible conflicts. When we insert a possible EOL, the single
etisserant@0:  * look ahead token becomes the EOL, which means the potential conflicts
etisserant@0:  * could no longer be resolved.
etisserant@0:  * Removing these conflicts would make the st syntax parser very convoluted,
etisserant@0:  * and adding the extraneous EOL would make it very cluttered.
etisserant@0:  * This option was therefore dropped in favour of another!
etisserant@0:  *
etisserant@0:  * I ended up implementing (2). Unfortunately the lexical analyser can
etisserant@0:  * not easily distinguish between il and st code, since function
etisserant@0:  * calls in il are very similar to function block calls in st.
etisserant@0:  * We therefore use an extra 'body' state. When the lexical parser
etisserant@0:  * finds that last END_VAR, it enters the body state. This state
etisserant@0:  * must figure out what language is being parsed from the first few
etisserant@0:  * tokens, and switch to the correct state (st or il) according to the
etisserant@0:  * language. This means that we insert quite a bit of knowledge of the
etisserant@0:  * syntax of the languages into the lexical parser. This is ugly, but it
etisserant@0:  * works, and at least it is possible to keep all the state changes together
etisserant@0:  * to make it easier to remove them later on if need be.
etisserant@0:  * The body state returns any matched text back to the buffer with unput(),
etisserant@0:  * to be later matched correctly by the apropriate language parser (st or il).
etisserant@0:  * The state machine has 6 possible states (INITIAL, config, decl, body, st, il)
etisserant@0:  * Possible state changes are:
lbessard@3:  *   INITIAL -> decl_state (when a FUNCTION, FUNCTION_BLOCK, or PROGRAM is found,
etisserant@0:  *                    and followed by a VAR declaration)
lbessard@3:  *   INITIAL -> il_st_state (when a FUNCTION, FUNCTION_BLOCK, or PROGRAM is found,
etisserant@0:  *                    and _not_ followed by a VAR declaration)
lbessard@3:  *   INITIAL -> config_state (when a CONFIGURATION is found)
lbessard@3:  *   decl_state    -> il_st_state (when the last END_VAR is found, i.e. the function body starts)
lbessard@3:  *   il_st_state   -> sfc_state (when it figures out it is parsing sfc language)
lbessard@3:  *   il_st_state   -> st_state (when it figures out it is parsing st language)
lbessard@3:  *   il_st_state   -> il_state (when it figures out it is parsing il language)
lbessard@3:  *   decl_state    -> INITIAL (when a END_FUNCTION, END_FUNCTION_BLOCK, or END_PROGRAM is found)
lbessard@3:  *   st_state      -> INITIAL (when a END_FUNCTION, END_FUNCTION_BLOCK, or END_PROGRAM is found)
lbessard@3:  *   sfc_state     -> INITIAL (when a END_FUNCTION, END_FUNCTION_BLOCK, or END_PROGRAM is found)
lbessard@3:  *   il_state      -> INITIAL (when a END_FUNCTION, END_FUNCTION_BLOCK, or END_PROGRAM is found)
lbessard@3:  *   config_state  -> INITIAL (when a END_CONFIGURATION is found)
etisserant@0:  */
etisserant@0: /* we are parsing a configuration. */
lbessard@3: %s config_state
etisserant@0: 
etisserant@0: /* we are parsing a function, program or function block declaration */
lbessard@3: %s decl_state
etisserant@0: 
etisserant@0: /* we will be parsing a function body. Whether il/st is remains unknown */
lbessard@3: %x il_st_state
etisserant@0: 
etisserant@0: /* we are parsing il code -> flex must return the EOL tokens!       */
lbessard@3: %s il_state
etisserant@0: 
etisserant@0: /* we are parsing st code -> flex must not return the EOL tokens!   */
lbessard@3: %s st_state
etisserant@0: 
etisserant@1: /* we are parsing sfc code -> flex must not return the EOL tokens!   */
lbessard@3: %s sfc_state
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /*******************/
etisserant@0: /* File #include's */
etisserant@0: /*******************/
etisserant@0: 
etisserant@0: /* We extend the IEC 61131-3 standard syntax to allow inclusion
etisserant@0:  * of other files, using the IEC 61131-3 pragma directive...
etisserant@0:  * The accepted syntax is:
etisserant@0:  *  {#include "<filename>"}
etisserant@0:  */
etisserant@0: 
etisserant@0: /* the "include" states are used for picking up the name of an include file */
etisserant@0: %x include_beg
etisserant@0: %x include_filename
etisserant@0: %x include_end
etisserant@0: 
etisserant@0: 
etisserant@0: file_include_pragma_filename	[^\"]*
etisserant@0: file_include_pragma_beg		"{#include"{st_whitespace_only}\"
etisserant@0: file_include_pragma_end		\"{st_whitespace_only}"}"
etisserant@0: file_include_pragma			{file_include_pragma_beg}{file_include_pragma_filename}{file_include_pragma_end}
etisserant@0: 
etisserant@0: 
etisserant@0: %{
etisserant@0: #define MAX_INCLUDE_DEPTH 16
etisserant@0: 
etisserant@0: typedef struct {
etisserant@0: 	  YY_BUFFER_STATE buffer_state;
etisserant@0: 	  int lineno;
etisserant@0: 	  const char *filename;
etisserant@0: 	} include_stack_t;
etisserant@0: 
etisserant@0: include_stack_t include_stack[MAX_INCLUDE_DEPTH];
etisserant@0: int include_stack_ptr = 0;
etisserant@0: 
etisserant@0: const char *INCLUDE_DIRECTORIES[] = {
etisserant@0: 	"",
etisserant@0: 	"lib/",
etisserant@0: 	"/lib/",
etisserant@0: 	"/usr/lib/",
etisserant@0: 	"/usr/lib/iec/",
etisserant@0: 	LIBDIRECTORY "/",
etisserant@0: 	NULL /* must end with NULL!! */
etisserant@0: 	};
etisserant@0: 
etisserant@0: %}
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /*****************************/
etisserant@0: /* Prelimenary constructs... */
etisserant@0: /*****************************/
etisserant@0: 
etisserant@0: 
etisserant@0: /* A pragma... */
etisserant@0: 
etisserant@0: pragma "{"[^}]*"}"
etisserant@0: 
etisserant@0: /* NOTE: this seemingly unnecessary complex definition is required
etisserant@0:  *       to be able to eat up comments such as:
etisserant@0:  *          '(* Testing... ! ***** ******)'
etisserant@0:  *       without using the trailing context command in flex (/{context})
etisserant@0:  *       since {comment} itself will later be used with
etisserant@0:  *       trailing context ({comment}/{context})
etisserant@0:  */
etisserant@0: not_asterisk				[^*]
etisserant@0: not_close_parenthesis_nor_asterisk	[^*)]
etisserant@0: asterisk				"*"
etisserant@0: comment_text		{not_asterisk}|(({asterisk}+){not_close_parenthesis_nor_asterisk})
etisserant@0: 
etisserant@0: comment		"(*"({comment_text}*)({asterisk}+)")"
etisserant@0: 
etisserant@0: 
etisserant@0: /*
etisserant@0: 3.1 Whitespace
etisserant@0:  (NOTE: Whitespace IS clearly defined, to include newline!!! See section 2.1.4!!!)
etisserant@0:  No definition of whitespace is given, in other words, the characters that may be used to seperate language tokens are not pecisely defined. One may nevertheless make an inteligent guess of using the space (' '), and other characters also commonly considered whitespace in other programming languages (horizontal tab, vertical tab, form feed, etc.).
etisserant@0:  The main question is whether the newline character should be considered whitespace. IL language statements use an EOL token (End Of Line) to distinguish between some language constructs. The EOL token itself is openly defined as "normally consist[ing] of the 'paragraph separator' ", leaving the final choice open to each implemention. If we choose the newline character to represent the EOL token, it may then not be considered whitespace.
etisserant@0:  On the other hand, some examples that come in a non-normative annex of the specification allow function declarations to span multiple3.1 Whitespace
etisserant@0:  (NOTE: Whitespace IS clearly defined, to include newline!!! See section 2.1.4!!!)
etisserant@0:  No definition of whitespace is given, in other words, the characters that may be used to seperate language tokens are not pecisely defined. One may nevertheless make an inteligent guess of using the space (' '), and other characters also commonly considered whitespace in other programming languages (horizontal tab, vertical tab, form feed, etc.).
etisserant@0:  The main question is whether the newline character should be considered whitespace. IL language statements use an EOL token (End Of Line) to distinguish between some language constructs. The EOL token itself is openly defined as "normally consist[ing] of the 'paragraph separator' ", leaving the final choice open to each implemention. If we choose the newline character to represent the EOL token, it may then not be considered whitespace.
etisserant@0:  On the other hand, some examples that come in a non-normative annex of the specification allow function declarations to span multiple lines, which means that the newline character is being considered as whitespace.
etisserant@0:  Our implementation works around this issue by including the new line character in the whitespace while parsing function declarations and the ST language, and parsing it as the EOL token only while parsing IL language statements. This requires the use of a state machine in the lexical parser that needs at least some knowledge of the syntax itself.
etisserant@0: */
etisserant@0: /* NOTE: Our definition of whitespace will only work in ASCII!
etisserant@0:  *
etisserant@0:  *       Since the IL language needs to know the location of newline
etisserant@0:  *       (token EOL -> '\n' ), we need one definition of whitespace
etisserant@0:  *       for each language...
etisserant@0:  */
etisserant@0: /*
etisserant@0:  * NOTE: we cannot use
etisserant@0:  *         st_whitespace	[:space:]*
etisserant@0:  *       since we use {st_whitespace} as trailing context. In our case
etisserant@0:  *       this would not constitute "dangerous trailing context", but the
etisserant@0:  *       lexical generator (i.e. flex) does not know this (since it does
etisserant@0:  *       not know which characters belong to the set [:space:]), and will
etisserant@0:  *       generate a "dangerous trailing context" warning!
etisserant@0:  *       We use this alternative just to stop the flex utility from
etisserant@0:  *       generating the invalid (in this case) warning...
etisserant@0:  */
etisserant@0: 
etisserant@0: st_whitespace_only	[ \f\n\r\t\v]*
etisserant@0: il_whitespace_only	[ \f\r\t\v]*
etisserant@0: 
etisserant@0: st_whitespace_text	{st_whitespace_only}|{comment}|{pragma}
etisserant@0: il_whitespace_text	{il_whitespace_only}|{comment}|{pragma}
etisserant@0: 
etisserant@0: st_whitespace	{st_whitespace_text}*
etisserant@0: il_whitespace	{il_whitespace_text}*
etisserant@0: 
etisserant@0: st_whitespace_text_no_pragma	{st_whitespace_only}|{comment}
etisserant@0: il_whitespace_text_no_pragma	{il_whitespace_only}|{comment}
etisserant@0: 
etisserant@0: st_whitespace_no_pragma	{st_whitespace_text_no_pragma}*
etisserant@0: il_whitespace_no_pragma	{il_whitespace_text_no_pragma}*
etisserant@0: 
etisserant@0: qualified_identifier	{identifier}(\.{identifier})?
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /*****************************************/
etisserant@0: /* B.1.1 Letters, digits and identifiers */
etisserant@0: /*****************************************/
etisserant@0: /* NOTE: The following definitions only work if the host computer
etisserant@0:  *       is using the ASCII maping. For e.g., with EBCDIC [A-Z]
etisserant@0:  *       contains non-alphabetic characters!
etisserant@0:  *       The correct way of doing it would be to use
etisserant@0:  *       the [:upper:] etc... definitions.
etisserant@0:  *
etisserant@0:  *       Unfortunately, further on we need all printable
etisserant@0:  *       characters (i.e. [:print:]), but excluding '$'.
etisserant@0:  *       Flex does not allow sets to be composed by excluding
etisserant@0:  *       elements. Sets may only be constructed by adding new
etisserant@0:  *       elements, which means that we have to revert to
etisserant@0:  *       [\x20\x21\x23\x25\x26\x28-x7E] for the definition
etisserant@0:  *       of the printable characters with the required exceptions.
etisserant@0:  *       The above also implies the use of ASCII, but now we have
etisserant@0:  *       no way to work around it|
etisserant@0:  *
etisserant@0:  *       The conclusion is that our parser is limited to ASCII
etisserant@0:  *       based host computers!!
etisserant@0:  */
etisserant@0: letter		[A-Za-z]
etisserant@0: digit		[0-9]
etisserant@0: octal_digit	[0-7]
etisserant@0: hex_digit	{digit}|[A-F]
etisserant@0: identifier	({letter}|(_({letter}|{digit})))((_?({letter}|{digit}))*)
etisserant@0: 
etisserant@0: 
etisserant@0: /*******************/
etisserant@0: /* B.1.2 Constants */
etisserant@0: /*******************/
etisserant@0: 
etisserant@0: /******************************/
etisserant@0: /* B.1.2.1   Numeric literals */
etisserant@0: /******************************/
etisserant@0: integer         {digit}((_?{digit})*)
etisserant@0: binary_integer  2#{bit}((_?{bit})*)
etisserant@0: bit		[0-1]
etisserant@0: octal_integer   8#{octal_digit}((_?{octal_digit})*)
etisserant@0: hex_integer     16#{hex_digit}((_?{hex_digit})*)
etisserant@0: exponent        [Ee]([+-]?){integer}
etisserant@0: /* The correct definition for real would be:
etisserant@0:  * real		{integer}\.{integer}({exponent}?)
etisserant@0:  *
etisserant@0:  * Unfortunately, the spec also defines fixed_point (B 1.2.3.1) as:
etisserant@0:  * fixed_point		{integer}\.{integer}
etisserant@0:  *
etisserant@0:  * This means that {integer}\.{integer} could be interpreted
etisserant@0:  * as either a fixed_point or a real.
etisserant@0:  * I have opted to interpret {integer}\.{integer} as a fixed_point.
etisserant@0:  * In order to do this, the definition of real has been changed to:
etisserant@0:  * real		{integer}\.{integer}{exponent}
etisserant@0:  *
etisserant@0:  * This means that the syntax parser now needs to define a real to be
etisserant@0:  * either a real_token or a fixed_point_token!
etisserant@0:  */
etisserant@0: real		{integer}\.{integer}{exponent}
etisserant@0: 
etisserant@0: 
etisserant@0: /*******************************/
etisserant@0: /* B.1.2.2   Character Strings */
etisserant@0: /*******************************/
etisserant@0: /*
etisserant@0: common_character_representation :=
etisserant@0: <any printable character except '$', '"' or "'">
etisserant@0: |'$$'
etisserant@0: |'$L'|'$N'|'$P'|'$R'|'$T'
etisserant@0: |'$l'|'$n'|'$p'|'$r'|'$t'
etisserant@0: 
etisserant@0: NOTE: 	$ = 0x24
etisserant@0: 	" = 0x22
etisserant@0: 	' = 0x27
etisserant@0: 
etisserant@0: 	printable chars in ASCII: 0x20-0x7E
etisserant@0: */
etisserant@0: 
etisserant@0: esc_char_u		$L|$N|$P|$R|$T
etisserant@0: esc_char_l		$l|$n|$p|$r|$t
etisserant@0: esc_char		$$|{esc_char_u}|{esc_char_l}
etisserant@0: double_byte_char	(${hex_digit}{hex_digit}{hex_digit}{hex_digit})
etisserant@0: single_byte_char	(${hex_digit}{hex_digit})
etisserant@0: 
etisserant@0: /* WARNING:
etisserant@0:  * This definition is only valid in ASCII...
etisserant@0:  *
etisserant@0:  * Flex includes the function print_char() that defines
etisserant@0:  * all printable characters portably (i.e. whatever character
etisserant@0:  * encoding is currently being used , ASCII, EBCDIC, etc...)
etisserant@0:  * Unfortunately, we cannot generate the definition of
etisserant@0:  * common_character_representation portably, since flex
etisserant@0:  * does not allow definition of sets by subtracting
etisserant@0:  * elements in one set from another set.
etisserant@0:  * This means we must build up the defintion of
etisserant@0:  * common_character_representation using only set addition,
etisserant@0:  * which leaves us with the only choice of defining the
etisserant@0:  * characters non-portably...
etisserant@0:  */
etisserant@0: common_character_representation		[\x20\x21\x23\x25\x26\x28-\x7E]|{esc_char}
etisserant@0: double_byte_character_representation 	$\"|'|{double_byte_char}|{common_character_representation}
etisserant@0: single_byte_character_representation 	$'|\"|{single_byte_char}|{common_character_representation}
etisserant@0: 
etisserant@0: 
etisserant@0: double_byte_character_string	\"({double_byte_character_representation}*)\"
etisserant@0: single_byte_character_string	'({single_byte_character_representation}*)'
etisserant@0: 
etisserant@0: 
etisserant@0: /************************/
etisserant@0: /* B 1.2.3.1 - Duration */
etisserant@0: /************************/
etisserant@0: fixed_point		{integer}\.{integer}
etisserant@0: 
etisserant@0: fixed_point_d		{fixed_point}d
etisserant@0: integer_d		{integer}d
etisserant@0: 
etisserant@0: fixed_point_h		{fixed_point}h
etisserant@0: integer_h		{integer}h
etisserant@0: 
etisserant@0: fixed_point_m		{fixed_point}m
etisserant@0: integer_m		{integer}m
etisserant@0: 
etisserant@0: fixed_point_s		{fixed_point}s
etisserant@0: integer_s		{integer}s
etisserant@0: 
etisserant@0: fixed_point_ms		{fixed_point}ms
etisserant@0: integer_ms		{integer}ms
etisserant@0: 
etisserant@0: 
etisserant@0: /********************************************/
etisserant@0: /* B.1.4.1   Directly Represented Variables */
etisserant@0: /********************************************/
etisserant@0: /* The correct definition, if the standard were to be followed... */
mario@11: 
mario@11: location_prefix			[IQM]
mario@11: size_prefix			[XBWDL]
mario@11: direct_variable_standard	%{location_prefix}({size_prefix}?){integer}((.{integer})*)
mario@11: 
etisserant@0: 
etisserant@0: /* For the MatPLC, we will accept %<identifier>
etisserant@0:  * as a direct variable, this being mapped onto the MatPLC point
etisserant@0:  * named <identifier>
etisserant@0:  */
etisserant@0: /* TODO: we should not restrict it to only the accepted syntax
etisserant@0:  * of <identifier> as specified by the standard. MatPLC point names
etisserant@0:  * have a more permissive syntax.
etisserant@0:  *
etisserant@0:  * e.g. "P__234"
etisserant@0:  *    Is a valid MatPLC point name, but not a valid <identifier> !!
etisserant@0:  *    The same happens with names such as "333", "349+23", etc...
etisserant@0:  *    How can we handle these more expressive names in our case?
etisserant@0:  *    Remember that some direct variable may remain anonymous, with
etisserant@0:  *    declarations such as:
etisserant@0:  *    VAR
etisserant@0:  *       AT %I3 : BYTE := 255;
etisserant@0:  *    END_VAR
mario@11:  *    in which case we are currently using "%I3" as the variable
mario@11:  *    name.
mario@11:  */
mario@11: direct_variable_matplc		%{identifier}
mario@11: 
mario@11: direct_variable			{direct_variable_standard}|{direct_variable_matplc}
etisserant@0: 
etisserant@0: /******************************************/
etisserant@0: /* B 1.4.3 - Declaration & Initialisation */
etisserant@0: /******************************************/
etisserant@0: incompl_location	%[IQM]\*
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: %%
etisserant@0: 	/* fprintf(stderr, "flex: state %d\n", YY_START); */
etisserant@0: 
etisserant@0: 	/*****************************************************/
etisserant@0: 	/*****************************************************/
etisserant@0: 	/*****************************************************/
etisserant@0: 	/*****                                           *****/
etisserant@0: 	/*****                                           *****/
etisserant@0: 	/*****   F I R S T    T H I N G S    F I R S T   *****/
etisserant@0: 	/*****                                           *****/
etisserant@0: 	/*****                                           *****/
etisserant@0: 	/*****************************************************/
etisserant@0: 	/*****************************************************/
etisserant@0: 	/*****************************************************/
etisserant@0: 
mario@13: 	if (get_goto_body_state()) {
mario@6: 	  yy_push_state(il_st_state);
mario@13: 	  rst_goto_body_state();
mario@6: 	}
lbessard@3: 
etisserant@0: 	/*********************************/
etisserant@0: 	/* Handle the pragmas!     */
etisserant@0: 	/*********************************/
etisserant@0: 
etisserant@0: 	/* We start off by searching for the pragmas we handle in the lexical parser. */
etisserant@0: <INITIAL>{file_include_pragma}	unput_text(0); yy_push_state(include_beg);
etisserant@0: 
etisserant@0: 	/* Any other pragma we find, we just pass it up to the syntax parser...   */
mario@13: 	/* Note that the <il_st_state> state is exclusive, so we have to include it here too. */
etisserant@0: {pragma}	{/* return the pragmma without the enclosing '{' and '}' */
mario@13: 		 yytext[strlen(yytext)-1] = '\0';
etisserant@0: 		 yylval.ID=strdup(yytext+1);
etisserant@0: 		 return pragma_token;
etisserant@0: 		}
lbessard@3: <il_st_state>{pragma} {/* return the pragmma without the enclosing '{' and '}' */
mario@13: 		 yytext[strlen(yytext)-1] = '\0';
etisserant@0: 		 yylval.ID=strdup(yytext+1);
etisserant@0: 		 return pragma_token;
etisserant@0: 		}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*********************************/
etisserant@0: 	/* Handle the file includes!     */
etisserant@0: 	/*********************************/
etisserant@0: <include_beg>{file_include_pragma_beg}	BEGIN(include_filename);
etisserant@0: 
etisserant@0: <include_filename>{file_include_pragma_filename}	{
etisserant@0: 			  /* got the include file name */
etisserant@0: 			  int i;
etisserant@0: 
etisserant@0: 			  if (include_stack_ptr >= MAX_INCLUDE_DEPTH) {
etisserant@0: 			    fprintf(stderr, "Includes nested too deeply\n");
etisserant@0: 			    exit( 1 );
etisserant@0: 			  }
etisserant@0: 
etisserant@0: 			  (include_stack[include_stack_ptr]).buffer_state = YY_CURRENT_BUFFER;
etisserant@0: 			  (include_stack[include_stack_ptr]).lineno = yylineno;
etisserant@0: 			  (include_stack[include_stack_ptr]).filename = current_filename;
etisserant@0: 			  include_stack_ptr++;
etisserant@0: 			  yylineno = 1;
etisserant@0: 			  current_filename = strdup(yytext);
etisserant@0: 
etisserant@0: 			  for (i = 0, yyin = NULL; (INCLUDE_DIRECTORIES[i] != NULL) && (yyin == NULL); i++) {
etisserant@0: 			    char *full_name = strdup2(INCLUDE_DIRECTORIES[i], yytext);
etisserant@0: 			    if (full_name == NULL) {
etisserant@0: 			      fprintf(stderr, "Out of memory!\n");
etisserant@0: 			      exit( 1 );
etisserant@0: 			    }
etisserant@0: 			    yyin = fopen(full_name, "r");
etisserant@0: 			    free(full_name);
etisserant@0: 			  }
etisserant@0: 
etisserant@0: 			  if (!yyin) {
etisserant@0: 			    fprintf(stderr, "Error opening included file %s\n", yytext);
etisserant@0: 			    exit( 1 );
etisserant@0: 			  }
etisserant@0: 
etisserant@0: 			  /* switch input buffer to new file... */
etisserant@0: 			  yy_switch_to_buffer(yy_create_buffer(yyin, YY_BUF_SIZE));
etisserant@0: 			  /* switch to whatever state was active before the include file */
etisserant@0: 			  yy_pop_state();
etisserant@0: 			  /* now process the new file... */
etisserant@0: 			}
etisserant@0: 
etisserant@0: 
etisserant@0: <<EOF>>			{
etisserant@0: 			  if (--include_stack_ptr < 0) {
etisserant@0: 			    yyterminate();
etisserant@0: 			  } else {
etisserant@0: 			    yy_delete_buffer(YY_CURRENT_BUFFER);
etisserant@0: 			    yy_switch_to_buffer((include_stack[include_stack_ptr]).buffer_state);
etisserant@0: 			    yylineno = include_stack[include_stack_ptr].lineno;
etisserant@0: 			      /* removing constness of char *. This is safe actually,
etisserant@0: 			       * since the only real const char * that is stored on the stack is
etisserant@1: 			       * the first one (i.e. the one that gets stored in include_stack[0],
etisserant@0: 			       * which is never free'd!
etisserant@0: 			       */
etisserant@0: 			    free((char *)current_filename);
etisserant@0: 			    current_filename = include_stack[include_stack_ptr].filename;
etisserant@0: 			    yy_push_state(include_end);
etisserant@0: 			  }
etisserant@0: 			}
etisserant@0: 
etisserant@0: <include_end>{file_include_pragma_end}	yy_pop_state();
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*********************************/
etisserant@0: 	/* Handle all the state changes! */
etisserant@0: 	/*********************************/
etisserant@0: 
lbessard@3: 	/* INITIAL -> decl_state */
etisserant@0: <INITIAL>{
lbessard@3: 	/* NOTE: how about functions that do not declare variables, and go directly to the il_st_state???
etisserant@0: 	 *      - According to Section 2.5.1.3 (Function Declaration), item 2 in the list, a FUNCTION
etisserant@0: 	 *        must have at least one input argument, so a correct declaration will have at least
etisserant@0: 	 *        one VAR_INPUT ... VAR_END construct!
etisserant@0: 	 *      - According to Section 2.5.2.2 (Function Block Declaration), a FUNCTION_BLOCK
etisserant@0: 	 *        must have at least one input argument, so a correct declaration will have at least
etisserant@0: 	 *        one VAR_INPUT ... VAR_END construct!
etisserant@0: 	 *      - According to Section 2.5.3 (Programs), a PROGRAM must have at least one input
etisserant@0: 	 *        argument, so a correct declaration will have at least one VAR_INPUT ... VAR_END
etisserant@0: 	 *        construct!
etisserant@0: 	 *
etisserant@0: 	 *       All the above means that we needn't worry about PROGRAMs, FUNCTIONs or
lbessard@3: 	 *       FUNCTION_BLOCKs that do not have at least one VAR_END before the il_st_state.
etisserant@0: 	 *       If the code has an error, and no VAR_END before the body, we will simply
lbessard@3: 	 *       continue in the <decl_state> state, untill the end of the FUNCTION, FUNCTION_BLOCK
etisserant@0: 	 *       or PROGAM.
etisserant@0: 	 */
lbessard@3: FUNCTION				BEGIN(decl_state); return FUNCTION;
lbessard@3: FUNCTION_BLOCK				BEGIN(decl_state); return FUNCTION_BLOCK;
lbessard@3: PROGRAM					BEGIN(decl_state); return PROGRAM;
lbessard@3: CONFIGURATION				BEGIN(config_state); return CONFIGURATION;
etisserant@0: }
etisserant@0: 
lbessard@3: 	/* INITIAL -> il_st_state */
etisserant@0: 	/* required if the function, program, etc.. has no VAR block! */
mario@6: 	/* We comment it out since the standard does not allow this.  */
mario@6: 	/* NOTE: Even if we were to include the following code, it    */
mario@6: 	/*       would have no effect whatsoever since the above      */
mario@6: 	/*       rules will take precendence!                         */
mario@6: 	/*
etisserant@0: <INITIAL>{
lbessard@3: FUNCTION	BEGIN(il_st_state); return FUNCTION;
lbessard@3: FUNCTION_BLOCK	BEGIN(il_st_state); return FUNCTION_BLOCK;
lbessard@3: PROGRAM		BEGIN(il_st_state); return PROGRAM;
etisserant@0: }
mario@6: 	*/
mario@6: 
mario@6: 	/* decl_state -> (il_st_state | sfc_state) */
lbessard@3: <decl_state>{
mario@6: END_VAR{st_whitespace}VAR		{unput_text(strlen("END_VAR")); 
mario@6: 					 return END_VAR;
mario@6: 					}
mario@6: END_VAR{st_whitespace}INITIAL_STEP	{unput_text(strlen("END_VAR")); 
mario@9: 					 yy_push_state(sfc_state); 
mario@6: 					 return END_VAR;
mario@6: 					}
mario@6: END_VAR{st_whitespace}			{unput_text(strlen("END_VAR")); 
mario@6: 					 cmd_goto_body_state(); 
mario@6: 					 return END_VAR;
mario@6: 					}
etisserant@0: }
etisserant@0: 
mario@6: 	/* il_st_state -> (il_state | st_state) */
lbessard@3: <il_st_state>{
mario@13: {st_whitespace_no_pragma}			/* Eat any whitespace */
lbessard@3: {qualified_identifier}{st_whitespace}":="	unput_text(0); BEGIN(st_state);
lbessard@3: {qualified_identifier}"["			unput_text(0); BEGIN(st_state);
lbessard@3: 
lbessard@3: RETURN						unput_text(0); BEGIN(st_state);
lbessard@3: IF						unput_text(0); BEGIN(st_state);
lbessard@3: CASE						unput_text(0); BEGIN(st_state);
lbessard@3: FOR						unput_text(0); BEGIN(st_state);
lbessard@3: WHILE						unput_text(0); BEGIN(st_state);
lbessard@3: REPEAT						unput_text(0); BEGIN(st_state);
lbessard@3: EXIT						unput_text(0); BEGIN(st_state);
mario@6: 	/* ':=' occurs only in transitions, and not Function or FB bodies! */
mario@6: :=						unput_text(0); BEGIN(st_state);  
etisserant@0: 
etisserant@0: 
etisserant@0: {identifier}	{int token = get_identifier_token(yytext);
etisserant@0: 		 if (token == prev_declared_fb_name_token) {
etisserant@0: 		   /* the code has a call to a function block */
lbessard@3: 		   BEGIN(st_state);
etisserant@0: 		 } else {
lbessard@3: 		   BEGIN(il_state);
etisserant@0: 		 }
etisserant@0: 		 unput_text(0);
etisserant@0: 		}
lbessard@3: .		unput_text(0); BEGIN(il_state);
lbessard@3: }	/* end of il_st_state lexical parser */
lbessard@3: 
lbessard@3: 	/* (il_state | st_state) -> $previous_state (decl_state or sfc_state) */
lbessard@3: <il_state,st_state>{
lbessard@3: END_FUNCTION		yy_pop_state(); unput_text(0);
lbessard@3: END_FUNCTION_BLOCK	yy_pop_state(); unput_text(0);
lbessard@3: END_PROGRAM		yy_pop_state(); unput_text(0);
lbessard@3: END_TRANSITION		yy_pop_state(); unput_text(0);
mario@6: END_ACTION		yy_pop_state(); unput_text(0);
lbessard@3: }
lbessard@3: 
lbessard@4: 	/* sfc_state -> INITIAL */
lbessard@4: <sfc_state>{
lbessard@4: END_FUNCTION		yy_pop_state(); unput_text(0);
lbessard@4: END_FUNCTION_BLOCK	yy_pop_state(); unput_text(0);
lbessard@4: END_PROGRAM		yy_pop_state(); unput_text(0);
lbessard@4: }
lbessard@4: 
lbessard@4: 	/* decl_state -> INITIAL */
lbessard@4: <decl_state>{
etisserant@0: END_FUNCTION		BEGIN(INITIAL); return END_FUNCTION;
etisserant@0: END_FUNCTION_BLOCK	BEGIN(INITIAL); return END_FUNCTION_BLOCK;
etisserant@0: END_PROGRAM		BEGIN(INITIAL); return END_PROGRAM;
lbessard@3: }
etisserant@0: 	/* config -> INITIAL */
etisserant@0: END_CONFIGURATION	BEGIN(INITIAL); return END_CONFIGURATION;
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 	/***************************************/
etisserant@0: 	/* Next is to to remove all whitespace */
etisserant@0: 	/***************************************/
etisserant@0: 	/* NOTE: pragmas are handled right at the beginning... */
etisserant@0: 
mario@13: <INITIAL,config_state,decl_state,st_state,sfc_state>{st_whitespace_no_pragma}	/* Eat any whitespace */
mario@13: <il_state>{il_whitespace_no_pragma}		/* Eat any whitespace */
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*****************************************/
etisserant@0: 	/* B.1.1 Letters, digits and identifiers */
etisserant@0: 	/*****************************************/
etisserant@0: 	/* NOTE: 'R1', 'IN', etc... are IL operators, and therefore tokens
etisserant@0: 	 *       On the other hand, the spec does not define them as keywords,
etisserant@0: 	 *       which means they may be re-used for variable names, etc...!
etisserant@0: 	 *       The syntax parser already caters for the possibility of these
etisserant@0: 	 *       tokens being used for variable names in their declarations.
etisserant@0: 	 *       When they are declared, they will be added to the variable symbol table!
etisserant@0: 	 *       Further appearances of these tokens must no longer be parsed
etisserant@0: 	 *       as R1_tokens etc..., but rather as variable_name_tokens!
etisserant@0: 	 *
etisserant@0: 	 *       That is why the first thing we do with identifiers, even before
etisserant@0: 	 *       checking whether they may be a 'keyword', is to check whether
etisserant@0: 	 *       they have been previously declared as a variable name,
etisserant@0: 	 *
mario@13: 	 *       However, we have a dilema! Should we here also check for
mario@13: 	 *       prev_declared_derived_function_name_token?
mario@13: 	 *       If we do, then the 'MOD' default library function (defined in
mario@13: 	 *       the standard) will always be returned as a function name, and
mario@13: 	 *       it will therefore not be possible to use it as an operator as 
mario@13: 	 *       in the following ST expression 'X := Y MOD Z;' !
mario@13: 	 *       If we don't, then even it will not be possible to use 'MOD'
mario@13: 	 *       as a funtion as in 'X := MOD(Y, Z);'
mario@13: 	 *       We solve this by NOT testing for function names here, and
mario@13: 	 *       handling this function and keyword clash in bison!
etisserant@0: 	 */
etisserant@0: {identifier} 	{int token = get_identifier_token(yytext);
etisserant@0: 		 if ((token == prev_declared_variable_name_token) ||
mario@13: //		     (token == prev_declared_derived_function_name_token) || // DO NOT add this condition!
etisserant@0: 		     (token == prev_declared_fb_name_token)) {
etisserant@0: 		 /*
etisserant@0: 		 if (token != identifier_token)
etisserant@0: 		 */
mario@13: 		 /* NOTE: if we replace the above uncommented conditions with
mario@13:                   *       the simple test of (token != identifier_token), then 
mario@13:                   *       'MOD' et al must be removed from the 
mario@13:                   *       library_symbol_table as a default function name!
etisserant@0: 		  */
etisserant@0: 		   yylval.ID=strdup(yytext);
etisserant@0: 		   return token;
etisserant@0: 		 }
etisserant@0: 		 /* otherwise, leave it for the other lexical parser rules... */
etisserant@0: 		 REJECT;
etisserant@0: 		}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/******************************************************/
etisserant@0: 	/******************************************************/
etisserant@0: 	/******************************************************/
etisserant@0: 	/*****                                            *****/
etisserant@0: 	/*****                                            *****/
etisserant@0: 	/*****   N O W    D O   T H E   K E Y W O R D S   *****/
etisserant@0: 	/*****                                            *****/
etisserant@0: 	/*****                                            *****/
etisserant@0: 	/******************************************************/
etisserant@0: 	/******************************************************/
etisserant@0: 	/******************************************************/
etisserant@0: 
etisserant@0: 
etisserant@0: EN	return EN;
etisserant@0: ENO	return ENO;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/******************************/
etisserant@0: 	/* B 1.2.1 - Numeric Literals */
etisserant@0: 	/******************************/
etisserant@0: TRUE		return TRUE;
etisserant@0: BOOL#1  	return TRUE;
etisserant@0: FALSE		return FALSE;
etisserant@0: BOOL#0  	return FALSE;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/************************/
etisserant@0: 	/* B 1.2.3.1 - Duration */
etisserant@0: 	/************************/
etisserant@0: t#		return T_SHARP;
etisserant@0: T#		return T_SHARP;
etisserant@0: TIME		return TIME;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/************************************/
etisserant@0: 	/* B 1.2.3.2 - Time of day and Date */
etisserant@0: 	/************************************/
etisserant@0: TIME_OF_DAY	return TIME_OF_DAY;
etisserant@0: TOD		return TIME_OF_DAY;
etisserant@0: DATE		return DATE;
etisserant@0: d#		return D_SHARP;
etisserant@0: D#		return D_SHARP;
etisserant@0: DATE_AND_TIME	return DATE_AND_TIME;
etisserant@0: DT		return DATE_AND_TIME;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/***********************************/
etisserant@0: 	/* B 1.3.1 - Elementary Data Types */
etisserant@0: 	/***********************************/
etisserant@0: BYTE		return BYTE;
etisserant@0: WORD		return WORD;
etisserant@0: DWORD		return DWORD;
etisserant@0: LWORD		return LWORD;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************/
etisserant@0: 	/* B 1.3.2 - Generic data types */
etisserant@0: 	/********************************/
etisserant@0: 	/* Strangely, the following symbols do not seem to be required! */
etisserant@0: 	/* But we include them so they become reserved words, and do not
etisserant@0: 	 * get passed up to bison as an identifier...
etisserant@0: 	 */
etisserant@0: ANY		return ANY;
etisserant@0: ANY_DERIVED	return ANY_DERIVED;
etisserant@0: ANY_ELEMENTARY	return ANY_ELEMENTARY;
etisserant@0: ANY_MAGNITUDE	return ANY_MAGNITUDE;
etisserant@0: ANY_NUM		return ANY_NUM;
etisserant@0: ANY_REAL	return ANY_REAL;
etisserant@0: ANY_INT		return ANY_INT;
etisserant@0: ANY_BIT		return ANY_BIT;
etisserant@0: ANY_STRING	return ANY_STRING;
etisserant@0: ANY_DATE	return ANY_DATE;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************/
etisserant@0: 	/* B 1.3.3 - Derived data types */
etisserant@0: 	/********************************/
etisserant@0: ":="		return ASSIGN;
etisserant@0: ".."		return DOTDOT;
etisserant@0: TYPE		return TYPE;
etisserant@0: END_TYPE	return END_TYPE;
etisserant@0: ARRAY		return ARRAY;
etisserant@0: OF		return OF;
etisserant@0: STRUCT		return STRUCT;
etisserant@0: END_STRUCT	return END_STRUCT;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*********************/
etisserant@0: 	/* B 1.4 - Variables */
etisserant@0: 	/*********************/
etisserant@0: REAL		return REAL;
etisserant@0: LREAL		return LREAL;
etisserant@0: 
etisserant@0: SINT		return SINT;
etisserant@0: INT		return INT;
etisserant@0: DINT		return DINT;
etisserant@0: LINT		return LINT;
etisserant@0: 
etisserant@0: USINT		return USINT;
etisserant@0: UINT		return UINT;
etisserant@0: UDINT		return UDINT;
etisserant@0: ULINT		return ULINT;
etisserant@0: 
etisserant@0: 
etisserant@0: WSTRING		return WSTRING;
etisserant@0: STRING		return STRING;
etisserant@0: BOOL		return BOOL;
etisserant@0: 
etisserant@0: TIME		return TIME;
etisserant@0: DATE		return DATE;
etisserant@0: DT		return DT;
etisserant@0: TOD		return TOD;
etisserant@0: DATE_AND_TIME	return DATE_AND_TIME;
etisserant@0: TIME_OF_DAY	return TIME_OF_DAY;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/******************************************/
etisserant@0: 	/* B 1.4.3 - Declaration & Initialisation */
etisserant@0: 	/******************************************/
etisserant@0: VAR_INPUT	return VAR_INPUT;
etisserant@0: VAR_OUTPUT	return VAR_OUTPUT;
etisserant@0: VAR_IN_OUT	return VAR_IN_OUT;
etisserant@0: VAR_EXTERNAL	return VAR_EXTERNAL;
etisserant@0: VAR_GLOBAL	return VAR_GLOBAL;
etisserant@0: END_VAR		return END_VAR;
etisserant@0: RETAIN		return RETAIN;
etisserant@0: NON_RETAIN	return NON_RETAIN;
etisserant@0: R_EDGE		return R_EDGE;
etisserant@0: F_EDGE		return F_EDGE;
etisserant@0: AT		return AT;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/***********************/
etisserant@0: 	/* B 1.5.1 - Functions */
etisserant@0: 	/***********************/
etisserant@0: FUNCTION	return FUNCTION;
etisserant@0: END_FUNCTION	return END_FUNCTION;
etisserant@0: VAR		return VAR;
etisserant@0: CONSTANT	return CONSTANT;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*****************************/
etisserant@0: 	/* B 1.5.2 - Function Blocks */
etisserant@0: 	/*****************************/
etisserant@0: FUNCTION_BLOCK		return FUNCTION_BLOCK;
etisserant@0: END_FUNCTION_BLOCK	return END_FUNCTION_BLOCK;
etisserant@0: VAR_TEMP		return VAR_TEMP;
etisserant@0: VAR			return VAR;
etisserant@0: NON_RETAIN		return NON_RETAIN;
etisserant@0: END_VAR			return END_VAR;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/**********************/
etisserant@0: 	/* B 1.5.3 - Programs */
etisserant@0: 	/**********************/
etisserant@0: PROGRAM		return PROGRAM;
etisserant@0: END_PROGRAM	return END_PROGRAM;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************************/
etisserant@0: 	/* B 1.6 Sequential Function Chart elements */
etisserant@0: 	/********************************************/
etisserant@0: 	/* NOTE: the following identifiers/tokens clash with the R and S IL operators, as well
etisserant@0: 	.* as other identifiers that may be used as variable names inside IL and ST programs.
etisserant@0: 	 * They will have to be handled when we include parsing of SFC... For now, simply
etisserant@0: 	 * ignore them!
etisserant@0: 	 */
etisserant@1: 	 
lbessard@3: <sfc_state>{
etisserant@0: ACTION		return ACTION;
etisserant@0: END_ACTION	return END_ACTION;
etisserant@0: 
etisserant@0: TRANSITION	return TRANSITION;
etisserant@0: END_TRANSITION	return END_TRANSITION;
etisserant@0: FROM		return FROM;
etisserant@0: TO		return TO;
etisserant@0: PRIORITY	return PRIORITY;
etisserant@0: 
etisserant@0: INITIAL_STEP	return INITIAL_STEP;
etisserant@0: STEP		return STEP;
etisserant@0: END_STEP	return END_STEP;
etisserant@0: 
etisserant@0: L		return L;
etisserant@0: D		return D;
etisserant@0: SD		return SD;
etisserant@0: DS		return DS;
etisserant@0: SL		return SL;
etisserant@0: 
etisserant@0: N		return N;
etisserant@0: P		return P;
etisserant@0: 
etisserant@0: R		return R;
etisserant@0: S		return S;
etisserant@1: }
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************/
etisserant@0: 	/* B 1.7 Configuration elements */
etisserant@0: 	/********************************/
etisserant@0: CONFIGURATION		return CONFIGURATION;
etisserant@0: END_CONFIGURATION	return END_CONFIGURATION;
etisserant@0: TASK			return TASK;
etisserant@0: RESOURCE		return RESOURCE;
etisserant@0: ON			return ON;
etisserant@0: END_RESOURCE		return END_RESOURCE;
etisserant@0: VAR_CONFIG		return VAR_CONFIG;
etisserant@0: VAR_ACCESS		return VAR_ACCESS;
etisserant@0: END_VAR			return END_VAR;
etisserant@0: WITH			return WITH;
etisserant@0: PROGRAM			return PROGRAM;
etisserant@0: RETAIN			return RETAIN;
etisserant@0: NON_RETAIN		return NON_RETAIN;
etisserant@0: PRIORITY		return PRIORITY;
etisserant@0: SINGLE			return SINGLE;
etisserant@0: INTERVAL		return INTERVAL;
etisserant@0: READ_WRITE		return READ_WRITE;
etisserant@0: READ_ONLY		return READ_ONLY;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/***********************************/
etisserant@0: 	/* B 2.1 Instructions and Operands */
etisserant@0: 	/***********************************/
lbessard@3: <il_state>\n		return EOL;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*******************/
etisserant@0: 	/* B 2.2 Operators */
etisserant@0: 	/*******************/
etisserant@0: 	/* NOTE: we can't have flex return the same token for
etisserant@0: 	 *       ANDN and &N, neither for AND and &, since
etisserant@0: 	 *       AND and ANDN are considered valid variable
etisserant@0: 	 *       function or functionblock type names!
etisserant@0: 	 *       This means that the parser may decide that the
etisserant@0: 	 *       AND or ANDN strings found in the source code
etisserant@0: 	 *       are being used as variable names
etisserant@0: 	 *       and not as operators, and will therefore transform
etisserant@0: 	 *       these tokens into indentifier tokens!
etisserant@0: 	 *       We can't have the parser thinking that the source
etisserant@0: 	 *       code contained the string AND (which may be interpreted
etisserant@0: 	 *       as a vairable name) when in reality the source code
etisserant@0: 	 *       merely contained the character &, so we use two
etisserant@0: 	 *       different tokens for & and AND (and similarly
etisserant@0: 	 *       ANDN and &N)!
etisserant@0: 	 */
etisserant@0: LD		return LD;
etisserant@0: LDN		return LDN;
etisserant@0: ST		return ST;
etisserant@0: STN		return STN;
etisserant@0: NOT		return NOT;
etisserant@0: S		return S;
etisserant@0: R		return R;
etisserant@0: S1		return S1;
etisserant@0: R1		return R1;
etisserant@0: CLK		return CLK;
etisserant@0: CU		return CU;
etisserant@0: CD		return CD;
etisserant@0: PV		return PV;
etisserant@0: IN		return IN;
etisserant@0: PT		return PT;
etisserant@0: AND		return AND;
etisserant@0: &		return AND2;
etisserant@0: OR		return OR;
etisserant@0: XOR		return XOR;
etisserant@0: ANDN		return ANDN;
etisserant@0: &N		return ANDN2;
etisserant@0: ORN		return ORN;
etisserant@0: XORN		return XORN;
etisserant@0: ADD		return ADD;
etisserant@0: SUB		return SUB;
etisserant@0: MUL		return MUL;
etisserant@0: DIV		return DIV;
etisserant@0: MOD		return MOD;
etisserant@0: GT		return GT;
etisserant@0: GE		return GE;
etisserant@0: EQ		return EQ;
etisserant@0: LT		return LT;
etisserant@0: LE		return LE;
etisserant@0: NE		return NE;
etisserant@0: CAL		return CAL;
etisserant@0: CALC		return CALC;
etisserant@0: CALCN		return CALCN;
etisserant@0: RET		return RET;
etisserant@0: RETC		return RETC;
etisserant@0: RETCN		return RETCN;
etisserant@0: JMP		return JMP;
etisserant@0: JMPC		return JMPC;
etisserant@0: JMPCN		return JMPCN;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/***********************/
etisserant@0: 	/* B 3.1 - Expressions */
etisserant@0: 	/***********************/
etisserant@0: "**"		return OPER_EXP;
etisserant@0: "<>"		return OPER_NE;
etisserant@0: ">="		return OPER_GE;
etisserant@0: "<="		return OPER_LE;
etisserant@0: AND		return AND;
etisserant@0: XOR		return XOR;
etisserant@0: OR		return OR;
etisserant@0: NOT		return NOT;
etisserant@0: MOD		return MOD;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*****************************************/
etisserant@0: 	/* B 3.2.2 Subprogram Control Statements */
etisserant@0: 	/*****************************************/
etisserant@0: :=		return ASSIGN;
etisserant@0: =>		return SENDTO;
etisserant@0: RETURN		return RETURN;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************/
etisserant@0: 	/* B 3.2.3 Selection Statements */
etisserant@0: 	/********************************/
etisserant@0: IF		return IF;
etisserant@0: THEN		return THEN;
etisserant@0: ELSIF		return ELSIF;
etisserant@0: ELSE		return ELSE;
etisserant@0: END_IF		return END_IF;
etisserant@0: 
etisserant@0: CASE		return CASE;
etisserant@0: OF		return OF;
etisserant@0: ELSE		return ELSE;
etisserant@0: END_CASE	return END_CASE;
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************/
etisserant@0: 	/* B 3.2.4 Iteration Statements */
etisserant@0: 	/********************************/
etisserant@0: FOR		return FOR;
etisserant@0: TO		return TO;
etisserant@0: BY		return BY;
etisserant@0: DO		return DO;
etisserant@0: END_FOR		return END_FOR;
etisserant@0: 
etisserant@0: WHILE		return WHILE;
etisserant@0: DO		return DO;
etisserant@0: END_WHILE	return END_WHILE;
etisserant@0: 
etisserant@0: REPEAT		return REPEAT;
etisserant@0: UNTIL		return UNTIL;
etisserant@0: END_REPEAT	return END_REPEAT;
etisserant@0: 
etisserant@0: EXIT		return EXIT;
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************************************/
etisserant@0: 	/********************************************************/
etisserant@0: 	/********************************************************/
etisserant@0: 	/*****                                              *****/
etisserant@0: 	/*****                                              *****/
etisserant@0: 	/*****  N O W    W O R K    W I T H    V A L U E S  *****/
etisserant@0: 	/*****                                              *****/
etisserant@0: 	/*****                                              *****/
etisserant@0: 	/********************************************************/
etisserant@0: 	/********************************************************/
etisserant@0: 	/********************************************************/
etisserant@0: 
etisserant@0: 
etisserant@0: 	/********************************************/
etisserant@0: 	/* B.1.4.1   Directly Represented Variables */
etisserant@0: 	/********************************************/
etisserant@0: {direct_variable}	{yylval.ID=strdup(yytext); return direct_variable_token;}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/******************************************/
etisserant@0: 	/* B 1.4.3 - Declaration & Initialisation */
etisserant@0: 	/******************************************/
etisserant@0: {incompl_location}	{yylval.ID=strdup(yytext); return incompl_location_token;}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/************************/
etisserant@0: 	/* B 1.2.3.1 - Duration */
etisserant@0: 	/************************/
etisserant@0: {fixed_point}		{yylval.ID=strdup(yytext); return fixed_point_token;}
etisserant@0: 
etisserant@0: {fixed_point_d}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return fixed_point_d_token;}
etisserant@0: {integer_d}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return integer_d_token;}
etisserant@0: 
etisserant@0: {fixed_point_h}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return fixed_point_h_token;}
etisserant@0: {integer_h}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return integer_h_token;}
etisserant@0: 
etisserant@0: {fixed_point_m}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return fixed_point_m_token;}
etisserant@0: {integer_m}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return integer_m_token;}
etisserant@0: 
etisserant@0: {fixed_point_s}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return fixed_point_s_token;}
etisserant@0: {integer_s}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-1] = '\0'; return integer_s_token;}
etisserant@0: 
etisserant@0: {fixed_point_ms}	{yylval.ID=strdup(yytext); yylval.ID[yyleng-2] = '\0'; return fixed_point_ms_token;}
etisserant@0: {integer_ms}		{yylval.ID=strdup(yytext); yylval.ID[yyleng-2] = '\0'; return integer_ms_token;}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*******************************/
etisserant@0: 	/* B.1.2.2   Character Strings */
etisserant@0: 	/*******************************/
etisserant@0: {double_byte_character_string} {yylval.ID=strdup(yytext); return double_byte_character_string_token;}
etisserant@0: {single_byte_character_string} {yylval.ID=strdup(yytext); return single_byte_character_string_token;}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/******************************/
etisserant@0: 	/* B.1.2.1   Numeric literals */
etisserant@0: 	/******************************/
etisserant@0: {integer}		{yylval.ID=strdup(yytext); return integer_token;}
etisserant@0: {real}			{yylval.ID=strdup(yytext); return real_token;}
etisserant@0: {binary_integer}	{yylval.ID=strdup(yytext); return binary_integer_token;}
etisserant@0: {octal_integer} 	{yylval.ID=strdup(yytext); return octal_integer_token;}
etisserant@0: {hex_integer} 		{yylval.ID=strdup(yytext); return hex_integer_token;}
etisserant@0: 
etisserant@0: 
etisserant@0: 	/*****************************************/
etisserant@0: 	/* B.1.1 Letters, digits and identifiers */
etisserant@0: 	/*****************************************/
lbessard@3: <st_state>{identifier}/({st_whitespace})"=>"	{yylval.ID=strdup(yytext); return sendto_identifier_token;}
lbessard@3: <il_state>{identifier}/({il_whitespace})"=>"	{yylval.ID=strdup(yytext); return sendto_identifier_token;}
etisserant@0: {identifier} 				{yylval.ID=strdup(yytext);
etisserant@0: 					 /*printf("returning identifier...: %s, %d\n", yytext, get_identifier_token(yytext));*/
etisserant@0: 					 return get_identifier_token(yytext);}
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: 	/************************************************/
etisserant@0: 	/************************************************/
etisserant@0: 	/************************************************/
etisserant@0: 	/*****                                      *****/
etisserant@0: 	/*****                                      *****/
etisserant@0: 	/*****   T H E    L E F T O V E R S . . .   *****/
etisserant@0: 	/*****                                      *****/
etisserant@0: 	/*****                                      *****/
etisserant@0: 	/************************************************/
etisserant@0: 	/************************************************/
etisserant@0: 	/************************************************/
etisserant@0: 
etisserant@0: 	/* do the single character tokens...
etisserant@0: 	 *
etisserant@0: 	 *  e.g.:  ':'  '('  ')'  '+'  '*'  ...
etisserant@0: 	 */
etisserant@0: .	{return yytext[0];}
etisserant@0: 
etisserant@0: 
etisserant@0: %%
etisserant@0: 
etisserant@0: 
etisserant@0: /***********************************/
etisserant@0: /* Utility function definitions... */
etisserant@0: /***********************************/
etisserant@0: 
etisserant@0: /* print the include file stack to stderr... */
etisserant@0: void print_include_stack(void) {
etisserant@0:   int i;
etisserant@0: 
etisserant@0:   if ((include_stack_ptr - 1) >= 0)
etisserant@0:     fprintf (stderr, "in file "); 
etisserant@0:   for (i = include_stack_ptr - 1; i >= 0; i--)
etisserant@0:     fprintf (stderr, "included from file %s:%d\n", include_stack[i].filename, include_stack[i].lineno);
etisserant@0: }
etisserant@0: 
etisserant@0: 
etisserant@0: /* return all the text in the current token back to the input stream, except the first n chars. */
etisserant@0: void unput_text(unsigned int n) {
etisserant@0:   /* it seems that flex has a bug in that it will not correctly count the line numbers
etisserant@0:    * if we return newlines back to the input stream. These newlines will be re-counted
etisserant@0:    * a second time when they are processed again by flex.
etisserant@0:    * We therefore determine how many newlines are in the text we are returning,
etisserant@0:    * and decrement the line counter acordingly...
etisserant@0:    */
etisserant@0:   unsigned int i;
etisserant@0: 
etisserant@0:   for (i = n; i < strlen(yytext); i++)
etisserant@0:     if (yytext[i] == '\n')
etisserant@0:       yylineno--;
etisserant@0: 
etisserant@0:   /* now return all the text back to the input stream... */
etisserant@0:   yyless(n);
etisserant@0: }
etisserant@0: 
etisserant@0: 
etisserant@0: /* Called by flex when it reaches the end-of-file */
etisserant@0: int yywrap(void)
etisserant@0: {
etisserant@0:   /* We reached the end of the input file... */
etisserant@0: 
etisserant@0:   /* Should we continue with another file? */
etisserant@0:   /* If so:
etisserant@0:    *   open the new file...
etisserant@0:    *   return 0;
etisserant@0:    */
etisserant@0: 
etisserant@0:   /* to we stop processing...
etisserant@0:    *
etisserant@0:    *   return 1;
etisserant@0:    */
etisserant@0: 
etisserant@0: 
etisserant@0:   return 1;  /* Stop scanning at end of input file. */
etisserant@0: }
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: /*************************************/
etisserant@0: /* Include a main() function to test */
etisserant@0: /* the token parsing by flex....     */
etisserant@0: /*************************************/
etisserant@0: #ifdef TEST_MAIN
etisserant@0: 
etisserant@0: #include "../util/symtable.hh"
etisserant@0: 
etisserant@0: yystype yylval;
etisserant@0: YYLTYPE yylloc;
etisserant@0: 
etisserant@0: const char *current_filename;
etisserant@0: 
mario@15: 
mario@15: 
etisserant@0: int get_identifier_token(const char *identifier_str) {return 0;}
etisserant@0: 
etisserant@0: 
etisserant@0: 
etisserant@0: int main(int argc, char **argv) {
etisserant@0: 
etisserant@0:   FILE *in_file;
etisserant@0:   int res;
etisserant@0: 
etisserant@0:   if (argc == 1) {
etisserant@0:     /* Work as an interactive (command line) parser... */
etisserant@0:     while((res=yylex()))
etisserant@0:       fprintf(stderr, "(line %d)token: %d\n", yylineno, res);
etisserant@0:   } else {
etisserant@0:     /* Work as non-interactive (file) parser... */
etisserant@0:     if((in_file = fopen(argv[1], "r")) == NULL) {
etisserant@0:       char *errmsg = strdup2("Error opening main file ", argv[1]);
etisserant@0:       perror(errmsg);
etisserant@0:       free(errmsg);
etisserant@0:       return -1;
etisserant@0:     }
etisserant@0: 
etisserant@0:     /* parse the file... */
etisserant@0:     yyin = in_file;
etisserant@0:     current_filename = argv[1];
etisserant@0:     while(1) {
etisserant@0:       res=yylex();
etisserant@0:       fprintf(stderr, "(line %d)token: %d (%s)\n", yylineno, res, yylval.ID);
etisserant@0:     }
etisserant@0:   }
etisserant@0: 
etisserant@0:   return 0;
etisserant@0: 
etisserant@0: }
etisserant@0: #endif