/* $Id$ */
/* $Name$ */
-#ifndef IN_LEX
+#if LEX_SECTION == 1
#include "gedcom_internal.h"
#include "multilex.h"
static int level_diff=MAXGEDCLEVEL;
static size_t line_len = 0;
-static char ptr_buf[MAXGEDCPTRLEN * UTF_FACTOR + 1];
-static char tag_buf[MAXGEDCTAGLEN * UTF_FACTOR + 1];
-static char str_buf[MAXGEDCLINELEN * UTF_FACTOR + 1];
+static struct conv_buffer* ptr_buffer = NULL;
+static struct conv_buffer* tag_buffer = NULL;
+static struct conv_buffer* str_buffer = NULL;
+
+#define INITIAL_PTR_BUFFER_LEN MAXGEDCPTRLEN * UTF_FACTOR + 1
+#define INITIAL_TAG_BUFFER_LEN MAXGEDCTAGLEN * UTF_FACTOR + 1
+#define INITIAL_STR_BUFFER_LEN MAXGEDCLINELEN * UTF_FACTOR + 1
#ifdef LEXER_TEST
YYSTYPE gedcom_lval;
int line_no = 1;
-int compat_at = 0;
int gedcom_lex();
fprintf(stderr, "(%d) %s\n", type, msg);
}
-int test_loop(ENCODING enc, char* code)
+int test_loop(ENCODING enc, const char* code)
{
int tok, res;
init_encodings();
level_diff);
}
-static void error_tag_too_long(char *tag)
+static void error_tag_too_long(const char *tag)
{
gedcom_error(_("Tag '%s' too long, max %d characters allowed"),
tag, MAXGEDCTAGLEN);
}
-static void error_invalid_character(char *str, char ch)
+static void error_invalid_character(const char *str, char ch)
{
gedcom_error(_("Invalid character for encoding: '%s' (0x%02x)"), str, ch);
}
-static void error_pointer_too_long(char *ptr)
+static void error_pointer_too_long(const char *ptr)
{
gedcom_error(_("Pointer '%s' too long, max %d characters allowed"),
ptr, MAXGEDCPTRLEN);
gedcom_error(_("'@' character should be written as '@@' in values"));
}
-static void error_unexpected_character(char* str, char ch)
+static void error_unexpected_character(const char* str, char ch)
{
gedcom_error(_("Unexpected character: '%s' (0x%02x)"), str, ch);
}
-#else /* of #ifndef IN_LEX */
+/* This is to bypass the iconv conversion (if the input is UTF-8 coming
+ from the program) */
+static int dummy_conv = 0;
+
+#elif LEX_SECTION == 2
#define TO_INTERNAL(STR,OUTBUF) \
- to_internal(STR, yyleng, OUTBUF, sizeof(OUTBUF))
+ (dummy_conv ? STR : to_internal(STR, yyleng, OUTBUF))
#define INIT_LINE_LEN \
line_len = 0;
#define MKTAGACTION(THETAG) \
{ CHECK_LINE_LEN; \
- gedcom_lval.tag.string = TO_INTERNAL(yytext, tag_buf); \
+ gedcom_lval.tag.string = TO_INTERNAL(yytext, tag_buffer); \
gedcom_lval.tag.value = TAG_##THETAG; \
BEGIN(NORMAL); \
+ line_no++; \
return TAG_##THETAG; \
}
#define ACTION_DIGITS \
- { int level = atoi(TO_INTERNAL(yytext, str_buf)); \
+ { int level = atoi(TO_INTERNAL(yytext, str_buffer)); \
CHECK_LINE_LEN; \
if ((level < 0) || (level > MAXGEDCLEVEL)) { \
error_level_out_of_range(); \
+ line_no++; \
return BADTOKEN; \
} \
level_diff = level - current_level; \
else { \
/* should never happen (error to GEDCOM spec) */ \
error_level_too_high(level_diff); \
+ line_no++; \
return BADTOKEN; \
} \
}
#define ACTION_ALPHANUM \
{ if (strlen(yytext) > MAXGEDCTAGLEN * encoding_width) { \
error_tag_too_long(yytext); \
+ line_no++; \
return BADTOKEN; \
} \
CHECK_LINE_LEN; \
- gedcom_lval.tag.string = TO_INTERNAL(yytext, tag_buf); \
+ gedcom_lval.tag.string = TO_INTERNAL(yytext, tag_buffer); \
gedcom_lval.tag.value = USERTAG; \
BEGIN(NORMAL); \
+ line_no++; \
return USERTAG; \
}
#define ACTION_DELIM \
{ CHECK_LINE_LEN; \
- gedcom_lval.string = TO_INTERNAL(yytext, str_buf); \
+ gedcom_lval.string = TO_INTERNAL(yytext, str_buffer); \
return DELIM; \
}
#define ACTION_ANY \
{ char* tmp; \
CHECK_LINE_LEN; \
- tmp = TO_INTERNAL(yytext, str_buf); \
+ tmp = TO_INTERNAL(yytext, str_buffer); \
if (!tmp) { \
/* Something went wrong during conversion... */ \
error_invalid_character(yytext, yytext[0]); \
#define ACTION_ESCAPE \
{ CHECK_LINE_LEN; \
- gedcom_lval.string = TO_INTERNAL(yytext, str_buf); \
+ gedcom_lval.string = TO_INTERNAL(yytext, str_buffer); \
return ESCAPE; \
}
error_pointer_too_long(yytext); \
return BADTOKEN; \
} \
- gedcom_lval.string = TO_INTERNAL(yytext, ptr_buf); \
+ gedcom_lval.string = TO_INTERNAL(yytext, ptr_buffer); \
return POINTER; \
}
#define ACTION_TERMINATOR \
{ CHECK_LINE_LEN; \
INIT_LINE_LEN; \
- line_no++; \
BEGIN(INITIAL); \
}
} \
else { \
char* ptr; int size; \
- /* Reset our state */ \
- current_level = -1; \
- level_diff = MAXGEDCLEVEL; \
- /* ... then terminate lex */ \
+ /* ... terminate lex */ \
yyterminate(); \
/* Get rid of f*cking compiler warning from lex generated code */ \
/* yyterminate does return(), so program will never come here */ \
}
#define ACTION_NORMAL_AT \
- { if (compat_at) { \
+ { if (compat_mode(C_NO_DOUBLE_AT)) { \
int i, j; \
char *yycopy = strdup(yytext); \
if (yycopy) { \
return BADTOKEN; \
}
-#endif /* IN_LEX */
+#elif LEX_SECTION == 3
+
+int yywrap()
+{
+ return 1;
+}
+
+static void free_conv_buffers()
+{
+ free_conv_buffer(ptr_buffer);
+ free_conv_buffer(tag_buffer);
+ free_conv_buffer(str_buffer);
+}
+
+static void yylex_cleanup()
+{
+ /* fix memory leak in lex */
+ yy_delete_buffer(yy_current_buffer);
+ yy_current_buffer = NULL;
+ free_conv_buffers();
+}
+
+static void init_conv_buffers()
+{
+ if (!ptr_buffer) {
+ ptr_buffer = create_conv_buffer(INITIAL_PTR_BUFFER_LEN);
+ tag_buffer = create_conv_buffer(INITIAL_TAG_BUFFER_LEN);
+ str_buffer = create_conv_buffer(INITIAL_STR_BUFFER_LEN);
+ }
+}
+
+static int exitfuncregistered = 0;
+
+void yymyinit(FILE *f)
+{
+ if (! exitfuncregistered && atexit(yylex_cleanup) == 0)
+ exitfuncregistered = 1;
+ init_conv_buffers();
+ yyin = f;
+ yyrestart(f);
+ /* Reset our state */
+ current_level = -1;
+ level_diff = MAXGEDCLEVEL;
+ BEGIN(INITIAL);
+}
+
+#endif
projects / gedcom-parse.git / blobdiff