X-Git-Url: https://git.dlugolecki.net.pl/?a=blobdiff_plain;f=gedcom%2Fgedcom_lex_common.c;h=149a8a8bda9630855cb7fed8115ac38bd1fdfcec;hb=eb6f3993afd2330bf4ae536bc8679d791b681dc4;hp=d0d329cab5e031d81a20bcab027e8bed6384ada1;hpb=2468a5aefafdfbd1982fe3214e3866e072892506;p=gedcom-parse.git diff --git a/gedcom/gedcom_lex_common.c b/gedcom/gedcom_lex_common.c index d0d329c..149a8a8 100644 --- a/gedcom/gedcom_lex_common.c +++ b/gedcom/gedcom_lex_common.c @@ -26,6 +26,7 @@ #include "gedcom_internal.h" #include "multilex.h" #include "encoding.h" +#include "encoding_state.h" #include "gedcom.h" #include "gedcom.tabgen.h" #include "compat.h" @@ -34,15 +35,19 @@ static size_t encoding_width; static int current_level = -1; static int level_diff=MAXGEDCLEVEL; static size_t line_len = 0; +static int tab_space = 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(); @@ -88,9 +93,10 @@ int test_loop(ENCODING enc, const char* code) /* These are defined as functions here, because xgettext has trouble extracting the strings out of long pre-processor defined */ -static void error_line_too_long() +static void error_line_too_long(const char *line) { - gedcom_error(_("Line too long, max %d characters allowed"), MAXGEDCLINELEN); + gedcom_error(_("Line too long, max %d characters allowed: %s"), + MAXGEDCLINELEN, line); } static void error_level_leading_zero() @@ -131,15 +137,24 @@ static void error_at_character() gedcom_error(_("'@' character should be written as '@@' in values")); } +static void error_tab_character() +{ + gedcom_error(_("Tab character is not allowed in values")); +} + static void error_unexpected_character(const char* str, char ch) { gedcom_error(_("Unexpected character: '%s' (0x%02x)"), str, ch); } +/* 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; @@ -148,18 +163,25 @@ static void error_unexpected_character(const char* str, char ch) { if (line_len != (size_t)-1) { \ line_len += strlen(yytext); \ if (line_len > MAXGEDCLINELEN * encoding_width) { \ - error_line_too_long(); \ + error_line_too_long(yytext); \ line_len = (size_t)-1; \ return BADTOKEN; \ } \ } \ } +#define GENERATE_TAB_SPACE \ + { gedcom_lval.string = " "; \ + tab_space--; \ + return DELIM; \ + } + #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; \ } @@ -196,10 +218,16 @@ static void error_unexpected_character(const char* str, char ch) But because this means that one token is converted into a series of tokens, there is some initial code following immediately here - that returns "pending" tokens. */ + that returns "pending" tokens. + + Also, for compatibility tabs are converted into spaces, which is + also handled here */ #define ACTION_BEFORE_REGEXPS \ - { if (level_diff < 1) { \ + { if (compat_mode(C_TAB_CHARACTER) && tab_space-- > 0) { \ + GENERATE_TAB_SPACE; \ + } \ + else if (level_diff < 1) { \ level_diff++; \ return CLOSE; \ } \ @@ -227,10 +255,11 @@ static void error_unexpected_character(const char* str, char ch) #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; \ @@ -248,6 +277,7 @@ static void error_unexpected_character(const char* str, char ch) else { \ /* should never happen (error to GEDCOM spec) */ \ error_level_too_high(level_diff); \ + line_no++; \ return BADTOKEN; \ } \ } @@ -256,19 +286,21 @@ static void error_unexpected_character(const char* str, char ch) #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; \ } @@ -276,7 +308,7 @@ static void error_unexpected_character(const char* str, char ch) #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]); \ @@ -298,7 +330,7 @@ static void error_unexpected_character(const char* str, char ch) #define ACTION_ESCAPE \ { CHECK_LINE_LEN; \ - gedcom_lval.string = TO_INTERNAL(yytext, str_buf); \ + gedcom_lval.string = TO_INTERNAL(yytext, str_buffer); \ return ESCAPE; \ } @@ -309,7 +341,7 @@ static void error_unexpected_character(const char* str, char ch) 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; \ } @@ -323,7 +355,8 @@ static void error_unexpected_character(const char* str, char ch) #define ACTION_TERMINATOR \ { CHECK_LINE_LEN; \ INIT_LINE_LEN; \ - line_no++; \ + if (line_no == 1) \ + set_read_encoding_terminator(TO_INTERNAL(yytext, str_buffer)); \ BEGIN(INITIAL); \ } @@ -349,7 +382,7 @@ static void error_unexpected_character(const char* str, char ch) } #define ACTION_NORMAL_AT \ - { if (compat_at) { \ + { if (compat_mode(C_NO_DOUBLE_AT)) { \ int i, j; \ char *yycopy = strdup(yytext); \ if (yycopy) { \ @@ -368,6 +401,17 @@ static void error_unexpected_character(const char* str, char ch) } \ } +#define ACTION_TAB \ + { if (compat_mode(C_TAB_CHARACTER)) { \ + tab_space = 8; \ + GENERATE_TAB_SPACE; \ + } \ + else { \ + error_tab_character(); \ + return BADTOKEN; \ + } \ + } + #define ACTION_UNEXPECTED \ { error_unexpected_character(yytext, yytext[0]); \ return BADTOKEN; \ @@ -380,11 +424,28 @@ 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; @@ -393,11 +454,13 @@ 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