X-Git-Url: https://git.dlugolecki.net.pl/?a=blobdiff_plain;f=gedcom.y;h=5416d50b67527e7193bbedb9159ee81dd5f9e86e;hb=4a9af1fa889f85ce33ae94abec4ff5df002036be;hp=68b2f488755b53b04c286bebbc3a0f7ddb522e65;hpb=cf91051d49e34888991303c62de71ab4f0bc6db1;p=gedcom-parse.git diff --git a/gedcom.y b/gedcom.y index 68b2f48..5416d50 100644 --- a/gedcom.y +++ b/gedcom.y @@ -1,3 +1,12 @@ +/* This program is free software; you can redistribute it and/or modify * + * it under the terms of the GNU General Public License as published by * + * the Free Software Foundation; either version 2 of the License, or * + * (at your option) any later version. * + + (C) 2001 by The Genes Development Team + Original author: Peter Verthez (Peter.Verthez@advalvas.be) +*/ + /* $Id$ */ /* $Name$ */ @@ -122,13 +131,16 @@ %{ #include "gedcom.h" +#include "multilex.h" +#include "encoding.h" int count_level = 0; int fail = 0; int compat_enabled = 1; +int gedcom_high_level_debug = 0; int compatibility = 0; -MECHANISM curr_mechanism=IMMED_FAIL; -char string_buf[MAXGEDCLINELEN+1]; +MECHANISM error_mechanism=IMMED_FAIL; +char string_buf[MAXGEDCLINELEN*4+1]; char *string_buf_ptr; enum _COMPAT { @@ -149,13 +161,13 @@ int compat_mode(int flags); #define HANDLE_ERROR \ { \ - if (curr_mechanism == IMMED_FAIL) { \ + if (error_mechanism == IMMED_FAIL) { \ YYABORT; \ } \ - else if (curr_mechanism == DEFER_FAIL) { \ + else if (error_mechanism == DEFER_FAIL) { \ yyerrok; fail = 1; \ } \ - else if (curr_mechanism == IGNORE_ERRORS) { \ + else if (error_mechanism == IGNORE_ERRORS) { \ yyerrok; \ } \ } @@ -210,6 +222,9 @@ int compat_mode(int flags); %} %union { + int level; + char *pointer; + char *tag; char *string; } @@ -217,145 +232,145 @@ int compat_mode(int flags); %expect 300 %token BADTOKEN -%token OPEN +%token OPEN %token CLOSE %token ESCAPE %token DELIM %token ANYCHAR -%token POINTER -%token USERTAG -%token TAG_ABBR -%token TAG_ADDR -%token TAG_ADR1 -%token TAG_ADR2 -%token TAG_ADOP -%token TAG_AFN -%token TAG_AGE -%token TAG_AGNC -%token TAG_ALIA -%token TAG_ANCE -%token TAG_ANCI -%token TAG_ANUL -%token TAG_ASSO -%token TAG_AUTH -%token TAG_BAPL -%token TAG_BAPM -%token TAG_BARM -%token TAG_BASM -%token TAG_BIRT -%token TAG_BLES -%token TAG_BLOB -%token TAG_BURI -%token TAG_CALN -%token TAG_CAST -%token TAG_CAUS -%token TAG_CENS -%token TAG_CHAN -%token TAG_CHAR -%token TAG_CHIL -%token TAG_CHR -%token TAG_CHRA -%token TAG_CITY -%token TAG_CONC -%token TAG_CONF -%token TAG_CONL -%token TAG_CONT -%token TAG_COPR -%token TAG_CORP -%token TAG_CREM -%token TAG_CTRY -%token TAG_DATA -%token TAG_DATE -%token TAG_DEAT -%token TAG_DESC -%token TAG_DESI -%token TAG_DEST -%token TAG_DIV -%token TAG_DIVF -%token TAG_DSCR -%token TAG_EDUC -%token TAG_EMIG -%token TAG_ENDL -%token TAG_ENGA -%token TAG_EVEN -%token TAG_FAM -%token TAG_FAMC -%token TAG_FAMF -%token TAG_FAMS -%token TAG_FCOM -%token TAG_FILE -%token TAG_FORM -%token TAG_GEDC -%token TAG_GIVN -%token TAG_GRAD -%token TAG_HEAD -%token TAG_HUSB -%token TAG_IDNO -%token TAG_IMMI -%token TAG_INDI -%token TAG_LANG -%token TAG_LEGA -%token TAG_MARB -%token TAG_MARC -%token TAG_MARL -%token TAG_MARR -%token TAG_MARS -%token TAG_MEDI -%token TAG_NAME -%token TAG_NATI -%token TAG_NATU -%token TAG_NCHI -%token TAG_NICK -%token TAG_NMR -%token TAG_NOTE -%token TAG_NPFX -%token TAG_NSFX -%token TAG_OBJE -%token TAG_OCCU -%token TAG_ORDI -%token TAG_ORDN -%token TAG_PAGE -%token TAG_PEDI -%token TAG_PHON -%token TAG_PLAC -%token TAG_POST -%token TAG_PROB -%token TAG_PROP -%token TAG_PUBL -%token TAG_QUAY -%token TAG_REFN -%token TAG_RELA -%token TAG_RELI -%token TAG_REPO -%token TAG_RESI -%token TAG_RESN -%token TAG_RETI -%token TAG_RFN -%token TAG_RIN -%token TAG_ROLE -%token TAG_SEX -%token TAG_SLGC -%token TAG_SLGS -%token TAG_SOUR -%token TAG_SPFX -%token TAG_SSN -%token TAG_STAE -%token TAG_STAT -%token TAG_SUBM -%token TAG_SUBN -%token TAG_SURN -%token TAG_TEMP -%token TAG_TEXT -%token TAG_TIME -%token TAG_TITL -%token TAG_TRLR -%token TAG_TYPE -%token TAG_VERS -%token TAG_WIFE -%token TAG_WILL - -%type anystdtag -%type anytoptag +%token POINTER +%token USERTAG +%token TAG_ABBR +%token TAG_ADDR +%token TAG_ADR1 +%token TAG_ADR2 +%token TAG_ADOP +%token TAG_AFN +%token TAG_AGE +%token TAG_AGNC +%token TAG_ALIA +%token TAG_ANCE +%token TAG_ANCI +%token TAG_ANUL +%token TAG_ASSO +%token TAG_AUTH +%token TAG_BAPL +%token TAG_BAPM +%token TAG_BARM +%token TAG_BASM +%token TAG_BIRT +%token TAG_BLES +%token TAG_BLOB +%token TAG_BURI +%token TAG_CALN +%token TAG_CAST +%token TAG_CAUS +%token TAG_CENS +%token TAG_CHAN +%token TAG_CHAR +%token TAG_CHIL +%token TAG_CHR +%token TAG_CHRA +%token TAG_CITY +%token TAG_CONC +%token TAG_CONF +%token TAG_CONL +%token TAG_CONT +%token TAG_COPR +%token TAG_CORP +%token TAG_CREM +%token TAG_CTRY +%token TAG_DATA +%token TAG_DATE +%token TAG_DEAT +%token TAG_DESC +%token TAG_DESI +%token TAG_DEST +%token TAG_DIV +%token TAG_DIVF +%token TAG_DSCR +%token TAG_EDUC +%token TAG_EMIG +%token TAG_ENDL +%token TAG_ENGA +%token TAG_EVEN +%token TAG_FAM +%token TAG_FAMC +%token TAG_FAMF +%token TAG_FAMS +%token TAG_FCOM +%token TAG_FILE +%token TAG_FORM +%token TAG_GEDC +%token TAG_GIVN +%token TAG_GRAD +%token TAG_HEAD +%token TAG_HUSB +%token TAG_IDNO +%token TAG_IMMI +%token TAG_INDI +%token TAG_LANG +%token TAG_LEGA +%token TAG_MARB +%token TAG_MARC +%token TAG_MARL +%token TAG_MARR +%token TAG_MARS +%token TAG_MEDI +%token TAG_NAME +%token TAG_NATI +%token TAG_NATU +%token TAG_NCHI +%token TAG_NICK +%token TAG_NMR +%token TAG_NOTE +%token TAG_NPFX +%token TAG_NSFX +%token TAG_OBJE +%token TAG_OCCU +%token TAG_ORDI +%token TAG_ORDN +%token TAG_PAGE +%token TAG_PEDI +%token TAG_PHON +%token TAG_PLAC +%token TAG_POST +%token TAG_PROB +%token TAG_PROP +%token TAG_PUBL +%token TAG_QUAY +%token TAG_REFN +%token TAG_RELA +%token TAG_RELI +%token TAG_REPO +%token TAG_RESI +%token TAG_RESN +%token TAG_RETI +%token TAG_RFN +%token TAG_RIN +%token TAG_ROLE +%token TAG_SEX +%token TAG_SLGC +%token TAG_SLGS +%token TAG_SOUR +%token TAG_SPFX +%token TAG_SSN +%token TAG_STAE +%token TAG_STAT +%token TAG_SUBM +%token TAG_SUBN +%token TAG_SURN +%token TAG_TEMP +%token TAG_TEXT +%token TAG_TIME +%token TAG_TITL +%token TAG_TRLR +%token TAG_TYPE +%token TAG_VERS +%token TAG_WIFE +%token TAG_WILL + +%type anystdtag +%type anytoptag %type line_item %type mand_line_item %type note_line_item @@ -388,13 +403,11 @@ record : fam_rec head_sect : OPEN DELIM TAG_HEAD { START(HEAD) } head_subs - { if (compat_mode(C_FTREE)) { - CHECK3(SOUR, GEDC, CHAR); - } - else { - CHECK4(SOUR, SUBM, GEDC, CHAR); - } - } + { if (compat_mode(C_FTREE)) + CHECK3(SOUR, GEDC, CHAR) + else + CHECK4(SOUR, SUBM, GEDC, CHAR) + } CLOSE { } ; @@ -420,6 +433,7 @@ head_sub : head_sour_sect { OCCUR2(SOUR, 1, 1) } /* HEAD.SOUR */ head_sour_sect : OPEN DELIM TAG_SOUR mand_line_item { set_compatibility($4); + gedcom_debug_print("===Source: '%s'\n", $4); START(SOUR) } head_sour_subs @@ -440,14 +454,17 @@ head_sour_sub : head_sour_vers_sect { OCCUR2(VERS, 0, 1) } head_sour_vers_sect : OPEN DELIM TAG_VERS mand_line_item { START(VERS)} no_std_subs { CHECK0 } CLOSE - { } + { gedcom_debug_print("===Source version: '%s'\n", $4); + } ; head_sour_name_sect : OPEN DELIM TAG_NAME mand_line_item { START(NAME) } no_std_subs { CHECK0 } CLOSE - { } + { gedcom_debug_print("===Source name: '%s'\n", $4); + } ; head_sour_corp_sect : OPEN DELIM TAG_CORP mand_line_item - { START(CORP) } + { gedcom_debug_print("===Source corp name: '%s'\n", $4); + START(CORP) } head_sour_corp_subs { CHECK0 } CLOSE @@ -564,7 +581,8 @@ head_gedc_form_sect : OPEN DELIM TAG_FORM mand_line_item /* HEAD.CHAR */ head_char_sect : OPEN DELIM TAG_CHAR mand_line_item - { START(CHAR) } + { if (open_conv_to_internal($4) == 0) YYERROR; + START(CHAR) } head_char_subs { CHECK0 } CLOSE @@ -2106,10 +2124,15 @@ opt_line_item : /* empty */ { } | DELIM line_item { } ; -line_item : anychar { CLEAR_BUFFER(string_buf); +line_item : anychar { size_t i; + CLEAR_BUFFER(string_buf); string_buf_ptr = string_buf; /* The following also takes care of '@@' */ - *string_buf_ptr++ = $1[0]; + if (!strncmp($1, "@@", 3)) + *string_buf_ptr++ = '@'; + else + for (i=0; i < strlen($1); i++) + *string_buf_ptr++ = $1[i]; $$ = string_buf; } | ESCAPE { CLEAR_BUFFER(string_buf); @@ -2123,8 +2146,13 @@ line_item : anychar { CLEAR_BUFFER(string_buf); YYERROR; } else { + size_t i; /* The following also takes care of '@@' */ - *string_buf_ptr++ = $2[0]; + if (!strncmp($2, "@@", 3)) + *string_buf_ptr++ = '@'; + else + for (i=0; i < strlen($2); i++) + *string_buf_ptr++ = $2[i]; $$ = string_buf; } } @@ -2375,17 +2403,37 @@ void pop_countarray() } /* Enabling debug mode */ -void gedcom_enable_debug() +/* level 0: no debugging */ +/* level 1: only internal */ +/* level 2: also bison */ +void gedcom_set_debug_level(int level) { + if (level > 0) { + gedcom_high_level_debug = 1; + } + if (level > 1) { #if YYDEBUG != 0 - gedcom_debug = 1; + gedcom_debug = 1; #endif + } +} + +int gedcom_debug_print(char* s, ...) +{ + int res; + if (gedcom_high_level_debug) { + va_list ap; + va_start(ap, s); + res = vfprintf(stderr, s, ap); + va_end(ap); + } + return(res); } /* Setting the error mechanism */ void gedcom_set_error_handling(MECHANISM mechanism) { - curr_mechanism = mechanism; + error_mechanism = mechanism; } /* Compatibility handling */ @@ -2413,3 +2461,4 @@ int compat_mode(int compat_flags) { return (compat_flags & compatibility); } +