diff options
Diffstat (limited to 'WWW/Library/Implementation/UCDefs.h')
| -rw-r--r-- | WWW/Library/Implementation/UCDefs.h | 79 |
1 files changed, 79 insertions, 0 deletions
diff --git a/WWW/Library/Implementation/UCDefs.h b/WWW/Library/Implementation/UCDefs.h new file mode 100644 index 00000000..0c8966ea --- /dev/null +++ b/WWW/Library/Implementation/UCDefs.h @@ -0,0 +1,79 @@ +/* Definitions for EXP_CHARTRANS */ + +#ifndef UCDEFS_H +#define UCDEFS_H + +typedef struct _LYUCcharset { + int UChndl; + char *MIMEname; + int enc; + int repertoire; + int codepoints; + int cpranges; /* which ranges have valid displayable chars + (including nbsp and shy) */ + int like8859; /* for which ranges is it like 8859-1 */ +} LYUCcharset; + +#define UCT_ENC_7BIT 0 +#define UCT_ENC_8BIT 1 +#define UCT_ENC_8859 2 /* ??? */ +#define UCT_ENC_2022 3 +#define UCT_ENC_MAYBE2022 4 +#define UCT_ENC_CJK 5 +#define UCT_ENC_16BIT 6 +#define UCT_ENC_UTF8 7 + + +#define UCT_REP_SUBSETOF_LAT1 0x01 +#define UCT_REP_SUPERSETOF_LAT1 0x02 +#define UCT_REP_IS_LAT1 UCT_REP_SUBSETOF_LAT1 | UCT_REP_SUPERSETOF_LAT1 +/* assume everything we deal with is included in the UCS2 reperoire, + so a flag for _REP_SUBSETOF_UCS2 would be redundant */ + +/* more general description how the code points relate to 8859-1 and UCS: */ +#define UCT_CP_SUBSETOF_LAT1 0x01 /* implies UCT_CP_SUBSETOF_UCS2 */ +#define UCT_CP_SUPERSETOF_LAT1 0x02 +#define UCT_CP_SUBSETOF_UCS2 0x04 + +#define UCT_CP_IS_LAT1 UCT_CP_SUBSETOF_LAT1 | UCT_CP_SUPERSETOF_LAT1 + +/* more specific bitflags for practically important code point ranges: */ +#define UCT_R_LOWCTRL 0x08 /* 0x00-0x1F, for completeness */ +#define UCT_R_7BITINV 0x10 /* invariant, displayable 7bit chars */ +#define UCT_R_7BITNAT 0x20 /* displayable 7bit, national */ +#define UCT_R_HIGHCTRL 0x40 +#define UCT_R_8859SPECL 0x80 /* special chars in 8859-x sets: nbsp and shy*/ +#define UCT_R_HIGH8BIT 0x100 /* rest of 0xA0-0xFF range */ + +#define UCT_R_ASCII UCT_R_7BITINV | UCT_R_7BITNAT /*displayable US-ASCII*/ +#define UCT_R_LAT1 UCT_R_ASCII | UCT_R_8859SPECL | UCT_R_HIGH8BIT +#define UCT_R_8859S UCT_R_ASCII | UCT_R_8859SPECL +#define UCT_R_8BIT UCT_R_LAT1 | UCT_R_HIGHCTRL /* full 8bit range */ + +/* For the following some coments are in HTAnchor.c. */ + +#define UCT_STAGE_MIME 0 +#define UCT_STAGE_PARSER 1 /* What the parser (SGML.c) gets to see */ +#define UCT_STAGE_STRUCTURED 2 /* What the structured stream (HTML) gets fed*/ +#define UCT_STAGE_HTEXT 3 /* What gets fed to the HText_* functions */ +#define UCT_STAGEMAX 4 + +#define UCT_SETBY_NONE 0 +#define UCT_SETBY_DEFAULT 1 +#define UCT_SETBY_STRUCTURED 2 /* structured stream stage */ +#define UCT_SETBY_PARSER 3 /* set by SGML parser or similar */ +#define UCT_SETBY_MIME 4 /* set explicitly by MIME charset parameter */ + +typedef struct _UCStageInfo +{ + int lock; /* by what it has been set */ + int LYhndl; + LYUCcharset C; +} UCStageInfo; + +typedef struct _UCAnchorInfo +{ + struct _UCStageInfo s[UCT_STAGEMAX]; +} UCAnchorInfo; + +#endif /* UCDEFS_H */ |