From 4f49b16079b27d7ce633619e4cddfdf2bffbc952 Mon Sep 17 00:00:00 2001 From: cashmere Date: Fri, 4 Sep 2026 09:05:24 +0200 Subject: initial commit --- src/wikiformat.c | 3179 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 3179 insertions(+) create mode 100644 src/wikiformat.c (limited to 'src/wikiformat.c') diff --git a/src/wikiformat.c b/src/wikiformat.c new file mode 100644 index 0000000..ef83cd9 --- /dev/null +++ b/src/wikiformat.c @@ -0,0 +1,3179 @@ +/* +** Copyright (c) 2007 D. Richard Hipp +** +** This program is free software; you can redistribute it and/or +** modify it under the terms of the Simplified BSD License (also +** known as the "2-Clause License" or "FreeBSD License".) + +** This program is distributed in the hope that it will be useful, +** but without any warranty; without even the implied warranty of +** merchantability or fitness for a particular purpose. +** +** Author contact information: +** drh@hwaci.com +** http://www.hwaci.com/drh/ +** +******************************************************************************* +** +** This file contains code to do formatting of wiki text. +*/ +#include "config.h" +#include +#include "wikiformat.h" + +#if INTERFACE +/* +** Allowed wiki transformation operations +*/ +#define WIKI_HTMLONLY 0x0001 /* HTML markup only. No wiki */ +#define WIKI_INLINE 0x0002 /* Do not surround with

..

*/ +/* available for reuse: 0x0004 --- formerly WIKI_NOBLOCK */ +#define WIKI_BUTTONS 0x0008 /* Allow sub-menu buttons */ +#define WIKI_NOBADLINKS 0x0010 /* Ignore broken hyperlinks */ +#define WIKI_LINKSONLY 0x0020 /* No markup. Only decorate links */ +#define WIKI_NEWLINE 0x0040 /* Honor \n - break lines at each \n */ +#define WIKI_MARKDOWNLINKS 0x0080 /* Resolve hyperlinks as in markdown */ +#define WIKI_SAFE 0x0100 /* Make the result safe for embedding */ +#define WIKI_TARGET_BLANK 0x0200 /* Hyperlinks go to a new window */ +#define WIKI_NOBRACKET 0x0400 /* Omit extra [..] around hyperlinks */ +#define WIKI_ADMIN 0x0800 /* Ignore g.perm.Hyperlink */ +#define WIKI_MARK 0x1000 /* Add .. around problems */ + +/* +** Return values from wiki_convert +*/ +#define RENDER_LINK 0x0001 /* One or more hyperlinks rendered */ +#define RENDER_ENTITY 0x0002 /* One or more HTML entities (ex: <) */ +#define RENDER_TAG 0x0004 /* One or more HTML tags */ +#define RENDER_BLOCKTAG 0x0008 /* One or more HTML block tags (ex:

) */ +#define RENDER_BLOCK 0x0010 /* Block wiki (paragraphs, etc.) */ +#define RENDER_MARK 0x0020 /* Output contains .. */ +#define RENDER_BADLINK 0x0100 /* Bad hyperlink syntax seen */ +#define RENDER_BADTARGET 0x0200 /* Bad hyperlink target */ +#define RENDER_BADTAG 0x0400 /* Bad HTML tag or tag syntax */ +#define RENDER_BADENTITY 0x0800 /* Bad HTML entity syntax */ +#define RENDER_BADHTML 0x1000 /* Bad HTML seen */ +#define RENDER_ERROR 0x8000 /* Some other kind of error */ +/* Composite values: */ +#define RENDER_ANYERROR 0x9f00 /* Mask for any kind of error */ + +#endif /* INTERFACE */ + + +/* +** These are the only markup attributes allowed. +*/ +enum allowed_attr_t { + ATTR_ALIGN = 1, + ATTR_ALT, + ATTR_BGCOLOR, + ATTR_BORDER, + ATTR_CELLPADDING, + ATTR_CELLSPACING, + ATTR_CITE, + ATTR_CLASS, + ATTR_CLEAR, + ATTR_COLOR, + ATTR_COLSPAN, + ATTR_COMPACT, + ATTR_DATETIME, + ATTR_FACE, + ATTR_FOR, + ATTR_HEIGHT, + ATTR_HIGH, + ATTR_HREF, + ATTR_HSPACE, + ATTR_ID, + ATTR_LINKS, + ATTR_LOW, + ATTR_MAX, + ATTR_MEDIA, + ATTR_MIN, + ATTR_NAME, + ATTR_OPEN, + ATTR_OPTIMUM, + ATTR_ROWSPAN, + ATTR_SIZE, + ATTR_SIZES, + ATTR_SRC, + ATTR_SRCSET, + ATTR_START, + ATTR_STYLE, + ATTR_TARGET, + ATTR_TITLE, + ATTR_TYPE, + ATTR_VALIGN, + ATTR_VALUE, + ATTR_VSPACE, + ATTR_WIDTH +}; + +typedef uint64_t amsk_t; +#define AMSK_ALIGN ((amsk_t)1 << 0) +#define AMSK_ALT ((amsk_t)1 << 1) +#define AMSK_BGCOLOR ((amsk_t)1 << 2) +#define AMSK_BORDER ((amsk_t)1 << 3) +#define AMSK_CELLPADDING ((amsk_t)1 << 4) +#define AMSK_CELLSPACING ((amsk_t)1 << 5) +#define AMSK_CLASS ((amsk_t)1 << 6) +#define AMSK_CLEAR ((amsk_t)1 << 7) +#define AMSK_COLOR ((amsk_t)1 << 8) +#define AMSK_COLSPAN ((amsk_t)1 << 9) +#define AMSK_COMPACT ((amsk_t)1 << 10) +#define AMSK_FACE ((amsk_t)1 << 11) +#define AMSK_HEIGHT ((amsk_t)1 << 12) +#define AMSK_HREF ((amsk_t)1 << 13) +#define AMSK_HSPACE ((amsk_t)1 << 14) +#define AMSK_ID ((amsk_t)1 << 15) +#define AMSK_LINKS ((amsk_t)1 << 16) +#define AMSK_NAME ((amsk_t)1 << 17) +#define AMSK_OPEN ((amsk_t)1 << 18) +#define AMSK_ROWSPAN ((amsk_t)1 << 19) +#define AMSK_SIZE ((amsk_t)1 << 20) +#define AMSK_SRC ((amsk_t)1 << 21) +#define AMSK_START ((amsk_t)1 << 22) +#define AMSK_STYLE ((amsk_t)1 << 23) +#define AMSK_TARGET ((amsk_t)1 << 24) +#define AMSK_TITLE ((amsk_t)1 << 25) +#define AMSK_TYPE ((amsk_t)1 << 26) +#define AMSK_VALIGN ((amsk_t)1 << 27) +#define AMSK_VALUE ((amsk_t)1 << 28) +#define AMSK_VSPACE ((amsk_t)1 << 29) +#define AMSK_WIDTH ((amsk_t)1 << 30) +#define AMSK_CITE ((amsk_t)1 << 31) +#define AMSK_DATETIME ((amsk_t)1 << 32) + +static const struct AllowedAttribute { + const char *zName; + amsk_t iMask; +} aAttribute[] = { + /* These indexes MUST line up with their + corresponding allowed_attr_t enum values. + */ + { 0, 0 }, + { "align", AMSK_ALIGN }, + { "alt", AMSK_ALT }, + { "bgcolor", AMSK_BGCOLOR }, + { "border", AMSK_BORDER }, + { "cellpadding", AMSK_CELLPADDING }, + { "cellspacing", AMSK_CELLSPACING }, + { "cite", AMSK_CITE }, + { "class", AMSK_CLASS }, + { "clear", AMSK_CLEAR }, + { "color", AMSK_COLOR }, + { "colspan", AMSK_COLSPAN }, + { "compact", AMSK_COMPACT }, + { "datetime", AMSK_DATETIME }, + { "face", AMSK_FACE }, + { "for", 0 }, + { "height", AMSK_HEIGHT }, + { "high", 0 }, + { "href", AMSK_HREF }, + { "hspace", AMSK_HSPACE }, + { "id", AMSK_ID }, + { "links", AMSK_LINKS }, + { "low", 0 }, + { "max", 0 }, + { "media", 0 }, + { "min", 0 }, + { "name", AMSK_NAME }, + { "open", AMSK_OPEN }, + { "optimum", 0 }, + { "rowspan", AMSK_ROWSPAN }, + { "size", AMSK_SIZE }, + { "sizes", 0 }, + { "src", AMSK_SRC }, + { "srcset", 0 }, + { "start", AMSK_START }, + { "style", AMSK_STYLE }, + { "target", AMSK_TARGET }, + { "title", AMSK_TITLE }, + { "type", AMSK_TYPE }, + { "valign", AMSK_VALIGN }, + { "value", AMSK_VALUE }, + { "vspace", AMSK_VSPACE }, + { "width", AMSK_WIDTH }, +}; + +/* +** Use binary search to locate a tag in the aAttribute[] table. +*/ +static int findAttr(const char *z){ + int i, c, first, last; + first = 1; + last = count(aAttribute) - 1; + while( first<=last ){ + i = (first+last)/2; + c = fossil_strcmp(aAttribute[i].zName, z); + if( c==0 ){ + return i; + }else if( c<0 ){ + first = i+1; + }else{ + last = i-1; + } + } + return 0; +} + + + +/* +** Allowed markup. +** +** Except for MARKUP_INVALID, this must all be in alphabetical order +** and in numerical sequence. The first markup type must be zero. +** The value for MARKUP_XYZ must correspond to the entry +** in aMarkup[]. +*/ +enum markup_t { + MARKUP_INVALID = 0, + MARKUP_A, + MARKUP_ABBR, + MARKUP_ADDRESS, + MARKUP_HTML5_ARTICLE, + MARKUP_HTML5_ASIDE, + MARKUP_B, + MARKUP_BIG, + MARKUP_BLOCKQUOTE, + MARKUP_BR, + MARKUP_CAPTION, + MARKUP_CENTER, + MARKUP_CITE, + MARKUP_CODE, + MARKUP_COL, + MARKUP_COLGROUP, + MARKUP_DD, + MARKUP_DEL, + MARKUP_DETAILS, + MARKUP_DFN, + MARKUP_DIV, + MARKUP_DL, + MARKUP_DT, + MARKUP_EM, + MARKUP_FIGCAPTION, + MARKUP_FIGURE, + MARKUP_FONT, + MARKUP_HTML5_FOOTER, + MARKUP_H1, + MARKUP_H2, + MARKUP_H3, + MARKUP_H4, + MARKUP_H5, + MARKUP_H6, + MARKUP_HTML5_HEADER, + MARKUP_HR, + MARKUP_I, + MARKUP_IMG, + MARKUP_INS, + MARKUP_KBD, + MARKUP_LABEL, + MARKUP_LI, + MARKUP_MARK, + MARKUP_METER, + MARKUP_HTML5_NAV, + MARKUP_NOBR, + MARKUP_NOWIKI, + MARKUP_OL, + MARKUP_P, + MARKUP_PICTURE, + MARKUP_PRE, + MARKUP_PROGRESS, + MARKUP_Q, + MARKUP_S, + MARKUP_SAMP, + MARKUP_HTML5_SECTION, + MARKUP_SMALL, + MARKUP_SOURCE, + MARKUP_SPAN, + MARKUP_STRIKE, + MARKUP_STRONG, + MARKUP_SUB, + MARKUP_SUMMARY, + MARKUP_SUP, + MARKUP_TABLE, + MARKUP_TBODY, + MARKUP_TD, + MARKUP_TFOOT, + MARKUP_TH, + MARKUP_THEAD, + MARKUP_TIME, + MARKUP_TITLE, + MARKUP_TR, + MARKUP_TT, + MARKUP_U, + MARKUP_UL, + MARKUP_VAR, + MARKUP_VERBATIM, + MARKUP_WBR +}; + +/* +** The various markup is divided into the following types: +*/ +#define MUTYPE_SINGLE 0x0001 /* ,
, or

*/ +#define MUTYPE_BLOCK 0x0002 /* Forms a new paragraph. ex:

,

*/ +#define MUTYPE_FONT 0x0004 /* Font changes. ex: , , */ +#define MUTYPE_LIST 0x0010 /* Lists.
    ,