diff options
| -rwxr-xr-x | httpd/cgi-bin/check | 231 |
1 files changed, 35 insertions, 196 deletions
diff --git a/httpd/cgi-bin/check b/httpd/cgi-bin/check index f2d13e4..29373b9 100755 --- a/httpd/cgi-bin/check +++ b/httpd/cgi-bin/check @@ -8,7 +8,7 @@ # This source code is available under the license at: # http://www.w3.org/Consortium/Legal/copyright-software # -# $Id: check,v 1.48 1999-12-01 00:59:31 gerald Exp $ +# $Id: check,v 1.49 1999-12-01 01:10:18 gerald Exp $ # # Load modules @@ -29,11 +29,17 @@ use constant UNDEF => undef; # Constant definitions ############################################################################# -my $cvsrevision = '$Revision: 1.48 $'; -my $cvsdate = '$Date: 1999-12-01 00:59:31 $'; +my $cvsrevision = '$Revision: 1.49 $'; +my $cvsdate = '$Date: 1999-12-01 01:10:18 $'; my $logfile = "/var/log/httpd/val-svc"; +my $base_path = '/usr/local/src/validator/'; +my $fpis_db = $base_path . 'htdocs/config/fpis.cfg'; +my $frag_db = $base_path . 'htdocs/config/frag.cfg'; +my $elem_db = $base_path . 'htdocs/config/eref.cfg'; + +my $element_ref = 'http://www.htmlhelp.com/reference/html40/'; my $uri_def_uri = "http://www.w3.org/Addressing/#terms"; my $faqloc = "http://www.cs.duke.edu/~dsb/kgv-faq/"; my $faqerrloc = "${faqloc}errors.html"; @@ -77,86 +83,11 @@ my $gifborder = " border=0"; my @options = qw(weblint pw outline ss sp noatt); -############################################################################# -# Array of FPIs -> plain text version strings -############################################################################# - -my %pub_ids = ( - '-//IETF//DTD HTML Level 0//EN//2.0', 'HTML 0.0', - '-//IETF//DTD HTML Strict Level 0//EN//2.0', 'Strict HTML 0.0', - - '-//IETF//DTD HTML 2.0 Level 1//EN', 'HTML 1.0', - '-//IETF//DTD HTML 2.0 Strict Level 1//EN', 'Strict HTML 1.0', - - '-//IETF//DTD HTML 2.0 Strict//EN', 'Strict HTML 2.0', - '-//IETF//DTD HTML 2.0//EN', 'HTML 2.0', - '-//IETF//DTD HTML 2.1E//EN', 'HTML 2.1E', - - '-//AS//DTD HTML 3.0 asWedit + extensions//EN', 'HTML 3.0 (AdvaSoft version)', - '-//IETF//DTD HTML 3.0//EN', 'HTML 3.0 (Beta)', - '-//W3O//DTD W3 HTML Strict 3.0//EN//', 'Strict HTML 3.0 (Beta)', - - '-//Sun Microsystems Corp.//DTD HotJava HTML//EN', 'Hotjava-HTML', -'-//Sun Microsystems Corp.//DTD HotJava Strict HTML//EN', 'Strict Hotjava-HTML', - '-//WebTechs//DTD Mozilla HTML 2.0//EN', 'Netscape-HTML', - '-//Netscape Comm. Corp. Strict//DTD HTML//EN', 'Strict Netscape-HTML', - '-//Microsoft//DTD Internet Explorer 2.0 HTML//EN', 'MSIE-HTML', - '-//Microsoft//DTD Internet Explorer 2.0 HTML Strict//EN', 'Strict MSIE-HTML', - '-//Microsoft//DTD Internet Explorer 3.0 HTML//EN', 'MSIE 3.0 HTML', - '-//Microsoft//DTD Internet Explorer 3.0 HTML Strict//EN', 'Strict MSIE 3.0 HTML', - '-//OReilly and Associates//DTD HTML Extended 1.0//EN', 'O\'Reilly HTML Extended v1.0', - '-//OReilly and Associates//DTD HTML Extended Relaxed 1.0//EN', 'O\'Reilly HTML Extended Relaxed v1.0', - - '-//IETF//DTD HTML V2.2//EN', 'HTML 2.2', - '-//W3C//DTD HTML 1996-01//EN', 'HTML 1996-01', - '-//W3C//DTD HTML 3.2 Final//EN', '<a href="http://www.w3.org/TR/REC-html32">HTML 3.2</a>', - '-//W3C//DTD HTML Experimental 970421//EN', '<a href="http://www.w3.org/TR/NOTE-html-970421.html">HTML 3.2 + Style</a>', - '+//Silmaril//DTD HTML Pro v0r11 19970101//EN', '<a href="http://www.ucc.ie/doc/www/html/dtds/htmlpro.html">HTML Pro</a>', - '-//Spyglass//DTD HTML 2.0 Extended//EN', 'Spyglass HTML 2.0 Extended', - 'http://www.w3.org/MarkUp/Cougar/Cougar.dtd', '<a href="http://www.w3.org/MarkUp/Cougar/">HTML Level "Cougar"</a>', - '-//W3C//DTD HTML 4.0//EN', '<a href="http://www.w3.org/TR/REC-html40/">HTML 4.0</a> Strict', - '-//W3C//DTD HTML 4.0 Transitional//EN', '<a href="http://www.w3.org/TR/REC-html40/">HTML 4.0</a> Transitional', - '-//W3C//DTD HTML 4.0 Frameset//EN', '<a href="http://www.w3.org/TR/PR-html40/">HTML 4.0</a> Frameset', - '-//W3C//DTD HTML 4.01//EN', '<a href="http://www.w3.org/TR/1999/PR-html40-19990824/">HTML 4.01</a> Strict', - '-//W3C//DTD HTML 4.01 Transitional//EN', '<a href="http://www.w3.org/TR/1999/PR-html40-19990824/">HTML 4.01</a> Transitional', - '-//W3C//DTD HTML 4.01 Frameset//EN', '<a href="http://www.w3.org/TR/1999/PR-html40-19990824/">HTML 4.01</a> Frameset', - '-//W3C//DTD XHTML 1.0 Strict//EN', '<a href="http://www.w3.org/TR/1999/PR-xhtml1-19990824/">XHTML 1.0</a> Strict', - '-//W3C//DTD XHTML 1.0 Transitional//EN', '<a href="http://www.w3.org/TR/1999/PR-xhtml1-19990824/">XHTML 1.0</a> Transitional', - '-//W3C//DTD XHTML 1.0 Frameset//EN', '<a href="http://www.w3.org/TR/1999/PR-xhtml1-19990824/">XHTML 1.0</a> Frameset', - 'XML', '<a href="http://www.w3.org/TR/REC-xml">XML</a>' - -); - -############################################################################# -# Array of errors -> fragment identifiers for error explanation links -############################################################################# - -my %frag = ( - 'entity end not allowed in comment', 'unterm-comment-1', - 'name start character invalid only s and comment allowed in comment declaration', 'unterm-comment-2', - 'name character invalid only s and comment allowed in comment declaration', 'unterm-comment-2', - 'unknown declaration type FOO', 'bad-comment', - 'character FOO not allowed in attribute specification list', 'attr-char', - 'an attribute value must be a literal unless it contains only name characters', 'attr-quoted', - 'syntax of attribute value does not conform to declared value', 'bad-attr-char', - 'length of attribute value must not exceed LITLEN less NORMSEP', 'name-length', - 'element FOO undefined', 'undef-tag', - 'element FOO not allowed here', 'not-allowed', - 'there is no attribute FOO', 'undef-attr', - 'FOO is not a member of the group specified in the declared value of this attribute', 'undef-attr-val', - 'FOO is not a member of a group specified for any attribute', 'bad-abbrev-attr', - 'end tag for FOO omitted but its declaration does not permit this', 'no-end-tag', - 'end tag for element FOO which is not open', 'floating-close', - 'end tag for FOO which is not finished', 'omitted-content', - 'start tag for FOO omitted but its declaration does not permit this', 'no-start-tag', - 'general entity FOO not defined and no default entity', 'bad-entity', - 'non SGML character number', 'bad-char', - 'cannot generate system identifier for entity FOO', 'bad-pub-id' - -# 'error', 'frag', -# 'character data is not allowed here', 'frag', - -); +# +# Read configuration files. +my $frag = &read_cfg($frag_db); # FPIs -> plain text version string +my $pub_ids = &read_cfg($fpis_db); # Errors -> fragment identifier +my $element_uri = &read_cfg($elem_db); # Element -> URI fragment ############################################################################# # Set up some signal handlers in case we get killed before exiting naturally @@ -486,7 +417,7 @@ else { $fpi = $doctype; } } -$version = $pub_ids{$fpi} || "unknown"; +$version = $pub_ids->{$fpi} || "unknown"; if ( $guessed_doctype ) { push( @fake_errors, "$sp:<OSFD>0:2:1:E: Missing DOCTYPE declaration at start of document (<a href=\"http://www.htmlhelp.org/tools/validator/doctype.html\">explanation...</a>)\n" ); @@ -633,9 +564,9 @@ if ( $? || $guessed_doctype ) { print qq{<span class=error>Error: $msg</span>}; - if ( defined $frag{$msgindex} ) { + if ( defined $frag->{$msgindex} ) { print qq{ (<a - href="$faqerrloc#$frag{$msgindex}">explanation...</a>)}; + href="$faqerrloc#$frag->{$msgindex}">explanation...</a>)}; } else { # remember msgindexes without frags, to get the KGV FAQ updated. $undef_frag{$msgindex} = 1; @@ -962,7 +893,7 @@ EOF { my $close = ''; $close = "/" if $1 eq ")"; # ")" -> close-tag "<" . $close . "<a href=\"" . - &html_element_ref($2) . + $element_ref . $element_uri->{lc($2)} . "\">$2<\/a>>" }egx; $printme =~ s,^A, A,; # indent attributes a bit @@ -1259,113 +1190,21 @@ EOHD } -sub html_element_ref { - - # returns a URI pointing to docs on the HTML element given as $1 - - my $element = shift; - - # the following hash was produced with: - # - # GET http://www.htmlhelp.com/reference/html40/alist.html > /tmp/e - # egrep '^ <li' /tmp/e | \ - # perl -pe 's|.*href="([^"]*)">(.*)</a>.*|\t"\L$2\E" => "$1",|' - - my %html_element_hash = ( - "a" => "special/a.html", - "abbr" => "phrase/abbr.html", - "acronym" => "phrase/acronym.html", - "address" => "block/address.html", - "applet" => "special/applet.html", - "area" => "special/area.html", - "b" => "fontstyle/b.html", - "base" => "head/base.html", - "basefont" => "special/basefont.html", - "bdo" => "special/bdo.html", - "big" => "fontstyle/big.html", - "blockquote" => "block/blockquote.html", - "body" => "html/body.html", - "br" => "special/br.html", - "button" => "forms/button.html", - "caption" => "tables/caption.html", - "center" => "block/center.html", - "cite" => "phrase/cite.html", - "code" => "phrase/code.html", - "col" => "tables/col.html", - "colgroup" => "tables/colgroup.html", - "dd" => "lists/dd.html", - "del" => "phrase/del.html", - "dfn" => "phrase/dfn.html", - "dir" => "lists/dir.html", - "div" => "block/div.html", - "dl" => "lists/dl.html", - "dt" => "lists/dt.html", - "em" => "phrase/em.html", - "fieldset" => "forms/fieldset.html", - "font" => "special/font.html", - "form" => "forms/form.html", - "frame" => "frames/frame.html", - "frameset" => "frames/frameset.html", - "h1" => "block/h1.html", - "h2" => "block/h2.html", - "h3" => "block/h3.html", - "h4" => "block/h4.html", - "h5" => "block/h5.html", - "h6" => "block/h6.html", - "head" => "head/head.html", - "hr" => "block/hr.html", - "html" => "html/html.html", - "i" => "fontstyle/i.html", - "iframe" => "special/iframe.html", - "img" => "special/img.html", - "input" => "forms/input.html", - "ins" => "phrase/ins.html", - "isindex" => "block/isindex.html", - "kbd" => "phrase/kbd.html", - "label" => "forms/label.html", - "legend" => "forms/legend.html", - "li" => "lists/li.html", - "link" => "head/link.html", - "map" => "special/map.html", - "menu" => "lists/menu.html", - "meta" => "head/meta.html", - "noframes" => "frames/noframes.html", - "noscript" => "block/noscript.html", - "object" => "special/object.html", - "ol" => "lists/ol.html", - "optgroup" => "forms/optgroup.html", - "option" => "forms/option.html", - "p" => "block/p.html", - "param" => "special/param.html", - "pre" => "block/pre.html", - "q" => "special/q.html", - "s" => "fontstyle/s.html", - "samp" => "phrase/samp.html", - "script" => "special/script.html", - "select" => "forms/select.html", - "small" => "fontstyle/small.html", - "span" => "special/span.html", - "strike" => "fontstyle/strike.html", - "strong" => "phrase/strong.html", - "style" => "head/style.html", - "sub" => "special/sub.html", - "sup" => "special/sup.html", - "table" => "tables/table.html", - "tbody" => "tables/tbody.html", - "td" => "tables/td.html", - "textarea" => "forms/textarea.html", - "tfoot" => "tables/tfoot.html", - "th" => "tables/th.html", - "thead" => "tables/thead.html", - "title" => "head/title.html", - "tr" => "tables/tr.html", - "tt" => "fontstyle/tt.html", - "u" => "fontstyle/u.html", - "ul" => "lists/ul.html", - "var" => "phrase/var.html" - ); - - return "http://www.htmlhelp.com/reference/html40/" . - $html_element_hash{"\L$element"}; -} +# +# Read TAB-delimited configuration files. Returns a hash reference. +sub read_cfg { + my $file = shift; + my %cfg; + + open CFG, $file or die "open($file) returned: $!\n"; + while (<CFG>) { + next if /^\s*$/; + next if /^\s*#/; + chomp; + my($k, $v) = split /\t+/, $_; + $cfg{$k} = $v; + } + close CFG; + return \%cfg; +} |