#!/bin/bash # # CGI script for post-processing CSS3 modules # # CGI defines the following variables # (see http://hoohoo.ncsa.uiuc.edu/cgi/) # # SERVER_SOFTWARE # SERVER_NAME # GATEWAY_INTERFACE # SERVER_PROTOCOL # SERVER_PORT # REQUEST_METHOD # PATH_INFO # PATH_TRANSLATED # SCRIPT_NAME # QUERY_STRING # REMOTE_HOST # REMOTE_ADDR # AUTH_TYPE # REMOTE_USER # REMOTE_IDENT # CONTENT_TYPE # CONTENT_LENGTH # HTTP_ACCEPT # HTTP_USER_AGENT # HTTP_* # # The following variables are expected on input or in QUERY_STRING: # # - method # is either "text," "url" or "file" # - text # should contain HTML source; it is only used if method is "text" # - file # should contain HTML source; it is only used if method is "file" # - url # is a URL to dereference to get HTML source; only used if method is "url" # - output # is "err", "html", or "auto" # - login # optional, used if method is url and the URL is password-protected # - password # optional, used if method is url and the URL is password-protected # - date # optional, default is today's date (format: YYYY-MM-DD) # - ids # if on, adds IDs to all P and LI elements # - group # The group that is writing the document, e.g., "CSS" ort "FX". This # is used to select an appropriate status section. # - omitdchtml # Normally, the postprocessor adds META/LINK elements according to # DC-HTML, to aid software for citation management in extracting # bibliographic data. If omitdchtml is on, no such elements are # added. See http://dublincore.org/documents/2008/08/04/dc-html/ # - status # The status of the document (WD, CR, REC, ED, etc.). If not set, # the script looks for it in the "this version" URL. And if not # found, it defaults to "ED" # # The HTML source is transformed with a couple of programs. The output # is either the transformed HTML (if "output" is "html") or the errors # and warnings that were generated during the transformation (if # "output" is "err"). If "output" is "auto," the output is the # transformed HTML if there are no errors, and an HTTP 400 error # otherwise. # # If group is "CSS" or "FX", the program also expands pairs of quotes # as in 'property' and ''CSS code''. # # Copyright © 2007-2012 W3C® (MIT, ERCIM, Keio) # See http://www.w3.org/Consortium/Legal/ipr-notice # # Author: Bert Bos # Created: 28 July 2000 # For testing: # #debugging=true [ $debugging ] && exec 2>&1 [ $debugging ] && echo "Content-Type: text/plain" [ $debugging ] && echo # error -- output an error message and exit # function error { echo "Status: $1 $2" echo "Content-Type: text/plain" echo echo "Error: $2" exit } # url_check -- check if the URL is "safe" # function url_check { case $1 in ftp:*|http:*) return 0;; *) return 1;; # Forbid "file:" URLs (security hole!) esac } # ---------------------------- main ---------------------------- PREFIX=/tmp/process.cgi.$$. TMPERR=/tmp/process.cgi.err.$$ TMPOUT4=/tmp/process.cgi.out4.$$ TMPOUT3=/tmp/process.cgi.out3.$$ TMPOUT2=/tmp/process.cgi.out2.$$ TMPOUT1=/tmp/process.cgi.out1.$$ TMPOUT=/tmp/process.cgi.out.$$ NORM=/tmp/process.cgi.norm.$$ NORMAUX=/tmp/process.cgi.normaux.$$ INFORM=/tmp/process.cgi.inform.$$ INFORMAUX=/tmp/process.cgi.informaux.$$ trap 'rm -f $TMPERR $TMPOUT4 $TMPOUT3 $TMPOUT2 $TMPOUT1 $TMPOUT $NORM $NORMAUX $INFORM $INFORMAUX $BIBASC ${PREFIX}*' 0 bibfile=biblio.ref bibfileBase=http://dev.w3.org/CSS/biblio/ statusdir=/afs/w3.org/pub/WWW/Style/Group/css3-src rights="http://www.w3.org/Consortium/Legal/ipr-notice#Copyright" PATH=$PWD/bin:$PATH [ $debugging ] && echo parseform = `which parseform` [ $debugging ] && echo hxnum = `which hxnum` [ $debugging ] && echo hxtoc = `which hxtoc` [ $debugging ] && echo hxindex = `which hxindex` [ $debugging ] && echo hxcite = `which hxcite` [ $debugging ] && echo hxmkbib = `which hxmkbib` [ $debugging ] && echo special-include = `which special-include` [ $debugging ] && echo wget = `which wget` [ $debugging ] && echo hxaddid = `which hxaddid` [ $debugging ] && echo hxpipe = `which hxpipe` [ $debugging ] && echo hxunpipe = `which hxunpipe` [ $debugging ] && echo hxref = `which hxref` [ $debugging ] && echo hxnormalize = `which hxnormalize` [ $debugging ] && echo hxextract = `which hxextract` [ $debugging ] && echo quotestoclasses = `which quotestoclasses` [ $debugging ] && echo hxselect = `which hxselect` [ $debugging ] && echo propdeftodb = `which propdeftodb` [ $debugging ] && echo propdefdbtotable = `which propdefdbtotable` # Sanity check: do we understand this CGI version? # [ $GATEWAY_INTERFACE = "CGI/1.1" ] || error 501 "CGI version mismatch" [ $debugging ] && echo "-------------------------------" [ $debugging ] && set [ $debugging ] && echo "-------------------------------" # Parse the form data (parseform understands GET, POST and file upload). # Extract the form fields. # parseform $PREFIX [ -f ${PREFIX}output ] && output=$(tr -c -d a-zA-Z0-9 <${PREFIX}output) [ -f ${PREFIX}method ] && method=$(tr -c -d a-zA-Z0-9 <${PREFIX}method) [ -f ${PREFIX}group ] && group=$(tr -c -d a-zA-Z0-9 <${PREFIX}group) [ -f ${PREFIX}login ] && login=$(<${PREFIX}login) [ -f ${PREFIX}password ] && password=$(<${PREFIX}password) [ -f ${PREFIX}date ] && date=$(tr -c -d 0-9 <${PREFIX}date) [ -f ${PREFIX}ids ] && ids=$(<${PREFIX}ids) [ -f ${PREFIX}status ] && status=$(<${PREFIX}status) [ -f ${PREFIX}omitdchtml ] && omitdchtml=$(<${PREFIX}omitdchtml) [ $debugging ] && echo method: \"$method\" [ $debugging ] && echo output: \"$output\" [ $debugging ] && echo group: \"$group\" [ $debugging ] && echo login: \"$login\" [ $debugging ] && echo password: \"$password\" [ $debugging ] && echo date: \"$date\" [ $debugging ] && echo ids: \"$ids\" [ $debugging ] && echo status: \"$status\" [ $debugging ] && echo omitdchtml: \"$omitdchtml\" group=${group:-none} output=${output:-auto} # Depending on the method, store the content of the "file," "text" or # (indirectly) "url" variables in a temporary file. # case $method in file) input=${PREFIX}file;; text) input=${PREFIX}text;; url) [ -f ${PREFIX}url ] || error 403 "Incorrect query: missing \"url\" field" f=$(< ${PREFIX}url) url_check $f || error 403 "Only \"ftp:\" or \"http:\" URLs are allowed" h=`wget -nv --http-user="$login" --http-passwd="$password" -O $TMPOUT3 $f \ 2>&1` || error 403 "Failed to retrieve from that URL: ${h/$'\n'/ }" input=$TMPOUT3;; *) error 403 "Incorrect query: missing or illegal field \"method\"";; esac [ $debugging ] && echo [ $debugging ] && echo input: \"$input\" # Get a copy of the bibliography database # wget -N --no-remove-listing -nv $bibfileBase/$bibfile # Heuristically derive the status, shortname and the format of the # document, by looking for certain keywords and a "dated" URL in the # first few lines of the file. # # Bug: The sed command to find the shortname assumes the

is not # on the same line as the DOCTYPE, so that it can skip the latter. # Files with CR instead of LF (old Mac files) will fail this. A "tr # '\r' '\n'" can fix it, but seems unnecessary, in view of the rarity # of such files. # if head -1 $input | fgrep -q ''; then is_html5=true else is_html5=false fi [ $debugging ] && echo format: \"$format\" [ $debugging ] && echo html5: $is_html5 if ! $is_html5; then biblioinc=$statusdir/biblio.inc copyright=$statusdir/copyright.inc cite='[%L]' else biblioinc=$statusdir/biblio5.inc copyright=$statusdir/copyright.inc cite='[%L]' fi # If not yet set, find the status in the "This version" URL, or default to "ED" [[ -z "$status" ]] && status=`sed -r -n -e ' # Remove comments, PIs and DOCTYPEs, including multi-line ones :start; s///g s/]*>//g s/<\?[^>]*>//g ///g s/]*>//g s/<\?[^>]*>//g ///g / comment # Extract table of descriptors and insert it at the place of a # comment # { mv $TMPOUT $TMPOUT1 hxunent -b $TMPOUT1 | hxaddid -x dfn >$TMPOUT2 hxpipe $TMPOUT2 |\ CLASS1=propdef CLASS2=propdef-extra propdeftodb |\ sort -bdf |\ CLASS=properties propdefdbtotable |\ hxunpipe >$TMPOUT3 hxpipe $TMPOUT2 |\ CLASS1=descdef CLASS2=descdef-extra propdeftodb |\ sort -bdf |\ CLASS=descriptors propdefdbtotable |\ hxunpipe >$TMPOUT4 special-include properties $TMPOUT3 <$TMPOUT2 |\ special-include descriptors $TMPOUT4 >$TMPOUT } 2>>$TMPERR #(sh propdeftotab $TMPOUT1 |\ # sh tabtoproptable $TMPOUT1 >$TMPOUT) 2>>$TMPERR # # Extract table of descriptors and insert it at the place of a # comment # #mv $TMPOUT $TMPOUT1 #(sh propdeftotab -c descdef $TMPOUT1 |\ # sh tabtoproptable $TMPOUT1 descriptors >$TMPOUT) 2>>$TMPERR # If DC-HTML mark-up is requested, (heuristically) find the # editors/authors and insert corresponding META and LINK elements. # (We also need to duplicate the title for Google Scholar. :-( ) # # To do: Remove or merge any existing META and LINK elements for # dcterms.title, dcterms.creator, dcterms.publisher, dcterms.type, # dcterms.issued and dcterms.identifier, even if the prefix isn't # "dcterms." # if [ "$omitdchtml" != "on" ]; then hxnormalize -x -i 0 -l 99999 $TMPOUT >$TMPOUT1 2>/dev/null hxselect head <$TMPOUT1 >$TMPOUT4 # Isolate the DDs that contain the editors or authors. Then use the # hcard mark-up, if any, to find the names; or apply some heuristics # to strip affiliations, e-mail addresses, etc. # h=$(sed -r \ -e '1,/]*> *([Ee]dit|[Aa]uth)ors?[ ]*:/d' \ -e '/<\/dl/,$d' \ -e '/
]*>//g' \ -e 's/,? *<.*>//g' \ -e 's/,? *[a-zA-Z0-9_.-]*@[a-zA-Z0-9_.-]*//g' \ -e 's/\([^)]*\) *$//' \ -e 's/, *....*$//' \ -e '/^ *$/d' \ -e 's/ */ /g' \ -e 's/^ //' \ -e 's/ $//' <<<"$h") fi [ -n "$names" ] || echo "Warning: Could not find editors or authors." >&2 [ $debugging ] && echo names: \"${names//$'\n'/, }\" isodate=${cdate:0:4}-${cdate:4:2}-${cdate:6:2} # Generate the LINK and META elements. Then insert those in the HTML # file and also add the DC-HTML specification to the PROFILE. HTML5 # doesn't yet define schema.dcterms and dcterms.rights as keywords, # so they are commented-out for now. # rm -f $TMPOUT3 if $is_html5; then echo '' >>$TMPOUT3; fi echo "" >>$TMPOUT3 echo "" >>$TMPOUT3 echo "" >>$TMPOUT3 sed \ -e 's/"/\"/g' \ -e 's/^//' <<<"$names" >>$TMPOUT3 if ! egrep -q ']*name="dcterms\.publisher"' $TMPOUT4; then echo '' >>$TMPOUT3 elif ! fgrep -q '&2 fi if ! egrep -q ']*name="dcterms\.identifier"' $TMPOUT4; then echo "" >>$TMPOUT3 else echo "Warning: Document already has a META with name=dcterms.identifier" >&2 fi if egrep -q ']*name="dcterms\.(title|type|date|creator)"' $TMPOUT4; then echo "Warning: Removed existing META elements with dcterms.title, dcterms.type, dcterms.date and dcterms.creator" >&2 fi if ! $is_html5; then # Add PROFILE attribute dcterms='http://dublincore.org/documents/2008/08/04/dc-html/' if head -30 $TMPOUT1 | fgrep -i ']*\$/{N;b start;} s||| s|]*profile=['\"]|&$dcterms |" $TMPOUT1 >$TMPOUT mv $TMPOUT $TMPOUT1 fi fi sed -e " s/]*name=\"dcterms.title\"[^>]*>// s/]*name=\"dcterms.type\"[^>]*>// s/]*name=\"dcterms.date\"[^>]*>// s/]*name=\"dcterms.creator\"[^>]*>// /<\/title>/r $TMPOUT3" $TMPOUT1 >$TMPOUT fi 2>>$TMPERR # Extract bibliography. Convert the bibliography from UTF-8 to # numerical character entities before use. Remove references from the # list of informal references that are already in the normative list. # comm -23 <(sort -u $INFORMAUX) <(sort -u $NORMAUX) >$TMPOUT4 mv $TMPOUT4 $INFORMAUX hxmkbib -a $NORMAUX $bibfile $biblioinc 2>>$TMPERR | xml2asc >$NORM hxmkbib -a $INFORMAUX $bibfile $biblioinc 2>>$TMPERR | xml2asc >$INFORM # Insert the bibliography. Make cross references to defining instances # of terms. # mv $TMPOUT $TMPOUT1 (special-include normative $NORM $TMPOUT1 |\ special-include informative $INFORM |\ hxunent -b |\ hxref -l $format |\ LC_ALL=en_US hxindex $format -t -n |\ hxnormalize $format -i 1 -l 78) >$TMPOUT 2>>$TMPERR [ $debugging ] && wc $TMPOUT [ $debugging ] && wc $TMPERR # Depending on "output," return either the errors and warnings or the # generated HTML # if [ $output == "err" ] || ( [ $output == "auto" ] && [ -s $TMPERR ] ); then echo "Content-Type: text/plain" echo "X-W3C-Postprocessor-Errors:" $(egrep -i -v '^Warning:' $TMPERR | wc -l) echo "X-W3C-Postprocessor-Warnings:" $(egrep -i '^Warning:' $TMPERR | wc -l) echo if [ $REQUEST_METHOD != "HEAD" ]; then if [ -s $TMPERR ]; then cat $TMPERR; else echo "No errors"; fi echo "(Processed in $SECONDS seconds)" fi else echo "Content-Type: text/html" echo "X-W3C-Postprocessor-Errors:" $(egrep -i -v '^Warning:' $TMPERR | wc -l) echo "X-W3C-Postprocessor-Warnings:" $(egrep -i '^Warning:' $TMPERR | wc -l) echo if [ $REQUEST_METHOD != "HEAD" ]; then cat $TMPOUT; fi fi # Local variables: # mode: ksh # End: