upload.sh

   1 #!/bin/sh
   2
   3 # This is a simple script that will produce my blog on repo.or.cz
   4
   5 # The idea is to have source-<timestamp>.txt files as input, having the
   6 # stories, and this script turning them into nice HTML, committing
   7 # everything, and then pushing it to my repository.
   8
   9 # The blog will then be served using gitweb.
  10
  11 # To make it easier on me, if a file "source.txt" exists, it is
  12 # automatically renamed using the current timestamp.
  13
  14 # TODO: have a configurable maximum number of entries per page, and links
  15 #       to older pages
  16
  17 # make sure we're in the correct working directory
  18 cd "$(dirname "$0")"
  19
  20 GITWEBURL="$(git config gitweb.url)"
  21 test -z "$GITWEBURL" && {
  22         echo "Please set gitweb.url in the Git config first!" >&2
  23         exit 1
  24 }
  25
  26 URLPREFIX="$(dirname "$GITWEBURL")"/
  27 REMOTEREPOSITORY="$(basename "$GITWEBURL")"
  28 BRANCH=blog
  29 URL="$REMOTEREPOSITORY?a=blob_plain;hb=$BRANCH;f="
  30 ORIGURL=$URL
  31 NEW=new
  32 OUTPUT=index.html
  33 RSS=blog.rss
  34 TEST=test.html
  35 TITLE="Dscho's blog"
  36
  37 LC_ALL=C
  38 export LC_ALL
  39
  40 move_new_entry_back () {
  41         test -f source-$now.txt &&
  42         mv source-$now.txt $NEW &&
  43         git rm --cached -f source-$now.txt
  44 }
  45
  46 die () {
  47         move_new_entry_back
  48         echo "$*" >&2
  49         exit 1
  50 }
  51
  52 nth () {
  53         # add illogical suffix
  54         case "$1" in
  55         *1?|*[04-9]) echo "$1th";;
  56         *1) echo "$1st";;
  57         *2) echo "$1nd";;
  58         *3) echo "$1rd";;
  59         esac
  60 }
  61
  62 make_chinese_hour () {
  63         case $1 in
  64         23|00) echo Rat;;
  65         01|02) echo Buffalo;;
  66         03|04) echo Tiger;;
  67         05|06) echo Rabbit;;
  68         07|08) echo Dragon;;
  69         09|10) echo Snake;;
  70         11|12) echo Horse;;
  71         13|14) echo Goat;;
  72         15|16) echo Monkey;;
  73         17|18) echo Rooster;;
  74         19|20) echo Dog;;
  75         21|22) echo Pig;;
  76         esac
  77 }
  78
  79 digit_to_roman () {
  80         case $1 in
  81         1) echo $2;;
  82         2) echo $2$2;;
  83         3) echo $2$2$2;;
  84         4) echo $2$3;;
  85         5) echo $3;;
  86         6) echo $3$2;;
  87         7) echo $3$2$2;;
  88         8) echo $3$2$2$2;;
  89         9) echo $2$4;;
  90         esac
  91 }
  92
  93 make_roman_number () {
  94         case $1 in
  95         '') ;;
  96         ?) digit_to_roman $1 I V X;;
  97         ??) echo $(digit_to_roman ${1%?} X L C)$(make_roman_number ${1#?});;
  98         ???) echo $(digit_to_roman ${1%??} C D M)$(make_roman_number ${1#?});;
  99         ????) echo $(digit_to_roman ${1%???} M)$(make_roman_number ${1#?});;
 100         esac
 101 }
 102
 103 make_date () {
 104         printf "%s, %s of %s, Anno Domini %s, at the hour of the %s\n" \
 105                 $(date +%A -d @$1) \
 106                 $(nth $(date +%e -d @$1)) \
 107                 $(date +%B -d @$1) \
 108                 $(make_roman_number $(date +%Y -d @$1)) \
 109                 $(make_chinese_hour $(date +%H -d @$1))
 110 }
 111
 112 # make an argument for sed, to replace $1..$1 by <$2>..</$2>
 113 markup_substitution () {
 114         case "$1" in
 115         ?) echo "s/$1\\([^$1]*\\)$1/<$2>\\\\1<\/$2>/g";;
 116         ??)
 117                 tmp="[^${1%?}]*"
 118                 tmp2="\\|${1%?}[^${1#?}]$tmp"
 119                 tmp3="\\($tmp\\($tmp2\\($tmp2\\($tmp2\\)\\)\\)\\)"
 120                 echo "s/$1$tmp3$1/<$2>\\\\1<\/$2>/g"
 121         ;;
 122         esac
 123 }
 124
 125 # transform markup in stdin to HTML
 126 markup () {
 127         image_pattern="\\[\\[Image:\([^]]*\)"
 128         image_pattern2="$image_pattern\(\\|[^\]]*\)\?\]\]"
 129         case "$*" in
 130         *invert-bash*) bash_bg=white; bash_fg=black;;
 131         *) bash_bg=black; bash_fg=white;;
 132         esac
 133         sed -e 's!^$!</p><p>!' \
 134                 -e 's!IMHO!in my humble opinion!g' \
 135                 -e 's!BTW!By the way,!g' \
 136                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 137                 -e 's!:-)!\&#x263a;!g' \
 138                 -e "s!$image_pattern2!<center><img src=$URL\1></center>!g" \
 139                 -e 's!<bash>!<table\
 140                                 border=1 bgcolor='$bash_bg'>\
 141                         <tr><td bgcolor=lightblue colspan=3>\
 142                                 \&nbsp;\
 143                         </td></tr>\
 144                         <tr><td>\
 145                                 <table cellspacing=5 border=0\
 146                                          style="color:'$bash_fg';">\
 147                                 <tr><td>\
 148                                         <pre>!' \
 149                 -e 's!</bash>!          </pre>\
 150                                 </td></tr>\
 151                                 </table>\
 152                         </td></tr>\
 153                         </table>!' \
 154                 -e "$(markup_substitution "''" i)" \
 155                 -e "$(markup_substitution "_" u)"
 156 }
 157
 158 # output lines containing <timestamp> <filename> <title>
 159 get_blog_entries () {
 160         for file in $(ls -r source-*.txt)
 161         do
 162                 basename=${file%.txt}
 163                 timestamp=${basename#source-}
 164                 title="$(sed 1q < $file | markup)"
 165                 echo "$timestamp $file $title"
 166         done
 167 }
 168
 169 # make HTML page
 170 make_html () {
 171         body_style="width:800px"
 172         body_style="$body_style;background-image:url(${URL}paper.jpg)"
 173         body_style="$body_style;background-repeat:repeat-y"
 174         body_style="$body_style;background-attachment:scroll"
 175         body_style="$body_style;padding:0px;"
 176         text_style="width:610px"
 177         text_style="$text_style;margin-left:120px"
 178         text_style="$text_style;margin-top:50px"
 179         text_style="$text_style;align:left"
 180         text_style="$text_style;vertical-align:top;"
 181         cat << EOF
 182 <html>
 183         <head>
 184                 <title>$TITLE</title>
 185                 <meta http-equiv="Content-Type"
 186                         content="text/html; charset=UTF-8"/>
 187         </head>
 188         <body style="$body_style">
 189                 <div style="$text_style">
 190                         <h1>$TITLE</h1>
 191 EOF
 192         indent='\t\t\t'
 193
 194         # make toc
 195         toc_width=400px
 196         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 197         {
 198                 echo "<div style=\"$toc_style\">"
 199                 echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 200                 echo "<tr><th>Table of contents:</th></tr>"
 201                 echo "<tr><td>"
 202                 echo '<p><ol>'
 203                 get_blog_entries |
 204                 while read timestamp filename title
 205                 do
 206                         date="$(date +"%d %b %Y" -d @$timestamp)"
 207                         echo "<li><a href=#$timestamp>$date $title</a>"
 208                 done
 209                 echo '</ol></p>'
 210                 echo '</td></tr></table>'
 211
 212                 # Links
 213                 test -f links.html && {
 214                         echo "<br>"
 215                         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 216                         echo "<tr><th>Links:</th></tr>"
 217                         echo "<tr><td>"
 218                         cat links.html
 219                         echo "</td></tr></table>"
 220                 }
 221
 222                 # Google AdSense
 223                 test -z "$DRYRUN" && {
 224                         echo "<br>"
 225                         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 226                         echo "<tr><td align=center>"
 227                         cat << EOF
 228 <script type="text/javascript"><!--
 229 google_ad_client = "pub-5106407705643819";
 230 /* 300x250, created 1/22/09 */
 231 google_ad_slot = "6468207338";
 232 google_ad_width = 300;
 233 google_ad_height = 250;
 234 //-->
 235 </script>
 236 <script type="text/javascript"
 237 src="http://pagead2.googlesyndication.com/pagead/show_ads.js">
 238 </script>
 239 EOF
 240                         echo "</td></tr></table>"
 241                 }
 242
 243                 # RSS feed
 244                 rss_style="background-color:orange;text-decoration:none"
 245                 rss_style="$rss_style;color:white;"
 246                 echo '<br>'
 247                 echo '<div style="text-align:right;">'
 248                 echo "<a href=\"$ORIGURL$RSS\""
 249                 echo '   title="Subscribe to my RSS feed"'
 250                 echo '   class="rss" rel="nofollow"'
 251                 echo "   style=\"$rss_style\">RSS</a>"
 252                 echo '</div>'
 253
 254                 echo '</div>'
 255         } | sed -s "s/^/$indent/"
 256
 257
 258         # timestamps will not need padding to sort correctly, for some time...
 259         get_blog_entries |
 260         while read timestamp filename title
 261         do
 262                 echo "<h6>$(make_date $timestamp)</h6>"
 263                 echo "<a name=$timestamp>"
 264                 echo "<h2>$title</h2>"
 265                 echo ""
 266                 echo "<p>"
 267                 sed 1d < $filename | markup
 268                 echo "</p>"
 269         done |
 270         sed -e "s/^./$indent&/" \
 271                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 272
 273         cat << EOF
 274                 </div>
 275         </body>
 276 </html>
 277 EOF
 278 }
 279
 280 generate_rss () {
 281         echo '<?xml version="1.0" encoding="utf-8"?>'
 282         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 283         echo '<channel>'
 284         echo "<title>Dscho's blog</title>"
 285         echo "<link>$URLPREFIX${URL}index.html</link>"
 286         self="$URLPREFIX$ORIGURL$RSS"
 287         selfattribs='rel="self" type="application/rss+xml"'
 288         echo "<atom:link href=\"$self\" $selfattribs/>"
 289         echo '<description>A few stories told by Dscho</description>'
 290         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 291         echo '<language>en-us</language>'
 292
 293         get_blog_entries |
 294         while read timestamp filename title
 295         do
 296                 echo '<item>'
 297                 echo "<title>$title</title>"
 298                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 299                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 300                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 301                 description="$(cat < $filename | markup invert-bash)"
 302                 echo "<description><![CDATA[$description]]></description>"
 303                 echo "</item>"
 304         done
 305
 306         echo '</channel>'
 307         echo '</rss>'
 308 }
 309
 310 # never, ever have spaces in the file names
 311 commit_new_images () {
 312         generate_rss > $RSS &&
 313         git add $RSS ||
 314         die "Could not generate $RSS"
 315
 316         images=
 317         for image in $(cat source-* |
 318                 tr ' ]|' '\n' |
 319                 sed -n 's/.*\[\[Image://p' |
 320                 sort |
 321                 uniq)
 322         do
 323                 git add $image || die "Could not git add image $image"
 324                 images="$images $image"
 325         done
 326
 327         git update-index --refresh &&
 328         git diff-files --quiet -- $images &&
 329         git diff --cached --quiet HEAD -- $images ||
 330         git commit -s -m "Commit some images on $(make_date $now)" $images
 331 }
 332
 333
 334
 335 # parse command line option
 336 case "$1" in
 337 *dry*) DRYRUN=1; shift;;
 338 *show*) firefox "$(pwd)"/$TEST; exit;;
 339 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 340 esac
 341
 342 test "$#" = 0 ||
 343 die "Usage: $0 [--dry-run]"
 344
 345 # make sure we're on the correct branch
 346 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 347 die "Not on branch $BRANCH"
 348
 349 # make sure there are no uncommitted changes
 350 git update-index --refresh &&
 351 git diff-files --quiet ||
 352 die "Have unstaged changes!"
 353
 354 # rename the new blog entry if it exists
 355 now=$(date +%s)
 356 test ! -f $NEW || {
 357         mv -i $NEW source-$now.txt &&
 358         git add source-$now.txt
 359 } ||
 360 die "Could not rename source.txt"
 361
 362 # commit the images that are referenced and not yet committed
 363 test ! -z "$DRYRUN" ||
 364 commit_new_images ||
 365 die "Could not commit new images"
 366
 367 # to find the images reliably, we have to use the commit name, not the branch
 368 # we use the latest commit touching an image file.
 369 IMAGEFILES="$(git ls-files |
 370         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\)$')"
 371 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 372 test -z "$REV" && REV=$BRANCH
 373 URL="$REMOTEREPOSITORY?a=blob_plain;hb=$REV;f="
 374
 375 if test ! -z "$DRYRUN"
 376 then
 377         # Output to test.html and have local links into the current directory
 378         OUTPUT=$TEST
 379         URL=
 380 fi
 381
 382 make_html > $OUTPUT || die "Could not write $OUTPUT"
 383
 384 test ! -z "$DRYRUN" && {
 385         move_new_entry_back
 386         exit
 387 }
 388
 389 git add $OUTPUT &&
 390 git commit -s -m "Update $(make_date $now)" &&
 391 git push origin +$BRANCH