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 "new" exists, it is automatically
  12 # renamed using the current timestamp.
  13
  14 # How to use:
  15 #
  16 # $ mkdir my-blog
  17 # $ cd my-blog
  18 # $ git init
  19 #
  20 # Then symlink or copy this file (upload.sh); you can track it or add it
  21 # to .gitignore, does not matter.
  22 #
  23 # Add a remote "origin" (you might want to track only the appropriate branch
  24 # if the repository contains other branches, too), add a background image,
  25 # and then set the config variables gitweb.url, blog.title, blog.background
  26 # and blog.branch appropriately.
  27 #
  28 # Example:
  29 #
  30 # $ git remote add -t blog repo.or.cz:/srv/git/git/dscho.git/
  31 # $ git symbolic-ref HEAD refs/heads/blog
  32 # $ cp ~/images/background.jpg ./
  33 # $ git config gitweb.url http://repo.or.cz/w/git/dscho.git
  34 # $ git config blog.title "Dscho's blog"
  35 # $ git config blog.background background.jpg
  36 # $ git config blog.branch blog
  37 #
  38 # Now you can start writing posts, by creating a file called "new", and
  39 # calling ./upload.sh to commit the post together with the images and
  40 # push all.
  41 #
  42 # Note that no file names may contain spaces.
  43
  44 # TODO: document the "syntax" of the source-*.txt files
  45
  46
  47 # make sure we're in the correct working directory
  48 cd "$(dirname "$0")"
  49
  50 GITWEBURL="$(git config gitweb.url)"
  51 test -z "$GITWEBURL" && {
  52         echo "Please set gitweb.url in the Git config first!" >&2
  53         exit 1
  54 }
  55
  56 get_config () {
  57         value=$(git config blog.$1)
  58         test -z "$value" && value=$2
  59         echo $value
  60 }
  61
  62 BACKGROUNDIMG=$(get_config background paper.jpg)
  63 TITLE=$(get_config title "Dscho's Git log")
  64 MAXENTRIES=$(get_config maxPostsPerPage 10)
  65 BRANCH=$(get_config branch blog)
  66
  67 URLPREFIX="$(dirname "$GITWEBURL")"/
  68 REMOTEREPOSITORY="$(basename "$GITWEBURL")"
  69 case "$GITWEBURL" in
  70 *'?'*) BLOBPLAIN="$REMOTEREPOSITORY;a=blob_plain";;
  71 */) URLPREFIX=$GITWEBURL; BLOBPLAIN="a=blob_plain";;
  72 *) BLOBPLAIN="$REMOTEREPOSITORY?a=blob_plain";;
  73 esac
  74 URL="$BLOBPLAIN;hb=$BRANCH;f="
  75 ORIGURL=$URL
  76 NEW=new
  77 OUTPUT=index.html
  78 RSS=blog.rss
  79 TEST=test.html
  80 THIS=$0
  81
  82 LC_ALL=C
  83 export LC_ALL
  84
  85 move_new_entry_back () {
  86         test -f source-$now.txt &&
  87         mv source-$now.txt $NEW &&
  88         git rm --cached -f source-$now.txt
  89 }
  90
  91 die () {
  92         move_new_entry_back
  93         echo "$*" >&2
  94         exit 1
  95 }
  96
  97 strip_prefix () {
  98         echo "${1#$2}"
  99 }
 100
 101 chomp () {
 102         strip_prefix "${1%$3}" "$2"
 103 }
 104
 105 nth () {
 106         # add illogical suffix
 107         case "$1" in
 108         *1?|*[04-9]) echo "$1th";;
 109         *1) echo "$1st";;
 110         *2) echo "$1nd";;
 111         *3) echo "$1rd";;
 112         esac
 113 }
 114
 115 make_chinese_hour () {
 116         case $1 in
 117         23|00) echo Rat;;
 118         01|02) echo Buffalo;;
 119         03|04) echo Tiger;;
 120         05|06) echo Rabbit;;
 121         07|08) echo Dragon;;
 122         09|10) echo Snake;;
 123         11|12) echo Horse;;
 124         13|14) echo Goat;;
 125         15|16) echo Monkey;;
 126         17|18) echo Rooster;;
 127         19|20) echo Dog;;
 128         21|22) echo Pig;;
 129         esac
 130 }
 131
 132 digit_to_roman () {
 133         case $1 in
 134         1) echo $2;;
 135         2) echo $2$2;;
 136         3) echo $2$2$2;;
 137         4) echo $2$3;;
 138         5) echo $3;;
 139         6) echo $3$2;;
 140         7) echo $3$2$2;;
 141         8) echo $3$2$2$2;;
 142         9) echo $2$4;;
 143         esac
 144 }
 145
 146 make_roman_number () {
 147         case $1 in
 148         '') ;;
 149         ?) digit_to_roman $1 I V X;;
 150         ??) echo $(digit_to_roman ${1%?} X L C)$(make_roman_number ${1#?});;
 151         ???) echo $(digit_to_roman ${1%??} C D M)$(make_roman_number ${1#?});;
 152         ????) echo $(digit_to_roman ${1%???} M)$(make_roman_number ${1#?});;
 153         esac
 154 }
 155
 156 make_date () {
 157         printf "%s, %s of %s, Anno Domini %s, at the hour of the %s\n" \
 158                 $(date +%A -d @$1) \
 159                 $(nth $(date +%e -d @$1)) \
 160                 $(date +%B -d @$1) \
 161                 $(make_roman_number $(date +%Y -d @$1)) \
 162                 $(make_chinese_hour $(date +%H -d @$1))
 163 }
 164
 165 # make an argument for sed, to replace $1..$1 by <$2>..</$2>
 166 markup_substitution () {
 167         # sed does not know minimal match with .*, only greedy one
 168         middle=
 169         middleend=
 170         delim=$1
 171         right_no=3
 172         while test ! -z "$delim"
 173         do
 174                 right_no=$(($right_no+1))
 175                 test $right_no -gt 9 &&
 176                 die "Invalid markup pattern: $1"
 177                 first=$(expr "$delim" : '\(.\)')
 178                 delim=${delim#$first}
 179                 test -z "$delim" && {
 180                         middle="$middle\\([^$first]\\|$first[A-Za-z0-9]\\)"
 181                         break
 182                 }
 183                 middle="$middle\\([^$first]\\|$first"
 184                 middleend="$middleend\\)"
 185         done
 186
 187         left="\\(^\\|[^A-Za-z0-9]\\)$1\\($middle$middleend*\\)"
 188         right="$1\\($\\|[^A-Za-z0-9]\\)"
 189         echo "s/$left$right/\\\\1<$2>\\\\2<\/$2>\\\\$right_no/g"
 190 }
 191
 192 space80='                          '
 193 space80="$space80$space80$space80  "
 194 # transform markup in stdin to HTML
 195 markup () {
 196         case "$*" in
 197         *invert-bash*) bash_bg=white; bash_fg=black;;
 198         *) bash_bg=black; bash_fg=white;;
 199         esac
 200         sed -e 's!^$!</p><p>!' \
 201                 -e "$(markup_substitution "''" i)" \
 202                 -e "$(markup_substitution "_" u)" \
 203                 -e 's!IMHO!in my humble opinion!g' \
 204                 -e 's!BTW!By the way,!g' \
 205                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 206                 -e 's!:-)!\&#x263a;!g' \
 207                 -e "s!\\[\\[\(Image\|SVG\):.*!$THIS handle &!e" \
 208                 -e 's!<bash>!<table\
 209                                 border=1 bgcolor='$bash_bg'>\
 210                         <tr><td bgcolor=lightblue colspan=3>\
 211                                 <pre>'"$space80"'</pre>\
 212                         </td></tr>\
 213                         <tr><td>\
 214                                 <table cellspacing=5 border=0\
 215                                          style="color:'$bash_fg';">\
 216                                 <tr><td>\
 217                                         <pre>!' \
 218                 -e 's!</bash>!</pre>\
 219                                 </td></tr>\
 220                                 </table>\
 221                         </td></tr>\
 222                         </table>!' \
 223
 224 }
 225
 226 # output lines containing <timestamp> <filename> <title>
 227 get_blog_entries () {
 228         for file in $(ls -r source-*.txt)
 229         do
 230                 timestamp=$(chomp $file source- .txt)
 231                 title="$(sed 1q < $file | markup)"
 232                 echo "$timestamp $file $title"
 233         done
 234 }
 235
 236 get_last_removed_entry () {
 237         git log --pretty=format: --name-only --diff-filter=D HEAD |
 238         while read line
 239         do
 240                 case "$line" in
 241                 source-*.txt) file=$line;;
 242                 '') test -z "$file" || {
 243                         echo "$file"
 244                         break
 245                 };;
 246                 esac
 247         done
 248 }
 249
 250 box_count=0
 251 begin_box () {
 252         test $box_count = 0 || echo "<br>"
 253         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 254         echo "<tr><th>$1</th></tr>"
 255         echo "<tr><td>"
 256 }
 257
 258 end_box () {
 259         echo "</td></tr></table>"
 260         box_count=$(($box_count+1))
 261 }
 262
 263 # make HTML page
 264 make_html () {
 265         body_style="width:800px"
 266         body_style="$body_style;background-image:url($URL$BACKGROUNDIMG)"
 267         body_style="$body_style;background-repeat:repeat-y"
 268         body_style="$body_style;background-attachment:scroll"
 269         body_style="$body_style;padding:0px;"
 270         text_style="width:610px"
 271         text_style="$text_style;margin-left:120px"
 272         text_style="$text_style;margin-top:50px"
 273         text_style="$text_style;align:left"
 274         text_style="$text_style;vertical-align:top;"
 275         cat << EOF
 276 <html>
 277         <head>
 278                 <title>$TITLE</title>
 279                 <meta http-equiv="Content-Type"
 280                         content="text/html; charset=UTF-8"/>
 281         </head>
 282         <body style="$body_style">
 283                 <div style="$text_style">
 284                         <h1>$TITLE</h1>
 285 EOF
 286         indent='\t\t\t'
 287
 288         # make toc
 289         toc_width=400px
 290         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 291         {
 292                 echo "<div style=\"$toc_style\">"
 293                 begin_box "Table of contents:"
 294                 echo '<p><ul>'
 295                 get_blog_entries |
 296                 while read timestamp filename title
 297                 do
 298                         date="$(date +"%d %b %Y" -d @$timestamp)"
 299                         echo "<li><a href=#$timestamp>$date $title</a>"
 300                 done
 301                 echo '</ul></p>'
 302                 file=
 303                 last_removed_entry=$(get_last_removed_entry)
 304                 test -z "$last_removed_entry" || {
 305                         commit=$(git log --pretty=format:%H --diff-filter=AM \
 306                                         -- $last_removed_entry |
 307                                 head -n 1)
 308                         previous="$BLOBPLAIN;hb=$commit"
 309                         echo "<a href=$previous;f=index.html>Older posts</a>"
 310                 }
 311                 end_box
 312
 313                 # RSS feed
 314                 rss_style="background-color:orange;text-decoration:none"
 315                 rss_style="$rss_style;color:white;font-family:sans-serif;"
 316                 echo '<br>'
 317                 echo '<div style="text-align:right;">'
 318                 echo "<a href=\"$ORIGURL$RSS\""
 319                 echo '   title="Subscribe to my RSS feed"'
 320                 echo '   class="rss" rel="nofollow"'
 321                 echo "   style=\"$rss_style\">RSS</a>"
 322                 echo '</div>'
 323
 324                 # About
 325                 test -f about.html && {
 326                         begin_box "About this blog:"
 327                         cat about.html
 328                         end_box
 329                 }
 330
 331                 # Links
 332                 test -f links.html && {
 333                         begin_box "Links:"
 334                         cat links.html
 335                         end_box
 336                 }
 337
 338                 # Google AdSense
 339                 test -z "$DRYRUN" && test -f google.adsense && {
 340                         begin_box "Google Ads:"
 341                         cat google.adsense
 342                         end_box
 343                 }
 344
 345                 echo '</div>'
 346         } | sed -s "s/^/$indent/"
 347
 348
 349         # timestamps will not need padding to sort correctly, for some time...
 350         get_blog_entries |
 351         while read timestamp filename title
 352         do
 353                 echo "<h6>$(make_date $timestamp)</h6>"
 354                 echo "<a name=$timestamp>"
 355                 echo "<h2>$title</h2>"
 356                 echo ""
 357                 echo "<p>"
 358                 sed 1d < $filename | markup
 359                 echo "</p>"
 360         done |
 361         sed -e "s/^./$indent&/" \
 362                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 363
 364         cat << EOF
 365                 </div>
 366         </body>
 367 </html>
 368 EOF
 369 }
 370
 371 generate_rss () {
 372         echo '<?xml version="1.0" encoding="utf-8"?>'
 373         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 374         echo '<channel>'
 375         echo "<title>Dscho's blog</title>"
 376         echo "<link>$URLPREFIX${URL}index.html</link>"
 377         self="$URLPREFIX$ORIGURL$RSS"
 378         selfattribs='rel="self" type="application/rss+xml"'
 379         echo "<atom:link href=\"$self\" $selfattribs/>"
 380         echo '<description>A few stories told by Dscho</description>'
 381         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 382         echo '<language>en-us</language>'
 383
 384         get_blog_entries |
 385         while read timestamp filename title
 386         do
 387                 # remove all tags
 388                 title=$(echo "$title" | sed 's/<[^>]*>//g')
 389                 echo '<item>'
 390                 echo "<title>$title</title>"
 391                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 392                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 393                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 394                 description="$(cat < $filename | markup invert-bash)"
 395                 echo "<description><![CDATA[$description]]></description>"
 396                 echo "</item>"
 397         done
 398
 399         echo '</channel>'
 400         echo '</rss>'
 401 }
 402
 403 get_image_files () {
 404         git ls-files |
 405         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\|adsense\)$'
 406 }
 407
 408 remove_old_entries () {
 409         count=$(ls source-*.txt | wc -l)
 410         test $MAXENTRIES -ge $count && return
 411
 412         for file in source-*.txt
 413         do
 414                 test $MAXENTRIES -lt $count || break
 415                 git rm $file > /dev/null || return 1
 416                 count=$(($count-1))
 417                 echo $file
 418         done
 419
 420         # remove no-longer referenced images
 421         image_files=$(get_image_files)
 422         referenced_files="$(cat source-*.txt |
 423                 tr ']|' '\n' |
 424                 sed -ne 's/\[\[\(Image\|SVG\)://p') $BACKGROUNDIMG"
 425         for file in $(echo $image_files $referenced_files $referenced_files |
 426                         tr ' ' '\n' | sort | uniq -u)
 427         do
 428                 git rm $file > /dev/null || return 1
 429                 echo $file
 430         done
 431 }
 432
 433 # never, ever have spaces in the file names
 434 commit_new_images () {
 435         files="$(remove_old_entries) $RSS $BACKGROUNDIMG" ||
 436         die "Could not remove old entries"
 437
 438
 439         generate_rss > $RSS &&
 440         git add $RSS ||
 441         die "Could not generate $RSS"
 442
 443         for image in $(cat source-* |
 444                 tr ' ]|' '\n' |
 445                 sed -n 's/.*\[\[\(Image\|SVG\)://p' |
 446                 sort |
 447                 uniq)
 448         do
 449                 git add $image || die "Could not git add image $image"
 450                 files="$files $image"
 451         done
 452
 453         git update-index --refresh &&
 454         git diff-files --quiet -- $files &&
 455         git diff --cached --quiet HEAD -- $files ||
 456         git commit -s -m "Housekeeping on $(make_date $now)" $files
 457 }
 458
 459 get_image_url () {
 460         test ! -z "$DRYRUN" && echo "$1" && return
 461         rev=$(git rev-list -1 HEAD -- $1)
 462         test -z "$rev" && die "No revision found for $1"
 463         echo "$BLOBPLAIN;hb=$rev;f=$1"
 464 }
 465
 466 handle_svg_file () {
 467         # for some reason, Firefox adds scrollbars, so nudge the width a bit
 468         width=$(sed -ne 's/.* width="\([^"]*\).*/\1/p' -e '/<metadata/q' < "$1")
 469         test -z "$width" || width=" width=$(($width+5))"
 470         url=$(get_image_url "$1")
 471         cat << EOF
 472 <center>
 473         <table border=0>
 474                 <tr>
 475                         <td align=center>
 476                                 <embed type="image/svg+xml"
 477                                         src="$url"$width />
 478                         </td>
 479                 </tr>
 480                 <tr>
 481                         <td align=center>
 482                                 <a href=$url>$1</a>
 483                         </td>
 484                 </tr>
 485         </table>
 486 </center>
 487 EOF
 488 }
 489
 490 handle_image_file () {
 491         echo "<center><img src=$(get_image_url "${1%% *}") ${1#* }></center>"
 492 }
 493
 494
 495
 496 # parse command line option
 497 case "$1" in
 498 *dry*) DRYRUN=1; export DRYRUN; shift;;
 499 *show*) firefox "$(pwd)"/$TEST; exit;;
 500 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 501 handle)
 502         shift
 503         case "$1" in
 504         "[[SVG:"*) handle_svg_file "$(chomp "$*" '\[\[SVG:' '\]\]')";;
 505         "[[Image:"*) handle_image_file "$(chomp "$*" '\[\[Image:' '\]\]')";;
 506         esac
 507         exit
 508 ;;
 509 '') ;;
 510 *) die "Unknown command: $1";;
 511 esac
 512
 513 test "$#" = 0 ||
 514 die "Usage: $0 [--dry-run]"
 515
 516 # make sure we're on the correct branch
 517 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 518 die "Not on branch $BRANCH"
 519
 520 # make sure there are no uncommitted changes
 521 git update-index --refresh &&
 522 git diff-files --quiet ||
 523 die "Have unstaged changes!"
 524
 525 # rename the new blog entry if it exists
 526 now=$(date +%s)
 527 test ! -f $NEW || {
 528         mv -i $NEW source-$now.txt &&
 529         git add source-$now.txt
 530 } ||
 531 die "Could not rename source.txt"
 532
 533 # commit the images that are referenced and not yet committed
 534 test ! -z "$DRYRUN" ||
 535 commit_new_images ||
 536 die "Could not commit new images"
 537
 538 # to find the images reliably, we have to use the commit name, not the branch
 539 # we use the latest commit touching an image file.
 540 IMAGEFILES="$(get_image_files)"
 541 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 542 test -z "$REV" && REV=$BRANCH
 543 URL="$BLOBPLAIN;hb=$REV;f="
 544
 545 if test ! -z "$DRYRUN"
 546 then
 547         # Output to test.html and have local links into the current directory
 548         OUTPUT=$TEST
 549         URL=
 550 fi
 551
 552 make_html > $OUTPUT || die "Could not write $OUTPUT"
 553
 554 test ! -z "$DRYRUN" && {
 555         move_new_entry_back
 556         exit
 557 }
 558
 559 git add $OUTPUT &&
 560 git commit -s -m "Update $(make_date $now)" &&
 561 git push origin +$BRANCH