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=$(get_config title "Dscho's blog")
  64 MAXENTRIES=$(get_config maxPostsPerPage 10)
  65 BRANCH=$(get_config branch blog)
  66
  67 URLPREFIX="$(dirname "$GITWEBURL")"/
  68 REMOTEREPOSITORY="$(basename "$GITWEBURL")"
  69 URL="$REMOTEREPOSITORY?a=blob_plain;hb=$BRANCH;f="
  70 ORIGURL=$URL
  71 NEW=new
  72 OUTPUT=index.html
  73 RSS=blog.rss
  74 TEST=test.html
  75 THIS=$0
  76
  77 LC_ALL=C
  78 export LC_ALL
  79
  80 move_new_entry_back () {
  81         test -f source-$now.txt &&
  82         mv source-$now.txt $NEW &&
  83         git rm --cached -f source-$now.txt
  84 }
  85
  86 die () {
  87         move_new_entry_back
  88         echo "$*" >&2
  89         exit 1
  90 }
  91
  92 strip_prefix () {
  93         echo "${1#$2}"
  94 }
  95
  96 chomp () {
  97         strip_prefix "${1%$3}" "$2"
  98 }
  99
 100 nth () {
 101         # add illogical suffix
 102         case "$1" in
 103         *1?|*[04-9]) echo "$1th";;
 104         *1) echo "$1st";;
 105         *2) echo "$1nd";;
 106         *3) echo "$1rd";;
 107         esac
 108 }
 109
 110 make_chinese_hour () {
 111         case $1 in
 112         23|00) echo Rat;;
 113         01|02) echo Buffalo;;
 114         03|04) echo Tiger;;
 115         05|06) echo Rabbit;;
 116         07|08) echo Dragon;;
 117         09|10) echo Snake;;
 118         11|12) echo Horse;;
 119         13|14) echo Goat;;
 120         15|16) echo Monkey;;
 121         17|18) echo Rooster;;
 122         19|20) echo Dog;;
 123         21|22) echo Pig;;
 124         esac
 125 }
 126
 127 digit_to_roman () {
 128         case $1 in
 129         1) echo $2;;
 130         2) echo $2$2;;
 131         3) echo $2$2$2;;
 132         4) echo $2$3;;
 133         5) echo $3;;
 134         6) echo $3$2;;
 135         7) echo $3$2$2;;
 136         8) echo $3$2$2$2;;
 137         9) echo $2$4;;
 138         esac
 139 }
 140
 141 make_roman_number () {
 142         case $1 in
 143         '') ;;
 144         ?) digit_to_roman $1 I V X;;
 145         ??) echo $(digit_to_roman ${1%?} X L C)$(make_roman_number ${1#?});;
 146         ???) echo $(digit_to_roman ${1%??} C D M)$(make_roman_number ${1#?});;
 147         ????) echo $(digit_to_roman ${1%???} M)$(make_roman_number ${1#?});;
 148         esac
 149 }
 150
 151 make_date () {
 152         printf "%s, %s of %s, Anno Domini %s, at the hour of the %s\n" \
 153                 $(date +%A -d @$1) \
 154                 $(nth $(date +%e -d @$1)) \
 155                 $(date +%B -d @$1) \
 156                 $(make_roman_number $(date +%Y -d @$1)) \
 157                 $(make_chinese_hour $(date +%H -d @$1))
 158 }
 159
 160 # make an argument for sed, to replace $1..$1 by <$2>..</$2>
 161 markup_substitution () {
 162         case "$1" in
 163         ?) echo "s/$1\\([^$1]*\\)$1/<$2>\\\\1<\/$2>/g";;
 164         ??)
 165                 tmp="[^${1%?}]*"
 166                 tmp2="\\|${1%?}[^${1#?}]$tmp"
 167                 tmp3="\\($tmp\\($tmp2\\($tmp2\\($tmp2\\)\\)\\)\\)"
 168                 echo "s/$1$tmp3$1/<$2>\\\\1<\/$2>/g"
 169         ;;
 170         esac
 171 }
 172
 173 # transform markup in stdin to HTML
 174 markup () {
 175         case "$*" in
 176         *invert-bash*) bash_bg=white; bash_fg=black;;
 177         *) bash_bg=black; bash_fg=white;;
 178         esac
 179         sed -e 's!^$!</p><p>!' \
 180                 -e "$(markup_substitution "''" i)" \
 181                 -e "$(markup_substitution "_" u)" \
 182                 -e 's!IMHO!in my humble opinion!g' \
 183                 -e 's!BTW!By the way,!g' \
 184                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 185                 -e 's!:-)!\&#x263a;!g' \
 186                 -e "s!\\[\\[\(Image\|SVG\):.*!$THIS handle &!e" \
 187                 -e 's!<bash>!<table\
 188                                 border=1 bgcolor='$bash_bg'>\
 189                         <tr><td bgcolor=lightblue colspan=3>\
 190                                 \&nbsp;\
 191                         </td></tr>\
 192                         <tr><td>\
 193                                 <table cellspacing=5 border=0\
 194                                          style="color:'$bash_fg';">\
 195                                 <tr><td>\
 196                                         <pre>!' \
 197                 -e 's!</bash>!          </pre>\
 198                                 </td></tr>\
 199                                 </table>\
 200                         </td></tr>\
 201                         </table>!' \
 202
 203 }
 204
 205 # output lines containing <timestamp> <filename> <title>
 206 get_blog_entries () {
 207         for file in $(ls -r source-*.txt)
 208         do
 209                 timestamp=$(chomp $file source- .txt)
 210                 title="$(sed 1q < $file | markup)"
 211                 echo "$timestamp $file $title"
 212         done
 213 }
 214
 215 get_last_removed_entry () {
 216         git log --pretty=format: --name-only --diff-filter=D HEAD |
 217         while read line
 218         do
 219                 case "$line" in
 220                 source-*.txt) file=$line;;
 221                 '') test -z "$file" || {
 222                         echo "$file"
 223                         break
 224                 };;
 225                 esac
 226         done
 227 }
 228
 229 box_count=0
 230 begin_box () {
 231         test $box_count = 0 || echo "<br>"
 232         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 233         echo "<tr><th>$1</th></tr>"
 234         echo "<tr><td>"
 235 }
 236
 237 end_box () {
 238         echo "</td></tr></table>"
 239         box_count=$(($box_count+1))
 240 }
 241
 242 # make HTML page
 243 make_html () {
 244         body_style="width:800px"
 245         body_style="$body_style;background-image:url($URL$BACKGROUNDIMG)"
 246         body_style="$body_style;background-repeat:repeat-y"
 247         body_style="$body_style;background-attachment:scroll"
 248         body_style="$body_style;padding:0px;"
 249         text_style="width:610px"
 250         text_style="$text_style;margin-left:120px"
 251         text_style="$text_style;margin-top:50px"
 252         text_style="$text_style;align:left"
 253         text_style="$text_style;vertical-align:top;"
 254         cat << EOF
 255 <html>
 256         <head>
 257                 <title>$TITLE</title>
 258                 <meta http-equiv="Content-Type"
 259                         content="text/html; charset=UTF-8"/>
 260         </head>
 261         <body style="$body_style">
 262                 <div style="$text_style">
 263                         <h1>$TITLE</h1>
 264 EOF
 265         indent='\t\t\t'
 266
 267         # make toc
 268         toc_width=400px
 269         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 270         {
 271                 echo "<div style=\"$toc_style\">"
 272                 begin_box "Table of contents:"
 273                 echo '<p><ul>'
 274                 get_blog_entries |
 275                 while read timestamp filename title
 276                 do
 277                         date="$(date +"%d %b %Y" -d @$timestamp)"
 278                         echo "<li><a href=#$timestamp>$date $title</a>"
 279                 done
 280                 echo '</ul></p>'
 281                 file=
 282                 last_removed_entry=$(get_last_removed_entry)
 283                 test -z "$last_removed_entry" || {
 284                         commit=$(git log --pretty=format:%H --diff-filter=AM \
 285                                         -- $last_removed_entry |
 286                                 head -n 1)
 287                         previous="$REMOTEREPOSITORY?a=blob_plain;hb=$commit"
 288                         echo "<a href=$previous;f=index.html>Older posts</a>"
 289                 }
 290                 end_box
 291
 292                 # RSS feed
 293                 rss_style="background-color:orange;text-decoration:none"
 294                 rss_style="$rss_style;color:white;font-family:sans-serif;"
 295                 echo '<br>'
 296                 echo '<div style="text-align:right;">'
 297                 echo "<a href=\"$ORIGURL$RSS\""
 298                 echo '   title="Subscribe to my RSS feed"'
 299                 echo '   class="rss" rel="nofollow"'
 300                 echo "   style=\"$rss_style\">RSS</a>"
 301                 echo '</div>'
 302
 303                 # About
 304                 test -f about.html && {
 305                         begin_box "About this blog:"
 306                         cat about.html
 307                         end_box
 308                 }
 309
 310                 # Links
 311                 test -f links.html && {
 312                         begin_box "Links:"
 313                         cat links.html
 314                         end_box
 315                 }
 316
 317                 # Google AdSense
 318                 test -z "$DRYRUN" && test -f google.adsense && {
 319                         begin_box "Google Ads:"
 320                         cat google.adsense
 321                         end_box
 322                 }
 323
 324                 echo '</div>'
 325         } | sed -s "s/^/$indent/"
 326
 327
 328         # timestamps will not need padding to sort correctly, for some time...
 329         get_blog_entries |
 330         while read timestamp filename title
 331         do
 332                 echo "<h6>$(make_date $timestamp)</h6>"
 333                 echo "<a name=$timestamp>"
 334                 echo "<h2>$title</h2>"
 335                 echo ""
 336                 echo "<p>"
 337                 sed 1d < $filename | markup
 338                 echo "</p>"
 339         done |
 340         sed -e "s/^./$indent&/" \
 341                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 342
 343         cat << EOF
 344                 </div>
 345         </body>
 346 </html>
 347 EOF
 348 }
 349
 350 generate_rss () {
 351         echo '<?xml version="1.0" encoding="utf-8"?>'
 352         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 353         echo '<channel>'
 354         echo "<title>Dscho's blog</title>"
 355         echo "<link>$URLPREFIX${URL}index.html</link>"
 356         self="$URLPREFIX$ORIGURL$RSS"
 357         selfattribs='rel="self" type="application/rss+xml"'
 358         echo "<atom:link href=\"$self\" $selfattribs/>"
 359         echo '<description>A few stories told by Dscho</description>'
 360         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 361         echo '<language>en-us</language>'
 362
 363         get_blog_entries |
 364         while read timestamp filename title
 365         do
 366                 echo '<item>'
 367                 echo "<title>$title</title>"
 368                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 369                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 370                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 371                 description="$(cat < $filename | markup invert-bash)"
 372                 echo "<description><![CDATA[$description]]></description>"
 373                 echo "</item>"
 374         done
 375
 376         echo '</channel>'
 377         echo '</rss>'
 378 }
 379
 380 get_image_files () {
 381         git ls-files |
 382         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\|adsense\)$'
 383 }
 384
 385 remove_old_entries () {
 386         count=$(ls source-*.txt | wc -l)
 387         test $MAXENTRIES -ge $count && return
 388
 389         for file in source-*.txt
 390         do
 391                 test $MAXENTRIES -lt $count || break
 392                 git rm $file > /dev/null || return 1
 393                 count=$(($count-1))
 394                 echo $file
 395         done
 396
 397         # remove no-longer referenced images
 398         image_files=$(get_image_files)
 399         referenced_files="$(cat source-*.txt |
 400                 tr ']|' '\n' |
 401                 sed -ne 's/\[\[\(Image\|SVG\)://p') $BACKGROUNDIMG"
 402         for file in $(echo $image_files $referenced_files $referenced_files |
 403                         tr ' ' '\n' | sort | uniq -u)
 404         do
 405                 git rm $file > /dev/null || return 1
 406                 echo $file
 407         done
 408 }
 409
 410 # never, ever have spaces in the file names
 411 commit_new_images () {
 412         files="$(remove_old_entries) $RSS $BACKGROUNDIMG" ||
 413         die "Could not remove old entries"
 414
 415
 416         generate_rss > $RSS &&
 417         git add $RSS ||
 418         die "Could not generate $RSS"
 419
 420         for image in $(cat source-* |
 421                 tr ' ]|' '\n' |
 422                 sed -n 's/.*\[\[\(Image\|SVG\)://p' |
 423                 sort |
 424                 uniq)
 425         do
 426                 git add $image || die "Could not git add image $image"
 427                 files="$files $image"
 428         done
 429
 430         git update-index --refresh &&
 431         git diff-files --quiet -- $files &&
 432         git diff --cached --quiet HEAD -- $files ||
 433         git commit -s -m "Commit some images on $(make_date $now)" $files
 434 }
 435
 436 get_image_url () {
 437         test ! -z "$DRYRUN" && echo "$1" && return
 438         rev=$(git rev-list -1 HEAD -- $1)
 439         test -z "$rev" && die "No revision found for $1"
 440         echo "$REMOTEREPOSITORY?a=blob_plain;hb=$rev;f=$1"
 441 }
 442
 443 handle_svg_file () {
 444         # for some reason, Firefox adds scrollbars, so nudge the width a bit
 445         width=$(sed -ne 's/.* width="\([^"]*\).*/\1/p' -e '/<metadata/q' < "$1")
 446         test -z "$width" || width=" width=$(($width+5))"
 447         url=$(get_image_url "$1")
 448         cat << EOF
 449 <center>
 450         <table border=0>
 451                 <tr>
 452                         <td align=center>
 453                                 <embed type="image/svg+xml"
 454                                         src="$url"$width />
 455                         </td>
 456                 </tr>
 457                 <tr>
 458                         <td align=center>
 459                                 <a href=$url>$1</a>
 460                         </td>
 461                 </tr>
 462         </table>
 463 </center>
 464 EOF
 465 }
 466
 467 handle_image_file () {
 468         echo "<center><img src=$(get_image_url "${1%% *}") ${1#* }></center>"
 469 }
 470
 471
 472
 473 # parse command line option
 474 case "$1" in
 475 *dry*) DRYRUN=1; export DRYRUN; shift;;
 476 *show*) firefox "$(pwd)"/$TEST; exit;;
 477 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 478 handle)
 479         shift
 480         case "$1" in
 481         "[[SVG:"*) handle_svg_file "$(chomp "$*" '\[\[SVG:' '\]\]')";;
 482         "[[Image:"*) handle_image_file "$(chomp "$*" '\[\[Image:' '\]\]')";;
 483         esac
 484         exit
 485 ;;
 486 '') ;;
 487 *) die "Unknown command: $1";;
 488 esac
 489
 490 test "$#" = 0 ||
 491 die "Usage: $0 [--dry-run]"
 492
 493 # make sure we're on the correct branch
 494 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 495 die "Not on branch $BRANCH"
 496
 497 # make sure there are no uncommitted changes
 498 git update-index --refresh &&
 499 git diff-files --quiet ||
 500 die "Have unstaged changes!"
 501
 502 # rename the new blog entry if it exists
 503 now=$(date +%s)
 504 test ! -f $NEW || {
 505         mv -i $NEW source-$now.txt &&
 506         git add source-$now.txt
 507 } ||
 508 die "Could not rename source.txt"
 509
 510 # commit the images that are referenced and not yet committed
 511 test ! -z "$DRYRUN" ||
 512 commit_new_images ||
 513 die "Could not commit new images"
 514
 515 # to find the images reliably, we have to use the commit name, not the branch
 516 # we use the latest commit touching an image file.
 517 IMAGEFILES="$(get_image_files)"
 518 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 519 test -z "$REV" && REV=$BRANCH
 520 URL="$REMOTEREPOSITORY?a=blob_plain;hb=$REV;f="
 521
 522 if test ! -z "$DRYRUN"
 523 then
 524         # Output to test.html and have local links into the current directory
 525         OUTPUT=$TEST
 526         URL=
 527 fi
 528
 529 make_html > $OUTPUT || die "Could not write $OUTPUT"
 530
 531 test ! -z "$DRYRUN" && {
 532         move_new_entry_back
 533         exit
 534 }
 535
 536 git add $OUTPUT &&
 537 git commit -s -m "Update $(make_date $now)" &&
 538 git push origin +$BRANCH