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 "$GITWEBURL")"
  69 case "$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         case "$1" in
 168         ?) echo "s/$1\\([^$1]*\\)$1/<$2>\\\\1<\/$2>/g";;
 169         ??)
 170                 tmp="[^${1%?}]*"
 171                 tmp2="\\|${1%?}[^${1#?}]$tmp"
 172                 tmp3="\\($tmp\\($tmp2\\($tmp2\\($tmp2\\)\\)\\)\\)"
 173                 echo "s/$1$tmp3$1/<$2>\\\\1<\/$2>/g"
 174         ;;
 175         esac
 176 }
 177
 178 space80='        '
 179 space80="$space80$space80$space80$space80$space80$space80$space80$space80"
 180 # transform markup in stdin to HTML
 181 markup () {
 182         case "$*" in
 183         *invert-bash*) bash_bg=white; bash_fg=black;;
 184         *) bash_bg=black; bash_fg=white;;
 185         esac
 186         sed -e 's!^$!</p><p>!' \
 187                 -e "$(markup_substitution "''" i)" \
 188                 -e "$(markup_substitution "_" u)" \
 189                 -e 's!IMHO!in my humble opinion!g' \
 190                 -e 's!BTW!By the way,!g' \
 191                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 192                 -e 's!:-)!\&#x263a;!g' \
 193                 -e "s!\\[\\[\(Image\|SVG\):.*!$THIS handle &!e" \
 194                 -e 's!<bash>!<table\
 195                                 border=1 bgcolor='$bash_bg'>\
 196                         <tr><td bgcolor=lightblue colspan=3>\
 197                                 <pre>'"$space80"'</pre>\
 198                         </td></tr>\
 199                         <tr><td>\
 200                                 <table cellspacing=5 border=0\
 201                                          style="color:'$bash_fg';">\
 202                                 <tr><td>\
 203                                         <pre>!' \
 204                 -e 's!</bash>!</pre>\
 205                                 </td></tr>\
 206                                 </table>\
 207                         </td></tr>\
 208                         </table>!' \
 209
 210 }
 211
 212 # output lines containing <timestamp> <filename> <title>
 213 get_blog_entries () {
 214         for file in $(ls -r source-*.txt)
 215         do
 216                 timestamp=$(chomp $file source- .txt)
 217                 title="$(sed 1q < $file | markup)"
 218                 echo "$timestamp $file $title"
 219         done
 220 }
 221
 222 get_last_removed_entry () {
 223         git log --pretty=format: --name-only --diff-filter=D HEAD |
 224         while read line
 225         do
 226                 case "$line" in
 227                 source-*.txt) file=$line;;
 228                 '') test -z "$file" || {
 229                         echo "$file"
 230                         break
 231                 };;
 232                 esac
 233         done
 234 }
 235
 236 box_count=0
 237 begin_box () {
 238         test $box_count = 0 || echo "<br>"
 239         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 240         echo "<tr><th>$1</th></tr>"
 241         echo "<tr><td>"
 242 }
 243
 244 end_box () {
 245         echo "</td></tr></table>"
 246         box_count=$(($box_count+1))
 247 }
 248
 249 # make HTML page
 250 make_html () {
 251         body_style="width:800px"
 252         body_style="$body_style;background-image:url($URL$BACKGROUNDIMG)"
 253         body_style="$body_style;background-repeat:repeat-y"
 254         body_style="$body_style;background-attachment:scroll"
 255         body_style="$body_style;padding:0px;"
 256         text_style="width:610px"
 257         text_style="$text_style;margin-left:120px"
 258         text_style="$text_style;margin-top:50px"
 259         text_style="$text_style;align:left"
 260         text_style="$text_style;vertical-align:top;"
 261         cat << EOF
 262 <html>
 263         <head>
 264                 <title>$TITLE</title>
 265                 <meta http-equiv="Content-Type"
 266                         content="text/html; charset=UTF-8"/>
 267         </head>
 268         <body style="$body_style">
 269                 <div style="$text_style">
 270                         <h1>$TITLE</h1>
 271 EOF
 272         indent='\t\t\t'
 273
 274         # make toc
 275         toc_width=400px
 276         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 277         {
 278                 echo "<div style=\"$toc_style\">"
 279                 begin_box "Table of contents:"
 280                 echo '<p><ul>'
 281                 get_blog_entries |
 282                 while read timestamp filename title
 283                 do
 284                         date="$(date +"%d %b %Y" -d @$timestamp)"
 285                         echo "<li><a href=#$timestamp>$date $title</a>"
 286                 done
 287                 echo '</ul></p>'
 288                 file=
 289                 last_removed_entry=$(get_last_removed_entry)
 290                 test -z "$last_removed_entry" || {
 291                         commit=$(git log --pretty=format:%H --diff-filter=AM \
 292                                         -- $last_removed_entry |
 293                                 head -n 1)
 294                         previous="$BLOBPLAIN;hb=$commit"
 295                         echo "<a href=$previous;f=index.html>Older posts</a>"
 296                 }
 297                 end_box
 298
 299                 # RSS feed
 300                 rss_style="background-color:orange;text-decoration:none"
 301                 rss_style="$rss_style;color:white;font-family:sans-serif;"
 302                 echo '<br>'
 303                 echo '<div style="text-align:right;">'
 304                 echo "<a href=\"$ORIGURL$RSS\""
 305                 echo '   title="Subscribe to my RSS feed"'
 306                 echo '   class="rss" rel="nofollow"'
 307                 echo "   style=\"$rss_style\">RSS</a>"
 308                 echo '</div>'
 309
 310                 # About
 311                 test -f about.html && {
 312                         begin_box "About this blog:"
 313                         cat about.html
 314                         end_box
 315                 }
 316
 317                 # Links
 318                 test -f links.html && {
 319                         begin_box "Links:"
 320                         cat links.html
 321                         end_box
 322                 }
 323
 324                 # Google AdSense
 325                 test -z "$DRYRUN" && test -f google.adsense && {
 326                         begin_box "Google Ads:"
 327                         cat google.adsense
 328                         end_box
 329                 }
 330
 331                 echo '</div>'
 332         } | sed -s "s/^/$indent/"
 333
 334
 335         # timestamps will not need padding to sort correctly, for some time...
 336         get_blog_entries |
 337         while read timestamp filename title
 338         do
 339                 echo "<h6>$(make_date $timestamp)</h6>"
 340                 echo "<a name=$timestamp>"
 341                 echo "<h2>$title</h2>"
 342                 echo ""
 343                 echo "<p>"
 344                 sed 1d < $filename | markup
 345                 echo "</p>"
 346         done |
 347         sed -e "s/^./$indent&/" \
 348                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 349
 350         cat << EOF
 351                 </div>
 352         </body>
 353 </html>
 354 EOF
 355 }
 356
 357 generate_rss () {
 358         echo '<?xml version="1.0" encoding="utf-8"?>'
 359         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 360         echo '<channel>'
 361         echo "<title>Dscho's blog</title>"
 362         echo "<link>$URLPREFIX${URL}index.html</link>"
 363         self="$URLPREFIX$ORIGURL$RSS"
 364         selfattribs='rel="self" type="application/rss+xml"'
 365         echo "<atom:link href=\"$self\" $selfattribs/>"
 366         echo '<description>A few stories told by Dscho</description>'
 367         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 368         echo '<language>en-us</language>'
 369
 370         get_blog_entries |
 371         while read timestamp filename title
 372         do
 373                 # remove all tags
 374                 title=$(echo "$title" | sed 's/<[^>]*>//g')
 375                 echo '<item>'
 376                 echo "<title>$title</title>"
 377                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 378                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 379                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 380                 description="$(cat < $filename | markup invert-bash)"
 381                 echo "<description><![CDATA[$description]]></description>"
 382                 echo "</item>"
 383         done
 384
 385         echo '</channel>'
 386         echo '</rss>'
 387 }
 388
 389 get_image_files () {
 390         git ls-files |
 391         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\|adsense\)$'
 392 }
 393
 394 remove_old_entries () {
 395         count=$(ls source-*.txt | wc -l)
 396         test $MAXENTRIES -ge $count && return
 397
 398         for file in source-*.txt
 399         do
 400                 test $MAXENTRIES -lt $count || break
 401                 git rm $file > /dev/null || return 1
 402                 count=$(($count-1))
 403                 echo $file
 404         done
 405
 406         # remove no-longer referenced images
 407         image_files=$(get_image_files)
 408         referenced_files="$(cat source-*.txt |
 409                 tr ']|' '\n' |
 410                 sed -ne 's/\[\[\(Image\|SVG\)://p') $BACKGROUNDIMG"
 411         for file in $(echo $image_files $referenced_files $referenced_files |
 412                         tr ' ' '\n' | sort | uniq -u)
 413         do
 414                 git rm $file > /dev/null || return 1
 415                 echo $file
 416         done
 417 }
 418
 419 # never, ever have spaces in the file names
 420 commit_new_images () {
 421         files="$(remove_old_entries) $RSS $BACKGROUNDIMG" ||
 422         die "Could not remove old entries"
 423
 424
 425         generate_rss > $RSS &&
 426         git add $RSS ||
 427         die "Could not generate $RSS"
 428
 429         for image in $(cat source-* |
 430                 tr ' ]|' '\n' |
 431                 sed -n 's/.*\[\[\(Image\|SVG\)://p' |
 432                 sort |
 433                 uniq)
 434         do
 435                 git add $image || die "Could not git add image $image"
 436                 files="$files $image"
 437         done
 438
 439         git update-index --refresh &&
 440         git diff-files --quiet -- $files &&
 441         git diff --cached --quiet HEAD -- $files ||
 442         git commit -s -m "Housekeeping on $(make_date $now)" $files
 443 }
 444
 445 get_image_url () {
 446         test ! -z "$DRYRUN" && echo "$1" && return
 447         rev=$(git rev-list -1 HEAD -- $1)
 448         test -z "$rev" && die "No revision found for $1"
 449         echo "$BLOBPLAIN;hb=$rev;f=$1"
 450 }
 451
 452 handle_svg_file () {
 453         # for some reason, Firefox adds scrollbars, so nudge the width a bit
 454         width=$(sed -ne 's/.* width="\([^"]*\).*/\1/p' -e '/<metadata/q' < "$1")
 455         test -z "$width" || width=" width=$(($width+5))"
 456         url=$(get_image_url "$1")
 457         cat << EOF
 458 <center>
 459         <table border=0>
 460                 <tr>
 461                         <td align=center>
 462                                 <embed type="image/svg+xml"
 463                                         src="$url"$width />
 464                         </td>
 465                 </tr>
 466                 <tr>
 467                         <td align=center>
 468                                 <a href=$url>$1</a>
 469                         </td>
 470                 </tr>
 471         </table>
 472 </center>
 473 EOF
 474 }
 475
 476 handle_image_file () {
 477         echo "<center><img src=$(get_image_url "${1%% *}") ${1#* }></center>"
 478 }
 479
 480
 481
 482 # parse command line option
 483 case "$1" in
 484 *dry*) DRYRUN=1; export DRYRUN; shift;;
 485 *show*) firefox "$(pwd)"/$TEST; exit;;
 486 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 487 handle)
 488         shift
 489         case "$1" in
 490         "[[SVG:"*) handle_svg_file "$(chomp "$*" '\[\[SVG:' '\]\]')";;
 491         "[[Image:"*) handle_image_file "$(chomp "$*" '\[\[Image:' '\]\]')";;
 492         esac
 493         exit
 494 ;;
 495 '') ;;
 496 *) die "Unknown command: $1";;
 497 esac
 498
 499 test "$#" = 0 ||
 500 die "Usage: $0 [--dry-run]"
 501
 502 # make sure we're on the correct branch
 503 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 504 die "Not on branch $BRANCH"
 505
 506 # make sure there are no uncommitted changes
 507 git update-index --refresh &&
 508 git diff-files --quiet ||
 509 die "Have unstaged changes!"
 510
 511 # rename the new blog entry if it exists
 512 now=$(date +%s)
 513 test ! -f $NEW || {
 514         mv -i $NEW source-$now.txt &&
 515         git add source-$now.txt
 516 } ||
 517 die "Could not rename source.txt"
 518
 519 # commit the images that are referenced and not yet committed
 520 test ! -z "$DRYRUN" ||
 521 commit_new_images ||
 522 die "Could not commit new images"
 523
 524 # to find the images reliably, we have to use the commit name, not the branch
 525 # we use the latest commit touching an image file.
 526 IMAGEFILES="$(get_image_files)"
 527 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 528 test -z "$REV" && REV=$BRANCH
 529 URL="$BLOBPLAIN;hb=$REV;f="
 530
 531 if test ! -z "$DRYRUN"
 532 then
 533         # Output to test.html and have local links into the current directory
 534         OUTPUT=$TEST
 535         URL=
 536 fi
 537
 538 make_html > $OUTPUT || die "Could not write $OUTPUT"
 539
 540 test ! -z "$DRYRUN" && {
 541         move_new_entry_back
 542         exit
 543 }
 544
 545 git add $OUTPUT &&
 546 git commit -s -m "Update $(make_date $now)" &&
 547 git push origin +$BRANCH