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         # work around stupid dash interpreting backslashes when expanding vars
 190         bs="\\\\"
 191         test "\\" = "$(echo "$bs")" || bs="\\"
 192         echo "s/$left$right/${bs}1<$2>${bs}2<\/$2>$bs$right_no/g"
 193 }
 194
 195 space80='                          '
 196 space80="$space80$space80$space80  "
 197 # transform markup in stdin to HTML
 198 markup () {
 199         case "$*" in
 200         *invert-bash*) bash_bg=white; bash_fg=black;;
 201         *) bash_bg=black; bash_fg=white;;
 202         esac
 203         sed -e 's!^$!</p><p>!' \
 204                 -e "$(markup_substitution "''" i)" \
 205                 -e "$(markup_substitution "_" u)" \
 206                 -e 's!IMHO!in my humble opinion!g' \
 207                 -e 's!BTW!By the way,!g' \
 208                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 209                 -e 's!:-)!\&#x263a;!g' \
 210                 -e "s!\\[\\[\(Image\|SVG\):.*!$THIS handle &!e" \
 211                 -e 's!<bash>!<table\
 212                                 border=1 bgcolor='$bash_bg'>\
 213                         <tr><td bgcolor=lightblue colspan=3>\
 214                                 <pre>'"$space80"'</pre>\
 215                         </td></tr>\
 216                         <tr><td>\
 217                                 <table cellspacing=5 border=0\
 218                                          style="color:'$bash_fg';">\
 219                                 <tr><td>\
 220                                         <pre>!' \
 221                 -e 's!</bash>!</pre>\
 222                                 </td></tr>\
 223                                 </table>\
 224                         </td></tr>\
 225                         </table>!' \
 226
 227 }
 228
 229 # output lines containing <timestamp> <filename> <title>
 230 get_blog_entries () {
 231         for file in $(ls -r source-*.txt)
 232         do
 233                 timestamp=$(chomp $file source- .txt)
 234                 title="$(sed 1q < $file | markup)"
 235                 echo "$timestamp $file $title"
 236         done
 237 }
 238
 239 get_last_removed_entry () {
 240         git log --pretty=format: --name-only --diff-filter=D HEAD |
 241         while read line
 242         do
 243                 case "$line" in
 244                 source-*.txt) file=$line;;
 245                 '') test -z "$file" || {
 246                         echo "$file"
 247                         break
 248                 };;
 249                 esac
 250         done
 251 }
 252
 253 box_count=0
 254 begin_box () {
 255         test $box_count = 0 || echo "<br>"
 256         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 257         echo "<tr><th>$1</th></tr>"
 258         echo "<tr><td>"
 259 }
 260
 261 end_box () {
 262         echo "</td></tr></table>"
 263         box_count=$(($box_count+1))
 264 }
 265
 266 # make HTML page
 267 make_html () {
 268         body_style="width:800px"
 269         body_style="$body_style;background-image:url($URL$BACKGROUNDIMG)"
 270         body_style="$body_style;background-repeat:repeat-y"
 271         body_style="$body_style;background-attachment:scroll"
 272         body_style="$body_style;padding:0px;"
 273         text_style="width:610px"
 274         text_style="$text_style;margin-left:120px"
 275         text_style="$text_style;margin-top:50px"
 276         text_style="$text_style;align:left"
 277         text_style="$text_style;vertical-align:top;"
 278         cat << EOF
 279 <html>
 280         <head>
 281                 <title>$TITLE</title>
 282                 <meta http-equiv="Content-Type"
 283                         content="text/html; charset=UTF-8"/>
 284         </head>
 285         <body style="$body_style">
 286                 <div style="$text_style">
 287                         <h1>$TITLE</h1>
 288 EOF
 289         indent='\t\t\t'
 290
 291         # make toc
 292         toc_width=400px
 293         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 294         {
 295                 echo "<div style=\"$toc_style\">"
 296                 begin_box "Table of contents:"
 297                 echo '<p><ul>'
 298                 get_blog_entries |
 299                 while read timestamp filename title
 300                 do
 301                         date="$(date +"%d %b %Y" -d @$timestamp)"
 302                         echo "<li><a href=#$timestamp>$date $title</a>"
 303                 done
 304                 echo '</ul></p>'
 305                 file=
 306                 last_removed_entry=$(get_last_removed_entry)
 307                 test -z "$last_removed_entry" || {
 308                         commit=$(git log --pretty=format:%H --diff-filter=AM \
 309                                         -- $last_removed_entry |
 310                                 head -n 1)
 311                         previous="$BLOBPLAIN;hb=$commit"
 312                         echo "<a href=$previous;f=index.html>Older posts</a>"
 313                 }
 314                 end_box
 315
 316                 # RSS feed
 317                 rss_style="background-color:orange;text-decoration:none"
 318                 rss_style="$rss_style;color:white;font-family:sans-serif;"
 319                 echo '<br>'
 320                 echo '<div style="text-align:right;">'
 321                 echo "<a href=\"$ORIGURL$RSS\""
 322                 echo '   title="Subscribe to my RSS feed"'
 323                 echo '   class="rss" rel="nofollow"'
 324                 echo "   style=\"$rss_style\">RSS</a>"
 325                 echo '</div>'
 326
 327                 # About
 328                 test -f about.html && {
 329                         begin_box "About this blog:"
 330                         cat about.html
 331                         end_box
 332                 }
 333
 334                 # Links
 335                 test -f links.html && {
 336                         begin_box "Links:"
 337                         cat links.html
 338                         end_box
 339                 }
 340
 341                 # Google AdSense
 342                 test -z "$DRYRUN" && test -f google.adsense && {
 343                         begin_box "Google Ads:"
 344                         cat google.adsense
 345                         end_box
 346                 }
 347
 348                 echo '</div>'
 349         } | sed -s "s/^/$indent/"
 350
 351
 352         # timestamps will not need padding to sort correctly, for some time...
 353         get_blog_entries |
 354         while read timestamp filename title
 355         do
 356                 echo "<h6>$(make_date $timestamp)</h6>"
 357                 echo "<a name=$timestamp>"
 358                 echo "<h2>$title</h2>"
 359                 echo ""
 360                 echo "<p>"
 361                 sed 1d < $filename | markup
 362                 echo "</p>"
 363         done |
 364         sed -e "s/^./$indent&/" \
 365                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 366
 367         cat << EOF
 368                 </div>
 369         </body>
 370 </html>
 371 EOF
 372 }
 373
 374 generate_rss () {
 375         echo '<?xml version="1.0" encoding="utf-8"?>'
 376         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 377         echo '<channel>'
 378         echo "<title>Dscho's blog</title>"
 379         echo "<link>$URLPREFIX${URL}index.html</link>"
 380         self="$URLPREFIX$ORIGURL$RSS"
 381         selfattribs='rel="self" type="application/rss+xml"'
 382         echo "<atom:link href=\"$self\" $selfattribs/>"
 383         echo '<description>A few stories told by Dscho</description>'
 384         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 385         echo '<language>en-us</language>'
 386
 387         get_blog_entries |
 388         while read timestamp filename title
 389         do
 390                 # remove all tags
 391                 title=$(echo "$title" | sed 's/<[^>]*>//g')
 392                 echo '<item>'
 393                 echo "<title>$title</title>"
 394                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 395                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 396                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 397                 description="$(cat < $filename | markup invert-bash)"
 398                 echo "<description><![CDATA[$description]]></description>"
 399                 echo "</item>"
 400         done
 401
 402         echo '</channel>'
 403         echo '</rss>'
 404 }
 405
 406 get_image_files () {
 407         git ls-files |
 408         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\|adsense\)$'
 409 }
 410
 411 remove_old_entries () {
 412         count=$(ls source-*.txt | wc -l)
 413         test $MAXENTRIES -ge $count && return
 414
 415         for file in source-*.txt
 416         do
 417                 test $MAXENTRIES -lt $count || break
 418                 git rm $file > /dev/null || return 1
 419                 count=$(($count-1))
 420                 echo $file
 421         done
 422
 423         # remove no-longer referenced images
 424         image_files=$(get_image_files)
 425         referenced_files="$(cat source-*.txt |
 426                 tr ']|' '\n' |
 427                 sed -ne 's/\[\[\(Image\|SVG\)://p') $BACKGROUNDIMG"
 428         for file in $(echo $image_files $referenced_files $referenced_files |
 429                         tr ' ' '\n' | sort | uniq -u)
 430         do
 431                 git rm $file > /dev/null || return 1
 432                 echo $file
 433         done
 434 }
 435
 436 # never, ever have spaces in the file names
 437 commit_new_images () {
 438         files="$(remove_old_entries) $RSS $BACKGROUNDIMG" ||
 439         die "Could not remove old entries"
 440
 441
 442         generate_rss > $RSS &&
 443         git add $RSS ||
 444         die "Could not generate $RSS"
 445
 446         for image in $(cat source-* |
 447                 tr ' ]|' '\n' |
 448                 sed -n 's/.*\[\[\(Image\|SVG\)://p' |
 449                 sort |
 450                 uniq)
 451         do
 452                 git add $image || die "Could not git add image $image"
 453                 files="$files $image"
 454         done
 455
 456         git update-index --refresh &&
 457         git diff-files --quiet -- $files &&
 458         git diff --cached --quiet HEAD -- $files ||
 459         git commit -s -m "Housekeeping on $(make_date $now)" $files
 460 }
 461
 462 get_image_url () {
 463         test ! -z "$DRYRUN" && echo "$1" && return
 464         rev=$(git rev-list -1 HEAD -- $1)
 465         test -z "$rev" && die "No revision found for $1"
 466         echo "$BLOBPLAIN;hb=$rev;f=$1"
 467 }
 468
 469 handle_svg_file () {
 470         # for some reason, Firefox adds scrollbars, so nudge the width a bit
 471         width=$(sed -ne 's/.* width="\([^"]*\).*/\1/p' -e '/<metadata/q' < "$1")
 472         test -z "$width" || width=" width=$(($width+5))"
 473         url=$(get_image_url "$1")
 474         cat << EOF
 475 <center>
 476         <table border=0>
 477                 <tr>
 478                         <td align=center>
 479                                 <embed type="image/svg+xml"
 480                                         src="$url"$width />
 481                         </td>
 482                 </tr>
 483                 <tr>
 484                         <td align=center>
 485                                 <a href=$url>$1</a>
 486                         </td>
 487                 </tr>
 488         </table>
 489 </center>
 490 EOF
 491 }
 492
 493 handle_image_file () {
 494         echo "<center><img src=$(get_image_url "${1%% *}") ${1#* }></center>"
 495 }
 496
 497
 498
 499 # parse command line option
 500 case "$1" in
 501 *dry*) DRYRUN=1; export DRYRUN; shift;;
 502 *show*) firefox "$(pwd)"/$TEST; exit;;
 503 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 504 handle)
 505         shift
 506         case "$1" in
 507         "[[SVG:"*) handle_svg_file "$(chomp "$*" '\[\[SVG:' '\]\]')";;
 508         "[[Image:"*) handle_image_file "$(chomp "$*" '\[\[Image:' '\]\]')";;
 509         esac
 510         exit
 511 ;;
 512 '') ;;
 513 *) die "Unknown command: $1";;
 514 esac
 515
 516 test "$#" = 0 ||
 517 die "Usage: $0 [--dry-run]"
 518
 519 # make sure we're on the correct branch
 520 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 521 die "Not on branch $BRANCH"
 522
 523 # make sure there are no uncommitted changes
 524 git update-index --refresh &&
 525 git diff-files --quiet ||
 526 die "Have unstaged changes!"
 527
 528 # rename the new blog entry if it exists
 529 now=$(date +%s)
 530 test ! -f $NEW || {
 531         mv -i $NEW source-$now.txt &&
 532         git add source-$now.txt
 533 } ||
 534 die "Could not rename source.txt"
 535
 536 # commit the images that are referenced and not yet committed
 537 test ! -z "$DRYRUN" ||
 538 commit_new_images ||
 539 die "Could not commit new images"
 540
 541 # to find the images reliably, we have to use the commit name, not the branch
 542 # we use the latest commit touching an image file.
 543 IMAGEFILES="$(get_image_files)"
 544 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 545 test -z "$REV" && REV=$BRANCH
 546 URL="$BLOBPLAIN;hb=$REV;f="
 547
 548 if test ! -z "$DRYRUN"
 549 then
 550         # Output to test.html and have local links into the current directory
 551         OUTPUT=$TEST
 552         URL=
 553 fi
 554
 555 make_html > $OUTPUT || die "Could not write $OUTPUT"
 556
 557 test ! -z "$DRYRUN" && {
 558         move_new_entry_back
 559         exit
 560 }
 561
 562 git add $OUTPUT &&
 563 git commit -s -m "Update $(make_date $now)" &&
 564 git push origin +$BRANCH