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.description "A few stories told by Dscho"
  36 # $ git config blog.background background.jpg
  37 # $ git config blog.branch blog
  38 #
  39 # Now you can start writing posts, by creating a file called "new", and
  40 # calling ./upload.sh to commit the post together with the images and
  41 # push all.
  42 #
  43 # Note that no file names may contain spaces.
  44
  45 # TODO: document the "syntax" of the source-*.txt files
  46
  47
  48 # make sure we're in the correct working directory
  49 cd "$(dirname "$0")"
  50
  51 GITWEBURL="$(git config gitweb.url)"
  52 test -z "$GITWEBURL" && {
  53         echo "Please set gitweb.url in the Git config first!" >&2
  54         exit 1
  55 }
  56
  57 get_config () {
  58         value=$(git config blog.$1)
  59         test -z "$value" && value=$2
  60         echo $value
  61 }
  62
  63 BACKGROUNDIMG=$(get_config background paper.jpg)
  64 TITLE=$(get_config title "Dscho's Git log")
  65 MAXENTRIES=$(get_config maxPostsPerPage 10)
  66 BRANCH=$(get_config branch blog)
  67 DESC=$(get_config description "A few stories told by Dscho")
  68
  69 URLPREFIX="$(dirname "$GITWEBURL")"/
  70 REMOTEREPOSITORY="$(basename "$GITWEBURL")"
  71 case "$GITWEBURL" in
  72 *'?'*) BLOBPLAIN="$REMOTEREPOSITORY;a=blob_plain";;
  73 */) URLPREFIX=$GITWEBURL; BLOBPLAIN="a=blob_plain";;
  74 *) BLOBPLAIN="$REMOTEREPOSITORY?a=blob_plain";;
  75 esac
  76 URL="$BLOBPLAIN;hb=$BRANCH;f="
  77 ORIGURL=$URL
  78 NEW=new
  79 OUTPUT=index.html
  80 RSS=blog.rss
  81 TEST=test.html
  82 THIS=$0
  83
  84 LC_ALL=C
  85 export LC_ALL
  86
  87 move_new_entry_back () {
  88         test -f source-$now.txt &&
  89         mv source-$now.txt $NEW &&
  90         git rm --cached -f source-$now.txt
  91 }
  92
  93 die () {
  94         move_new_entry_back
  95         echo "$*" >&2
  96         exit 1
  97 }
  98
  99 strip_prefix () {
 100         echo "${1#$2}"
 101 }
 102
 103 chomp () {
 104         strip_prefix "${1%$3}" "$2"
 105 }
 106
 107 nth () {
 108         # add illogical suffix
 109         case "$1" in
 110         *1?|*[04-9]) echo "$1th";;
 111         *1) echo "$1st";;
 112         *2) echo "$1nd";;
 113         *3) echo "$1rd";;
 114         esac
 115 }
 116
 117 make_chinese_hour () {
 118         case $1 in
 119         23|00) echo Rat;;
 120         01|02) echo Buffalo;;
 121         03|04) echo Tiger;;
 122         05|06) echo Rabbit;;
 123         07|08) echo Dragon;;
 124         09|10) echo Snake;;
 125         11|12) echo Horse;;
 126         13|14) echo Goat;;
 127         15|16) echo Monkey;;
 128         17|18) echo Rooster;;
 129         19|20) echo Dog;;
 130         21|22) echo Pig;;
 131         esac
 132 }
 133
 134 digit_to_roman () {
 135         case $1 in
 136         1) echo $2;;
 137         2) echo $2$2;;
 138         3) echo $2$2$2;;
 139         4) echo $2$3;;
 140         5) echo $3;;
 141         6) echo $3$2;;
 142         7) echo $3$2$2;;
 143         8) echo $3$2$2$2;;
 144         9) echo $2$4;;
 145         esac
 146 }
 147
 148 make_roman_number () {
 149         case $1 in
 150         '') ;;
 151         ?) digit_to_roman $1 I V X;;
 152         ??) echo $(digit_to_roman ${1%?} X L C)$(make_roman_number ${1#?});;
 153         ???) echo $(digit_to_roman ${1%??} C D M)$(make_roman_number ${1#?});;
 154         ????) echo $(digit_to_roman ${1%???} M)$(make_roman_number ${1#?});;
 155         esac
 156 }
 157
 158 make_date () {
 159         printf "%s, %s of %s, Anno Domini %s, at the hour of the %s\n" \
 160                 $(date +%A -d @$1) \
 161                 $(nth $(date +%e -d @$1)) \
 162                 $(date +%B -d @$1) \
 163                 $(make_roman_number $(date +%Y -d @$1)) \
 164                 $(make_chinese_hour $(date +%H -d @$1))
 165 }
 166
 167 # make an argument for sed, to replace $1..$1 by <$2>..</$2>
 168 markup_substitution () {
 169         # sed does not know minimal match with .*, only greedy one
 170         middle=
 171         middleend=
 172         delim=$1
 173         right_no=3
 174         while test ! -z "$delim"
 175         do
 176                 right_no=$(($right_no+1))
 177                 test $right_no -gt 9 &&
 178                 die "Invalid markup pattern: $1"
 179                 first=$(expr "$delim" : '\(.\)')
 180                 delim=${delim#$first}
 181                 test -z "$delim" && {
 182                         middle="$middle\\([^$first]\\|$first[A-Za-z0-9]\\)"
 183                         break
 184                 }
 185                 middle="$middle\\([^$first]\\|$first"
 186                 middleend="$middleend\\)"
 187         done
 188
 189         left="\\(^\\|[^A-Za-z0-9]\\)$1\\($middle$middleend*\\)"
 190         right="$1\\($\\|[^A-Za-z0-9]\\)"
 191         # work around stupid dash interpreting backslashes when expanding vars
 192         bs="\\\\"
 193         test "\\" = "$(echo "$bs")" || bs="\\"
 194         echo "s/$left$right/${bs}1<$2>${bs}2<\/$2>$bs$right_no/g"
 195 }
 196
 197 space80='                          '
 198 space80="$space80$space80$space80  "
 199 # transform markup in stdin to HTML
 200 markup () {
 201         case "$*" in
 202         *invert-bash*) bash_bg=white; bash_fg=black;;
 203         *) bash_bg=black; bash_fg=white;;
 204         esac
 205         sed -e 's!^$!</p><p>!' \
 206                 -e "$(markup_substitution "''" i)" \
 207                 -e "$(markup_substitution "_" u)" \
 208                 -e 's!IMHO!in my humble opinion!g' \
 209                 -e 's!BTW!By the way,!g' \
 210                 -e 's!repo.or.cz!<a href=http://&>&</a>!g' \
 211                 -e 's!:-)!\&#x263a;!g' \
 212                 -e "s!\\[\\[\(Image\|SVG\):.*!$THIS handle &!e" \
 213                 -e 's!<bash>!<table\
 214                                 border=1 bgcolor='$bash_bg'>\
 215                         <tr><td bgcolor=lightblue colspan=3>\
 216                                 <pre>'"$space80"'</pre>\
 217                         </td></tr>\
 218                         <tr><td>\
 219                                 <table cellspacing=5 border=0\
 220                                          style="color:'$bash_fg';">\
 221                                 <tr><td>\
 222                                         <pre>!' \
 223                 -e 's!</bash>!</pre>\
 224                                 </td></tr>\
 225                                 </table>\
 226                         </td></tr>\
 227                         </table>!' \
 228
 229 }
 230
 231 # output lines containing <timestamp> <filename> <title>
 232 get_blog_entries () {
 233         for file in $(ls -r source-*.txt)
 234         do
 235                 timestamp=$(chomp $file source- .txt)
 236                 title="$(sed 1q < $file | markup)"
 237                 echo "$timestamp $file $title"
 238         done
 239 }
 240
 241 get_last_removed_entry () {
 242         git log --pretty=format: --name-only --diff-filter=D HEAD |
 243         while read line
 244         do
 245                 case "$line" in
 246                 source-*.txt) file=$line;;
 247                 '') test -z "$file" || {
 248                         echo "$file"
 249                         break
 250                 };;
 251                 esac
 252         done
 253 }
 254
 255 box_count=0
 256 begin_box () {
 257         test $box_count = 0 || echo "<br>"
 258         echo "<table width=$toc_width bgcolor=#e0e0e0 border=1>"
 259         echo "<tr><th>$1</th></tr>"
 260         echo "<tr><td>"
 261 }
 262
 263 end_box () {
 264         echo "</td></tr></table>"
 265         box_count=$(($box_count+1))
 266 }
 267
 268 # make HTML page
 269 make_html () {
 270         body_style="width:800px"
 271         body_style="$body_style;background-image:url($URL$BACKGROUNDIMG)"
 272         body_style="$body_style;background-repeat:repeat-y"
 273         body_style="$body_style;background-attachment:scroll"
 274         body_style="$body_style;padding:0px;"
 275         text_style="width:610px"
 276         text_style="$text_style;margin-left:120px"
 277         text_style="$text_style;margin-top:50px"
 278         text_style="$text_style;align:left"
 279         text_style="$text_style;vertical-align:top;"
 280         cat << EOF
 281 <html>
 282         <head>
 283                 <title>$TITLE</title>
 284                 <meta http-equiv="Content-Type"
 285                         content="text/html; charset=UTF-8"/>
 286         </head>
 287         <body style="$body_style">
 288                 <div style="$text_style">
 289                         <h1>$TITLE</h1>
 290 EOF
 291         indent='\t\t\t'
 292
 293         # make toc
 294         toc_width=400px
 295         toc_style="position:absolute;top:50px;left:810px;width=$toc_width"
 296         {
 297                 echo "<div style=\"$toc_style\">"
 298                 begin_box "Table of contents:"
 299                 echo '<p><ul>'
 300                 get_blog_entries |
 301                 while read timestamp filename title
 302                 do
 303                         date="$(date +"%d %b %Y" -d @$timestamp)"
 304                         echo "<li><a href=#$timestamp>$date $title</a>"
 305                 done
 306                 echo '</ul></p>'
 307                 file=
 308                 last_removed_entry=$(get_last_removed_entry)
 309                 test -z "$last_removed_entry" || {
 310                         commit=$(git log --pretty=format:%H --diff-filter=AM \
 311                                         -- $last_removed_entry |
 312                                 head -n 1)
 313                         previous="$BLOBPLAIN;hb=$commit"
 314                         echo "<a href=$previous;f=index.html>Older posts</a>"
 315                 }
 316                 end_box
 317
 318                 # RSS feed
 319                 rss_style="background-color:orange;text-decoration:none"
 320                 rss_style="$rss_style;color:white;font-family:sans-serif;"
 321                 echo '<br>'
 322                 echo '<div style="text-align:right;">'
 323                 echo "<a href=\"$ORIGURL$RSS\""
 324                 echo '   title="Subscribe to my RSS feed"'
 325                 echo '   class="rss" rel="nofollow"'
 326                 echo "   style=\"$rss_style\">RSS</a>"
 327                 echo '</div>'
 328
 329                 # About
 330                 test -f about.html && {
 331                         begin_box "About this blog:"
 332                         cat about.html
 333                         end_box
 334                 }
 335
 336                 # Links
 337                 test -f links.html && {
 338                         begin_box "Links:"
 339                         cat links.html
 340                         end_box
 341                 }
 342
 343                 # Google AdSense
 344                 test -z "$DRYRUN" && test -f google.adsense && {
 345                         begin_box "Google Ads:"
 346                         cat google.adsense
 347                         end_box
 348                 }
 349
 350                 echo '</div>'
 351         } | sed -s "s/^/$indent/"
 352
 353
 354         # timestamps will not need padding to sort correctly, for some time...
 355         get_blog_entries |
 356         while read timestamp filename title
 357         do
 358                 echo "<h6>$(make_date $timestamp)</h6>"
 359                 echo "<a name=$timestamp>"
 360                 echo "<h2>$title</h2>"
 361                 echo ""
 362                 echo "<p>"
 363                 sed 1d < $filename | markup
 364                 echo "</p>"
 365         done |
 366         sed -e "s/^./$indent&/" \
 367                 -e "/<pre>/,/<\/pre>/s/^$indent//"
 368
 369         cat << EOF
 370                 </div>
 371         </body>
 372 </html>
 373 EOF
 374 }
 375
 376 generate_rss () {
 377         echo '<?xml version="1.0" encoding="utf-8"?>'
 378         echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'
 379         echo '<channel>'
 380         echo "<title>$TITLE</title>"
 381         echo "<link>$URLPREFIX${URL}index.html</link>"
 382         self="$URLPREFIX$ORIGURL$RSS"
 383         selfattribs='rel="self" type="application/rss+xml"'
 384         echo "<atom:link href=\"$self\" $selfattribs/>"
 385         echo "<description>$DESC</description>"
 386         echo "<lastBuildDate>$(date --rfc-2822)</lastBuildDate>"
 387         echo '<language>en-us</language>'
 388
 389         get_blog_entries |
 390         while read timestamp filename title
 391         do
 392                 # remove all tags
 393                 title=$(echo "$title" | sed 's/<[^>]*>//g')
 394                 echo '<item>'
 395                 echo "<title>$title</title>"
 396                 echo "<link>$URLPREFIX${URL}index.html#$timestamp</link>"
 397                 echo "<guid>$URLPREFIX${URL}index.html#$timestamp</guid>"
 398                 echo "<pubDate>$(date --rfc-2822 -d @$timestamp)</pubDate>"
 399                 description="$(cat < $filename | markup invert-bash)"
 400                 echo "<description><![CDATA[$description]]></description>"
 401                 echo "</item>"
 402         done
 403
 404         echo '</channel>'
 405         echo '</rss>'
 406 }
 407
 408 get_image_files () {
 409         git ls-files |
 410         grep -v '\.\(rss\|html\|gitignore\|in\|sh\|txt\|adsense\)$'
 411 }
 412
 413 remove_old_entries () {
 414         count=$(ls source-*.txt | wc -l)
 415         test $MAXENTRIES -ge $count && return
 416
 417         for file in source-*.txt
 418         do
 419                 test $MAXENTRIES -lt $count || break
 420                 git rm $file > /dev/null || return 1
 421                 count=$(($count-1))
 422                 echo $file
 423         done
 424
 425         # remove no-longer referenced images
 426         image_files=$(get_image_files)
 427         referenced_files="$(cat source-*.txt |
 428                 tr ']|' '\n' |
 429                 sed -ne 's/\[\[\(Image\|SVG\)://p') $BACKGROUNDIMG"
 430         for file in $(echo $image_files $referenced_files $referenced_files |
 431                         tr ' ' '\n' | sort | uniq -u)
 432         do
 433                 git rm $file > /dev/null || return 1
 434                 echo $file
 435         done
 436 }
 437
 438 # never, ever have spaces in the file names
 439 commit_new_images () {
 440         files="$(remove_old_entries) $RSS $BACKGROUNDIMG" ||
 441         die "Could not remove old entries"
 442
 443
 444         generate_rss > $RSS &&
 445         git add $RSS ||
 446         die "Could not generate $RSS"
 447
 448         for image in $(cat source-* |
 449                 tr ' ]|' '\n' |
 450                 sed -n 's/.*\[\[\(Image\|SVG\)://p' |
 451                 sort |
 452                 uniq)
 453         do
 454                 git add $image || die "Could not git add image $image"
 455                 files="$files $image"
 456         done
 457
 458         git update-index --refresh &&
 459         git diff-files --quiet -- $files &&
 460         git diff --cached --quiet HEAD -- $files ||
 461         git commit -s -m "Housekeeping on $(make_date $now)" $files
 462 }
 463
 464 get_image_url () {
 465         test ! -z "$DRYRUN" && echo "$1" && return
 466         rev=$(git rev-list -1 HEAD -- $1)
 467         test -z "$rev" && die "No revision found for $1"
 468         echo "$BLOBPLAIN;hb=$rev;f=$1"
 469 }
 470
 471 handle_svg_file () {
 472         # for some reason, Firefox adds scrollbars, so nudge the width a bit
 473         width=$(sed -ne 's/.* width="\([^"]*\).*/\1/p' -e '/<metadata/q' < "$1")
 474         test -z "$width" || width=" width=$(($width+5))"
 475         url=$(get_image_url "$1")
 476         cat << EOF
 477 <center>
 478         <table border=0>
 479                 <tr>
 480                         <td align=center>
 481                                 <embed type="image/svg+xml"
 482                                         src="$url"$width />
 483                         </td>
 484                 </tr>
 485                 <tr>
 486                         <td align=center>
 487                                 <a href=$url>$1</a>
 488                         </td>
 489                 </tr>
 490         </table>
 491 </center>
 492 EOF
 493 }
 494
 495 handle_image_file () {
 496         echo "<center><img src=$(get_image_url "${1%% *}") ${1#* }></center>"
 497 }
 498
 499
 500
 501 # parse command line option
 502 case "$1" in
 503 *dry*) DRYRUN=1; export DRYRUN; shift;;
 504 *show*) firefox "$(pwd)"/$TEST; exit;;
 505 *remote*) firefox $URLPREFIX$URL$OUTPUT; exit;;
 506 handle)
 507         shift
 508         case "$1" in
 509         "[[SVG:"*) handle_svg_file "$(chomp "$*" '\[\[SVG:' '\]\]')";;
 510         "[[Image:"*) handle_image_file "$(chomp "$*" '\[\[Image:' '\]\]')";;
 511         esac
 512         exit
 513 ;;
 514 '') ;;
 515 *) die "Unknown command: $1";;
 516 esac
 517
 518 test "$#" = 0 ||
 519 die "Usage: $0 [--dry-run]"
 520
 521 # make sure we're on the correct branch
 522 test refs/heads/$BRANCH = $(git symbolic-ref HEAD) ||
 523 die "Not on branch $BRANCH"
 524
 525 # make sure there are no uncommitted changes
 526 git update-index --refresh &&
 527 git diff-files --quiet ||
 528 die "Have unstaged changes!"
 529
 530 # rename the new blog entry if it exists
 531 now=$(date +%s)
 532 test ! -f $NEW || {
 533         mv -i $NEW source-$now.txt &&
 534         git add source-$now.txt
 535 } ||
 536 die "Could not rename source.txt"
 537
 538 # commit the images that are referenced and not yet committed
 539 test ! -z "$DRYRUN" ||
 540 commit_new_images ||
 541 die "Could not commit new images"
 542
 543 # to find the images reliably, we have to use the commit name, not the branch
 544 # we use the latest commit touching an image file.
 545 IMAGEFILES="$(get_image_files)"
 546 REV=$(git rev-list -1 HEAD -- $IMAGEFILES)
 547 test -z "$REV" && REV=$BRANCH
 548 URL="$BLOBPLAIN;hb=$REV;f="
 549
 550 if test ! -z "$DRYRUN"
 551 then
 552         # Output to test.html and have local links into the current directory
 553         OUTPUT=$TEST
 554         URL=
 555 fi
 556
 557 make_html > $OUTPUT || die "Could not write $OUTPUT"
 558
 559 test ! -z "$DRYRUN" && {
 560         move_new_entry_back
 561         exit
 562 }
 563
 564 git add $OUTPUT &&
 565 git commit -s -m "Update $(make_date $now)" &&
 566 git push origin +$BRANCH