pldnotify.awk

   1 #!/bin/awk -f
   2 # $Revision$, $Date$
   3 #
   4 # Copyright (C) 2000-2010 PLD-Team <feedback@pld-linux.org>
   5 # Authors:
   6 #       Sebastian Zagrodzki <zagrodzki@pld-linux.org>
   7 #       Jacek Konieczny <jajcus@pld-linux.org>
   8 #       Andrzej Krzysztofowicz <ankry@pld-linux.org>
   9 #       Jakub Bogusz <qboosh@pld-linux.org>
  10 #       Elan Ruusamäe <glen@pld-linux.org>
  11 #
  12 # See cvs log pldnotify.awk for list of contributors
  13 #
  14 # TODO:
  15 # - "SourceXDownload" support (use given URLs if present instead of cut-down SourceX URLs)
  16 # - "SourceXActiveFTP" support
  17
  18 function fixedsub(s1,s2,t,      ind) {
  19 # substitutes fixed strings (not regexps)
  20         if (ind = index(t,s1)) {
  21                 t = substr(t, 1, ind-1) s2 substr(t, ind+length(s1))
  22         }
  23         return t
  24 }
  25
  26 function ispre(s) {
  27         if ((s~"pre")||(s~"PRE")||(s~"beta")||(s~"BETA")||(s~"alpha")||(s~"ALPHA")||(s~"rc")||(s~"RC")) {
  28                 if (DEBUG) print "pre-version"
  29                 return 1
  30         } else {
  31                 return 0
  32         }
  33 }
  34
  35 function compare_ver(v1,v2) {
  36 # compares version numbers
  37         while (match(v1,/[a-zA-Z][0-9]|[0-9][a-zA-Z]/))
  38                 v1=(substr(v1,1,RSTART) "." substr(v1,RSTART+RLENGTH-1))
  39         while (match(v2,/[a-zA-Z][0-9]|[0-9][a-zA-Z]/))
  40                 v2=(substr(v2,1,RSTART) "." substr(v2,RSTART+RLENGTH-1))
  41         sub("^0*","",v1)
  42         sub("^0*","",v2)
  43         gsub("\.0*",".",v1)
  44         gsub("\.0*",".",v2)
  45         if (DEBUG) print "v1 == " v1
  46         if (DEBUG) print "v2 == " v2
  47         count=split(v1,v1a,"\.")
  48         count2=split(v2,v2a,"\.")
  49
  50         if (count<count2) mincount=count
  51         else mincount=count2
  52
  53         for (i=1; i<=mincount; i++) {
  54                 if (v1a[i]=="") v1a[i]=0
  55                 if (v2a[i]=="") v2a[i]=0
  56                 if (DEBUG) print "i == " i
  57                 if (DEBUG) print "v1[i] == " v1a[i]
  58                 if (DEBUG) print "v2[i] == " v2a[i]
  59                 if ((v1a[i]~/[0-9]/)&&(v2a[i]~/[0-9]/)) {
  60                         if (length(v2a[i])>length(v1a[i]))
  61                                 return 1
  62                         else if (v2a[i]>v1a[i])
  63                                 return 1
  64                         else if (length(v1a[i])>length(v2a[i]))
  65                                 return 0
  66                         else if (v1a[i]>v2a[i])
  67                                 return 0
  68                 } else if ((v1a[i]~/[A-Za-z]/)&&(v2a[i]~/[A-Za-z]/)) {
  69                         if (v2a[i]>v1a[i])
  70                                 return 1
  71                         else if (v1a[i]>v2a[i])
  72                                 return 0
  73                 } else if (ispre(v1a[i]) == 1)
  74                         return 1
  75                 else
  76                         return 0
  77         }
  78         if ((count2==mincount)&&(count!=count2)) {
  79                 for (i=count2+1; i<=count; i++)
  80                         if (ispre(v1a[i]) == 1)
  81                                 return 1
  82                 return 0
  83         } else if (count!=count2) {
  84                 for (i=count+1; i<=count2; i++)
  85                         if (ispre(v2a[i]) == 1)
  86                                 return 0
  87                 return 1
  88         }
  89         return 0
  90 }
  91
  92 function compare_ver_dec(v1,v2) {
  93 # compares version numbers as decimal floats
  94         while (match(v1,/[0-9][a-zA-Z]/))
  95                 v1=(substr(v1,1,RSTART) "." substr(v1,RSTART+RLENGTH-1))
  96         while (match(v2,/[0-9][a-zA-Z]/))
  97                 v2=(substr(v2,1,RSTART) "." substr(v2,RSTART+RLENGTH-1))
  98         sub("^0*","",v1)
  99         sub("^0*","",v2)
 100         if (DEBUG) print "v1 == " v1
 101         if (DEBUG) print "v2 == " v2
 102         count=split(v1,v1a,"\.")
 103         count2=split(v2,v2a,"\.")
 104
 105         if (count<count2) mincount=count
 106         else mincount=count2
 107
 108         for (i=1; i<=mincount; i++) {
 109                 if (v1a[i]=="") v1a[i]=0
 110                 if (v2a[i]=="") v2a[i]=0
 111                 if (DEBUG) print "i == " i
 112                 if (DEBUG) print "v1[i] == " v1a[i]
 113                 if (DEBUG) print "v2[i] == " v2a[i]
 114                 if ((v1a[i]~/[0-9]/)&&(v2a[i]~/[0-9]/)) {
 115                         if (i==2) {
 116                                 if (0+("." v2a[i])>0+("." v1a[i]))
 117                                         return 1
 118                                 else if (0+("." v1a[i])>0+("." v2a[i]))
 119                                         return 0
 120                         } else {
 121                                 if (length(v2a[i])>length(v1a[i]))
 122                                         return 1
 123                                 else if (v2a[i]>v1a[i])
 124                                         return 1
 125                                 else if (length(v1a[i])>length(v2a[i]))
 126                                         return 0
 127                                 else if (v1a[i]>v2a[i])
 128                                         return 0
 129                         }
 130                 } else if ((v1a[i]~/[A-Za-z]/)&&(v2a[i]~/[A-Za-z]/)) {
 131                         if (v2a[i]>v1a[i])
 132                                 return 1
 133                         else if (v1a[i]>v2a[i])
 134                                 return 0
 135                 } else if (ispre(v1a[i]) == 1)
 136                         return 1
 137                 else
 138                         return 0
 139         }
 140         if ((count2==mincount)&&(count!=count2)) {
 141                 for (i=count2+1; i<=count; i++)
 142                         if (ispre(v1a[i]) == 1)
 143                                 return 1
 144                 return 0
 145         } else if (count!=count2) {
 146                 for (i=count+1; i<=count2; i++)
 147                         if (ispre(v2a[i]) == 1)
 148                                 return 0
 149                 return 1
 150         }
 151         return 0
 152 }
 153
 154 function link_seen(link) {
 155         for (seenlink in frameseen) {
 156                 if (seenlink == link) {
 157                         if (DEBUG) print "Link: [" link "] seen already, skipping..."
 158                         return 1
 159                 }
 160         }
 161         frameseen[link]=1
 162         return 0
 163 }
 164
 165 function mktemp(   _cmd, _tmpfile) {
 166         _cmd = "mktemp /tmp/XXXXXX"
 167         _cmd | getline _tmpfile
 168         close(_cmd)
 169         return _tmpfile
 170 }
 171
 172 # get all <A HREF=..> tags from specified URL
 173 function get_links(url,filename,   errno,link,oneline,retval,odp,wholeodp,lowerodp,tmpfile,cmd) {
 174
 175         wholeerr=""
 176
 177         tmpfile = mktemp()
 178         tmpfileerr = mktemp()
 179
 180         if (url ~ /^http:\/\/(download|dl).(sf|sourceforge).net\//) {
 181                 gsub("^http://(download|dl).(sf|sourceforge).net/", "", url)
 182                 gsub("/.*", "", url)
 183                 url = "http://sourceforge.net/projects/" url "/files/"
 184                 if (DEBUG) print "sf url, mungled url to: " url
 185         }
 186
 187         if (url ~ /^http:\/\/(.*)\.googlecode\.com\/files\//) {
 188                 gsub("^http://", "", url)
 189                 gsub("\..*", "", url)
 190                 url = "http://code.google.com/p/" url "/downloads/list"
 191                 if (DEBUG) print "googlecode url, mungled url to: " url
 192         }
 193
 194         if (url ~ /^http:\/\/pecl.php.net\/get\//) {
 195                 gsub("-.*", "", filename)
 196                 url = "http://pecl.php.net/package/" filename
 197                 if (DEBUG) print "pecl.php.net url, mungled url to: " url
 198         }
 199
 200         if (url ~ /^(http|ftp):\/\/mysql.*\/Downloads\/MySQL-5.1\//) {
 201                 url = "http://dev.mysql.com/downloads/mysql/5.1.html#source"
 202                  if (DEBUG) print "mysql 5.1 url, mungled url to: " url
 203         }
 204
 205         if (url ~/^(http|https):\/\/launchpad\.net\/(.*)\//) {
 206                 gsub("^(http|https):\/\/launchpad\.net\/", "", url)
 207                 gsub("\/.*/", "", url)
 208                 url = "https://code.launchpad.net/" url "/+download"
 209                 if (DEBUG) print "main launchpad url, mungled url to: " url
 210         }
 211
 212         if (url ~/^(http|https):\/\/edge\.launchpad\.net\/(.*)\//) {
 213                 gsub("^(http|https):\/\/edge\.launchpad\.net\/", "", url)
 214                 gsub("\/.*/", "", url)
 215                 url = "https://edge.launchpad.net/" url "/+download"
 216                 if (DEBUG) print "edge launchpad url, mungled url to: " url
 217         }
 218
 219         if (url ~/^(http|https):\/\/github.com\/.*\/(.*)\/tarball\//) {
 220                 gsub("\/tarball\/.*", "/downloads", url)
 221                 if (DEBUG) print "github tarball url, mungled url to: " url
 222         }
 223
 224
 225         if (DEBUG) print "Retrieving: " url
 226         cmd = "wget --user-agent \"Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2) Gecko/20100129 PLD/3.0 (Th) Iceweasel/3.6\" -nv -O - \"" url "\" -t 2 -T 45 --passive-ftp --no-check-certificate > " tmpfile " 2> " tmpfileerr
 227         if (DEBUG) print "Execute: " cmd
 228         errno = system(cmd)
 229         if (DEBUG) print "Execute done"
 230
 231         if (errno==0) {
 232                 wholeodp = ""
 233                 if ( DEBUG ) print "Reading succeess response..."
 234                 while (getline oneline < tmpfile)
 235                         wholeodp=(wholeodp " " oneline)
 236                 # if ( DEBUG ) print "Response: " wholeodp
 237         } else {
 238                 if ( DEBUG ) print "Reading failure response..."
 239                 wholeerr = ""
 240                 while (getline oneline < tmpfileerr)
 241                         wholeerr=(wholeerr " " oneline)
 242                 if ( DEBUG ) print "Error Response: " wholeerr
 243         }
 244
 245         system("rm -f " tmpfile)
 246         system("rm -f " tmpfileerr)
 247
 248         urldir=url;
 249         sub(/[^\/]+$/,"",urldir)
 250
 251         if ( errno==0) {
 252                 while (match(wholeodp, /<([aA]|[fF][rR][aA][mM][eE])[ \t][^>]*>/) > 0) {
 253                         if (DEBUG) print "Processing links..."
 254                         odp=substr(wholeodp,RSTART,RLENGTH);
 255                         wholeodp=substr(wholeodp,RSTART+RLENGTH);
 256
 257                         lowerodp=tolower(odp);
 258                         if (lowerodp ~ /<frame[ \t]/) {
 259                                 sub(/[sS][rR][cC]=[ \t]*/,"src=",odp);
 260                                 match(odp,/src="[^"]+"/)
 261                                 newurl=substr(odp,RSTART+5,RLENGTH-6)
 262                                 if (DEBUG) print "Frame: " newurl
 263                                 if (newurl !~ /\//) {
 264                                         newurl=(urldir newurl)
 265                                         if (DEBUG) print "Frame->: " newurl
 266                                 }
 267
 268                                 if (link_seen(newurl)) {
 269                                         newurl=""
 270                                         continue
 271                                 }
 272
 273                                 retval=(retval " " get_links(newurl))
 274                         } else if (lowerodp ~ /href=[ \t]*"[^"]*"/) {
 275                                 sub(/[hH][rR][eE][fF]=[ \t]*"/,"href=\"",odp)
 276                                 match(odp,/href="[^"]*"/)
 277                                 link=substr(odp,RSTART,RLENGTH)
 278                                 odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
 279                                 link=substr(link,7,length(link)-7)
 280
 281                                 if (link_seen(link)) {
 282                                         link=""
 283                                         continue
 284                                 }
 285
 286                                 retval=(retval " " link)
 287                                 if (DEBUG) print "href(\"\"): " link
 288                         } else if (lowerodp ~ /href=[ \t]*'[^']*'/) {
 289                                 sub(/[hH][rR][eE][fF]=[ \t]*'/,"href='",odp)
 290                                 match(odp,/href='[^']*'/)
 291                                 link=substr(odp,RSTART,RLENGTH)
 292                                 odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
 293                                 link=substr(link,7,length(link)-7)
 294
 295                                 if (link_seen(link)) {
 296                                         link=""
 297                                         continue
 298                                 }
 299
 300                                 retval=(retval " " link)
 301                                 if (DEBUG) print "href(''): " link
 302                         } else if (lowerodp ~ /href=[ \t]*[^ \t>]*/) {
 303                                 sub(/[hH][rR][eE][fF]=[ \t]*/,"href=",odp)
 304                                 match(odp,/href=[^ \t>]*/)
 305                                 link=substr(odp,RSTART,RLENGTH)
 306                                 odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
 307                                 link=substr(link,6,length(link)-5)
 308
 309                                 if (link_seen(link)) {
 310                                         link=""
 311                                         continue
 312                                 }
 313
 314                                 retval=(retval " " link)
 315                                 if (DEBUG) print "href(): " link
 316                         } else {
 317                                 # <a ...> but not href - skip
 318                                 if (DEBUG) print "skipping <a > without href: " odp
 319                         }
 320                 }
 321         } else {
 322                 retval=("WGET ERROR: " errno ": " wholeerr)
 323         }
 324
 325
 326         if (DEBUG) print "Returning: " retval
 327         return retval
 328 }
 329
 330 function subst_defines(var,defs) {
 331 # substitute all possible RPM macros
 332         while ((var ~ /%{.*}/) || (var ~ /%[A-Za-z0-9_]+/)) {
 333                 oldvar=var
 334                 for (j in defs) {
 335                         gsub("%{" j "}", defs[j], var)
 336                         gsub("%" j , defs[j], var)
 337                 }
 338                 if (var==oldvar) {
 339                         if ( DEBUG ) for (i in defs) print i " == " defs[i]
 340                         return var
 341                 }
 342         }
 343         return var
 344 }
 345
 346 function find_mirror(url) {
 347
 348         while (succ = (getline line < "mirrors")) {
 349             if (succ==-1) { return url }
 350                 nf=split(line,fields,"|")
 351                 if (nf>1){
 352                         origin=fields[1]
 353                         mirror=fields[2]
 354                         mname=fields[3]
 355                         prefix=substr(url,1,length(origin))
 356                         if (prefix==origin){
 357                                 if ( DEBUG ) print "Mirror fount at " mname
 358                                 close("mirrors")
 359                                 return mirror substr(url,length(origin)+1)
 360                         }
 361                 }
 362         }
 363
 364         return url
 365 }
 366
 367 function process_source(number,lurl,name,version) {
 368 # fetches file list, and compares version numbers
 369         if ( DEBUG ) print "Processing " lurl
 370
 371         if ( index(lurl,version)==0 ) {
 372                 if (DEBUG) print "There is no version number."
 373                 return 0
 374         }
 375
 376         sub("://",":",lurl)
 377         sub("/",":/",lurl)
 378         gsub("[^/]*$",":&",lurl)
 379         split(lurl,url,":")
 380         acc=url[1]
 381         host=url[2]
 382         dir=url[3]
 383         filename=url[4]
 384
 385         if (index(dir,version)) {
 386                 # directory name as version maching mode:
 387                 # if /something/version/name-version.tarball then check
 388                 # in /something/ looking for newer directory
 389                 dir=substr(dir,1,index(dir,version)-1)
 390                 sub("[^/]*$","",dir)
 391                 sub("(\.tar\.(bz|bz2|gz)|zip)$","",filename)
 392         }
 393
 394         if ( DEBUG ) print "Will check a directory: " dir
 395         if ( DEBUG ) print "and a file: " filename
 396
 397         filenameexp=filename
 398         gsub("\+","\\+",filenameexp)
 399         sub(version,"[A-Za-z0-9.]+",filenameexp)
 400         gsub("\.","\\.",filenameexp)
 401         if ( DEBUG ) print "Expression: " filenameexp
 402         match(filename,version)
 403         prever=substr(filename,1,RSTART-1)
 404         postver=substr(filename,RSTART+RLENGTH)
 405         if ( DEBUG ) print "Before number: " prever
 406         if ( DEBUG ) print "and after: " postver
 407         newurl=find_mirror(acc "://" host dir)
 408         #print acc "://" host dir
 409         #newurl=url[1]"://"url[2]url[3]url[4]
 410         #newurl=acc "://" host dir filename
 411         if ( DEBUG ) print "Looking at " newurl
 412
 413         references=0
 414         finished=0
 415         oldversion=version
 416         odp=get_links(newurl,filename)
 417         if( odp ~ "ERROR: ") {
 418                 print name "(" number ") " odp
 419         } else {
 420                 if (DEBUG) print "WebPage downloaded"
 421                 c=split(odp,linki)
 422                 for (nr=1; nr<=c; nr++) {
 423                         addr=linki[nr]
 424                         if (DEBUG) print "Found link: " addr
 425                         if ((addr ~ filenameexp) && !(addr ~ "[-_.0-9A-Za-z~]" filenameexp)) {
 426                                 match(addr,filenameexp)
 427                                 newfilename=substr(addr,RSTART,RLENGTH)
 428                                 if (DEBUG) print "Hypothetical new: " newfilename
 429                                 newfilename=fixedsub(prever,"",newfilename)
 430                                 newfilename=fixedsub(postver,"",newfilename)
 431                                 if (DEBUG) print "Version: " newfilename
 432                                 if (newfilename ~ /\.(asc|sig|pkg|bin|binary|built)$/) continue
 433                                 # strip ending (happens when in directiory name as version matching mode)
 434                                 sub("(\.tar\.(bz|bz2|gz)|zip)$","",newfilename)
 435                                 if (NUMERIC) {
 436                                         if ( compare_ver_dec(version, newfilename)==1 ) {
 437                                                 if (DEBUG) print "Yes, there is new one"
 438                                                 version=newfilename
 439                                                 finished=1
 440                                         }
 441                                 } else if ( compare_ver(version, newfilename)==1 ) {
 442                                         if (DEBUG) print "Yes, there is new one"
 443                                         version=newfilename
 444                                         finished=1
 445                                 }
 446                         }
 447                 }
 448                 if (finished==0)
 449                         print name "(" number ") seems ok: " oldversion
 450                 else
 451                         print name "(" number ") [OLD] " oldversion " [NEW] " version
 452         }
 453 }
 454
 455 # upgrade check for pear package using PEAR CLI
 456 function pear_upgrade(name, ver) {
 457         pname = name;
 458         sub(/^php-pear-/, "", pname);
 459
 460         pearcmd = "pear remote-info " pname " | awk '/^Latest/{print $NF}'"
 461         if (DEBUG) {
 462                 print "pearcmd: " pearcmd
 463         }
 464         pearcmd | getline nver
 465         close(pearcmd)
 466
 467         if (compare_ver(ver, nver)) {
 468                 print name " [OLD] " ver " [NEW] " nver
 469         } else {
 470                 print name " seems ok: " ver
 471         }
 472
 473         return
 474 }
 475
 476 function process_data(name,ver,rel,src) {
 477         if (name ~ /^php-pear-/) {
 478                 return pear_upgrade(name, ver);
 479         }
 480
 481 # this function checks if substitutions were valid, and if true:
 482 # processes each URL and tries to get current file list
 483         for (i in src) {
 484                 if ( src[i] ~ /%{nil}/ ) {
 485                         gsub(/\%\{nil\}/, "", src[i])
 486                 }
 487                 if ( src[i] !~ /%{.*}/ && src[i] !~ /%[A-Za-z0-9_]/ )  {
 488                         if ( DEBUG ) print "Source: " src[i]
 489                         process_source(i,src[i],name,ver)
 490                 } else {
 491                         print FNAME ":" i ": impossible substitution: " src[i]
 492                 }
 493         }
 494 }
 495
 496 BEGIN {
 497         # if U want to use DEBUG, run script with "-v DEBUG=1"
 498         # or uncomment the line below
 499         # DEBUG = 1
 500
 501         errno=system("wget --help > /dev/null 2>&1")
 502         if (errno) {
 503                 print "No wget installed!"
 504                 exit 1
 505         }
 506         if (ARGC>=3 && ARGV[2]=="-n") {
 507                 NUMERIC=1
 508                 for (i=3; i<ARGC; i++) ARGV[i-1]=ARGV[i]
 509                 ARGC=ARGC-1
 510         }
 511 }
 512
 513 FNR==1 {
 514         if ( ARGIND != 1 ) {
 515                 # clean frameseen for each ARG
 516                 for (i in frameseen) {
 517                         delete frameseen[i]
 518                 }
 519                 frameseen[0] = 1
 520
 521                 process_data(NAME,VER,REL,SRC)
 522                 NAME="" ; VER="" ; REL=""
 523                 for (i in DEFS) delete DEFS[i]
 524                 for (i in SRC) delete SRC[i]
 525         }
 526         FNAME=FILENAME
 527         DEFS["_alt_kernel"]=""
 528         DEFS["20"]="\\ "
 529 }
 530
 531 /^[Uu][Rr][Ll]:/&&(URL=="") { URL=subst_defines($2,DEFS) ; DEFS["url"]=URL }
 532 /^[Nn]ame:/&&(NAME=="") { NAME=subst_defines($2,DEFS) ; DEFS["name"]=NAME }
 533 /^[Vv]ersion:/&&(VER=="") { VER=subst_defines($2,DEFS) ; DEFS["version"]=VER }
 534 /^[Rr]elease:/&&(REL=="") { REL=subst_defines($2,DEFS) ; DEFS["release"]=REL }
 535 /^[Ss]ource[0-9]*:/ { if (/(ftp|http|https):\/\//) SRC[FNR]=subst_defines($2,DEFS) }
 536 /%define/ { DEFS[$2]=subst_defines($3,DEFS) }
 537
 538 END {
 539         process_data(NAME,VER,REL,SRC)
 540 }