[packages/rpm-build-tools.git] / pldnotify.awk

#!/bin/awk -f
# $Revision$, $Date$
#
# Copyright (C) 2000-2008 PLD-Team <feedback@pld-linux.org>
# Authors:
#	Sebastian Zagrodzki <zagrodzki@pld-linux.org>
#	Jacek Konieczny <jajcus@pld-linux.org>
#	Andrzej Krzysztofowicz <ankry@pld-linux.org>
#	Jakub Bogusz <qboosh@pld-linux.org>
#	Elan Ruusamäe <glen@pld-linux.org>
#
# See cvs log pldnotify.awk for list of contributors
#
# TODO:
# - "SourceXDownload" support (use given URLs if present instead of cut-down SourceX URLs)
# - "SourceXActiveFTP" support

function fixedsub(s1,s2,t,	ind) {
# substitutes fixed strings (not regexps)
	if (ind = index(t,s1)) {
		t = substr(t, 1, ind-1) s2 substr(t, ind+length(s1))
	}
	return t
}

function ispre(s) {
	if ((s~"pre")||(s~"PRE")||(s~"beta")||(s~"BETA")||(s~"alpha")||(s~"ALPHA")||(s~"rc")||(s~"RC")) {
		if (DEBUG) print "pre-version"
		return 1
	} else {
		return 0
	}
}

function compare_ver(v1,v2) {
# compares version numbers
	while (match(v1,/[a-zA-Z][0-9]|[0-9][a-zA-Z]/))
		v1=(substr(v1,1,RSTART) "." substr(v1,RSTART+RLENGTH-1))
	while (match(v2,/[a-zA-Z][0-9]|[0-9][a-zA-Z]/))
		v2=(substr(v2,1,RSTART) "." substr(v2,RSTART+RLENGTH-1))
	sub("^0*","",v1)
	sub("^0*","",v2)
	gsub("\.0*",".",v1)
	gsub("\.0*",".",v2)
	if (DEBUG) print "v1 == " v1
	if (DEBUG) print "v2 == " v2
	count=split(v1,v1a,"\.")
	count2=split(v2,v2a,"\.")

	if (count<count2) mincount=count
	else mincount=count2

	for (i=1; i<=mincount; i++) {
		if (v1a[i]=="") v1a[i]=0
		if (v2a[i]=="") v2a[i]=0
		if (DEBUG) print "i == " i
		if (DEBUG) print "v1[i] == " v1a[i]
		if (DEBUG) print "v2[i] == " v2a[i]
		if ((v1a[i]~/[0-9]/)&&(v2a[i]~/[0-9]/)) {
			if (length(v2a[i])>length(v1a[i]))
				return 1
			else if (v2a[i]>v1a[i])
				return 1
			else if (length(v1a[i])>length(v2a[i]))
				return 0
			else if (v1a[i]>v2a[i])
				return 0
		} else if ((v1a[i]~/[A-Za-z]/)&&(v2a[i]~/[A-Za-z]/)) {
			if (v2a[i]>v1a[i])
				return 1
			else if (v1a[i]>v2a[i])
				return 0
		} else if (ispre(v1a[i]) == 1)
			return 1
		else
			return 0
	}
	if ((count2==mincount)&&(count!=count2)) {
		for (i=count2+1; i<=count; i++)
			if (ispre(v1a[i]) == 1)
				return 1
		return 0
	} else if (count!=count2) {
		for (i=count+1; i<=count2; i++)
			if (ispre(v2a[i]) == 1)
				return 0
		return 1
	}
	return 0
}

function compare_ver_dec(v1,v2) {
# compares version numbers as decimal floats
	while (match(v1,/[0-9][a-zA-Z]/))
		v1=(substr(v1,1,RSTART) "." substr(v1,RSTART+RLENGTH-1))
	while (match(v2,/[0-9][a-zA-Z]/))
		v2=(substr(v2,1,RSTART) "." substr(v2,RSTART+RLENGTH-1))
	sub("^0*","",v1)
	sub("^0*","",v2)
	if (DEBUG) print "v1 == " v1
	if (DEBUG) print "v2 == " v2
	count=split(v1,v1a,"\.")
	count2=split(v2,v2a,"\.")

	if (count<count2) mincount=count
	else mincount=count2

	for (i=1; i<=mincount; i++) {
		if (v1a[i]=="") v1a[i]=0
		if (v2a[i]=="") v2a[i]=0
		if (DEBUG) print "i == " i
		if (DEBUG) print "v1[i] == " v1a[i]
		if (DEBUG) print "v2[i] == " v2a[i]
		if ((v1a[i]~/[0-9]/)&&(v2a[i]~/[0-9]/)) {
			if (i==2) {
				if (0+("." v2a[i])>0+("." v1a[i]))
					return 1
				else if (0+("." v1a[i])>0+("." v2a[i]))
					return 0
			} else {
				if (length(v2a[i])>length(v1a[i]))
					return 1
				else if (v2a[i]>v1a[i])
					return 1
				else if (length(v1a[i])>length(v2a[i]))
					return 0
				else if (v1a[i]>v2a[i])
					return 0
			}
		} else if ((v1a[i]~/[A-Za-z]/)&&(v2a[i]~/[A-Za-z]/)) {
			if (v2a[i]>v1a[i])
				return 1
			else if (v1a[i]>v2a[i])
				return 0
		} else if (ispre(v1a[i]) == 1)
			return 1
		else
			return 0
	}
	if ((count2==mincount)&&(count!=count2)) {
		for (i=count2+1; i<=count; i++)
			if (ispre(v1a[i]) == 1)
				return 1
		return 0
	} else if (count!=count2) {
		for (i=count+1; i<=count2; i++)
			if (ispre(v2a[i]) == 1)
				return 0
		return 1
	}
	return 0
}

function link_seen(link) {
	for (seenlink in frameseen) {
		if (seenlink == link) {
			if (DEBUG) print "Link: [" link "] seen already, skipping..."
			return 1
		}
	}
	frameseen[link]=1
	return 0
}

function get_links(url,filename,errno,link,oneline,retval,odp,wholeodp,lowerodp,tmpfile) {
# get all <A HREF=..> tags from specified URL
	"mktemp /tmp/XXXXXX" | getline tmpfile
	close(tmpfile)

	if (url ~ /^http:\/\/(download|dl).(sf|sourceforge).net\//) {
		gsub("^http://(download|dl).(sf|sourceforge).net/", "", url)
		url = "http://prdownloads.sourceforge.net/" substr(url, 1, 1) "/" substr(url, 1, 2) "/" url
		if (DEBUG) print "sf url, mungled url to: " url
	}

	if (url ~ /^http:\/\/(.*)\.googlecode\.com\/files\//) {
		gsub("^http://", "", url)
		gsub("\..*", "", url)
		url = "http://code.google.com/p/" url "/downloads/list"
		if (DEBUG) print "googlecode url, mungled url to: " url
	}

	if (url ~ /^http:\/\/pecl.php.net\/get\//) {
		gsub("-.*", "", filename)
		url = "http://pecl.php.net/package/" filename
		if (DEBUG) print "pecl.php.net url, mungled url to: " url
	}

	if (url ~ /^(http|ftp):\/\/mysql.*\/Downloads\/MySQL-5.1\//) {
		url = "http://dev.mysql.com/downloads/mysql/5.1.html#source"
		 if (DEBUG) print "mysql 5.1 url, mungled url to: " url
	}


	if (DEBUG) print "Retrieving: " url
	errno=system("wget -O - \"" url "\" -t 2 -T 45 --passive-ftp > " tmpfile " 2>/dev/null" )

	if (errno==0) {
		while (getline oneline < tmpfile)
			wholeodp=(wholeodp " " oneline)
		if ( DEBUG ) print "Response: " wholeodp
	}

	close(tmpfile)
	system("rm -f " tmpfile)
	urldir=url;
	sub(/[^\/]+$/,"",urldir)

	if ( errno==0) {
		while (match(wholeodp, /<([aA]|[fF][rR][aA][mM][eE])[ \t][^>]*>/) > 0) {
			odp=substr(wholeodp,RSTART,RLENGTH);
			wholeodp=substr(wholeodp,RSTART+RLENGTH);

			lowerodp=tolower(odp);
			if (lowerodp ~ /<frame[ \t]/) {
				sub(/[sS][rR][cC]=[ \t]*/,"src=",odp);
				match(odp,/src="[^"]+"/)
				newurl=substr(odp,RSTART+5,RLENGTH-6)
				if (DEBUG) print "Frame: " newurl
				if (newurl !~ /\//) {
					newurl=(urldir newurl)
					if (DEBUG) print "Frame->: " newurl
				}

				if (link_seen(newurl)) {
					newurl=""
					continue
				}

				retval=(retval " " get_links(newurl))
			} else if (lowerodp ~ /href=[ \t]*"[^"]*"/) {
				sub(/[hH][rR][eE][fF]=[ \t]*"/,"href=\"",odp)
				match(odp,/href="[^"]*"/)
				link=substr(odp,RSTART,RLENGTH)
				odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
				link=substr(link,7,length(link)-7)

				if (link_seen(link)) {
					link=""
					continue
				}

				retval=(retval " " link)
				if (DEBUG) print "href(\"\"): " link
			} else if (lowerodp ~ /href=[ \t]*'[^']*'/) {
				sub(/[hH][rR][eE][fF]=[ \t]*'/,"href='",odp)
				match(odp,/href='[^']*'/)
				link=substr(odp,RSTART,RLENGTH)
				odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
				link=substr(link,7,length(link)-7)

				if (link_seen(link)) {
					link=""
					continue
				}

				retval=(retval " " link)
				if (DEBUG) print "href(''): " link
			} else if (lowerodp ~ /href=[ \t]*[^ \t>]*/) {
				sub(/[hH][rR][eE][fF]=[ \t]*/,"href=",odp)
				match(odp,/href=[^ \t>]*/)
				link=substr(odp,RSTART,RLENGTH)
				odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
				link=substr(link,6,length(link)-5)

				if (link_seen(link)) {
					link=""
					continue
				}

				retval=(retval " " link)
				if (DEBUG) print "href(): " link
			} else {
				# <a ...> but not href - skip
				if (DEBUG) print "skipping <a > without href: " odp
			}
		}
	} else {
		retval=("WGET ERROR: " errno)
	}


	if (DEBUG) print "Returning: " retval
	return retval
}

function subst_defines(var,defs) {
# substitute all possible RPM macros
	while ((var ~ /%{.*}/) || (var ~ /%[A-Za-z0-9_]+/)) {
		oldvar=var
		for (j in defs) {
			gsub("%{" j "}", defs[j], var)
			gsub("%" j , defs[j], var)
		}
		if (var==oldvar) {
			if ( DEBUG ) for (i in defs) print i " == " defs[i]
			return var
		}
	}
	return var
}

function find_mirror(url) {

	while (succ = (getline line < "mirrors")) {
	    if (succ==-1) { return url }
		nf=split(line,fields,"|")
		if (nf>1){
			origin=fields[1]
			mirror=fields[2]
			mname=fields[3]
			prefix=substr(url,1,length(origin))
			if (prefix==origin){
				if ( DEBUG ) print "Mirror fount at " mname
				close("mirrors")
				return mirror substr(url,length(origin)+1)
			}
		}
	}

	return url
}

function process_source(number,lurl,name,version) {
# fetches file list, and compares version numbers
	if ( DEBUG ) print "Processing " lurl

	if ( index(lurl,version)==0 ) {
		if (DEBUG) print "There is no version number."
		return 0
	}

	sub("://",":",lurl)
	sub("/",":/",lurl)
	gsub("[^/]*$",":&",lurl)
	split(lurl,url,":")
	acc=url[1]
	host=url[2]
	dir=url[3]
	filename=url[4]

	if (index(dir,version)) {
		dir=substr(dir,1,index(dir,version)-1)
		sub("[^/]*$","",dir)
		sub("(\.tar\.(bz|bz2|gz)|zip)$","",filename)
		if ( DEBUG ) print "Will check a directory: " dir
		if ( DEBUG ) print "and a file: " filename
	}

	filenameexp=filename
	gsub("\+","\\+",filenameexp)
	sub(version,"[A-Za-z0-9.]+",filenameexp)
	gsub("\.","\\.",filenameexp)
	if ( DEBUG ) print "Expression: " filenameexp
	match(filename,version)
	prever=substr(filename,1,RSTART-1)
	postver=substr(filename,RSTART+RLENGTH)
	if ( DEBUG ) print "Before number: " prever
	if ( DEBUG ) print "and after: " postver
	newurl=find_mirror(acc "://" host dir)
	#print acc "://" host dir
	#newurl=url[1]"://"url[2]url[3]url[4]
	#newurl=acc "://" host dir filename
	if ( DEBUG ) print "Looking at " newurl

	references=0
	finished=0
	oldversion=version
	odp=get_links(newurl,filename)
	if( odp ~ "ERROR: ") {
		print name "(" number ") " odp
	} else {
		if (DEBUG) print "WebPage downloaded"
		c=split(odp,linki)
		for (nr=1; nr<=c; nr++) {
			addr=linki[nr]
			if (DEBUG) print "Found link: " addr
			if ((addr ~ filenameexp) && !(addr ~ "[-_.0-9A-Za-z~]" filenameexp)) {
				match(addr,filenameexp)
				newfilename=substr(addr,RSTART,RLENGTH)
				if (DEBUG) print "Hypothetical new: " newfilename
				newfilename=fixedsub(prever,"",newfilename)
				newfilename=fixedsub(postver,"",newfilename)
				if (DEBUG) print "Version: " newfilename
				if (newfilename ~ /\.(pkg|bin|binary|built)$/) continue
				if (NUMERIC) {
					if ( compare_ver_dec(version, newfilename)==1 ) {
						if (DEBUG) print "Yes, there is new one"
						version=newfilename
						finished=1
					}
				} else if ( compare_ver(version, newfilename)==1 ) {
					if (DEBUG) print "Yes, there is new one"
					version=newfilename
					finished=1
				}
			}
		}
		if (finished==0)
			print name "(" number ") seems ok: " oldversion
		else
			print name "(" number ") [OLD] " oldversion " [NEW] " version
	}
}

# upgrade check for pear package using PEAR CLI
function pear_upgrade(name, ver) {
	pname = name;
	sub(/^php-pear-/, "", pname);

	pearcmd = "pear remote-info " pname " | awk '/^Latest/{print $NF}'"
	if (DEBUG) {
		print "pearcmd: " pearcmd
	}
	pearcmd | getline nver
	close(pearcmd)

	if (compare_ver(ver, nver)) {
		print name " [OLD] " ver " [NEW] " nver
	} else {
		print name " seems ok: " ver
	}

	return
}

function process_data(name,ver,rel,src) {
	if (name ~ /^php-pear-/) {
		return pear_upgrade(name, ver);
	}

# this function checks if substitutions were valid, and if true:
# processes each URL and tries to get current file list
	for (i in src) {
		if ( src[i] ~ /%{nil}/ ) {
			gsub(/\%\{nil\}/, "", src[i])
		}
		if ( src[i] !~ /%{.*}/ && src[i] !~ /%[A-Za-z0-9_]/ )  {
			if ( DEBUG ) print "Source: " src[i]
			process_source(i,src[i],name,ver)
		} else {
			print FNAME ":" i ": impossible substitution: " src[i]
		}
	}
}

BEGIN {
	# if U want to use DEBUG, run script with "-v DEBUG=1"
	# or uncomment the line below
	# DEBUG = 1

	errno=system("wget --help > /dev/null 2>&1")
	if (errno) {
		print "No wget installed!"
		exit 1
	}
	if (ARGC>=3 && ARGV[2]=="-n") {
		NUMERIC=1
		for (i=3; i<ARGC; i++) ARGV[i-1]=ARGV[i]
		ARGC=ARGC-1
	}
	frameseen[0] = 1
}

FNR==1 {
	if ( ARGIND != 1 ) {
		process_data(NAME,VER,REL,SRC)
		NAME="" ; VER="" ; REL=""
		for (i in DEFS) delete DEFS[i]
		for (i in SRC) delete SRC[i]
	}
	FNAME=FILENAME
	DEFS["_alt_kernel"]=""
	DEFS["20"]="\\ "
}

/^[Uu][Rr][Ll]:/&&(URL=="") { URL=subst_defines($2,DEFS) ; DEFS["url"]=URL }
/^[Nn]ame:/&&(NAME=="") { NAME=subst_defines($2,DEFS) ; DEFS["name"]=NAME }
/^[Vv]ersion:/&&(VER=="") { VER=subst_defines($2,DEFS) ; DEFS["version"]=VER }
/^[Rr]elease:/&&(REL=="") { REL=subst_defines($2,DEFS) ; DEFS["release"]=REL }
/^[Ss]ource[0-9]*:/ { if (/(ftp|http|https):\/\//) SRC[FNR]=subst_defines($2,DEFS) }
/%define/ { DEFS[$2]=subst_defines($3,DEFS) }

END {
	process_data(NAME,VER,REL,SRC)
}
Commit	Line	Data
6fb3305d	1	#!/bin/awk -f
d4089f30	2	# $Revision$, $Date$
b52a0a12	3	#
109ec22a	4	# Copyright (C) 2000-2008 PLD-Team <feedback@pld-linux.org>
b52a0a12 JB	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
3ca67c5b	17
2aba2068 SZ	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	}
6fb3305d	25
e3c5f324 JB	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	}
f0c5c231	34
70a5e8d4 SZ	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)
e2ebf07c JB	43	gsub("\.0*",".",v1)
e2ebf07c JB	44	gsub("\.0*",".",v2)
70a5e8d4 SZ	45	if (DEBUG) print "v1 == " v1
	46	if (DEBUG) print "v2 == " v2
	47	count=split(v1,v1a,"\.")
	48	count2=split(v2,v2a,"\.")
f0c5c231 ER	49
f0c5c231 ER	50	if (count<count2) mincount=count
70a5e8d4	51	else mincount=count2
f0c5c231	52
70a5e8d4	53	for (i=1; i<=mincount; i++) {
db83adc3 SZ	54	if (v1a[i]=="") v1a[i]=0
db83adc3 SZ	55	if (v2a[i]=="") v2a[i]=0
70a5e8d4 SZ	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
e3c5f324	73	} else if (ispre(v1a[i]) == 1)
70a5e8d4 SZ	74	return 1
	75	else
	76	return 0
	77	}
	78	if ((count2==mincount)&&(count!=count2)) {
	79	for (i=count2+1; i<=count; i++)
e3c5f324	80	if (ispre(v1a[i]) == 1)
70a5e8d4 SZ	81	return 1
	82	return 0
	83	} else if (count!=count2) {
	84	for (i=count+1; i<=count2; i++)
e3c5f324	85	if (ispre(v2a[i]) == 1)
70a5e8d4 SZ	86	return 0
	87	return 1
	88	}
db4f445e	89	return 0
70a5e8d4 SZ	90	}
70a5e8d4 SZ	91
f11b5a76	92	function compare_ver_dec(v1,v2) {
b8b1ac2a	93	# compares version numbers as decimal floats
f11b5a76	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,"\.")
f0c5c231 ER	104
f0c5c231 ER	105	if (count<count2) mincount=count
f11b5a76	106	else mincount=count2
f0c5c231	107
f11b5a76	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
e3c5f324	135	} else if (ispre(v1a[i]) == 1)
f11b5a76	136	return 1
	137	else
	138	return 0
	139	}
	140	if ((count2==mincount)&&(count!=count2)) {
	141	for (i=count2+1; i<=count; i++)
e3c5f324	142	if (ispre(v1a[i]) == 1)
f11b5a76	143	return 1
	144	return 0
	145	} else if (count!=count2) {
	146	for (i=count+1; i<=count2; i++)
e3c5f324	147	if (ispre(v2a[i]) == 1)
f11b5a76	148	return 0
	149	return 1
	150	}
	151	return 0
	152	}
	153
371e7443 AM	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
0ed4cd23	165	function get_links(url,filename,errno,link,oneline,retval,odp,wholeodp,lowerodp,tmpfile) {
70a5e8d4	166	# get all <A HREF=..> tags from specified URL
db4f445e	167	"mktemp /tmp/XXXXXX" \| getline tmpfile
9a8793f4	168	close(tmpfile)
356ec898	169
ea4c106c ER	170	if (url ~ /^http:\/\/(download\|dl).(sf\|sourceforge).net\//) {
ea4c106c ER	171	gsub("^http://(download\|dl).(sf\|sourceforge).net/", "", url)
a9dc9975	172	url = "http://prdownloads.sourceforge.net/" substr(url, 1, 1) "/" substr(url, 1, 2) "/" url
356ec898 ER	173	if (DEBUG) print "sf url, mungled url to: " url
356ec898 ER	174	}
f0c5c231	175
109ec22a	176	if (url ~ /^http:\/\/(.*)\.googlecode\.com\/files\//) {
3836b919 AM	177	gsub("^http://", "", url)
3836b919 AM	178	gsub("\..*", "", url)
b3df8960	179	url = "http://code.google.com/p/" url "/downloads/list"
3836b919 AM	180	if (DEBUG) print "googlecode url, mungled url to: " url
	181	}
	182
0ed4cd23 AM	183	if (url ~ /^http:\/\/pecl.php.net\/get\//) {
	184	gsub("-.*", "", filename)
	185	url = "http://pecl.php.net/package/" filename
	186	if (DEBUG) print "pecl.php.net url, mungled url to: " url
	187	}
	188
fbf290ff AM	189	if (url ~ /^(http\|ftp):\/\/mysql.*\/Downloads\/MySQL-5.1\//) {
	190	url = "http://dev.mysql.com/downloads/mysql/5.1.html#source"
	191	if (DEBUG) print "mysql 5.1 url, mungled url to: " url
	192	}
	193
	194
2e301ef5	195	if (DEBUG) print "Retrieving: " url
006dd08f	196	errno=system("wget -O - \"" url "\" -t 2 -T 45 --passive-ftp > " tmpfile " 2>/dev/null" )
f0c5c231	197
3f63995d	198	if (errno==0) {
7f460828	199	while (getline oneline < tmpfile)
fc589c7e JB	200	wholeodp=(wholeodp " " oneline)
fc589c7e JB	201	if ( DEBUG ) print "Response: " wholeodp
70a5e8d4	202	}
f0c5c231	203
3f63995d	204	close(tmpfile)
2e301ef5	205	system("rm -f " tmpfile)
55658186 JB	206	urldir=url;
55658186 JB	207	sub(/[^\/]+$/,"",urldir)
fc589c7e	208
3f63995d	209	if ( errno==0) {
fc589c7e JB	210	while (match(wholeodp, /<([aA]\|[fF][rR][aA][mM][eE])[ \t][^>]*>/) > 0) {
	211	odp=substr(wholeodp,RSTART,RLENGTH);
	212	wholeodp=substr(wholeodp,RSTART+RLENGTH);
	213
	214	lowerodp=tolower(odp);
	215	if (lowerodp ~ /<frame[ \t]/) {
	216	sub(/[sS][rR][cC]=[ \t]*/,"src=",odp);
	217	match(odp,/src="[^"]+"/)
	218	newurl=substr(odp,RSTART+5,RLENGTH-6)
157aa7fc	219	if (DEBUG) print "Frame: " newurl
55658186 JB	220	if (newurl !~ /\//) {
55658186 JB	221	newurl=(urldir newurl)
157aa7fc	222	if (DEBUG) print "Frame->: " newurl
55658186	223	}
371e7443 AM	224
	225	if (link_seen(newurl)) {
	226	newurl=""
	227	continue
	228	}
	229
2e301ef5	230	retval=(retval " " get_links(newurl))
fc589c7e	231	} else if (lowerodp ~ /href=[ \t]"[^"]"/) {
b5e77f8f	232	sub(/[hH][rR][eE][fF]=[ \t]*"/,"href=\"",odp)
95fcfc77	233	match(odp,/href="[^"]*"/)
b5e77f8f JB	234	link=substr(odp,RSTART,RLENGTH)
	235	odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
	236	link=substr(link,7,length(link)-7)
371e7443 AM	237
	238	if (link_seen(link)) {
	239	link=""
	240	continue
	241	}
	242
b5e77f8f	243	retval=(retval " " link)
95fcfc77	244	if (DEBUG) print "href(\"\"): " link
fc589c7e	245	} else if (lowerodp ~ /href=[ \t]'[^']'/) {
b5e77f8f	246	sub(/[hH][rR][eE][fF]=[ \t]*'/,"href='",odp)
95fcfc77	247	match(odp,/href='[^']*'/)
93c3b403	248	link=substr(odp,RSTART,RLENGTH)
b5e77f8f	249	odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
93c3b403	250	link=substr(link,7,length(link)-7)
371e7443 AM	251
	252	if (link_seen(link)) {
	253	link=""
	254	continue
	255	}
	256
93c3b403	257	retval=(retval " " link)
95fcfc77	258	if (DEBUG) print "href(''): " link
fc589c7e	259	} else if (lowerodp ~ /href=[ \t][^ \t>]/) {
4dde5541	260	sub(/[hH][rR][eE][fF]=[ \t]*/,"href=",odp)
95fcfc77	261	match(odp,/href=[^ \t>]*/)
8dfac79d	262	link=substr(odp,RSTART,RLENGTH)
b5e77f8f	263	odp=substr(odp,1,RSTART) substr(odp,RSTART+RLENGTH)
8dfac79d	264	link=substr(link,6,length(link)-5)
371e7443 AM	265
	266	if (link_seen(link)) {
	267	link=""
	268	continue
	269	}
	270
8dfac79d	271	retval=(retval " " link)
95fcfc77	272	if (DEBUG) print "href(): " link
6f57d45d	273	} else {
fc589c7e JB	274	# <a ...> but not href - skip
fc589c7e JB	275	if (DEBUG) print "skipping <a > without href: " odp
93c3b403	276	}
3f63995d SZ	277	}
3f63995d SZ	278	} else {
2e301ef5	279	retval=("WGET ERROR: " errno)
3f63995d	280	}
f0c5c231 ER	281
f0c5c231 ER	282
157aa7fc	283	if (DEBUG) print "Returning: " retval
db4f445e	284	return retval
70a5e8d4 SZ	285	}
70a5e8d4 SZ	286
6fb3305d	287	function subst_defines(var,defs) {
3255b4c7	288	# substitute all possible RPM macros
6fb3305d SZ	289	while ((var ~ /%{.*}/) \|\| (var ~ /%[A-Za-z0-9_]+/)) {
	290	oldvar=var
	291	for (j in defs) {
	292	gsub("%{" j "}", defs[j], var)
	293	gsub("%" j , defs[j], var)
	294	}
	295	if (var==oldvar) {
	296	if ( DEBUG ) for (i in defs) print i " == " defs[i]
	297	return var
	298	}
	299	}
	300	return var
	301	}
	302
c0bee1bc JK	303	function find_mirror(url) {
c0bee1bc JK	304
c54fbfc1 MK	305	while (succ = (getline line < "mirrors")) {
c54fbfc1 MK	306	if (succ==-1) { return url }
c0bee1bc	307	nf=split(line,fields,"\|")
a6f2cc98	308	if (nf>1){
c0bee1bc JK	309	origin=fields[1]
	310	mirror=fields[2]
	311	mname=fields[3]
	312	prefix=substr(url,1,length(origin))
	313	if (prefix==origin){
157aa7fc	314	if ( DEBUG ) print "Mirror fount at " mname
c0bee1bc	315	close("mirrors")
21b99dbc	316	return mirror substr(url,length(origin)+1)
c0bee1bc JK	317	}
	318	}
	319	}
	320
	321	return url
	322	}
	323
32aede2f	324	function process_source(number,lurl,name,version) {
3255b4c7	325	# fetches file list, and compares version numbers
157aa7fc	326	if ( DEBUG ) print "Processing " lurl
3f63995d SZ	327
3f63995d SZ	328	if ( index(lurl,version)==0 ) {
157aa7fc	329	if (DEBUG) print "There is no version number."
3f63995d SZ	330	return 0
	331	}
	332
6fb3305d SZ	333	sub("://",":",lurl)
	334	sub("/",":/",lurl)
	335	gsub("[^/]*$",":&",lurl)
	336	split(lurl,url,":")
	337	acc=url[1]
	338	host=url[2]
	339	dir=url[3]
70a5e8d4 SZ	340	filename=url[4]
70a5e8d4 SZ	341
3f63995d	342	if (index(dir,version)) {
70a5e8d4 SZ	343	dir=substr(dir,1,index(dir,version)-1)
	344	sub("[^/]*$","",dir)
	345	sub("(\.tar\.(bz\|bz2\|gz)\|zip)$","",filename)
157aa7fc AM	346	if ( DEBUG ) print "Will check a directory: " dir
157aa7fc AM	347	if ( DEBUG ) print "and a file: " filename
70a5e8d4	348	}
3f63995d	349
70a5e8d4 SZ	350	filenameexp=filename
70a5e8d4 SZ	351	gsub("\+","\\+",filenameexp)
0269da83	352	sub(version,"[A-Za-z0-9.]+",filenameexp)
e12bdf15	353	gsub("\.","\\.",filenameexp)
157aa7fc	354	if ( DEBUG ) print "Expression: " filenameexp
70a5e8d4 SZ	355	match(filename,version)
	356	prever=substr(filename,1,RSTART-1)
	357	postver=substr(filename,RSTART+RLENGTH)
157aa7fc AM	358	if ( DEBUG ) print "Before number: " prever
157aa7fc AM	359	if ( DEBUG ) print "and after: " postver
f0c5c231	360	newurl=find_mirror(acc "://" host dir)
c54fbfc1 MK	361	#print acc "://" host dir
	362	#newurl=url[1]"://"url[2]url[3]url[4]
	363	#newurl=acc "://" host dir filename
f0c5c231 ER	364	if ( DEBUG ) print "Looking at " newurl
f0c5c231 ER	365
6fb3305d	366	references=0
db83adc3	367	finished=0
db4f445e	368	oldversion=version
0ed4cd23	369	odp=get_links(newurl,filename)
3f63995d SZ	370	if( odp ~ "ERROR: ") {
3f63995d SZ	371	print name "(" number ") " odp
188dc51f	372	} else {
157aa7fc	373	if (DEBUG) print "WebPage downloaded"
188dc51f SZ	374	c=split(odp,linki)
	375	for (nr=1; nr<=c; nr++) {
	376	addr=linki[nr]
157aa7fc	377	if (DEBUG) print "Found link: " addr
3c68e899	378	if ((addr ~ filenameexp) && !(addr ~ "[-_.0-9A-Za-z~]" filenameexp)) {
188dc51f SZ	379	match(addr,filenameexp)
188dc51f SZ	380	newfilename=substr(addr,RSTART,RLENGTH)
eae5334c	381	if (DEBUG) print "Hypothetical new: " newfilename
2aba2068 SZ	382	newfilename=fixedsub(prever,"",newfilename)
2aba2068 SZ	383	newfilename=fixedsub(postver,"",newfilename)
157aa7fc	384	if (DEBUG) print "Version: " newfilename
155ac0a8	385	if (newfilename ~ /\.(pkg\|bin\|binary\|built)$/) continue
f11b5a76	386	if (NUMERIC) {
f11b5a76	387	if ( compare_ver_dec(version, newfilename)==1 ) {
157aa7fc	388	if (DEBUG) print "Yes, there is new one"
f11b5a76	389	version=newfilename
	390	finished=1
	391	}
	392	} else if ( compare_ver(version, newfilename)==1 ) {
157aa7fc	393	if (DEBUG) print "Yes, there is new one"
188dc51f SZ	394	version=newfilename
	395	finished=1
	396	}
70a5e8d4	397	}
6fb3305d	398	}
188dc51f	399	if (finished==0)
f1b8a975	400	print name "(" number ") seems ok: " oldversion
188dc51f SZ	401	else
188dc51f SZ	402	print name "(" number ") [OLD] " oldversion " [NEW] " version
6fb3305d SZ	403	}
6fb3305d SZ	404	}
eae5334c ER	405
	406	# upgrade check for pear package using PEAR CLI
	407	function pear_upgrade(name, ver) {
	408	pname = name;
	409	sub(/^php-pear-/, "", pname);
	410
	411	pearcmd = "pear remote-info " pname " \| awk '/^Latest/{print $NF}'"
	412	if (DEBUG) {
	413	print "pearcmd: " pearcmd
	414	}
	415	pearcmd \| getline nver
	416	close(pearcmd)
	417
	418	if (compare_ver(ver, nver)) {
	419	print name " [OLD] " ver " [NEW] " nver
	420	} else {
	421	print name " seems ok: " ver
	422	}
	423
	424	return
	425	}
f0c5c231	426
3255b4c7	427	function process_data(name,ver,rel,src) {
eae5334c ER	428	if (name ~ /^php-pear-/) {
	429	return pear_upgrade(name, ver);
	430	}
	431
3255b4c7 SZ	432	# this function checks if substitutions were valid, and if true:
3255b4c7 SZ	433	# processes each URL and tries to get current file list
6fb3305d	434	for (i in src) {
254a5f0b AM	435	if ( src[i] ~ /%{nil}/ ) {
	436	gsub(/\%\{nil\}/, "", src[i])
	437	}
6fb3305d	438	if ( src[i] !~ /%{.*}/ && src[i] !~ /%[A-Za-z0-9_]/ ) {
157aa7fc	439	if ( DEBUG ) print "Source: " src[i]
32aede2f	440	process_source(i,src[i],name,ver)
6fb3305d	441	} else {
c4a1f100	442	print FNAME ":" i ": impossible substitution: " src[i]
6fb3305d SZ	443	}
	444	}
	445	}
	446
	447	BEGIN {
	448	# if U want to use DEBUG, run script with "-v DEBUG=1"
	449	# or uncomment the line below
70a5e8d4	450	# DEBUG = 1
2e301ef5 SZ	451
	452	errno=system("wget --help > /dev/null 2>&1")
	453	if (errno) {
	454	print "No wget installed!"
	455	exit 1
	456	}
f11b5a76	457	if (ARGC>=3 && ARGV[2]=="-n") {
f11b5a76	458	NUMERIC=1
f0c5c231	459	for (i=3; i<ARGC; i++) ARGV[i-1]=ARGV[i]
f11b5a76	460	ARGC=ARGC-1
f11b5a76	461	}
371e7443	462	frameseen[0] = 1
6fb3305d SZ	463	}
	464
	465	FNR==1 {
	466	if ( ARGIND != 1 ) {
3255b4c7	467	process_data(NAME,VER,REL,SRC)
6fb3305d SZ	468	NAME="" ; VER="" ; REL=""
	469	for (i in DEFS) delete DEFS[i]
	470	for (i in SRC) delete SRC[i]
	471	}
	472	FNAME=FILENAME
f6c65bb7	473	DEFS["_alt_kernel"]=""
892e6dad	474	DEFS["20"]="\\ "
6fb3305d SZ	475	}
6fb3305d SZ	476
3255b4c7	477	/^[Uu][Rr][Ll]:/&&(URL=="") { URL=subst_defines($2,DEFS) ; DEFS["url"]=URL }
6fb3305d SZ	478	/^[Nn]ame:/&&(NAME=="") { NAME=subst_defines($2,DEFS) ; DEFS["name"]=NAME }
	479	/^[Vv]ersion:/&&(VER=="") { VER=subst_defines($2,DEFS) ; DEFS["version"]=VER }
	480	/^[Rr]elease:/&&(REL=="") { REL=subst_defines($2,DEFS) ; DEFS["release"]=REL }
a10d5bbe	481	/^[Ss]ource[0-9]*:/ { if (/(ftp\|http\|https):\/\//) SRC[FNR]=subst_defines($2,DEFS) }
6fb3305d SZ	482	/%define/ { DEFS[$2]=subst_defines($3,DEFS) }
	483
	484	END {
3255b4c7	485	process_data(NAME,VER,REL,SRC)
6fb3305d	486	}