extlnks.awk
ActivePublic
Actions

Authored by Green_Cardamom on Dec 21 2020, 8:08 PM.

Tags

None

Referenced Files

	F33968139: extlnks.awk
	Dec 22 2020, 3:57 PM

	F33963611: extlnks.awk
	Dec 21 2020, 8:26 PM

	F33963594: extlnks.awk
	Dec 21 2020, 8:08 PM

Subscribers

None

	#!/bin/awk -f
	@load "filefuncs"
	@load "readfile"
	BEGIN {
	# ################################################################################################
	#
	# External Links
	# Insert new External Links into an article at the right spot.
	# User:GreenC November 2015
	#
	# Pass variable id:
	#
	# awk -v id="/home/greenc/wi-awb/temp/wi-awb-1102154815/" -f extlnks.awk
	#
	# Directory "id" should already contain two files:
	#
	# article.txt .. wikitext of the article
	# out.librivox .. the template to be added
	#
	# Output file:
	#
	# article.extlnks.txt .. updated version with external link added
	#
	# ################################################################################################

	iaout = readfile(id "out.librivox")
	lv = 1

	if(exists(id "out.extlnk")) { # No ==External links== header found. Add it.
	location["last header"] = location_string(id, "==")
	location["persondata"] = location_string(id, "[{]{2}[ ]{0,1}[Pp]ersondata")
	location["authority"] = location_string(id, "[{]{2}[ ]{0,1}[Aa]uthority")
	location["start"] = location_string(id, "[{]{2}[ ]{0,1}[Ss][-]start")
	location["startbox"] = location_string(id, "[{]{2}[ ]{0,1}[Ss]tart [Bb]ox")
	location["metadata"] = location_string(id, "<!--[ ]{0,2}[Mm]etadata: see")
	location["default"] = location_string(id, "[{]{2}[ ]{0,1}[Dd][Ee][Ff][Aa][Uu][Ll][Tt][Ss][Oo][Rr][Tt]")
	PROCINFO["sorted_in"] = "@val_num_asc"
	point = 0
	for(idx in location) {
	if(idx ~ "last header" \|\| location[idx] == 0)
	continue
	point = location[idx]
	break
	}
	if(point != 0) {
	insert_header(id, point, "==External links==")
	print 1
	exit
	}
	else {
	print 0
	exit
	}
	}

	if(insert_link(id, iaout) == 1) { # 1 = successfully inserted the external link.
	print 1
	exit
	} else {
	print 0
	exit
	}
	}

	function location_string(id, str ,line,articleorig,i,re,loc) {

	articleorig = id "article.txt"
	i = loc = 0
	re = "^" str

	while ((getline line < articleorig) > 0) {
	i++
	if(match(line, re))
	loc = i
	}
	close(articleorig)
	return loc
	}

	function insert_header(id, point, str ,articleorig, articlenew,i,line) {

	articleorig = id "article.txt"
	articlenew = id "article.extlnks.txt"

	i = 0
	while ((getline line < articleorig) > 0) {
	i++
	if(i == point) {
	printf("%s\n",str) >> articlenew
	printf("%s\n",iaout) >> articlenew
	printf("%s\n",line) >> articlenew
	continue
	}
	printf("%s\n",line) >> articlenew
	}
	close(articlenew)
	}

	function insert_link(id,iaout, line,zone,articleorig,done) {

	zone = done = 0

	articleorig = id "article.txt"

	while ((getline line < articleorig) > 0) {
	if( (zone == 1 \|\| zone == 2) && ! done) {
	if(substr(line,1,1) == "*") {
	if(zone == 1) {
	printf("%s",iaout) >> id "article.extlnks.txt"
	printf("%s\n",line) >> id "article.extlnks.txt"
	}
	else if(zone == 2) {
	printf("%s\n",line) >> id "article.extlnks.txt"
	printf("%s",iaout) >> id "article.extlnks.txt"
	}
	zone = 0
	done = 1
	continue
	}
	else if(substr(line,1,1) == "" && lv) {
	printf("%s",iaout) >> id "article.extlnks.txt"
	printf("%s\n",line) >> id "article.extlnks.txt"
	zone = 0
	done = 1
	continue
	}
	else
	zone = 0
	}
	else
	zone = 0
	if(! lv && line ~ /={2,3}[ ]{0,2}[Ee]xternal [Ll]inks[ ]{0,2}={2,3}/)
	zone = 1
	if(lv && line ~ /[][ ][{][{][ ]*[Gg]utenberg author/)
	zone = 2
	if(lv && line ~ /[][ ][{][{][ ]*[Ii]nternet [Aa]rchive author/)
	zone = 1
	printf("%s\n",line) >> id "article.extlnks.txt"
	}
	close(articleorig)
	close(id "article.extlnks.txt")
	return done
	}

	#
	# Check for file existence. Return 1 if exists, 0 otherwise.
	# Requires GNU Awk:
	# @load "filefuncs"
	#
	function exists(name ,fd) {
	if ( stat(name, fd) == -1)
	return 0
	else
	return 1
	}

Event Timeline

Green_Cardamom created this paste.Dec 21 2020, 8:08 PM

Green_Cardamom edited the content of this paste. (Show Details)Dec 21 2020, 8:26 PM

Green_Cardamom edited the content of this paste. (Show Details)Dec 22 2020, 3:57 PM

extlnks.awkActivePublicActions

Event Timeline

extlnks.awk
ActivePublic
Actions