initial commit

readersplanet-dl.sh

@@ -0,0 +1,21 @@
+#!/bin/sh
+
+VAR=1511
+
+while true; do
+	echo "$VAR"
+	curl --cookie "osCsid=8a497efe811cfd788283cb087ebd195d" "http://www.readerserver.de/download.php?aboload=true&id=$VAR" > $VAR.pdf
+	if [ `file --brief --mime-type $VAR.pdf` != "application/pdf" ]; then
+		echo "no pdf"
+		#exit
+	fi
+	VAR=$(($VAR+1))
+done
+
+
+#osCsid=Host: .www.readerserver.de
+#Name: osCsid
+#Value: 8a497efe811cfd788283cb087ebd195d
+#Path: /
+#Secure: No
+#Expires: At the end of the session

test/find-atlan.py

@@ -0,0 +1,13 @@
+#!/bin/env python
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("ATLAN") or firstline.startswith("\fATLAN"):
+			newpath = os.path.join("atlan", filename)
+			os.rename(oldpath, newpath)

test/find-centauri.py

@@ -0,0 +1,13 @@
+#!/bin/env python
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("\fCentauri-Zyklus"):
+			newpath = os.path.join("centauri", filename)
+			os.rename(oldpath, newpath)

test/find-diegrößtesciencefictionseriederwelt.py

@@ -0,0 +1,14 @@
+#!/bin/env python
+# -*- coding: utf-8 -*-
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("\fD I E G R Ö S S T E S C I E N C E - F I C T I O N - S E R I E D E R W E L T"):
+			newpath = os.path.join("diegrößtesciencefictionseriederwelt2", filename)
+			os.rename(oldpath, newpath)

test/find-duplicates.py

@@ -0,0 +1,23 @@
+#/bin/env python
+
+from hashlib import md5
+import os
+
+# remove duplicates
+
+filedict = dict()
+
+for root, dirs, files in os.walk('.'):
+	for f in files:
+		path = os.path.join(root, f)
+		fo = open(path)
+		content = fo.read()
+		fo.close()
+		h = md5(content).hexdigest()
+		if filedict.get(h, None):
+			os.remove(path)
+		else:
+			filedict[h] = path
+
+print filedict
+

test/find-negasphäre.py

@@ -0,0 +1,14 @@
+#!/bin/env python
+# -*- coding: utf-8 -*-
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("\fN e g a s p h ä r e"):
+			newpath = os.path.join("negasphäre", filename)
+			os.rename(oldpath, newpath)

test/find-perry.py

@@ -0,0 +1,13 @@
+#!/bin/env python
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("\fNr. "):
+			newpath = os.path.join("perry", filename)
+			os.rename(oldpath, newpath)

test/find-terranova.py

@@ -0,0 +1,13 @@
+#!/bin/env python
+
+import os
+
+for root, dirs, files in os.walk('txt'):
+	for filename in files:
+		oldpath = os.path.join(root, filename)
+		f = open(oldpath)
+		firstline = f.readline()
+		f.close()
+		if firstline.startswith("\fT E R R A N O V A"):
+			newpath = os.path.join("terranova", filename)
+			os.rename(oldpath, newpath)

test/script.sh

@@ -0,0 +1,11 @@
+# for older ebooks:
+# pdftoppm -f 4 -l 4 -r 220 -x 164 -y 242 -W 1490 -H 2166 -png pdf/528.pdf foo
+
+pdftoppm -f 3      -r 190 -x 164 -y 278 -W 1490 -H 2130 -png pdf/528.pdf foo
+mogrify -gamma 0.25 -resize 600x800! +dither -type Grayscale -colors 8 foo-*.png
+
+pdftoppm -f 1 -l 1 -r 300 -png pdf/528.pdf foo
+mogrify -resize 600x800 -type Grayscale foo-01.png
+
+pdftoppm -f 2 -l 2 -r 300 -x 164 -y 278 -W 1490 -H 2130 -png pdf/528.pdf foo
+mogrify -resize 600x800 -type Grayscale foo-02.png

test/script2.sh

@@ -0,0 +1,29 @@
+#!/bin/sh
+
+#time gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/screen -dNOPAUSE -dQUIET -dBATCH -sOutputFile=output.pdf 064\ -\ Clark\ Darlton\ -\ Perry\ Rhodan\ -\ 64\ -\ Im\ Zeit-Gefängnis.pdf
+#root@192.168.111.1:/media/sd/PerryRhodan/Perry\\\ Rhodan\\\ -\\\ Romanzyklus\\\ -\\\ 0050-0099\\\ -\\\ Atlan\\\ und\\\ Arkon/
+
+[ $# = 2 ] || { echo "usage: $0 input.pdf output.pdf"; exit 1; }
+
+INPUT=$1
+OUTPUT=$2
+
+LEFT=0
+TOP=-30
+RIGHT=0
+BOTTOM=-20
+
+PART1=`mktemp`
+PART2=`mktemp`
+PART3=`mktemp`
+PART4=`mktemp`
+PART5=`mktemp`
+
+pdftk $INPUT cat 1-3 output $PART1
+pdfcrop $PART1 $PART3
+pdftk $INPUT cat 4-end output $PART2
+pdfcrop --margins "$LEFT $TOP $RIGHT $BOTTOM" --clip $PART2 $PART4
+pdftk $PART3 $PART4 cat output $PART5
+pdfcrop $PART5 "$OUTPUT"
+
+rm $PART1 $PART2 $PART3 $PART4 $PART5