0.8.0

leaf.php

@@ -2,10 +2,19 @@
 <?php
 
 // Leaf - Tools for Book Scans
-$version = "0.7.1";
+$version = "0.8.0";
 $time_start = microtime(true);
 date_default_timezone_set("America/Los_Angeles");
-$xt = 3; // threads
+
+// Detect cores
+
+$reasonable_cores = array(1,2,3,4,6,8,16);
+$cores = chop(shell_exec("system_profiler SPHardwareDataType 2> /dev/null | grep 'Total Number of Cores' | cut -d: -f2 | tr -d ' '"));
+if (in_array($cores,$reasonable_cores)) {
+	$xt = $cores;
+	} else {
+	$xt = 2;
+	}
 
 // Functions
 
@@ -152,22 +161,31 @@ USAGE: leaf [mode] [-options] directory
 Modes:
 clean                       remove scratch files
 crop                        define EXIF crop values using template files
-    -crops=<num>          specify how many files to use
+    -crops=<num> (6)        specify how many files to use
 desort                      remove image sequence prefix
 deskew                      detect rotation angles
-    -max=<num>            angles greater than this value are ignored, default .4
+    -max=<num> (.4)         angles greater than this value are ignored
-    -pad=<num>            pixels to pad around crop area, default 80
+    -pad=<num> (80)         pixels to pad around crop area
-    -contrast=<num>       contrast boost, default 20
+    -contrast=<num> (20)    levels attenuation
-    -size=<num>           size in pixels of dmap, default 2200
+    -size=<num> (2200)      dmap size in pixels
     -nomap                  don't use dmaps
 divide                      wrapper for imagemagick Divide_Src
     -map=<file>             specify brightness file
     -adjust=<params>        levels adjustment (ex. \"0%,98%,.9\")
-    -q=<quality>          quality out of 100
+    -q=<quality> (95)       quality out of 100
     -pages                  map file is for LR rotated pages
 dupes                       Find duplicate images using computed PHASH on thumbnails
-    -threshold=<num>      match threshold
+    -threshold=<num> (10)   match threshold
-    -walk=<num>           comparison scope (compare n image to n, n+1, n+2, etc)
+    -walk=<num> (5)         comparison scope (compare n image to n, n+1, n+2, etc)
+fixpdf                      de-rotate all pages in a pdf
+    -file=<pdf>             pdf file in target dir
+generate                    create final jpg images for pdf creation
+    -pixels=<num>           scale to x pixels on longest side
+    -inches=<num>           set dpi to x inches on longest side
+    -q=<0-100> (90)         jpeg quality
+    -adjust=<params>        levels adjustment (ex. \"0%,98%,.9\")
+    -px=<num>               png size multiplier
+    -padjust=<params>       png levels adjustment (otherwise calculated from adjust)
 makepdf                     combine images into a pdf with img2pdf
 profile                     apply xmp profile to images (requires exiv2 > 0.25)
     -file=<file>            xmp profile
@@ -176,13 +194,13 @@ resort                    reorder image sequence by adding a new image
     -x=<num>                position of inserted file
 review                      print a table of image dimension statistics
 rotate                      batch transform rotate (lossy)
-    -x=<angle>            rotation angle, default=90
+    -x=<angle> (90)         rotation angle
-    -q=<0-100>            jpeg quality, default read from source or 95
+    -q=<0-100> (auto)       jpeg quality, default read from source or 95
 setdpi                      set image dpi with exiftool
     -x=<dpi>                specify dpi
     -height=<inches>        calculate dpi from specified height
 sort                        sort files as AAABBB -> ABABAB
-    -m=<num>              specify midpoint (cover image)
+    -m=<num> (auto)         specify midpoint (cover image)
 strip                       strip exif crop values from images with exiftool
     
 ";
@@ -196,20 +214,62 @@ fin();
 echo Welcome("Generate final jpg images");
 /////////////////////////////////////////////////////////////////////////
 
+$dest = rtrim(args("dir"), '/')."_generated";
+
+if (count(glob($dest."/*.*"))) {
+	msg("Files already exist in ".$dest.". Move to trash?",2);
+	exec("/opt/local/bin/trash -a ".$dest);
+	mkdir($dest);
+	} elseif (!is_dir($dest)) {
+	mkdir($dest);
+	}
+
+if (args("pixels")) {
+	$pixels = args("pixels");
+	} else {
+	msg("No pixel value specified",1);
+	}
+
+if (args("inches")) {
+	$inches = args("inches");
+	} else {
+	msg("No inches value specified",1);
+	}
+	
+if (args("q")) {
+	$quality = args("q");
+	} else {
+	$quality = 90;
+	}
+
+if (args("px")) {
+	$multiplier = args("px");
+	} else {
+	$multiplier = 2;
+	}
+
 $files = glob(args("dir")."*.{jpg,JPG}", GLOB_BRACE);
 
 echo "Checking EXIF values: ";
 
 foreach ($files as $file) {
 	echo ".";
-	$lines = shell_exec("exiftool -s -s -s -f -Label -ConvertToGrayscale ".$file);
+	$lines = shell_exec("exiftool -s -s -s -f -Label -ConvertToGrayscale -ImageWidth -ImageHeight ".$file);
 	$parts = explode("\n", $lines);
-	if ($parts[0] == "bitmap") {
+	$width = $parts[2];
-		$bitmap[] = $file;
+	$height = $parts[3];
-		} elseif ($parts[1] == "True") {
+	$ratio = $height/$width;
-		$grey[] = $file;
+	if ($ratio > 3) {
+		$spine[] = array($file,$width,$height);
 		} else {
-		$color[] = $file;
+		$dims[$file] = $ratio;
+		}
+	if ($parts[0] == "bitmap") {
+		$bitmap[] = array($file,$width,$height);
+		} elseif ($parts[1] == "True") {
+		$grey[] = array($file,$width,$height);
+		} elseif ($ratio < 3) {
+		$color[] = array($file,$width,$height);
 		}
 	}
 
@@ -217,18 +277,151 @@ echo "\n\n";
 
 echo "Bitmap files: ";
 if (is_array($bitmap)) { echo count($bitmap); } else { echo "None"; }
-
+echo "\n";
-echo "\n\n";
 
 echo "Greyscale files: ";
 if (is_array($grey)) { echo count($grey); } else { echo "None"; }
-
+echo "\n";
-echo "\n\n";
 
 echo "Color files: ";
 if (is_array($color)) { echo count($color); } else { echo "None"; }
+echo "\n";
+
+echo "Spine files: ";
+if (is_array($spine)) { echo count($spine); } else { echo "None"; }
+echo "\n\n";
+
+if (count($spine) > 1) {
+	msg("More than one spine file found, meltdown",1);
+	}
+
+$ratioavg = number_format(array_sum($dims)/count($dims),3);
+
+echo "Average page ratio: ".$ratioavg;
+if ($ratioavg > 1) { // page is tall
+	$page_height = $pixels;
+	$page_width = floor($page_height/$ratioavg);
+	$paper_height = $inches;
+	$paper_width = number_format($paper_height/$ratioavg,2);
+	} else {
+	$page_width = $pixels;
+	$page_height = floor($page_width*$ratioavg);
+	$paper_width = $inches;
+	$paper_height = number_format($paper_width*$ratioavg,2);
+	}
+
+$dpi = $page_height / $paper_height;
+$canonicalpngdpi = $dpi*$multiplier;
+$pngdpi = $canonicalpngdpi*39.37007874016;
+$dims = $page_width."x".$page_height;
+$pngdims = ($page_width*$multiplier)."x".($page_height*$multiplier);
+
+echo "\n";
+echo "Page size: ".$paper_width."\" x ".$paper_height."\"";
+echo "\n";
+echo "Target pixel dimensions: ".$dims." (png=".$pngdims.")";
+echo "\n";
+echo "DPI = ".$dpi." (png=".$canonicalpngdpi.")";
+
+foreach ($color as $parts) {
+	$jpg[] = $parts;
+	}
+	
+foreach ($grey as $parts) {
+	$parts[] = 1;
+	$jpg[] = $parts;
+	}
+
+if (is_array($spine)) {
+	$spine[0][] = 2;
+	$jpg[] = $spine[0];
+	}
+
+foreach ($jpg as $parts) {
+
+	$file = $parts[0];
+	$output = $dest."/".basename($file);
+	
+	if (isset($parts[3]) && $parts[3] == 1) {
+		
+		$msg = "Processing greyscale page ".$file." @ ".$dims."!";
+		$cmd = "convert -resize ".$dims."\! -colorspace gray ";
+		
+		} elseif (isset($parts[3]) && $parts[3] == 2) {
+		
+		$sdims = $page_height."x".$page_height;
+		$msg = "Processing spine ".$file." @ ".$sdims;
+		$cmd = "convert -resize ".$sdims." ";
+		
+		} else {
+		
+		$msg = "Processing color page ".$file." @ ".$dims."!";
+		$cmd = "convert -resize ".$dims."\! ";
+		
+		}		
+	
+	if (args("adjust")) {
+		$msg .= " (".args("adjust").")";
+		$cmd .= "-level ".args("adjust")." ";
+		}
+	
+	$msg .= " [Q=".$quality."]";
+	$cmd .= "-quality ".$quality." ".$file." ".$output;
+	
+	$thread_a[] = array($msg, $cmd);
+	$thread_b[] = array(".","exiftool -overwrite_original -Xresolution=".$dpi." -Yresolution=".$dpi." -jfif:Xresolution=".$dpi." -jfif:Yresolution=".$dpi." ".$output);
+	
+	}
 
 echo "\n\n";
+msg("Beginning multithreaded convert with ".$xt." threads");
+multiexec($thread_a,$xt);
+echo "\n";
+msg("Setting DPI: ");
+multiexec($thread_b,$xt);
+
+foreach ($bitmap as $parts) {
+	
+	$file = $parts[0];
+	$ext = pathinfo($file,PATHINFO_EXTENSION);
+	$output = $dest."/".basename($file,$ext)."png";
+	
+	$msg = "Processing bitmap page ".$file." @ ".$pngdims."!";
+	$cmd = "convert -resize ".$pngdims."\! ";
+	
+	if (args("adjust")) {
+		$adjust = args("adjust");
+		} else {
+		$adjust = "0%,100%,1";
+		}
+		
+	$parts = explode(",", $adjust);
+	$black = rtrim($parts[0],"%");
+	$white = rtrim($parts[1],"%");
+	$gamma = $parts[2];
+	
+	if (args("padjust")) {
+		$radjust = args("padjust");
+		} else {
+		$radjust = ($black+60)."%".",".($white-15)."%".",".($gamma-.7);
+		}
+	
+	$msg .= " (".$radjust.")";
+	$cmd .= "-level ".$radjust." ";
+		
+	$cmd .= "-monochrome -colors 2 -depth 1 -negate ".$file." ".$output;
+	
+	$thread_c[] = array($msg, $cmd);
+	$thread_d[] = array(".","exiftool -overwrite_original -PixelsPerUnitX=".$pngdpi." -PixelsPerUnitY=".$pngdpi." ".$output);
+	
+	}
+
+echo "\n\n";
+msg("Beginning multithreaded convert with ".$xt." threads");
+multiexec($thread_c,$xt);
+echo "\n";
+msg("Setting DPI: ");
+multiexec($thread_d,$xt);
 
 fin();
 
@@ -250,6 +443,43 @@ if (!is_dir("scratch")) {
 
 fin();
 
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Note: Edges
+//////////////////////
+} elseif (args("app") == "edges") {
+echo Welcome("Edge detection");
+/////////////////////////////////////////////////////////////////////////
+
+$dest = rtrim(args("dir"), '/')."_edges";
+if (count(glob($dest."/*.*"))) {
+	msg("Files already exist in ".$dest.". Move to trash?",2);
+	exec("/opt/local/bin/trash -a ".$dest);
+	mkdir($dest);
+	} elseif (!is_dir($dest)) {
+	mkdir($dest);
+	}
+
+$files = glob(args("dir")."*.*");
+
+$size = "1400";
+$canny = "0x1+10%+30%";
+$hough = "9x9+200";
+
+foreach ($files as $file) {
+	$ext = pathinfo($file,PATHINFO_EXTENSION);
+	$output = $dest."/".basename($file,$ext)."jpg";	
+	$msg = "Detecting edges in ".$file;
+	$cmd = "convert ".$file."[".$size."x".$size."] -auto-orient \( +clone -canny ".$canny." -write /tmp/".basename($file)."_canny.png ";
+	$cmd .= "-background none -fill red -stroke red -strokewidth 1 -hough-lines ".$hough." -write /tmp/".basename($file)."_lines.png \) ";
+	$cmd .= "-composite ".$output;
+	$thread[] = array($msg, $cmd);
+	}
+
+msg("Beginning multithreaded convert with ".$xt." threads");
+multiexec($thread,$xt);
+
+fin();
+
 ////////////////////////////////////////////////////////////////////////////////////////////////
 // Note: Rotate
 //////////////////////
@@ -269,7 +499,9 @@ if (args("x")) {
 
 $dest = rtrim(args("dir"), '/')."_rotated";
 if (count(glob($dest."/*.*"))) {
-	msg("Files already exist in destination ".$dest,1);
+	msg("Files already exist in ".$dest.". Move to trash?",2);
+	exec("/opt/local/bin/trash -a ".$dest);
+	mkdir($dest);
 	} elseif (!is_dir($dest)) {
 	mkdir($dest);
 	}
@@ -412,7 +644,7 @@ if (ask("Create spreads? (Y/n)") != "n") {
 		exec("xattr -c ".implode(" ", $sheets));
 		$term = chop(shell_exec("env | grep 'TERM_PROGRAM=' | cut -f2- -d="));
 		if ($term == "iTerm.app") {
-			echo shell_exec("imgcat ".implode(" ", $sheets));
+			echo shell_exec("~/.iterm2/imgcat ".implode(" ", $sheets));
 			} else {	
 			exec("open ".implode(" ", $sheets)." -b com.apple.Preview");
 			}
@@ -775,13 +1007,59 @@ if (file_exists($dest)) {
 	if (file_exists($dest)) { msg("Freak accident",1); }
 	}
 echo "Creating pdf...\n\n";
-exec("img2pdf --verbose --viewer-page-layout twocolumnright --output ".$dest." ".$input);
+exec("img2pdf --verbose --viewer-page-layout twocolumnright --creator 'Leaf ".$version."' --output ".$dest." ".$input);
 echo "\n";
 msg("Press return to open in Acrobat",2);
 exec("open ".$dest." -b com.adobe.Acrobat.Pro");
 
 fin();
 
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Note: Fixpdf
+//////////////////////
+} elseif (args("app") == "fixpdf") {
+echo Welcome("Systematically de-rotate pages in pdf file with pdftk");
+/////////////////////////////////////////////////////////////////////////
+
+$file = args("dir").args("file");
+
+if (!file_exists($file)) {
+	msg("Problem reading file ".$file,1);
+	}
+
+$info = shell_exec("pdftk ".$file." dump_data");
+$chunks = explode("PageMediaBegin", $info);
+
+$items = "";
+$hits = 0;
+foreach ($chunks as $page => $chunk) {
+	preg_match("/(PageMediaRotation: )(\d+)(\n)/",$chunk,$matches);
+	if (isset($matches[2]) && $matches[2] > 0) {
+		echo "Page ".$page." is rotated ".$matches[2]."\n";
+		$hits++;
+		}
+	if ($page) {
+		$items .= $page."north ";
+		}
+	}
+	
+if (!$hits) {
+	msg("No rotated pages found.",1);
+	}
+
+$ext = pathinfo($file, PATHINFO_EXTENSION);
+$output = basename($file,".".$ext)."_derotated.pdf";
+echo "\n";
+echo "Writing ".$output;
+echo "\n";
+exec("pdftk ".$file." cat ".$items."output ".$output);
+
+echo "\n";
+msg("Remove original file?",2);
+exec("/opt/local/bin/trash -a ".$file."; mv ".$output." ".$file);
+
+fin();
+
 ////////////////////////////////////////////////////////////////////////////////////////////////
 // Note: Divide
 // We assume a .JPG file is an unmodified DCIM image and convert to a greyscale tif for overlay.
@@ -817,7 +1095,9 @@ if (!args("map")) {
 $dest = rtrim(args("dir"), '/')."_divided";
 
 if (count(glob($dest."/*.*"))) {
-	msg("Files already exist in destination ".$dest,1);
+	msg("Files already exist in ".$dest.". Move to trash?",2);
+	exec("/opt/local/bin/trash -a ".$dest);
+	mkdir($dest);
 	} elseif (!is_dir($dest)) {
 	mkdir($dest);
 	}
@@ -828,38 +1108,65 @@ if (args("q")) {
 	$quality = 95;
 	}
 
+if ($pages) {
+	$ext = pathinfo($map, PATHINFO_EXTENSION);
+	$mapleft = "scratch/".str_replace(".".$ext, "_left.".$ext, $map);
+	$mapright = "scratch/".str_replace(".".$ext, "_right.".$ext, $map);
+	exec("convert -rotate 90 -quality 95 ".$map." ".$mapleft);
+	exec("convert -rotate 270 -quality 95 ".$map." ".$mapright);
+	}
+
 echo "Building threads: ";
 
 $thread = array();
 foreach ($files as $file) {
+
 	$msg = "Dividing ".$file." with ".$map.", Q=".$quality;
-	list ($width, $height) = getimagesize($map);
+	$cmd = "convert ".$file." ";
-	list ($twidth, $theight) = getimagesize($file);
+	$output = $dest."/".basename($file);
-	if ($width != $twidth | $height != $theight) {
+	
-		$tmap = $map."'[".$twidth."x".$theight."!]'";
-		$msg .= " (resize map) ";
-		} else {
-		$tmap = $map;
-		}
-	if (args("adjust")) {
-		$msg .= " (".args("adjust").") ";
-		$cmd = "convert ".$file." ".$tmap." -compose Divide_Src -composite -level ".args("adjust")." -quality ".$quality." ".$dest."/".basename($file);
-		} else {
-		$cmd = "convert ".$file." ".$tmap." -compose Divide_Src -composite -quality ".$quality." ".$dest."/".basename($file);
-		}
 	if ($pages) {
+		
 		$label = trim(shell_exec("exiftool -s -s -s -Label ".$file));
+		
 		if ($label == "non-page") {
 			$msg .= " <Skipping non-page>";
-			$cmd = "true";
+			$cmd = "cp ".$file." ".$output;
 			echo "o";
-			} else {
+			$thread[] = array($msg, $cmd);
-			echo ".";
+			continue;
 			}
+		
+		if (substr(basename($file), 0, 3) % 2 == 0) {
+			$use = $mapright;
 			} else {
-		echo ".";
+			$use = $mapleft;
 			}
-	$cmd = "true";
+		list ($width, $height) = getimagesize($use);
+		list ($twidth, $theight) = getimagesize($file);
+		if ($width != $twidth | $height != $theight) {
+			$tmap = $use."'[".$twidth."x".$theight."!]'";
+			$msg .= " (resize map) ";
+			} else {
+			$tmap = $use;
+			}
+		
+		} else {
+		
+		$tmap = $map;
+		
+		}
+	
+	$cmd .= $tmap." -compose Divide_Src -composite -quality ".$quality." ";
+	
+	if (args("adjust")) {
+		$msg .= " (".args("adjust").") ";
+		$cmd .= "-level ".args("adjust")." ";
+		}
+
+	$cmd .= $output;
+
+	echo ".";
 	$thread[] = array($msg, $cmd);
 	}