//
if($argv[i]!="-")
+ Print("Rewrite font with CMAP ",$argv[i]);
cmap=LoadStringFromFile($argv[i])
glyphs=StrSplit(cmap,"\\n");
j=0
$__autoload['wsTools'] = dirname(__FILE__) . '/class.ws.tools.php';\r
$__autoload['wsHTML5Compiler'] = dirname(__FILE__) . '/html5/class.ws.html5.compiler.php';\r
$__autoload['wsHTML5Link'] = dirname(__FILE__) . '/html5/class.ws.html5.links.php';\r
-$__autoload['wsPDFFontExtractor'] = dirname(__FILE__) . '/class.ws.pdf.fontextractor.php';\r
+$__autoload['wsPDFFontExtractor'] = dirname(__FILE__) . '/fontextractor/class.ws.pdf.fontextractor.php';\r
+$__autoload['wsPDFFont'] = dirname(__FILE__) . '/fontextractor/class.ws.pdf.font.php';\r
?>
\ No newline at end of file
+++ /dev/null
-<?php
-
-class wsPDFFontExtractor {
-
- protected $in;
- protected $outpdf;
- protected $outweb;
- protected $fonts = array();
- protected $cmaps = array();
- protected $descendants = array();
- protected $descendantNames = array();
- protected $doc;
-
- public function __construct($in, $doc=null) {
- $this->doc = $doc;
- $this->in = $in . '/original.pdf';
- $this->outpdf = $in . '/fonts/pdf';
- $this->outweb = $in . '/fonts/web';
- }
-
- public function extract() {
- $this->clean();
-
- $this->getFonts();
- $this->getFontsDescriptors();
- $this->extractFonts();
- $this->extractCmaps();
- $this->convertToTTF();
- }
-
- public function convertToTTF() {
-
-
-
- $dr = opendir($this->outpdf);
- while ($file = readdir($dr)) {
- if ($file == '.' || $file == '..' || files::getExtension($file) == 'cmap') {
- continue;
- }
-
- $e = explode('.', $file);
- array_pop($e);
- $fname = implode('.', $e);
-
- $e = explode('-', $fname);
- array_pop($e);
- $fname = implode('-', $e);
-
- if (!isset($collections[$fname])) {
- $collections[$fname] = array();
- }
- $collections[$fname][] = $file;
- }
-
- foreach ($collections as $fontname => $files) {
- $descendant = array_keys($this->descendantNames, $fontname);
- if (count($descendant)) {
- $f = array_keys($this->descendants, $descendant[0]);
- $fontname = $f[0];
- }
-
-
- $fontforge = new cubeCommandLine('convert.pe');
- $fontforge->setPath(CONVERTER_PATH);
- foreach ($files as $file) {
- $fontforge->setArg(null, $this->outpdf . '/' . $file);
- }
- $cmapFile = $this->outpdf . '/' . $fontname . '.cmap';
- if (!file_exists($cmapFile)) {
- $fontforge->setArg(null, '-');
- } else {
- $fontforge->setArg(null, $cmapFile);
- }
- $fontforge->setArg(null, $this->outweb . '/' . $fontname . '.ttf');
- $fontforge->execute();
- $this->addToLog($fontforge);
- }
- }
-
- public function getFonts() {
- $pdfinfo = $this->getCommandLine('pdfinfo');
- $pdfinfo->setArg('f');
- $pdfinfo->setArg(null, $this->in);
- $pdfinfo->execute();
-
- $this->addToLog($pdfinfo);
-
- /*
- * /home/extranet/www/fluidbook/docs/100376/original.pdf:
-
- PDF-1.7
- Info object (9177 0 R):
- <<
- /CreationDate (D:20110831090005+02'00')
- /Creator (Adobe InDesign CS5 \(7.0\))
- /ModDate (D:20110901144502+02'00')
- /Producer (Adobe PDF Library 9.9)
- /Trapped /False
- >>
-
- Pages: 116
-
- Retrieving info from pages 1-116...
- Fonts (201):
- 1 ( 9180 0 R): Type1 'QGNYEE+AvantGarde-Book' (9193 0 R)
- 1 ( 9180 0 R): Type1 'QGNYEE+AvantGarde-ExtraLight' (9195 0 R)
- 2 ( 1 0 R): Type1 'QGNYEE+DIN-Light' (7605 0 R)
- 2 ( 1 0 R): Type1 'QGNYEE+AvantGarde-ExtraLightObl' (7608 0 R)
- 2 ( 1 0 R): Type1 'QGNYEE+AvantGarde-MediumObl' (7601 0 R)
-
- */
-
- if (preg_match_all("|\s+\d+\s+\(\s+\d+\s+\d+\s+\w+\):\s+\w+\s+'([A-Za-z-+]+)'\s+\((\d+)\s+\d+\s+\w+\)\s|", $pdfinfo->output, $matches)) {
- foreach ($matches[1] as $k => $name) {
- $this->fonts[$matches[2][$k]] = array('name' => $name);
- }
- }
- }
-
- protected function getFontsDescriptors() {
- $pdfshow = $this->getCommandLine('pdfshow');
- $pdfshow->setArg(null, $this->in);
- $pdfshow->setManualArg(implode(' ', array_keys($this->fonts)));
- $pdfshow->execute();
- $this->doc->addToLog($pdfshow);
-
- /*
- 8677 0 obj
- <<
- /BaseFont /YIKUMO+Futura-Condensed
- /Encoding /WinAnsiEncoding
- /FirstChar 32
- /FontDescriptor 7690 0 R
- /LastChar 119
- /Subtype /Type1
- /Type /Font
- /Widths [ 205 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
- 0 0 0 0 0 0 0 0 0 414 418 0 466 0 342 0 0 0 303 0 0 561 0
- 0 388 0 0 354 0 0 0 0 0 0 0 0 0 0 0 0 0 380 380 0 380 361
- 0 382 388 180 0 395 180 590 388 375 380 0 269 276 221 389
- 0 487 ]
- >>
- endobj
-
- 9193 0 obj
- <<
- /BaseFont /QGNYEE+AvantGarde-Book
- /Encoding /WinAnsiEncoding
- /FirstChar 32
- /FontDescriptor 9192 0 R
- /LastChar 201
- /Subtype /Type1
- /ToUnicode 9185 0 R
- /Type /Font
- /Widths [ 277 0 0 0 0 0 0 0 0 0 0 0 277 332 0 0 554 554 554
- 0 0 0 554 554 0 0 0 0 0 0 0 0 0 740 574 813 744 536 485 872
- 683 226 482 0 462 919 740 869 592 0 607 498 426 655 702 0
- 0 592 480 0 0 0 0 0 0 683 0 0 0 650 0 0 0 200 0 0 200 0 610
- 655 0 0 301 388 339 0 554 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1000
- 0 0 0 0 0 0 0 0 0 0 0 0 351 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
- 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
- 0 740 0 0 0 0 0 0 0 0 536 ]
- >>
- endobj
- *
- 10 0 obj
- <<
- /BaseFont /FuturaLT-Bold--Identity-H
- /DescendantFonts [ 11 0 R ]
- /Encoding /Identity-H
- /Name /F0
- /Subtype /Type0
- /ToUnicode 3881 0 R
- /Type /Font
- >>
- endobj
-
- */
-
- $lines = explode("\n", $pdfshow->output);
- $currentFont = null;
- foreach ($lines as $line) {
- $line = trim($line);
- if ($line == '' || $line == '<<' || $line == '>>') {
- continue;
- }
- if ($line == 'endobj') {
- $currentFont = null;
- continue;
- }
- $e = explode(' ', $line);
- if (count($e) == 3 && $e[2] == 'obj') {
- $currentFont = $e[0];
- continue;
- }
-
- if (is_null($currentFont)) {
- continue;
- }
-
- $fontname = $this->fonts[$currentFont]['name'];
-
- if ($e[0] == '/ToUnicode') {
- $this->fonts[$currentFont]['cmap'] = $e[1];
- $this->cmaps[$fontname] = $e[1];
- }
-
- if ($e[0] == '/DescendantFonts') {
- $this->fonts[$currentFont]['descendant'] = $e[2];
- $this->descendants[$fontname] = $e[2];
- }
- }
-
- $this->manageDescendants();
-
- $this->addToLog(print_r($this->fonts, true));
- }
-
- protected function manageDescendants() {
- if (!count($this->descendants)) {
- return;
- }
-
- $pdfshow = $this->getCommandLine('pdfshow');
- $pdfshow->setArg(null, $this->in);
- $pdfshow->setManualArg(implode(' ', $this->descendants));
- $pdfshow->execute();
- $this->doc->addToLog($pdfshow);
-
- /*
- 11 0 obj
- <<
- /BaseFont /EHLOJB+FuturaLT-Bold-OV-JVKOJB
- /CIDSystemInfo <<
- /Ordering (Identity)
- /Registry (Adobe)
- /Supplement 0
- >>
- /DW 343
- /FontDescriptor 3878 0 R
- /Name /F0
- /Subtype /CIDFontType0
- /Type /Font
- /W [ 1 [ 781 ] 4 [ 781 ] 9 [ 677 660 ] 12 [ 752 550 550 550 ]
- 17 [ 550 ] 20 [ 542 842 803 324 ] 26 [ 324 ] 28 [ 489 ] 30
- [ 495 ] 32 [ 983 881 ] 35 [ 887 ] 43 [ 661 887 680 616 ]
- 48 [ 533 ] 50 [ 778 ] 55 [ 761 ] 57 [ 789 ] 61 [ 707 ] 63
- [ 682 ] 77 [ 682 ] 87 [ 483 ] 93 [ 343 343 ] 97 [ 682 ] 106
- [ 610 610 610 ] 111 [ 686 ] 119 [ 388 ] 121 [ 689 ] 124 [
- 686 ] 126 [ 677 ] 134 [ 661 ] 136 [ 392 ] 138 [ 302 ] 140
- [ 302 302 ] 149 [ 989 ] 154 [ 661 ] 156 [ 686 ] 159 [ 662 ]
- 167 [ 686 ] 175 [ 682 ] 180 [ 343 ] 193 [ 372 ] 196 [ 453 ]
- 199 [ 513 ] 203 [ 686 686 ] 208 [ 360 ] 210 [ 686 ] 215 [
- 686 ] 217 [ 659 ] 232 [ 686 ] ]
- >>
- endobj
-
- */
-
- $currentDescendant = null;
- $lines = explode("\n", $pdfshow->output);
- foreach ($lines as $line) {
- $line = trim($line);
- $e = explode(' ', $line);
- if (count($e) == 3 && $e[2] == "obj") {
- $currentDescendant = $e[0];
- }
- if ($line == 'endobj') {
- $currentDescendant = null;
- }
-
- if (is_null($currentDescendant)) {
- continue;
- }
-
- if ($e[0] == '/BaseFont') {
- $this->descendantNames[$currentDescendant] = trim($e[1], "/");
- }
- }
-
- $this->addToLog('Descendant names : ' . print_r($this->descendantNames, true));
- }
-
- protected function extractFonts() {
- $descriptors = $this->getUniqueId('descriptor');
-
- $pdfextract = $this->getCommandLine('pdfextract');
- $pdfextract->setArg(null, $this->in);
- $pdfextract->execute();
- $this->addToLog($pdfextract);
-
-
- $formats = array('pnm', 'pam', 'pgm', 'jpg', 'gif', 'jpeg', 'png');
- foreach ($formats as $f) {
- `rm $this->outpdf/*.$f`;
- }
- }
-
- protected function extractCmaps() {
- $cmaps = array_unique($this->cmaps);
-
- $pdfshow = $this->getCommandLine('pdfshow');
- $pdfshow->setArg(null, $this->in);
- $pdfshow->setManualArg(implode(' ', $cmaps));
- $pdfshow->execute();
- $this->addToLog($pdfshow);
-
-
- /* 5266 0 obj
- <<
- /Filter /FlateDecode
- /Length 583
- >>
- stream
- /CIDInit /ProcSet findresource begin
- 12 dict begin
- begincmap
- /CIDSystemInfo
- << /Registry (Adobe)
- /Ordering (UCS) /Supplement 0 >> def
- /CMapName /Adobe-Identity-UCS def
- /CMapType 2 def
- 1 begincodespacerange
- <0000> <FFFF>
- endcodespacerange
- 81 beginbfchar
- <0000> <0020>
- <0001> <0021>
- <0005> <0025>
- <0007> <2019>
- <0008> <0028>
- <0009> <0029>
- <000C> <002C>
- <000D> <002D>
- <000E> <002E>
- <000F> <002F>
- endbfchar
- endcmap CMapName currentdict /CMap defineresource pop end end
- endstream
- endobj
-
- */
-
- $finalCmaps = array();
-
- $lines = explode("\n", $pdfshow->output);
- $currentCmap = null;
- $inMap = false;
- foreach ($lines as $line) {
- $line = trim($line);
- if ($line == '') {
- continue;
- }
- if ($line == 'endobj') {
- $currentCmap = null;
- continue;
- } elseif ($line == 'endbfchar') {
- $inMap = false;
- }
-
- $e = explode(' ', $line);
- if (count($e) == 3 && $e[2] == 'obj') {
- $currentCmap = $e[0];
- continue;
- }
-
- if (count($e) == 2 && $e[1] == 'beginbfchar') {
- $inMap = true;
- continue;
- }
-
- if (!$inMap || is_null($currentCmap)) {
- continue;
- }
-
- $fonts = array_keys($this->cmaps, $currentCmap);
-
- $from = trim($e[0], '<>');
- $to = trim($e[1], '<>');
-
- foreach ($fonts as $fontname) {
- if (!isset($finalCmaps[$fontname])) {
- $finalCmaps[$fontname] = array();
- }
-
- if (strlen($to) >= 8) {
- $toc = $to;
- $to = $this->manageLigatures($to);
- $this->addToLog('Correct ligature ' . $toc . ' -> ' . $to);
- }
-
- $finalCmaps[$fontname][$from] = $to;
- }
- }
-
- foreach ($finalCmaps as $fontname => $cmap) {
- arsort($cmap);
- $data = '';
- foreach ($cmap as $from => $to) {
- $data.=$from . "\t" . $to . "\n";
- }
-
- $this->addToLog('Write cmap for font ' . $fontname . ' in ' . $this->outpdf . '/' . $fontname . '.cmap');
-
- file_put_contents($this->outpdf . '/' . $fontname . '.cmap', $data);
- }
- }
-
- protected function manageLigatures($code) {
- $codes = str_split($code, 4);
- $str = '';
- foreach ($codes as $c) {
- $str.=chr(intval($c, 16));
- }
-
- $ligatures = array('AE' => '00C6', 'ae' => '00E6',
- 'OE' => '0152', 'oe' => '0152',
- 'IJ' => '0132', 'ij' => '0133',
- 'ff' => 'fb00',
- 'fi' => 'fb01',
- 'fl' => 'fb02',
- 'ffi' => 'fb03',
- 'ffl' => 'fb04');
-
- // Si un couple est une ligature standard
- if (isset($ligatures[$str])) {
- return $ligatures[$str];
- }
- // Sinon, on ne retourne que le premier code
- return $codes[0];
- }
-
- protected function getUniqueId($param) {
- $res = array();
- foreach ($this->fonts as $f) {
- if (isset($f[$param])) {
- $res[$id] = $f[$param];
- }
- }
- return array_unique($res);
- }
-
- protected function clean() {
- `rm -rf $this->outpdf`;
- `rm -rf $this->outweb`;
- mkdir($this->outweb, 0777, true);
- mkdir($this->outpdf, 0777, true);
- }
-
- protected function getCommandLine($program) {
- $res = new cubeCommandLine('/usr/local/mupdf/' . $program);
- $res->cd($this->outpdf);
- $res->setPath(CONVERTER_PATH);
- return $res;
- }
-
- protected function addToLog($tolog) {
- if (!is_null($this->doc)) {
- $this->doc->addToLog($tolog);
- } elseif ($tolog instanceof cubeCommandLine) {
- echo $tolog->commande . "\n\n";
- echo $tolog->output . "\n\n";
- } else {
- echo $tolog . "\n\n";
- }
- }
-
-}
-
-?>
--- /dev/null
+<?php
+
+class wsPDFFont {
+
+ protected $id;
+ protected $name;
+ protected $descendant = null;
+ protected $extractor;
+ protected $cmap_id = null;
+ protected $descriptor_id = null;
+ protected $font_file_id = null;
+ protected static $ligatures = array('AE' => '00C6', 'ae' => '00E6',
+ 'OE' => '0152', 'oe' => '0152',
+ 'IJ' => '0132', 'ij' => '0133',
+ 'ff' => 'fb00',
+ 'fi' => 'fb01',
+ 'fl' => 'fb02',
+ 'ffi' => 'fb03',
+ 'ffl' => 'fb04');
+
+ public function __construct(&$extractor, $id, $name='') {
+ $this->extractor = $extractor;
+ $this->id = $id;
+ $this->name = $name;
+
+ $this->getInfos();
+ $this->getDescriptor();
+ }
+
+ public function getDescriptor() {
+ if (is_null($this->getDescriptorId())) {
+ return;
+ }
+ /*
+ 213 0 obj
+ <<
+ /Ascent 951
+ /CapHeight 714
+ /CharSet (/period/A/B/D/E/G/L/M/O/U)
+ /Descent -216
+ /Flags 32
+ /FontBBox [ -164 -216 1031 951 ]
+ /FontFamily (HelveticaNeue MediumCond)
+ /FontFile3 212 0 R
+ /FontName /LYOXCO+HelveticaNeue-MediumCond
+ /FontStretch /Condensed
+ /FontWeight 500
+ /ItalicAngle 0
+ /StemV 108
+ /Type /FontDescriptor
+ /XHeight 538
+ >>
+ endobj
+
+ */
+
+ $lines = $this->showObject($this->getDescriptorId());
+ foreach ($lines as $line) {
+ $e = explode(" ", $line);
+ if (strpos($e[0], "/FontFile") !== false) {
+ $this->font_file_id = $e[1];
+ }
+ }
+ }
+
+ public function getInfos() {
+ $lines = $this->showObject($this->id);
+ // Test redirection
+ while (count($lines) == 1) {
+ $e = explode(' ', $lines[0]);
+ if ($e[0] == '[' && $e[4] == ']') {
+ // Change id and get infos again
+ $this->id=$e[1];
+ return $this->getInfos();
+ } else {
+ return;
+ }
+ }
+
+
+ foreach ($lines as $line) {
+ $e = explode(' ', $line);
+ if ($line == '' || $line == '<<' || $line == '>>') {
+ continue;
+ }
+
+ if ($e[0] == '/ToUnicode') {
+ $this->cmap_id = $e[1];
+ }
+
+ if ($e[0] == '/DescendantFonts') {
+ $this->descendant = new wsPDFFont($this->extractor, $e[1]);
+ }
+
+ if ($e[0] == '/BaseFont') {
+ $this->name = ltrim($e[1], '/');
+ }
+
+ if ($e[0] == '/FontDescriptor') {
+ $this->descriptor_id = $e[1];
+ }
+ }
+
+
+
+
+ /*
+ 8677 0 obj
+ <<
+ /BaseFont /YIKUMO+Futura-Condensed
+ /Encoding /WinAnsiEncoding
+ /FirstChar 32
+ /FontDescriptor 7690 0 R
+ /LastChar 119
+ /Subtype /Type1
+ /Type /Font
+ /Widths [ 205 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
+ 0 0 0 0 0 0 0 0 0 414 418 0 466 0 342 0 0 0 303 0 0 561 0
+ 0 388 0 0 354 0 0 0 0 0 0 0 0 0 0 0 0 0 380 380 0 380 361
+ 0 382 388 180 0 395 180 590 388 375 380 0 269 276 221 389
+ 0 487 ]
+ >>
+ endobj
+
+ 9193 0 obj
+ <<
+ /BaseFont /QGNYEE+AvantGarde-Book
+ /Encoding /WinAnsiEncoding
+ /FirstChar 32
+ /FontDescriptor 9192 0 R
+ /LastChar 201
+ /Subtype /Type1
+ /ToUnicode 9185 0 R
+ /Type /Font
+ /Widths [ 277 0 0 0 0 0 0 0 0 0 0 0 277 332 0 0 554 554 554
+ 0 0 0 554 554 0 0 0 0 0 0 0 0 0 740 574 813 744 536 485 872
+ 683 226 482 0 462 919 740 869 592 0 607 498 426 655 702 0
+ 0 592 480 0 0 0 0 0 0 683 0 0 0 650 0 0 0 200 0 0 200 0 610
+ 655 0 0 301 388 339 0 554 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1000
+ 0 0 0 0 0 0 0 0 0 0 0 0 351 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
+ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
+ 0 740 0 0 0 0 0 0 0 0 536 ]
+ >>
+ endobj
+ *
+ 10 0 obj
+ <<
+ /BaseFont /FuturaLT-Bold--Identity-H
+ /DescendantFonts [ 11 0 R ]
+ /Encoding /Identity-H
+ /Name /F0
+ /Subtype /Type0
+ /ToUnicode 3881 0 R
+ /Type /Font
+ >>
+ endobj
+ *
+ 11 0 obj
+ <<
+ /BaseFont /EHLOJB+FuturaLT-Bold-OV-JVKOJB
+ /CIDSystemInfo <<
+ /Ordering (Identity)
+ /Registry (Adobe)
+ /Supplement 0
+ >>
+ /DW 343
+ /FontDescriptor 3878 0 R
+ /Name /F0
+ /Subtype /CIDFontType0
+ /Type /Font
+ /W [ 1 [ 781 ] 4 [ 781 ] 9 [ 677 660 ] 12 [ 752 550 550 550 ]
+ 17 [ 550 ] 20 [ 542 842 803 324 ] 26 [ 324 ] 28 [ 489 ] 30
+ [ 495 ] 32 [ 983 881 ] 35 [ 887 ] 43 [ 661 887 680 616 ]
+ 48 [ 533 ] 50 [ 778 ] 55 [ 761 ] 57 [ 789 ] 61 [ 707 ] 63
+ [ 682 ] 77 [ 682 ] 87 [ 483 ] 93 [ 343 343 ] 97 [ 682 ] 106
+ [ 610 610 610 ] 111 [ 686 ] 119 [ 388 ] 121 [ 689 ] 124 [
+ 686 ] 126 [ 677 ] 134 [ 661 ] 136 [ 392 ] 138 [ 302 ] 140
+ [ 302 302 ] 149 [ 989 ] 154 [ 661 ] 156 [ 686 ] 159 [ 662 ]
+ 167 [ 686 ] 175 [ 682 ] 180 [ 343 ] 193 [ 372 ] 196 [ 453 ]
+ 199 [ 513 ] 203 [ 686 686 ] 208 [ 360 ] 210 [ 686 ] 215 [
+ 686 ] 217 [ 659 ] 232 [ 686 ] ]
+ >>
+ endobj
+
+ 248 0 obj
+ [ 247 0 R ]
+ endobj */
+ }
+
+ public function showObject($obj) {
+ return $this->extractor->showObject($obj);
+ }
+
+ public function getDescriptorId() {
+ if (!is_null($this->descendant)) {
+ return $this->descendant->getDescriptorId();
+ }
+ return $this->descriptor_id;
+ }
+
+ public function getCMAP() {
+ if (is_null($this->cmap_id)) {
+ if (!is_null($this->descendant)) {
+ return $this->descendant->getCMAP();
+ }
+ return array();
+ }
+ $res = array();
+ $lines = $this->showObject($this->cmap_id);
+
+ $inMap = false;
+ foreach ($lines as $line) {
+ $line = trim($line);
+ if ($line == '') {
+ continue;
+ }
+
+ $e = explode(' ', $line);
+
+ if (count($e) == 2 && $e[1] == 'beginbfchar') {
+ $inMap = true;
+ continue;
+ }
+
+ if (!$inMap) {
+ continue;
+ }
+
+ $from = trim($e[0], '<>');
+ $to = trim($e[1], '<>');
+
+ $res[$from] = $to;
+ }
+ return true;
+ }
+
+ protected function manageLigatures($code) {
+ $codes = str_split($code, 4);
+ $str = '';
+ foreach ($codes as $c) {
+ $str.=chr(intval($c, 16));
+ }
+
+
+
+ // Si un couple est une ligature standard
+ if (isset(self::$ligatures[$str])) {
+ return self::$ligatures[$str];
+ }
+ // Sinon, on ne retourne que le premier code
+ return $codes[0];
+ }
+
+ protected function getFontFileId() {
+ if (is_null($this->font_file_id)) {
+ if (!is_null($this->descendant)) {
+ return $this->descendant->getFontFileId();
+ }
+ return null;
+ }
+ return $this->font_file_id;
+ }
+
+ public function writeTTF($dir) {
+ $cmap = $this->writeCMAP();
+ $file = $this->writeFont();
+
+ $fontforge = new cubeCommandLine('convert.pe');
+ $fontforge->setPath(CONVERTER_PATH);
+ $fontforge->setArg(null, $file);
+ $fontforge->setArg(null, $cmap);
+ $fontforge->setArg(null, $dir . '/' . $this->name . '-' . $this->id . '.ttf');
+ $fontforge->execute();
+ $this->addToLog($fontforge);
+
+ unlink($cmap);
+ unlink($file);
+ }
+
+ protected function writeFont() {
+ $tmp = cubeFiles::tempnam();
+ $this->extractor->writeObject($this->getFontFileId(), $tmp);
+ return $tmp;
+ }
+
+ protected function writeCMAP() {
+ $tmp = cubeFiles::tempnam() . '.cmap';
+ $fp = fopen($tmp, 'wb');
+ $cmap = $this->getCMAP();
+ foreach ($cmap as $from => $to) {
+ fwrite($fp, $from . "\t" . $to . "\n");
+ }
+ fclose($fp);
+ return $tmp;
+ }
+
+ protected function addToLog($toLog) {
+ $this->extractor->addToLog($toLog);
+ }
+
+}
+
+?>
--- /dev/null
+<?php
+
+class wsPDFFontExtractor {
+
+ protected $in;
+ protected $outpdf;
+ protected $outweb;
+ protected $fonts = array();
+ protected $cmaps = array();
+ protected $descendants = array();
+ protected $descendantNames = array();
+ protected $doc;
+
+ public function __construct($in, $doc=null) {
+ $this->doc = $doc;
+ $this->in = $in . '/original.pdf';
+ $this->outpdf = $in . '/fonts/pdf';
+ $this->outweb = $in . '/fonts/web';
+ }
+
+ public function extract() {
+ $this->clean();
+
+ $this->getFonts();
+ $this->convertToTTF();
+
+ $this->addToLog(print_r($this,true));
+ }
+
+ public function convertToTTF() {
+ foreach ($this->fonts as $font) {
+ $font->writeTTF($this->outpdf);
+ }
+ }
+
+ public function getFonts() {
+ $pdfinfo = $this->getCommandLine('pdfinfo');
+ $pdfinfo->setArg('f');
+ $pdfinfo->setArg(null, $this->in);
+ $pdfinfo->execute();
+
+ $this->addToLog($pdfinfo);
+
+ /*
+ * /home/extranet/www/fluidbook/docs/100376/original.pdf:
+
+ PDF-1.7
+ Info object (9177 0 R):
+ <<
+ /CreationDate (D:20110831090005+02'00')
+ /Creator (Adobe InDesign CS5 \(7.0\))
+ /ModDate (D:20110901144502+02'00')
+ /Producer (Adobe PDF Library 9.9)
+ /Trapped /False
+ >>
+
+ Pages: 116
+
+ Retrieving info from pages 1-116...
+ Fonts (201):
+ 1 ( 9180 0 R): Type1 'QGNYEE+AvantGarde-Book' (9193 0 R)
+ 1 ( 9180 0 R): Type1 'QGNYEE+AvantGarde-ExtraLight' (9195 0 R)
+ 2 ( 1 0 R): Type1 'QGNYEE+DIN-Light' (7605 0 R)
+ 2 ( 1 0 R): Type1 'QGNYEE+AvantGarde-ExtraLightObl' (7608 0 R)
+ 2 ( 1 0 R): Type1 'QGNYEE+AvantGarde-MediumObl' (7601 0 R)
+
+ */
+
+ if (preg_match_all("|\s+\d+\s+\(\s+\d+\s+\d+\s+\w+\):\s+\w+\s+'([A-Za-z-+]+)'\s+\((\d+)\s+\d+\s+\w+\)\s|", $pdfinfo->output, $matches)) {
+ foreach ($matches[1] as $k => $name) {
+ $this->fonts[$matches[2][$k]] = new wsPDFFont($this, $matches[2][$k], $name);
+ }
+ }
+ }
+
+ protected function clean() {
+ `rm -rf $this->outpdf`;
+ `rm -rf $this->outweb`;
+ mkdir($this->outweb, 0777, true);
+ mkdir($this->outpdf, 0777, true);
+ }
+
+ protected function getCommandLine($program) {
+ $res = new cubeCommandLine('/usr/local/mupdf/' . $program);
+ $res->cd($this->outpdf);
+ $res->setPath(CONVERTER_PATH);
+ return $res;
+ }
+
+ public function showObject($object, $binary=null) {
+ $pdfshow = $this->getCommandLine('pdfshow');
+ $pdfshow->setArg(null, $this->in);
+ $pdfshow->setArg(null, $object);
+ $pdfshow->execute();
+
+ $this->addToLog($pdfshow);
+
+ $obj = $pdfshow->output;
+ $obj = explode("\n", $obj);
+ $res = array();
+ foreach ($obj as $o) {
+ $o=trim($o);
+ if($o==''){
+ continue;
+ }
+ $res[] = trim($o);
+
+ }
+ array_shift($res);
+ array_pop($res);
+
+ $this->addToLog(print_r($res,true));
+
+ return $res;
+ }
+
+ public function writeObject($obj, $to) {
+ $pdfshow = new cubeCommandLine('/usr/local/mupdf/pdfshow', $to, false);
+ $pdfshow->cd($this->outpdf);
+ $pdfshow->setPath(CONVERTER_PATH);
+ $pdfshow->setArg('b');
+ $pdfshow->setArg(null, $this->in);
+ $pdfshow->setArg(null, $obj);
+ $pdfshow->execute();
+
+ $this->addToLog($pdfshow->commande);
+ }
+
+ public function addToLog($tolog) {
+ if (!is_null($this->doc)) {
+ $this->doc->addToLog($tolog);
+ } elseif ($tolog instanceof cubeCommandLine) {
+ echo $tolog->commande . "\n\n";
+ echo $tolog->output . "\n\n";
+ } else {
+ echo $tolog . "\n\n";
+ }
+ }
+
+}
+
+?>
$formats = array('ttf', 'woff', 'svg');
foreach ($this->cssFont as $font => $infos) {
+ fb($infos, $font);
$index = $infos['index'];
foreach ($formats as $format) {
$res = '<div class="' . $class . '">';
foreach ($line['groups'] as $group) {
- $res.=$this->writeGroup($group, $line, $ox);
+ $res.=$this->writeGroup($group, $line);
}
$res.='</div>';
return $res;
foreach ($this->cssFont as $font => $infos) {
$index = $infos['index'];
$style = $infos['oblique'] ? 'oblique' : 'normal';
- $res[] = "@font-face{font-family:F" . $index . ";font-style:" . $style . ";src:url('F" . $index . ".ttf') format('truetype'),url('F" . $index . ".woff') format('woff'),url('F" . $index . ".svg#" . $font . "') format('svg')}";
+ $res[] = "@font-face{font-family:F" . $index . ";src:url('F" . $index . ".ttf') format('truetype'),url('F" . $index . ".woff') format('woff'),url('F" . $index . ".svg#" . $font . "') format('svg')}";
$res[] = '.f' . $index . '{font-family:F' . $index . ',Arial,Helvetica}';
}
projects / cubeextranet.git / commitdiff