From: Vincent Vanwaelscappel <vincent@cubedesigners.com>
Date: Thu, 21 Dec 2023 18:42:09 +0000 (+0100)
Subject: wip #6598 @2
X-Git-Url: http://git.cubedesigners.com/?a=commitdiff_plain;h=29b603e8654084d424a2e584593775a4573c20f7;p=fluidbook_processfarm.git

wip #6598 @2
---

diff --git a/Dockerfile b/Dockerfile
index 333570a..dbdc28f 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -9,7 +9,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends gnupg \
     && apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 4F4EA0AAE5267A6C \
     && apt-get update \
     && apt-get -y --no-install-recommends install x11-common pdftk poppler-utils ghostscript netpbm sshfs mupdf-tools php8.2-cli php8.2-fpm php8.2-curl php8.2-exif php8.2-xml php8.2-gd php8.2-ssh php8.2-zip libjpeg-turbo-progs zip unzip \
-        less nano wget bash lynx
+        less nano wget bash lynx rsync
 
 COPY --from=composer:2 /usr/bin/composer /usr/bin/composer
 COPY --chmod=755 startup /usr/bin/startup
diff --git a/app/process.php b/app/process.php
index 8db4271..87eb1ab 100644
--- a/app/process.php
+++ b/app/process.php
@@ -1,29 +1,55 @@
 <?php
 
+use Fluidbook\Farmer\PDFOperations;
 use Fluidbook\Farmer\ProcessFile;
 use Fluidbook\Farmer\ProcessToolboxDocumentFile;
 use Fluidbook\Farmer\ProcessToolboxPDFFile;
 
 require_once __DIR__ . "/vendor/autoload.php";
 
+if (isset($argv)) {
+    parse_str(implode('&', array_slice($argv, 1)), $_POST);
+}
+
 if (!isset($_POST['quality'])) {
     $_POST['quality'] = 85;
 }
 
+if (!isset($_POST['operation'])) {
+    $_POST['operation'] = 'unknown';
+}
+
 try {
-    if (!isset($_POST['pdf'])) {
-        if (isset($_POST['toolbox'])) {
-            $p = new ProcessToolboxDocumentFile($_POST['out'], $_POST['page'], $_POST['resolutionRatio'], $_POST['mobileRatio'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
+    if ($_POST['operation'] === 'fixpdf') {
+        $res = PDFOperations::fixPDF($_POST['pdf'], $_POST['out']);
+    } else if ($_POST['operation'] === 'cutpdf') {
+        $res = PDFOperations::cutPDF($_POST['pdf'], $_POST['mode'], $_POST['out']);
+    } else if ($_POST['operation'] === 'trimpdf') {
+        $res = PDFOperations::trimPDF($_POST['pdf'], $_POST['out']);
+    } else if ($_POST['operation'] === 'splitpdf') {
+        $res = PDFOperations::splitPDF($_POST['pdf'], $_POST['out']);
+    } else if ($_POST['operation'] === 'extractlinks') {
+        $res = PDFOperations::extractLinks($_POST['pdf'], $_POST['out']);
+    } else if ($_POST['operation'] === 'extracttexts') {
+        $res = PDFOperations::extractTexts($_POST['pdf'], $_POST['out'], $_POST['mode'] ?? 'standard', $_POST['method'] ?? 'fluidbook', $_POST['ignoreseparators'] ?? '');
+    } else if ($_POST['operation'] === 'extracthightlightsdata') {
+        $res = PDFOperations::extractHighlightsData($_POST['pdf'], $_POST['out'], $_POST['mode'], $_POST['ignoreseparators']);
+    } else {
+        if (!isset($_POST['pdf'])) {
+            if (isset($_POST['toolbox'])) {
+                $p = new ProcessToolboxDocumentFile($_POST['out'], $_POST['page'], $_POST['resolutionRatio'], $_POST['mobileRatio'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
+            } else {
+                $p = new ProcessFile($_POST['out'], $_POST['page'], $_POST['resolutionRatio'], $_POST['mobileRatio'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
+            }
         } else {
-            $p = new ProcessFile($_POST['out'], $_POST['page'], $_POST['resolutionRatio'], $_POST['mobileRatio'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
+            $p = new ProcessToolboxPDFFile($_POST['pdf'], $_POST['out'], $_POST['page'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
         }
-    } else {
-        $p = new ProcessToolboxPDFFile($_POST['pdf'], $_POST['out'], $_POST['page'], $_POST['format'], $_POST['resolution'], (int)$_POST['quality'], $_POST['withGraphics'], $_POST['withText'], $_POST['version'], $_POST['force']);
+        $res = $p->process();
     }
-    $res = $p->process();
-    ob_end_clean();
+
+    @ob_end_clean();
     die($res);
 } catch (Exception $e) {
-    ob_end_clean();
-    die('!!'.$e->getMessage());
+    @ob_end_clean();
+    die('!!' . $e->getMessage());
 }
\ No newline at end of file
diff --git a/app/src/Buffer.php b/app/src/Buffer.php
new file mode 100644
index 0000000..a660109
--- /dev/null
+++ b/app/src/Buffer.php
@@ -0,0 +1,54 @@
+<?php
+
+namespace Fluidbook\Farmer;
+
+use Cubist\Util\CommandLine\Rsync;
+use Cubist\Util\Files\Files;
+
+class Buffer
+{
+    protected static $_out = [];
+
+    public static function getOut($path)
+    {
+        self::getDirAndFile($path, $dir, $file);
+        if (!isset(self::$_out[$path])) {
+            self::$_out[$path] = ['local' => Files::mkdir('/tmp/out/' . sha1($path)), 'dir' => $dir];
+        }
+        return self::$_out[$path]['local'] . $file;
+    }
+
+    public static function getIn($path, $force = false)
+    {
+        self::getDirAndFile($path, $dir, $file);
+        if ($file) {
+            $e = explode('.', $file);
+            $ext = array_pop($e);
+        }
+        $in = Files::mkdir('/tmp/in/' . sha1($path)) . sha1($path) . '.' . $ext;
+        if (file_exists($path)) {
+            copy($path, $in);
+        }
+        return $in;
+    }
+
+    public static function syncOut()
+    {
+        foreach (self::$_out as $o) {
+            $rsync = new Rsync($o['local'], $o['dir']);
+            $rsync->execute();
+        }
+    }
+
+    protected static function getDirAndFile($path, &$dir, &$file)
+    {
+        $path = rtrim($path, '/');
+        $e = explode('/', $path);
+        $last = $e[count($e) - 1];
+        $file = '';
+        if (stristr($last, '.')) {
+            $file = array_pop($e);
+        }
+        $dir = implode('/', $e);
+    }
+}
\ No newline at end of file
diff --git a/app/src/PDFOperations.php b/app/src/PDFOperations.php
new file mode 100644
index 0000000..567f23c
--- /dev/null
+++ b/app/src/PDFOperations.php
@@ -0,0 +1,75 @@
+<?php
+
+namespace Fluidbook\Farmer;
+
+use Cubist\PDF\CommandLine\FWSTK;
+use Cubist\PDF\PDFTools;
+
+class PDFOperations
+{
+    public static function fixPDF($pdf, $out)
+    {
+        PDFTools::fixPDF(Buffer::getIn($pdf), Buffer::getOut($out));
+        Buffer::syncOut();
+        return $out;
+    }
+
+    public static function cutPDF($pdf, $mode, $out)
+    {
+        $fwstk = new FWSTK();
+        $fwstk->setArg('--input ' . Buffer::getIn($pdf));
+        $fwstk->setArg('--cut ' . $mode);
+        $fwstk->setArg('--output ' . Buffer::getOut($out));
+        $fwstk->execute();
+
+        Buffer::syncOut();
+
+        return $out;
+    }
+
+    public static function trimPDF($pdf, $out)
+    {
+        $fwstk = new FWSTK();
+        $fwstk->setArg('--input ' . Buffer::getIn($pdf));
+        $fwstk->setArg('--trim');
+        $fwstk->setArg('--output ' . Buffer::getOut($out));
+        $fwstk->execute();
+
+        Buffer::syncOut();
+
+        return $out;
+    }
+
+    public static function splitPDF($pdf, $out)
+    {
+        PDFTools::split(Buffer::getIn($pdf), Buffer::getOut($out));
+        Buffer::syncOut();
+        return $out;
+    }
+
+    /**
+     * @throws \Exception
+     */
+    public static function extractTexts($pdf, $out, $mode, $method, $ignoreSeparators)
+    {
+        PDFTools::extractTexts(Buffer::getIn($pdf), $out, $mode, $method, $ignoreSeparators);
+        Buffer::syncOut();
+        return $out;
+    }
+
+
+    public static function extractHighlightsData($pdf, $out, $mode, $ignoreSeparators)
+    {
+        PDFTools::extractHighlightsData(Buffer::getIn($pdf), $out, $mode, $ignoreSeparators);
+        Buffer::syncOut();
+        return $out;
+    }
+
+    public static function extractLinks($pdf, $out)
+    {
+        PDFTools::extractLinks(Buffer::getIn($pdf), $out);
+        Buffer::syncOut();
+        return $out;
+    }
+
+}
\ No newline at end of file
diff --git a/app/src/ProcessFile.php b/app/src/ProcessFile.php
index ca89658..6a252bd 100644
--- a/app/src/ProcessFile.php
+++ b/app/src/ProcessFile.php
@@ -113,10 +113,12 @@ class ProcessFile extends \Fluidbook\Tools\Jobs\ProcessFile
     public function getOut()
     {
         if ($this->getRegion() === 'UE') {
-            return str_replace(self::DISTANT_ROOT, self::LOCAL_ROOT, $this->getDistantOut());
+            $res = str_replace(self::DISTANT_ROOT, self::LOCAL_ROOT, $this->getDistantOut());
         } else {
-            return str_replace(self::US_DISTANT_ROOT, self::US_LOCAL_ROOT, $this->getDistantOut());
+            $res = str_replace(self::US_DISTANT_ROOT, self::US_LOCAL_ROOT, $this->getDistantOut());
         }
+
+        return Buffer::getOut($res);
     }
 
     /**
@@ -146,10 +148,12 @@ class ProcessFile extends \Fluidbook\Tools\Jobs\ProcessFile
     public function process()
     {
         if ($this->getRegion() === 'UE') {
-            return str_replace(self::LOCAL_ROOT, self::DISTANT_ROOT, $this->getPath($this->isForce()));
+            $res = str_replace(self::LOCAL_ROOT, self::DISTANT_ROOT, $this->getPath($this->isForce()));
         } else {
-            return str_replace(self::US_LOCAL_ROOT, self::US_DISTANT_ROOT, $this->getPath($this->isForce()));
+            $res = str_replace(self::US_LOCAL_ROOT, self::US_DISTANT_ROOT, $this->getPath($this->isForce()));
         }
+        Buffer::syncOut();
+        return $res;
     }
 
     public function getPdf(): string|null
@@ -185,8 +189,10 @@ class ProcessFile extends \Fluidbook\Tools\Jobs\ProcessFile
             if ($this->getPage() > $this->getPDFPagesNumber()) {
                 throw new \Exception('PDF document has ' . $this->getPDFPagesNumber() . ' pages and you are requesting the page ' . $this->getPage());
             }
-            PDFTools::split($this->getPdf(), $this->getOut() . '/pdf');
+            PDFOperations::splitPDF($this->getPdf(), $this->getOut() . '/pdf');
         }
-        return $res;
+        return Buffer::getIn($res);
     }
+
+
 }
\ No newline at end of file