WIP Display document preview from original file

app/Http/Controllers/FileController.php

@@ -7,14 +7,13 @@ use Illuminate\Http\JsonResponse;
 use Illuminate\Support\Facades\Storage;
 use Illuminate\Http\UploadedFile;
 use GuzzleHttp\Exception\BadResponseException;
-use App\SearchDisplace\SearchAndDisplaceXML;
-use Symfony\Component\Process\Process;
+use App\SearchDisplace\Convertor\Convertor;
 
 class FileController extends Controller
 {
     public function __construct()
     {
-        $this->directoryPath = storage_path("app/contracts/");
+        $this->storage = Storage::disk('local');
     }
 
     /**
@@ -46,14 +45,13 @@ class FileController extends Controller
                 'name' => $file->getClientOriginalName()
             ]);
 
-            $originalFile = $fileId . "/{$fileId}." . $file->extension();
+            $originalFile = $fileId . "/document." . $file->extension();
 
-            Storage::disk('local')->put("contracts/{$originalFile}", file_get_contents($file)); // keep the original file
+            $this->storage->put("contracts/{$originalFile}", file_get_contents($file)); // keep the original file
 
-            $process = new Process(['soffice', '--convert-to', 'xml', $this->directoryPath . $originalFile, '--outdir', $this->directoryPath . $fileId]);
-            $process->run();
+            $xml = Convertor::convert('xml', $this->storage->path("contracts/" . $originalFile));
 
-            Storage::delete("contracts/{$originalFile}");
+            $this->storage->delete("contracts/{$originalFile}");
 
             return response()->json([
                 'id'    => $fileId,
@@ -77,14 +75,12 @@ class FileController extends Controller
     public function convert(): JsonResponse
     {
         $file = (object) request()->input('file');
-        $searchers = request()->input('searchers');
+        $xml = $this->storage->path("contracts/{$file->id}/document_sdapplied.xml");
 
-        $xml = $this->directoryPath . "{$file->id}/{$file->id}.xml";
-
-        (new SearchAndDisplaceXML($xml, $searchers, $file->type))->execute();
+        $original = Convertor::convert($file->type, $xml, true);
 
         return response()->json([
-            'path' => "tmp/{$file->id}.{$file->type}"
+            'path' => "tmp/$original"
         ]);
     }
 
@@ -95,21 +91,20 @@ class FileController extends Controller
      */
     public function download(string $path)
     {
-        return Storage::download($path);
+        return $this->storage->download($path);
     }
 
     /**
-     * Delete a file currently in progress
+     * Clear contracts and tmp directory
      *
      * @param string $id
      * @return JsonResponse
      */
     public function delete(string $id): JsonResponse
     {
-        Storage::deleteDirectory("contracts/${id}");
-
-        $tmpFiles = Storage::allFiles("tmp");
-        $success = Storage::delete($tmpFiles);
+        $this->storage->deleteDirectory("contracts/${id}");
+        $tmpFiles = $this->storage->allFiles("tmp");
+        $success = $this->storage->delete($tmpFiles);
 
         return response()->json(['success' => $success]);
     }

app/Http/Controllers/SearchAndDisplaceController.php

@@ -3,10 +3,18 @@
 namespace App\Http\Controllers;
 
 use App\SearchDisplace\Documents\DocumentFile;
-use App\SearchDisplace\SearchAndDisplaceJSON;
+use Illuminate\Support\Facades\Storage;
+use App\SearchDisplace\SearchAndDisplaceXML;
 
 class SearchAndDisplaceController extends Controller
 {
+    protected $storage;
+
+    public function __construct()
+    {
+        $this->storage = Storage::disk('local');        
+    }
+    
     public function show($id)
     {
         $handler = new DocumentFile();
@@ -14,9 +22,10 @@ class SearchAndDisplaceController extends Controller
         try {
             $result = $handler->getAfterIngest($id);
 
-            // if ($result['status'] !== 'processing') {
-            //     $handler->destroy($id);
-            // }
+            // remove HTML file after receiving it's content
+            if ($result['status'] !== 'processing') {
+                $this->storage->delete("contracts/$id/document.html");
+            }
 
             return response()->json($result, 200);
         } catch (\Exception $exception) {
@@ -29,7 +38,7 @@ class SearchAndDisplaceController extends Controller
     public function store()
     {
         request()->validate([
-            'file' => 'required', // String or file.
+            'file' => 'required', // file.
             'searchers' => 'required|array',
             'searchers.*.key' => 'required',
             'searchers.*.type' => 'required|in:replace,displace',
@@ -37,16 +46,23 @@ class SearchAndDisplaceController extends Controller
             'searchOnly' => 'nullable|boolean'
         ]);
 
+        $file = request()->input('file');
         $searchOnly = request()->input('searchOnly') ?? false;
 
-        $searchAndDisplace = new SearchAndDisplaceJSON(
-            request()->input('file'),
-            request()->input('searchers'),
-            $searchOnly
-        );
+        $changes = (new SearchAndDisplaceXML($file, request()->input('searchers'), $searchOnly))->execute();
 
         try {
-            return response()->json($searchAndDisplace->execute(), 200);
+            $processedFile = "contracts/$file/document_sdapplied.html";
+
+            if($this->storage->exists($processedFile)) {
+                $processedFileContent = $this->storage->get($processedFile);
+                $this->storage->delete($processedFile);
+
+                return response()->json([
+                    'content' => $processedFileContent,
+                    'indexes' => []
+                ], 200);
+            }
         } catch (\Exception $exception) {
             return response()->json([
                 'message' => $exception->getMessage(),

app/SearchDisplace/Convertor/Convertor.php

@@ -0,0 +1,35 @@
+<?php
+
+namespace App\SearchDisplace\Convertor;
+
+use Symfony\Component\Process\Process;
+
+/**
+ * Convert documents from formats supported by Libre Office
+ */
+
+class Convertor {
+    /**
+     * @param $to desired file format
+     * @param $document absolute file path
+     * @param $tmp - if true file will be saved to tmp directory for download
+     * 
+     * @return string $path to converted file
+     */
+    public static function convert($to, $document, $tmp = false)
+    {
+        $path = pathinfo($document);
+        $dir = $path['dirname'];
+        $original = $dir . '/' . $path['basename'];
+        if(!$tmp) {
+            $folder = $dir . '/';
+        } else {
+            $folder = storage_path('app/tmp/');
+        }
+
+        $process = new Process(['soffice', '--convert-to', $to, $original, '--outdir', $folder]);
+        $process->run();
+
+        return $path['filename'] . '.' . $to;
+    }
+}

app/SearchDisplace/Documents/DocumentFile.php

@@ -3,6 +3,7 @@
 namespace App\SearchDisplace\Documents;
 
 use Illuminate\Support\Facades\Storage;
+use DOMDocument;
 
 class DocumentFile
 {
@@ -18,10 +19,10 @@ class DocumentFile
         $path = $this->getPath($id);
 
         // Ingest success.
-        if ($this->storage->exists("$path/document.json")) {
+        if ($this->storage->exists("$path/document.html")) {
             return [
                 'status' => 'success',
-                'content' => $this->getDocumentContent($path),
+                'content' => $this->getDocumentContent($path, $id),
             ];
         }
 
@@ -50,29 +51,27 @@ class DocumentFile
         return "contracts/$id";
     }
 
-    protected function getDocumentContent($path)
+    protected function getDocumentContent($path, $id)
     {
-        $content = json_decode($this->storage->get("$path/document.json"));
+        $document = $this->storage->path($path) . '/document.html';
 
-        return $this->convertToHTML($content);
+        $this->updateImagesPath($document, $id);
+
+        return file_get_contents($document);
     }
 
-    protected function convertToHTML($elements)
+    public static function updateImagesPath($document, $id)
     {
-        $html = '';
-        $url = url('/') . '/contracts-images';
-        foreach($elements as $key => $element) {
-            if($element->tag !== 'img') {
-                $html .= "<$element->tag style=\"$element->style\">$element->content</$element->tag>";
-            } else {
-                $src = $url . '/' . str_replace(' ', '%20', $element->src);
-                $html .= "<img $element->style src=\"$src\" alt=\"$element->details\">";
-            }
-
-            if($key !== array_key_last($elements))
-                $html .= '<br>';
+        $html = new DOMDocument();
+        $html->loadHTMLFile($document);
+        $url = url('/') . "/contracts-images/$id/";
+
+        foreach($html->getElementsByTagName('img') as $image) {
+            $src = $image->getAttribute('src');
+
+            $image->setAttribute('src', $url . $src);
         }
 
-        return $html;
+        $html->saveHTMLFile($document);
     }
 }

app/SearchDisplace/Ingest/HandleReceivedDocument.php

@@ -35,7 +35,7 @@ class HandleReceivedDocument
      */
     public function handle()
     {
-        if ($this->fileResultType === 'json') {
+        if ($this->fileResultType === 'html') {
             $this->handleDocument();
 
             return;
@@ -58,7 +58,7 @@ class HandleReceivedDocument
 
         // The .md extension signals the success status, the lack of signals the fail status.
         if ($this->status === 'success') {
-            $fileName = $fileName . '.json';
+            $fileName = $fileName . '.html';
         }
 
         $dir = "contracts/$this->id";

app/SearchDisplace/Ingest/SendDocument.php

@@ -23,10 +23,10 @@ class SendDocument
      * @param string $fileResultType
      * @throws \Exception
      */
-    public function execute($id, $document, string $fileResultType = 'json')
+    public function execute($id, $document, string $fileResultType = 'html')
     {
         try {
-            if ( ! in_array($fileResultType, ['md', 'original', 'json'])) {
+            if ($fileResultType !== 'html') {
                 throw new \Exception('Invalid file result type provided.');
             }
 

app/SearchDisplace/SearchAndDisplace.php

@@ -8,10 +8,10 @@ class SearchAndDisplace
 {
     protected $documentContent;
     protected $info;
-    protected $searchOnly = false;
+    protected $searchOnly;
     protected $isDocument = false;
 
-    public function __construct($documentContent, $info, $searchOnly = false, $isDocument = false)
+    public function __construct($documentContent, $info, $searchOnly, $isDocument = false)
     {
         $this->documentContent = $documentContent;
         $this->info = $info;

app/SearchDisplace/SearchAndDisplaceJSON.php

@@ -1,98 +0,0 @@
-<?php
-
-namespace App\SearchDisplace;
-
-use Illuminate\Support\Facades\Storage;
-use App\SearchDisplace\Output\HtmlOutput;
-
-class SearchAndDisplaceJSON
-{
-    protected $file;
-    protected $searchers;
-    protected $searchOnly;
-    protected $content;
-
-    public function __construct($file, $searchers, $searchOnly)
-    {
-        $searchOnly ? $this->content = $file : $this->file = $file . '/document.json';
-        $this->searchers = $searchers;
-
-        $this->storage = Storage::disk('local');
-    }
-
-    public function execute()
-    {
-        if(isset($this->content)) {
-            $search = new SearchAndDisplace(
-                        stripslashes($this->content),
-                        [
-                            'searchers' => $this->searchers,
-                        ],
-                        true
-                    );
-
-            return $search->execute();
-        }
-
-        if(! $this->storage->exists("contracts/$this->file")) {
-            return;
-        }
-
-        try {
-            $content = $this->getContent();
-        } catch (\Exception $exception) {
-            \Illuminate\Support\Facades\Log::info('EXCEPTION: ' . $exception->getMessage());
-            
-            return;
-        }
-
-        $searchDisplace = $this->applySD($content);
-
-        $convert = new HtmlOutput($searchDisplace);
-
-        return $convert->getData();
-    }
-
-    protected function getContent()
-    {
-        return json_decode($this->storage->get("contracts/$this->file"));
-    }
-
-    protected function applySD($elements)
-    {
-        $indexes = [];
-
-        foreach($elements as $element) {
-            if($element->tag === 'img')
-                continue;
-
-            $search = new SearchAndDisplace(
-                stripslashes($element->content),
-                [
-                    'searchers' => $this->searchers,
-                ],
-                false,
-                true
-            );
-    
-            $changed = $search->execute();
-
-            if($changed) {
-                foreach($changed['indexes'] as $key => $searcher) {
-                    foreach($searcher as $change) {
-                        if($change['start']) {
-                            $indexes[$key][] = $change;
-                            $element->paragraph = $key;
-                        }
-                    }
-                }
-                $element->content = $changed['content'];
-            }
-        }
-
-        return [
-            'content' => $elements,
-            'indexes' => $indexes
-        ];
-    }
-}

app/SearchDisplace/SearchAndDisplaceXML.php

@@ -2,57 +2,132 @@
 
 namespace App\SearchDisplace;
 
-use Symfony\Component\Process\Process;
+use App\SearchDisplace\Documents\DocumentFile;
+use Illuminate\Support\Facades\Storage;
+use App\SearchDisplace\Convertor\Convertor;
 
 class SearchAndDisplaceXML
 {
     protected $file;
     protected $searchers;
-    protected $type;
+    protected $storage;
+    protected $searchOnly;
 
-    public function __construct($file, $searchers, $type = 'odt')
+    public function __construct($file, $searchers, $searchOnly)
     {
-        $this->file = $file;
+        $this->fileDirectory = $file;
         $this->searchers = $searchers;
-        $this->type = $type;
+        $this->storage = Storage::disk('local'); 
+        $this->searchOnly = $searchOnly; 
     }
 
     public function execute()
     {
-        $this->applySD();
+        $sdXML = $this->applySD();
+        $pathinfo = pathinfo($sdXML);
 
-        $this->convertToOriginalFileType();
+        $this->convertSearchDisplacedXMLToHTML($sdXML);
+
+        DocumentFile::updateImagesPath($pathinfo['dirname'] . '/document_sdapplied.html', $this->fileDirectory);
+
+        return $pathinfo['filename'];
     }
     
-    protected function convertToOriginalFileType()
+    /**
+     * Convert (Search displaced) XML to HTML for browser preview
+     * 
+     * @return void
+     */
+    protected function convertSearchDisplacedXMLToHTML($file)
     {
-        (new Process(['soffice', '--convert-to', $this->type, $this->file, '--outdir', storage_path('app/tmp/')]))->run();
+        Convertor::convert('html', $file);
     }
 
+    /**
+     * Read XML document and send text contents to SD
+     * 
+     * @return void
+     */
     protected function applySD()
     {
         $dom = new \DOMDocument();
-        $dom->load($this->file);
-
-        foreach($dom->getElementsByTagName('span') as $p) {
-            $search = new SearchAndDisplace(
-                stripslashes($p->textContent),
-                [
-                    'searchers' => $this->searchers,
-                ],
-                false,
-                true
-            );
-    
-            $changed = $search->execute();
+        $filePath = $this->storage->path("contracts/$this->fileDirectory");
+        $dom->load($filePath . "/document.xml");
 
-            if(!$changed) {
-                continue;
-            }
+        foreach($dom->getElementsByTagName('p') as $p) {
+            if(count($p->childNodes) > 0 && isset($p->parentNode->tagName) && $p->parentNode->tagName !== "table:table-cell") {
+                foreach($p->childNodes as $child) {
+                    if(isset($child->tagName) && $child->tagName === "text:span") {
+                        $content = trim($child->textContent);
+                        if($content == '') {
+                            continue;
+                        }
 
-            $p->textContent = $changed['content'];
+                        $this->replace($content, $child, $dom);
+                    }
+                }
+            } else {
+                $content = trim($p->textContent);
+                if($content == '') {
+                    continue;
+                }
+
+                $this->replace($content, $p, $dom);
+            }
         }
 
-        $dom->save($this->file);
+        $dom->save($filePath . "/document_sdapplied.xml");
+
+        return $filePath . "/document_sdapplied.xml";
+    }
+
+    /**
+     * Apply SD on document's paragraph
+     * 
+     * @param string $content paragraph content
+     * @param $element DOM element
+     * 
+     * @return void
+     */
+    protected function replace($content, $element, $dom) {
+        $search = new SearchAndDisplace(
+                    stripslashes($content),
+                    [
+                        'searchers' => $this->searchers,
+                    ],
+                    $this->searchOnly,
+                    true
+                );
+
+        $changed = $search->execute();
+
+        if($changed) {
+            if($this->searchOnly) {
+                $content = $element->textContent;
+                $indexes = $changed;
+            } else {
+                $content = $changed['content'];
+                $indexes = $changed['indexes'];
+            }
+
+            foreach(array_keys($indexes) as $searcher) {
+                if(empty($indexes[$searcher])) {
+                    continue;
+                }
+
+                foreach($indexes[$searcher] as $change) {
+                    $first = substr($content, 0, $change['start']);
+                    $changed = "<mark>" . substr($content, $change['start'], $change['end'] - $change['start'] + 1) . "</mark>";
+                    $last = substr($content, $change['end'] + 1);
+        
+                    $element->textContent = $first;
+                    $changed = $dom->createElement("text:span", $changed);
+                    $last = $dom->createElement("text:span", $last);
+                    // $changed->setAttribute('style', 'background-color: red');
+                    $element->appendChild($changed);
+                    $element->appendChild($last);
+                }
+            }
+        }
     }
 }

resources/js/components/ProcessFile/ProcessFile.ts

@@ -1,4 +1,3 @@
-import marked from 'marked';
 import {Vue, Component, Prop, Watch} from 'vue-property-decorator';
 import {FileData} from '@/interfaces/FileData';
 import { eventBus } from '@/app';
@@ -357,7 +356,6 @@ export default class ProcessFile extends Vue {
 
     private async runSearchersWithoutDisplacing() {
         this.processing = true;
-        this.processedFileContent = '';
 
         let searchers: Array<{ key: string; type: string; value: string; }> = [];
 
@@ -370,10 +368,10 @@ export default class ProcessFile extends Vue {
         });
 
         try {
-            const response = await this.$api.filterDocument(this.fileContent, searchers, true);
+            const response = await this.$api.filterDocument(this.file.id, searchers, true);
 
-            this.processedFileContent = this.fileContent;
-            this.documentDiffIndexes = response;
+            this.processedFileContent = response.content;
+            // this.documentDiffIndexes = response;
             this.createDiffPreview();
 
             this.processing = false;