研学线路

app/Support/DocTextExtractor.php

@@ -3,6 +3,13 @@
 namespace App\Support;
 
 use Illuminate\Support\Str;
+use PhpOffice\PhpWord\Element\AbstractContainer;
+use PhpOffice\PhpWord\Element\Cell;
+use PhpOffice\PhpWord\Element\Row;
+use PhpOffice\PhpWord\Element\Table;
+use PhpOffice\PhpWord\Element\Text;
+use PhpOffice\PhpWord\Element\TextBreak;
+use PhpOffice\PhpWord\IOFactory;
 use RuntimeException;
 use Symfony\Component\Process\Process;
 use ZipArchive;
@@ -83,6 +90,11 @@ class DocTextExtractor
 
     public static function extractDoc(string $path): string
     {
+        $phpWordText = self::extractDocViaPhpWord($path);
+        if (is_string($phpWordText) && trim($phpWordText) !== '') {
+            return $phpWordText;
+        }
+
         $textutil = self::resolveBinary('textutil');
         if ($textutil !== null) {
             $out = self::runCommand([$textutil, '-convert', 'txt', '-stdout', $path], 45);
@@ -121,7 +133,96 @@ class DocTextExtractor
             @rmdir($tmpDir);
         }
 
-        throw new RuntimeException('无法解析 .doc 文件，请安装 LibreOffice（soffice），或改用 .docx 格式');
+        throw new RuntimeException('无法解析 .doc 申报表，请检查文件是否损坏，或另存为 .docx 后重试');
+    }
+
+    private static function extractDocViaPhpWord(string $path): ?string
+    {
+        try {
+            $phpWord = IOFactory::load($path, 'MsDoc');
+        } catch (\Throwable) {
+            return null;
+        }
+
+        $lines = [];
+        foreach ($phpWord->getSections() as $section) {
+            foreach ($section->getElements() as $element) {
+                $chunk = self::extractPhpWordElementText($element);
+                if ($chunk === '') {
+                    continue;
+                }
+                foreach (preg_split('/\R/u', $chunk) ?: [] as $line) {
+                    $line = self::normalizeLine((string) $line);
+                    if ($line !== '') {
+                        $lines[] = $line;
+                    }
+                }
+            }
+        }
+
+        $text = self::joinLines($lines);
+
+        return trim($text) !== '' ? $text : null;
+    }
+
+    private static function extractPhpWordElementText(object $element): string
+    {
+        if ($element instanceof Text) {
+            return (string) $element->getText();
+        }
+        if ($element instanceof TextBreak) {
+            return "\n";
+        }
+        if ($element instanceof Table) {
+            $parts = [];
+            foreach ($element->getRows() as $row) {
+                if ($row instanceof Row) {
+                    $parts[] = self::extractPhpWordRowText($row);
+                }
+            }
+
+            return implode("\n", array_filter($parts, fn ($part) => $part !== ''));
+        }
+        if ($element instanceof AbstractContainer) {
+            $parts = [];
+            foreach ($element->getElements() as $child) {
+                $parts[] = self::extractPhpWordElementText($child);
+            }
+
+            return implode('', $parts);
+        }
+        if (method_exists($element, 'getText')) {
+            $text = $element->getText();
+
+            return is_string($text) ? $text : '';
+        }
+
+        return '';
+    }
+
+    private static function extractPhpWordRowText(Row $row): string
+    {
+        $cells = [];
+        foreach ($row->getCells() as $cell) {
+            if ($cell instanceof Cell) {
+                $cells[] = self::extractPhpWordCellText($cell);
+            }
+        }
+
+        return implode("\n", array_values(array_filter(
+            array_map(fn ($cellText) => self::normalizeLine($cellText), $cells),
+            fn ($cellText) => $cellText !== ''
+        )));
+    }
+
+    private static function extractPhpWordCellText(Cell $cell): string
+    {
+        $parts = [];
+        foreach ($cell->getElements() as $element) {
+            $parts[] = self::extractPhpWordElementText($element);
+        }
+
+        return self::normalizeLine(implode('', $parts));
     }
 
     private static function textFromNode(\DOMXPath $xpath, \DOMNode $node): string

composer.json

@@ -11,7 +11,8 @@
         "laravel/sanctum": "^3.3",
         "laravel/tinker": "^2.8",
         "overtrue/pinyin": "^5.0",
-        "phpoffice/phpspreadsheet": "^5.3"
+        "phpoffice/phpspreadsheet": "^5.3",
+        "phpoffice/phpword": "^1.3"
     },
     "require-dev": {
         "fakerphp/faker": "^1.9.1",

composer.lock

@@ -4,7 +4,7 @@
         "Read more about it at https://getcomposer.org/doc/01-basic-usage.md#installing-dependencies",
         "This file is @generated automatically"
     ],
-    "content-hash": "2fd1262c3117ca0367f247bc6998ea9b",
+    "content-hash": "1a85fb70206d510a41d0911211e84521",
     "packages": [
         {
             "name": "brick/math",
@@ -2938,6 +2938,64 @@
             ],
             "time": "2025-03-16T02:16:27+00:00"
         },
+        {
+            "name": "phpoffice/math",
+            "version": "0.2.0",
+            "source": {
+                "type": "git",
+                "url": "https://github.com/PHPOffice/Math.git",
+                "reference": "fc2eb6d1a61b058d5dac77197059db30ee3c8329"
+            },
+            "dist": {
+                "type": "zip",
+                "url": "https://api.github.com/repos/PHPOffice/Math/zipball/fc2eb6d1a61b058d5dac77197059db30ee3c8329",
+                "reference": "fc2eb6d1a61b058d5dac77197059db30ee3c8329",
+                "shasum": "",
+                "mirrors": [
+                    {
+                        "url": "https://mirrors.aliyun.com/composer/dists/%package%/%reference%.%type%",
+                        "preferred": true
+                    }
+                ]
+            },
+            "require": {
+                "ext-dom": "*",
+                "ext-xml": "*",
+                "php": "^7.1|^8.0"
+            },
+            "require-dev": {
+                "phpstan/phpstan": "^0.12.88 || ^1.0.0",
+                "phpunit/phpunit": "^7.0 || ^9.0"
+            },
+            "type": "library",
+            "autoload": {
+                "psr-4": {
+                    "PhpOffice\\Math\\": "src/Math/"
+                }
+            },
+            "notification-url": "https://packagist.org/downloads/",
+            "license": [
+                "MIT"
+            ],
+            "authors": [
+                {
+                    "name": "Progi1984",
+                    "homepage": "https://lefevre.dev"
+                }
+            ],
+            "description": "Math - Manipulate Math Formula",
+            "homepage": "https://phpoffice.github.io/Math/",
+            "keywords": [
+                "MathML",
+                "officemathml",
+                "php"
+            ],
+            "support": {
+                "issues": "https://github.com/PHPOffice/Math/issues",
+                "source": "https://github.com/PHPOffice/Math/tree/0.2.0"
+            },
+            "time": "2024-08-12T07:30:45+00:00"
+        },
         {
             "name": "phpoffice/phpspreadsheet",
             "version": "5.3.0",
@@ -3050,6 +3108,121 @@
             },
             "time": "2025-11-24T15:47:10+00:00"
         },
+        {
+            "name": "phpoffice/phpword",
+            "version": "1.3.0",
+            "source": {
+                "type": "git",
+                "url": "https://github.com/PHPOffice/PHPWord.git",
+                "reference": "8392134ce4b5dba65130ba956231a1602b848b7f"
+            },
+            "dist": {
+                "type": "zip",
+                "url": "https://api.github.com/repos/PHPOffice/PHPWord/zipball/8392134ce4b5dba65130ba956231a1602b848b7f",
+                "reference": "8392134ce4b5dba65130ba956231a1602b848b7f",
+                "shasum": "",
+                "mirrors": [
+                    {
+                        "url": "https://mirrors.aliyun.com/composer/dists/%package%/%reference%.%type%",
+                        "preferred": true
+                    }
+                ]
+            },
+            "require": {
+                "ext-dom": "*",
+                "ext-json": "*",
+                "ext-xml": "*",
+                "php": "^7.1|^8.0",
+                "phpoffice/math": "^0.2"
+            },
+            "require-dev": {
+                "dompdf/dompdf": "^2.0",
+                "ext-gd": "*",
+                "ext-libxml": "*",
+                "ext-zip": "*",
+                "friendsofphp/php-cs-fixer": "^3.3",
+                "mpdf/mpdf": "^8.1",
+                "phpmd/phpmd": "^2.13",
+                "phpstan/phpstan-phpunit": "@stable",
+                "phpunit/phpunit": ">=7.0",
+                "symfony/process": "^4.4 || ^5.0",
+                "tecnickcom/tcpdf": "^6.5"
+            },
+            "suggest": {
+                "dompdf/dompdf": "Allows writing PDF",
+                "ext-gd2": "Allows adding images",
+                "ext-xmlwriter": "Allows writing OOXML and ODF",
+                "ext-xsl": "Allows applying XSL style sheet to headers, to main document part, and to footers of an OOXML template",
+                "ext-zip": "Allows writing OOXML and ODF"
+            },
+            "type": "library",
+            "autoload": {
+                "psr-4": {
+                    "PhpOffice\\PhpWord\\": "src/PhpWord"
+                }
+            },
+            "notification-url": "https://packagist.org/downloads/",
+            "license": [
+                "LGPL-3.0"
+            ],
+            "authors": [
+                {
+                    "name": "Mark Baker"
+                },
+                {
+                    "name": "Gabriel Bull",
+                    "email": "me@gabrielbull.com",
+                    "homepage": "http://gabrielbull.com/"
+                },
+                {
+                    "name": "Franck Lefevre",
+                    "homepage": "https://rootslabs.net/blog/"
+                },
+                {
+                    "name": "Ivan Lanin",
+                    "homepage": "http://ivan.lanin.org"
+                },
+                {
+                    "name": "Roman Syroeshko",
+                    "homepage": "http://ru.linkedin.com/pub/roman-syroeshko/34/a53/994/"
+                },
+                {
+                    "name": "Antoine de Troostembergh"
+                }
+            ],
+            "description": "PHPWord - A pure PHP library for reading and writing word processing documents (OOXML, ODF, RTF, HTML, PDF)",
+            "homepage": "https://phpoffice.github.io/PHPWord/",
+            "keywords": [
+                "ISO IEC 29500",
+                "OOXML",
+                "Office Open XML",
+                "OpenDocument",
+                "OpenXML",
+                "PhpOffice",
+                "PhpWord",
+                "Rich Text Format",
+                "WordprocessingML",
+                "doc",
+                "docx",
+                "html",
+                "odf",
+                "odt",
+                "office",
+                "pdf",
+                "php",
+                "reader",
+                "rtf",
+                "template",
+                "template processor",
+                "word",
+                "writer"
+            ],
+            "support": {
+                "issues": "https://github.com/PHPOffice/PHPWord/issues",
+                "source": "https://github.com/PHPOffice/PHPWord/tree/1.3.0"
+            },
+            "time": "2024-08-30T18:03:42+00:00"
+        },
         {
             "name": "phpoption/phpoption",
             "version": "1.9.3",