php完整处理word中表单数据的方法

使用php基础方式实现word中表单处理

<?php/*** zipFile 类用于处理 .docx 文件的解压、修改和重新打包*/
class zipFile
{/** @var ZipArchive ZIP 文件对象 */private $zipFile;/** @var string 临时目录路径 */private $tempDir;/** @var string 嵌入的 Excel 文件临时目录路径 */private $excelTempDir;/** @var string 原始 .docx 文件路径 */private $docxPath;/*** 构造函数* * @param string $docxPath .docx 文件路径* @throws Exception 如果无法打开 .docx 文件*/public function __construct($docxPath){$this->docxPath = $docxPath;$this->zipFile = new ZipArchive();if ($this->zipFile->open($docxPath) !== TRUE) {throw new Exception("无法打开 .docx 文件: $docxPath");}$this->tempDir = sys_get_temp_dir() . '/docx_' . uniqid();mkdir($this->tempDir, 0777, true);// ✅ 正确命名：用于存放解压的 Excel 内容$this->excelTempDir = $this->tempDir . '/embedded_excel';}/*** 解压整个 .docx 到临时目录* 并自动解压其中的 Workbook1.xlsx（如果存在）*/public function extract(){// 1. 解压 .docx 主文件$this->zipFile->extractTo($this->tempDir);$this->zipFile->close();// echo "✅ .docx 已解压到: {$this->tempDir}\n";// 2. 查找并解压嵌入的 Excel 文件$embeddedXlsxPath = $this->tempDir . '/word/embeddings/Workbook1.xlsx';if (!file_exists($embeddedXlsxPath)) {echo "⚠️ 未找到嵌入的 Workbook1.xlsx\n";return;}$excelZip = new ZipArchive();if ($excelZip->open($embeddedXlsxPath) !== TRUE) {throw new Exception("无法打开嵌入的 Workbook1.xlsx");}// 创建目录并解压 Excel 内容mkdir($this->excelTempDir, 0777, true);$excelZip->extractTo($this->excelTempDir);$excelZip->close();echo "✅ Workbook1.xlsx 已解压到: {$this->excelTempDir}\n";}/*** 获取解压后的 sheet1.xml 路径* * @return string*/public function getSheet1Path(){$path = $this->excelTempDir . '/xl/worksheets/sheet1.xml';if (!file_exists($path)) {throw new Exception("未找到 sheet1.xml: $path");}return $path;}/*** 获取 sharedStrings.xml 路径（用于字符串修改）* * @return string*/public function getSharedStringsPath(){$path = $this->excelTempDir . '/xl/sharedStrings.xml';if (!file_exists($path)) {throw new Exception("未找到 sharedStrings.xml");}return $path;}/*** 获取文件内容（.docx 内任意文件）* * @param string $path 文件路径（相对于解压目录）* @return string|null*/public function getFileContent($path){$fullPath = $this->tempDir . '/' . ltrim($path, '/');return file_exists($fullPath) ? file_get_contents($fullPath) : null;}/*** 写入修改后的文件* * @param string $path 文件路径（相对于解压目录）* @param string $content 文件内容*/public function putFileContent($path, $content){$fullPath = $this->tempDir . '/' . ltrim($path, '/');$dir = dirname($fullPath);if (!is_dir($dir)) {mkdir($dir, 0777, true);}file_put_contents($fullPath, $content);}/*** 修改 Excel 单元格值（仅限数字）* * @param string $cell 单元格地址，如 'B2'* @param mixed $newValue 新值*/public function modifyExcelCell($cell, $newValue){// 1. 修改 Excel 数据 (sheet1.xml)$sheetFile = $this->getSheet1Path();// 使用 DOMDocument 替代 simplexml_load_file$dom = new DOMDocument();$dom->load($sheetFile);if (!$dom) {throw new Exception("无法加载 sheet1.xml");}$xpath = new DOMXPath($dom);$xpath->registerNamespace('x', 'http://schemas.openxmlformats.org/spreadsheetml/2006/main');$nodes = $xpath->query("//x:c[@r='$cell']");if ($nodes->length == 0) {throw new Exception("未找到单元格: $cell");}$cNode = $nodes->item(0);$vNodes = $xpath->query(".//x:v", $cNode);if ($vNodes->length == 0) {throw new Exception("单元格 $cell 缺少 <v> 节点");}$vNode = $vNodes->item(0);$oldValue = $vNode->nodeValue;$vNode->nodeValue = $newValue;$dom->save($sheetFile);echo "✅ 已修改 Excel 单元格 $cell: $oldValue → $newValue\n";// 2. 同步更新图表缓存$this->updateChartCache($cell, $newValue);}//批量更新public function updateDataAndChart($dataMap){foreach ($dataMap as $cell => $value) {$this->modifyExcelCell($cell, $value);}}/*** 更新 chart1.xml 中的缓存值（用于柱状图、折线图等）* * @param string $cell 单元格地址，如 'B2'* @param mixed  $newValue 新值* @param string $chartXmlPath 图表文件路径，默认为 chart1.xml*//*** 根据单元格地址，在 chart1.xml 中找到对应的 <c:f> 并更新其 numCache* * @param string $cell 单元格地址，如 'B2'* @param mixed  $newValue 新值*/private function updateChartCache($cell, $newValue){$chartPath = $this->tempDir . '/word/charts/chart1.xml';if (!file_exists($chartPath)) {echo "⚠️ 未找到 chart1.xml，跳过图表缓存更新\n";return;}// 使用 DOMDocument 替代 simplexml_load_file$dom = new DOMDocument();$dom->load($chartPath);if (!$dom) {throw new Exception("无法加载 chart1.xml");}// 注册命名空间$xpath = new DOMXPath($dom);$xpath->registerNamespace('c', 'http://schemas.openxmlformats.org/drawingml/2006/chart');$xpath->registerNamespace('a', 'http://schemas.openxmlformats.org/drawingml/2006/main');// 构造目标引用，如：Sheet1!$B$2$escapedCell = preg_replace('/([A-Z]+)/', '$1$', $cell); // B2 -> B$2$targetRef = "Sheet1!\$$escapedCell"; // 注意：Sheet1!$B$2// 查找 <c:f> 内容为 Sheet1!$B$2 的节点$fNodes = $xpath->query("//c:f[text()='$targetRef']");if ($fNodes->length == 0) {echo "🔍 未在 chart1.xml 中找到引用 $targetRef\n";return;}foreach ($fNodes as $fNode) {// 找到父级 <c:ser> 或 <c:pt> 等结构$parent = $fNode->parentNode;// 查找对应的 <c:numCache>$numCacheNodes = $xpath->query(".//c:numCache", $parent);if ($numCacheNodes->length == 0) {// echo "⚠️ 找到引用 $targetRef，但无 <c:numCache> 缓存\n";continue;}$numCache = $numCacheNodes->item(0);// 查找 <c:pt idx="0"> 下的 <c:v>$ptNodes = $xpath->query(".//c:pt[@idx='0']/c:v", $numCache);if ($ptNodes->length > 0) {$ptNode = $ptNodes->item(0);$oldValue = $ptNode->nodeValue;$ptNode->nodeValue = $newValue;echo "📊 已更新图表缓存 [$targetRef]: $oldValue → $newValue\n";} else {// 如果没有 pt，尝试创建（高级功能，可选）echo "⚠️ 未找到 <c:pt> 节点，无法更新缓存: $targetRef\n";}}// 保存修改$dom->save($chartPath);}/*** 重新打包为 .docx* * @param string $outputPath 输出文件路径*/public function pack($outputPath){// 1. 先打包嵌入的 Excel，并自动清理临时目录$this->repackEmbeddedWorkbook();// 2. 再打包整个 .docx 文件$newZip = new ZipArchive();if ($newZip->open($outputPath, ZipArchive::CREATE | ZipArchive::OVERWRITE) !== TRUE) {throw new Exception("无法创建输出文件: $outputPath");}$files = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($this->tempDir, FilesystemIterator::SKIP_DOTS),RecursiveIteratorIterator::LEAVES_ONLY);foreach ($files as $file) {if (!$file->isDir()) {$filePath = $file->getRealPath();$relativePath = substr($filePath, strlen($this->tempDir) + 1);$newZip->addFile($filePath, $relativePath);}}$newZip->close();echo "✅ 已打包为: $outputPath\n";}/*** 将修改后的 embedded_excel/ 目录重新打包为 Workbook1.xlsx* 并删除临时解压目录（确保中间文件不残留）*/private function repackEmbeddedWorkbook(){$embeddedXlsxPath = $this->tempDir . '/word/embeddings/Workbook1.xlsx';// 如果没有 embedded_excel 目录，说明没有 Excel 或未解压if (!is_dir($this->excelTempDir)) {echo "⚠️ 无嵌入 Excel 临时目录，跳过打包 Workbook1.xlsx\n";return;}// 删除旧的 Workbook1.xlsx（如果存在）if (file_exists($embeddedXlsxPath)) {unlink($embeddedXlsxPath);}// 创建新的 Zip 存档$zip = new ZipArchive();if ($zip->open($embeddedXlsxPath, ZipArchive::CREATE | ZipArchive::OVERWRITE) !== TRUE) {throw new Exception("无法创建 Workbook1.xlsx");}$files = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($this->excelTempDir, FilesystemIterator::SKIP_DOTS),RecursiveIteratorIterator::LEAVES_ONLY);foreach ($files as $file) {if (!$file->isDir()) {$filePath = $file->getRealPath();$relativePath = substr($filePath, strlen($this->excelTempDir) + 1);$zip->addFile($filePath, $relativePath);}}$zip->close();echo "✅ Workbook1.xlsx 已重新打包\n";// 🔥 关键修复：立即删除 embedded_excel 临时目录$this->rrmdir($this->excelTempDir);echo "🗑️  已清理 Excel 临时目录: {$this->excelTempDir}\n";}/*** 列出解压目录中的所有文件* * @return array 文件路径列表*/public function listFiles(){$files = [];if (!is_dir($this->tempDir)) return $files;$iterator = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($this->tempDir, FilesystemIterator::SKIP_DOTS));foreach ($iterator as $file) {if ($file->isFile()) {$relativePath = substr($file->getPathname(), strlen($this->tempDir) + 1);$files[] = $relativePath;}}return $files;}/*** 清理临时文件*/public function cleanup(){if (is_dir($this->tempDir)) {$this->rrmdir($this->tempDir);}}/*** 递归删除目录* * @param string $dir 目录路径*/private function rrmdir($dir){$files = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($dir, FilesystemIterator::SKIP_DOTS),RecursiveIteratorIterator::CHILD_FIRST);foreach ($files as $fileinfo) {$todo = ($fileinfo->isDir() ? 'rmdir' : 'unlink');$todo($fileinfo->getRealPath());}rmdir($dir);}
}// ✅ 使用示例（修复后）
try {$processor = new zipFile('表单.docx');$processor->extract();// ✅ 方法一：使用封装好的 modifyExcelCell$processor->modifyExcelCell('B2', 500);$processor->pack('output.docx');$processor->cleanup();
} catch (Exception $e) {echo "❌ 错误: " . $e->getMessage() . "\n";
}