public
String convertToHtml(InputStream docxStream) throws Exception {
try (XWPFDocument document
= new XWPFDocument(docxStream)) {
StringBuilder htmlBuilder
= new
StringBuilder();
htmlBuilder.append(
"<body>
");
// 处理段落
for (XWPFParagraph paragraph
: document.getParagraphs()) {
htmlBuilder.append(processParagraph(paragraph));
}
// 处理表格
for (XWPFTable table
: document.getTables()) {
htmlBuilder.append(processTable(table));
}
htmlBuilder.append(
"</body>
");
return htmlBuilder.to
String();
}
}
private
String processParagraph(XWPFParagraph paragraph) throws Exception {
StringBuilder paraBuilder
= new
StringBuilder(
"<p>
");
for (XWPFRun run
: paragraph.getRuns()) {
String text
= run.getText(0);
// 处理图片
if (run.getEmbeddedPictures().size() > 0) {
for (XWPFPicture picture
: run.getEmbeddedPictures()) {
XWPFPictureData picData
= picture.getPictureData();
// 上传图片并获取URL
String imageUrl
= fileProcessingService.uploadImage(
picData.getData(),
picData.suggestFileExtension()
);
paraBuilder.append(
"<img src
=\
"
").append(imageUrl).append(
"\
" alt
=\
"\
">
");
}
}
if (text !
= null && !text.trim().isEmpty()) {
// 获取当前运行的样式
if (run.isBold()) paraBuilder.append(
"<strong>
");
if (run.isItalic()) paraBuilder.append(
"
<em>");
// 添加文本内容
paraBuilder.append(escapeHtml(text));
if (run.isItalic()) paraBuilder.append(
"</em>
");
if (run.isBold()) paraBuilder.append(
"</strong>
");
}
}
paraBuilder.append(
"</p>
");
return paraBuilder.to
String();
}
private
String processTable(XWPFTable table) throws Exception {
StringBuilder tableBuilder
= new
StringBuilder(
"<table border
=\
"1\
">
");
for (XWPFTableRow row
: table.getRows()) {
tableBuilder.append(
"<tr>
");
for (XWPFTableCell cell
: row.getTableCells()) {
tableBuilder.append(
"<td>
");
for (XWPFParagraph para
: cell.getParagraphs()) {
tableBuilder.append(processParagraph(para));
}
tableBuilder.append(
"</td>
");
}
tableBuilder.append(
"</tr>
");
}
tableBuilder.append(
"</table>
");
return tableBuilder.to
String();
}
private
String escapeHtml(
String text) {
return text.replace(
"&
",
"&
")
.replace(
"<
",
"<
")
.replace(
">
",
">
")
.replace(
"\
"
",
"
"
")
.replace(
"'
",
"'
");
}
private static final Map<
String,
String> COLOR_MAP
= new HashMap<>();
static {
Properties props
= new Properties();
try (InputStream input
= WordToHtmlConverter.class.getResourceAsStream(
"/colors.properties
")) {
if (input !
= null) {
props.load(input);
for (
String colorName
: props.
stringPropertyNames()) {
COLOR_MAP.put(colorName.toLowerCase(), props.getProperty(colorName));
}
} else {
// 如果文件不存在，使用一个默认的小映射表
initDefaultColorMap();
}
} catch (IOException e) {
initDefaultColorMap();
}
}
private static void initDefaultColorMap() {
//
初始化一个默认的小映射表
COLOR_MAP.put(
"black
",
"#000000
");
COLOR_MAP.put(
"white
",
"#FFFFFF
");
COLOR_MAP.put(
"red
",
"#FF0000
");
// ... 其他基本颜色
}
private
String getColorHex(
String colorName) {
if (colorName
== null || colorName.isEmpty()) {
return
null;
}
String lowerColor
= colorName.toLowerCase();
return COLOR_MAP.get(lowerColor);
}
需要将word中的文字 颜色\字体大小等格式写进HTML中
本文详细解释了Java中字符串的不同初始化方式及其特点，包括声明未初始化的引用变量、初始化为null及初始化为空字符串的区别，帮助读者理解如何正确使用这些初始化方式。
扫一扫
14
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
