maojian
/
opai-service-center
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

word转换pdf修复

master
maojian 2 months ago
parent
844d502fcf
commit
b454573984
3 changed files with 175 additions and 153 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 3
      document-convert-service/src/main/java/com/bw/convert/entity/Constants.java
  2. 233
      document-convert-service/src/main/java/com/bw/convert/service/impl/ConvertTaskServiceImpl.java
  3. 2
      opai-api/src/main/java/com/bw/opai/config/WebMvcConfig.java

3
document-convert-service/src/main/java/com/bw/convert/entity/Constants.java
View File

@ -40,7 +40,8 @@ public class Constants {
public static final String JOBID = "jobId"; public static final String JOBID = "jobId";
public static final String PROGRESS = "Progress"; public static final String PROGRESS = "Progress";
public static final String JOBDOCUMENTS = "JobDocuments"; public static final String JOBDOCUMENTS = "JobDocuments";
public static final String OutputDocuments = "OutputDocuments";
public static final String OUTPUTDOCUMENTS = "OutputDocuments";
public static final String INPUTFILES = "InputFiles";
public static final String FILES = "Files"; public static final String FILES = "Files";
public static final String IDRENAME = "#id"; public static final String IDRENAME = "#id";

233
document-convert-service/src/main/java/com/bw/convert/service/impl/ConvertTaskServiceImpl.java
View File

@ -157,6 +157,119 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
//识别中 -- 放回队列 //识别中 -- 放回队列
ConfigCache.resultQueue.put(task); ConfigCache.resultQueue.put(task);
}else if (progress == 100) { }else if (progress == 100) {
parseOcrRes(task,jobId);
}else {
//识别异常
log.error("文档转换异常：{}",resStr);
//发送失败结果
AppResultDoc entity = new AppResultDoc();
entity.setInput(task);
entity.setTaskId((String)task.get(Constants.TASKID));
entity.setAppId((Integer)task.get(Constants.ID));
long now = System.currentTimeMillis();
entity.setCreateTime(now);
Map<String, Object> result = new HashMap<String, Object>(16);
result.put(Constants.ERROR, "转换失败");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
} catch (Throwable e) {
// TODO: handle exception
log.error("创建文档解析任务异常。e:",e);
//发送失败结果
AppResultDoc entity = new AppResultDoc();
entity.setInput(task);
entity.setTaskId((String)task.get(Constants.TASKID));
entity.setAppId((Integer)task.get(Constants.ID));
long now = System.currentTimeMillis();
entity.setCreateTime(now);
Map<String, Object> result = new HashMap<String, Object>(16);
result.put(Constants.ERROR, "源文件解析异常");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
}
/**
* 将Base64编码的Word文档读取为文本
* @param base64Word Base64字符串
* @return Word文本内容
* @throws Exception
*/
public String readWordFromBase64(String base64Word) throws Exception {
byte[] bytes = Base64.getDecoder().decode(base64Word);
try (InputStream is = new ByteArrayInputStream(bytes)) {
// 尝试读取为docx格式
try {
XWPFDocument docx = new XWPFDocument(is);
XWPFWordExtractor extractor = new XWPFWordExtractor(docx);
return extractor.getText();
} catch (Exception e) {
e.printStackTrace();
return null;
}
}
}
/**
* Base64 字符串写入指定路径文件
*
* @param base64byte Base64 内容（支持 data:image/png;base64,xxx）
* @param fileType 文件类型，如 png/jpg/pdf
* @param filePath 保存目录路径，如 /data/upload/
* @param fileName 文件名称
* @return true 成功，false 失败
*/
public boolean base64ToFile(String base64byte, String filePath,String fileName) {
try {
if (base64byte == null || base64byte.isEmpty()) {
return false;
}
// 处理 base64 头
if (base64byte.contains(",")) {
base64byte = base64byte.substring(base64byte.indexOf(",") + 1);
}
byte[] bytes = Base64.getDecoder().decode(base64byte);
// 创建目录
File dir = new File(filePath);
if (!dir.exists() && !dir.mkdirs()) {
return false;
}
// 生成文件名
File file = new File(dir, fileName);
try (FileOutputStream fos = new FileOutputStream(file)) {
fos.write(bytes);
fos.flush();
}
return true;
} catch (Exception e) {
// 生产中建议打日志
log.error("Base64 写文件失败", e);
return false;
}
}
/**
* 解析查询结果
* @param task
* @param jobId
*/
private void parseOcrRes(Map<String, Object> task,String jobId) {
//文件类型 //文件类型
String fileType = (String) task.get(Constants.DOC_TYPE); String fileType = (String) task.get(Constants.DOC_TYPE);
//识别成功 -- 获取请求结果 //识别成功 -- 获取请求结果
@ -166,8 +279,17 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
String queryResStr = DownLoadUtil.doGet(lastQueryUrl); String queryResStr = DownLoadUtil.doGet(lastQueryUrl);
Map<String, Object> queryRes = JSONObject.parseObject(queryResStr); Map<String, Object> queryRes = JSONObject.parseObject(queryResStr);
List<Map<String, Object>> jobDocuments = (List<Map<String, Object>>) queryRes.get(Constants.JOBDOCUMENTS); List<Map<String, Object>> jobDocuments = (List<Map<String, Object>>) queryRes.get(Constants.JOBDOCUMENTS);
List<Map<String, Object>> outputDocuments = (List<Map<String, Object>>) jobDocuments.get(0).get(Constants.OutputDocuments);
List<Map<String, Object>> files = (List<Map<String, Object>>) outputDocuments.get(0).get(Constants.FILES);
List<Map<String, Object>> outputDocuments = null;
List<Map<String, Object>> files = null;
if(jobDocuments != null && jobDocuments.size() > 0) {
outputDocuments = (List<Map<String, Object>>) jobDocuments.get(0).get(Constants.OUTPUTDOCUMENTS);
files = (List<Map<String, Object>>) outputDocuments.get(0).get(Constants.FILES);
}else {
List<Map<String, Object>> InputFiles = (List<Map<String, Object>>) queryRes.get(Constants.INPUTFILES);
outputDocuments = (List<Map<String, Object>>)InputFiles.get(0).get(Constants.OUTPUTDOCUMENTS);
files = (List<Map<String, Object>>) outputDocuments.get(outputDocuments.size() - 1).get(Constants.FILES);
}
for (int i =0;i<files.size();i++) { for (int i =0;i<files.size();i++) {
Map<String, Object> map = files.get(i); Map<String, Object> map = files.get(i);
String fileContents = (String) map.get(Constants.FILECONTENTS); String fileContents = (String) map.get(Constants.FILECONTENTS);
@ -240,7 +362,7 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
} }
}else { }else {
//识别异常 //识别异常
log.error("html文档转换异常：{}",resStr);
log.error("html文档转换异常：{}",queryResStr);
//发送失败结果 //发送失败结果
AppResultDoc entity = new AppResultDoc(); AppResultDoc entity = new AppResultDoc();
entity.setInput(task); entity.setInput(task);
@ -291,7 +413,7 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity)); DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}else { }else {
//识别异常 //识别异常
log.error("文档转换异常：{}",resStr);
log.error("文档转换异常：{}",queryResStr);
//发送失败结果 //发送失败结果
AppResultDoc entity = new AppResultDoc(); AppResultDoc entity = new AppResultDoc();
entity.setInput(task); entity.setInput(task);
@ -309,112 +431,9 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
} }
} }
} }
}else {
//识别异常
log.error("文档转换异常：{}",resStr);
//发送失败结果
AppResultDoc entity = new AppResultDoc();
entity.setInput(task);
entity.setTaskId((String)task.get(Constants.TASKID));
entity.setAppId((Integer)task.get(Constants.ID));
long now = System.currentTimeMillis();
entity.setCreateTime(now);
Map<String, Object> result = new HashMap<String, Object>(16);
result.put(Constants.ERROR, "转换失败");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
} catch (Throwable e) {
// TODO: handle exception
log.error("创建文档解析任务异常。e:",e);
//发送失败结果
AppResultDoc entity = new AppResultDoc();
entity.setInput(task);
entity.setTaskId((String)task.get(Constants.TASKID));
entity.setAppId((Integer)task.get(Constants.ID));
long now = System.currentTimeMillis();
entity.setCreateTime(now);
Map<String, Object> result = new HashMap<String, Object>(16);
result.put(Constants.ERROR, "源文件解析异常");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
} }
/**
* 将Base64编码的Word文档读取为文本
* @param base64Word Base64字符串
* @return Word文本内容
* @throws Exception
*/
public String readWordFromBase64(String base64Word) throws Exception {
byte[] bytes = Base64.getDecoder().decode(base64Word);
try (InputStream is = new ByteArrayInputStream(bytes)) {
// 尝试读取为docx格式
try {
XWPFDocument docx = new XWPFDocument(is);
XWPFWordExtractor extractor = new XWPFWordExtractor(docx);
return extractor.getText();
} catch (Exception e) {
e.printStackTrace();
return null;
}
}
}
/**
* Base64 字符串写入指定路径文件
*
* @param base64byte Base64 内容（支持 data:image/png;base64,xxx）
* @param fileType 文件类型，如 png/jpg/pdf
* @param filePath 保存目录路径，如 /data/upload/
* @param fileName 文件名称
* @return true 成功，false 失败
*/
public boolean base64ToFile(String base64byte, String filePath,String fileName) {
try {
if (base64byte == null || base64byte.isEmpty()) {
return false;
}
// 处理 base64 头
if (base64byte.contains(",")) {
base64byte = base64byte.substring(base64byte.indexOf(",") + 1);
}
byte[] bytes = Base64.getDecoder().decode(base64byte);
// 创建目录
File dir = new File(filePath);
if (!dir.exists() && !dir.mkdirs()) {
return false;
}
// 生成文件名
File file = new File(dir, fileName);
try (FileOutputStream fos = new FileOutputStream(file)) {
fos.write(bytes);
fos.flush();
}
return true;
} catch (Exception e) {
// 生产中建议打日志
log.error("Base64 写文件失败", e);
return false;
}
}
public static void main(String[] args) throws Exception { public static void main(String[] args) throws Exception {
// OcrTaskServiceImpl ocrTaskServiceImpl = new OcrTaskServiceImpl(); // OcrTaskServiceImpl ocrTaskServiceImpl = new OcrTaskServiceImpl();
// String base64Word = "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"; // String base64Word = "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";

2
opai-api/src/main/java/com/bw/opai/config/WebMvcConfig.java
View File

@ -19,6 +19,8 @@ public class WebMvcConfig implements WebMvcConfigurer {
registry.addInterceptor(authInterceptor) registry.addInterceptor(authInterceptor)
// 拦截所有业务接口 // 拦截所有业务接口
.addPathPatterns("/apps/**") .addPathPatterns("/apps/**")
// 放行 datasave
.excludePathPatterns("/apps/datasave")
// 放行登录注册接口（否则就死循环了） // 放行登录注册接口（否则就死循环了）
.excludePathPatterns("/auth/**") .excludePathPatterns("/auth/**")
.excludePathPatterns("/static/**"); .excludePathPatterns("/static/**");

Write Preview
Loading…
Cancel
Save