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. 323
      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";

323
document-convert-service/src/main/java/com/bw/convert/service/impl/ConvertTaskServiceImpl.java
View File

@ -157,158 +157,7 @@ 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) {
//文件类型
String fileType = (String) task.get(Constants.DOC_TYPE);
//识别成功 -- 获取请求结果
String lastQueryUrl = queryUrl;
lastQueryUrl = lastQueryUrl.replace(Constants.IDRENAME,jobId);
log.info("queryUrl:{}",lastQueryUrl);
String queryResStr = DownLoadUtil.doGet(lastQueryUrl);
Map<String, Object> queryRes = JSONObject.parseObject(queryResStr);
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);
for (int i =0;i<files.size();i++) {
Map<String, Object> map = files.get(i);
String fileContents = (String) map.get(Constants.FILECONTENTS);
String fileName = (String) map.get(Constants.F_ILENAME);
if(fileType.contains(Constants.HTML)) {
/*******html********/
//拆分文件名，一部分是文件夹名称，一部分是文件名
String[] arr = fileName.split("\\\\");
String lastConvertFilePathPrefix = convertFilePathPrefix + arr[0];
boolean isInputFile = base64ToFile(fileContents, lastConvertFilePathPrefix, arr[1]);
if (isInputFile) {
log.info("html相关文件转换成功，fileName:{}",arr[1]);
if(i == files.size() -1 ) {
log.info("最后的文件转换成功，准备压缩文件夹，path={}", lastConvertFilePathPrefix);
try {
// 生成 zip 文件名
String zipFileName = UUID.randomUUID().toString().replace("-", "") + ".zip";
String zipFilePath = convertFilePathPrefix + zipFileName;
// 压缩文件夹
ZipUtil.zipDirectory(lastConvertFilePathPrefix, zipFilePath);
log.info("文件夹压缩成功，zipPath={}", zipFilePath);
// === 后续你如果要：上传 / 回传结果，可继续写 ===
String responseBody = DownLoadUtil.upLoadFile(zipFilePath, gofastUrl);
//删除目录
FileUtil.deleteDir(new File(lastConvertFilePathPrefix));
//删除压缩包
FileUtil.delFile(zipFilePath);
// 解析返回 JSON
JSONObject json = JSONObject.parseObject(responseBody);
// ⚠️ 根据你 GoFast 实际返回字段调整
String path = json.getString(Constants.PATH);
String url = gofastAccessPrefix + path;
//成功 发送结果
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.FILEURL, url);
result.put(Constants.FILENAME, zipFileName);
entity.setResult(result);
entity.setStatus(1);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
} catch (Exception e) {
log.error("HTML 文件夹压缩失败", e);
// 压缩失败，按失败处理
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<>(16);
result.put(Constants.ERROR, "HTML 文件夹压缩失败");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
}
}else {
//识别异常
log.error("html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));
break;
}
}else {
/******其他类型**********/
//输出文件
boolean isInputFile = base64ToFile(fileContents, convertFilePathPrefix, fileName);
if (isInputFile) {
//转换成功 --上传文件
String fileAddr = convertFilePathPrefix+fileName;
//上传
String responseBody = DownLoadUtil.upLoadFile(fileAddr, gofastUrl);
//删除源文件
FileUtil.delFile(fileAddr);
// 解析返回 JSON
JSONObject json = JSONObject.parseObject(responseBody);
// ⚠️ 根据你 GoFast 实际返回字段调整
String path = json.getString(Constants.PATH);
String url = gofastAccessPrefix + path;
//成功 发送结果
AppResultDoc entity = new AppResultDoc();
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.FILEURL, url);
result.put(Constants.FILENAME, fileName);
entity.setResult(result);
entity.setStatus(1);
entity.setDel(0);
entity.setInput(task);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}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));
}
}
}
parseOcrRes(task,jobId);
}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：{}",resStr);
@ -415,6 +264,176 @@ public class ConvertTaskServiceImpl implements ConvertTaskService {
} }
/**
* 解析查询结果
* @param task
* @param jobId
*/
private void parseOcrRes(Map<String, Object> task,String jobId) {
//文件类型
String fileType = (String) task.get(Constants.DOC_TYPE);
//识别成功 -- 获取请求结果
String lastQueryUrl = queryUrl;
lastQueryUrl = lastQueryUrl.replace(Constants.IDRENAME,jobId);
log.info("queryUrl:{}",lastQueryUrl);
String queryResStr = DownLoadUtil.doGet(lastQueryUrl);
Map<String, Object> queryRes = JSONObject.parseObject(queryResStr);
List<Map<String, Object>> jobDocuments = (List<Map<String, Object>>) queryRes.get(Constants.JOBDOCUMENTS);
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++) {
Map<String, Object> map = files.get(i);
String fileContents = (String) map.get(Constants.FILECONTENTS);
String fileName = (String) map.get(Constants.F_ILENAME);
if(fileType.contains(Constants.HTML)) {
/*******html********/
//拆分文件名，一部分是文件夹名称，一部分是文件名
String[] arr = fileName.split("\\\\");
String lastConvertFilePathPrefix = convertFilePathPrefix + arr[0];
boolean isInputFile = base64ToFile(fileContents, lastConvertFilePathPrefix, arr[1]);
if (isInputFile) {
log.info("html相关文件转换成功，fileName:{}",arr[1]);
if(i == files.size() -1 ) {
log.info("最后的文件转换成功，准备压缩文件夹，path={}", lastConvertFilePathPrefix);
try {
// 生成 zip 文件名
String zipFileName = UUID.randomUUID().toString().replace("-", "") + ".zip";
String zipFilePath = convertFilePathPrefix + zipFileName;
// 压缩文件夹
ZipUtil.zipDirectory(lastConvertFilePathPrefix, zipFilePath);
log.info("文件夹压缩成功，zipPath={}", zipFilePath);
// === 后续你如果要：上传 / 回传结果，可继续写 ===
String responseBody = DownLoadUtil.upLoadFile(zipFilePath, gofastUrl);
//删除目录
FileUtil.deleteDir(new File(lastConvertFilePathPrefix));
//删除压缩包
FileUtil.delFile(zipFilePath);
// 解析返回 JSON
JSONObject json = JSONObject.parseObject(responseBody);
// ⚠️ 根据你 GoFast 实际返回字段调整
String path = json.getString(Constants.PATH);
String url = gofastAccessPrefix + path;
//成功 发送结果
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.FILEURL, url);
result.put(Constants.FILENAME, zipFileName);
entity.setResult(result);
entity.setStatus(1);
entity.setDel(0);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
} catch (Exception e) {
log.error("HTML 文件夹压缩失败", e);
// 压缩失败，按失败处理
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<>(16);
result.put(Constants.ERROR, "HTML 文件夹压缩失败");
entity.setResult(result);
entity.setStatus(2);
entity.setDel(0);
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}
}
}else {
//识别异常
log.error("html文档转换异常：{}",queryResStr);
//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));
break;
}
}else {
/******其他类型**********/
//输出文件
boolean isInputFile = base64ToFile(fileContents, convertFilePathPrefix, fileName);
if (isInputFile) {
//转换成功 --上传文件
String fileAddr = convertFilePathPrefix+fileName;
//上传
String responseBody = DownLoadUtil.upLoadFile(fileAddr, gofastUrl);
//删除源文件
FileUtil.delFile(fileAddr);
// 解析返回 JSON
JSONObject json = JSONObject.parseObject(responseBody);
// ⚠️ 根据你 GoFast 实际返回字段调整
String path = json.getString(Constants.PATH);
String url = gofastAccessPrefix + path;
//成功 发送结果
AppResultDoc entity = new AppResultDoc();
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.FILEURL, url);
result.put(Constants.FILENAME, fileName);
entity.setResult(result);
entity.setStatus(1);
entity.setDel(0);
entity.setInput(task);
//回传给api服务保存
DownLoadUtil.doPost(saveUrl, JSONObject.toJSONString(entity));
}else {
//识别异常
log.error("文档转换异常：{}",queryResStr);
//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));
}
}
}
}
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