1 Star 0 Fork 0

Jmt / go-libs

加入 Gitee
与超过 1200万 开发者一起发现、参与优秀开源项目,私有仓库也完全免费 :)
免费加入
该仓库未声明开源许可证文件(LICENSE),使用请关注具体项目描述及其代码上游依赖。
克隆/下载
doc.go 2.36 KB
一键复制 编辑 原始数据 按行查看 历史
Jmt 提交于 2023-07-10 22:58 . add docconv
package docconv
import (
"bytes"
"fmt"
"io"
"io/ioutil"
"log"
"os"
"os/exec"
"time"
"github.com/richardlehane/mscfb"
"github.com/richardlehane/msoleps"
)
// ConvertDoc converts an MS Word .doc to text.
func ConvertDoc(r io.Reader) (string, map[string]string, error) {
f, err := NewLocalFile(r)
if err != nil {
return "", nil, fmt.Errorf("error creating local file: %v", err)
}
defer f.Done()
// Meta data
mc := make(chan map[string]string, 1)
go func() {
defer func() {
if e := recover(); e != nil {
log.Printf("panic when reading doc format: %v", e)
}
}()
meta := make(map[string]string)
doc, err := mscfb.New(f)
if err != nil {
log.Printf("ConvertDoc: could not read doc: %v", err)
return
}
props := msoleps.New()
for entry, err := doc.Next(); err == nil; entry, err = doc.Next() {
if msoleps.IsMSOLEPS(entry.Initial) {
if oerr := props.Reset(doc); oerr != nil {
log.Printf("ConvertDoc: could not reset props: %v", oerr)
break
}
for _, prop := range props.Property {
meta[prop.Name] = prop.String()
}
}
}
const defaultTimeFormat = "2006-01-02 15:04:05.999999999 -0700 MST"
// Convert parsed meta
if tmp, ok := meta["LastSaveTime"]; ok {
if t, err := time.Parse(defaultTimeFormat, tmp); err == nil {
meta["ModifiedDate"] = fmt.Sprintf("%d", t.Unix())
}
}
if tmp, ok := meta["CreateTime"]; ok {
if t, err := time.Parse(defaultTimeFormat, tmp); err == nil {
meta["CreatedDate"] = fmt.Sprintf("%d", t.Unix())
}
}
mc <- meta
}()
// Document body
bc := make(chan string, 1)
go func() {
// Save output to a file
outputFile, err := ioutil.TempFile("/tmp", "sajari-convert-")
if err != nil {
// TODO: Remove this.
log.Println("TempFile Out:", err)
return
}
defer os.Remove(outputFile.Name())
err = exec.Command("wvText", f.Name(), outputFile.Name()).Run()
if err != nil {
// TODO: Remove this.
log.Println("wvText:", err)
}
var buf bytes.Buffer
_, err = buf.ReadFrom(outputFile)
if err != nil {
// TODO: Remove this.
log.Println("wvText:", err)
}
bc <- buf.String()
}()
// TODO: Should errors in either of the above Goroutines stop things from progressing?
body := <-bc
meta := <-mc
// TODO: Check for errors instead of len(body) == 0?
if len(body) == 0 {
f.Seek(0, 0)
return ConvertDocx(f)
}
return body, meta, nil
}
1
https://gitee.com/tang-jinman/go-libs.git
git@gitee.com:tang-jinman/go-libs.git
tang-jinman
go-libs
go-libs
b81f544e35df

搜索帮助