fix bug: non-ascii file name not supported in detect type

tongwang · tongwang · commit 3fec5e360736 · 2020-02-27T15:41:48.000-05:00
diff --git a/tika/tika.py b/tika/tika.py
@@ -475,7 +475,7 @@ def detectType1(option, urlOrPath, serverEndpoint=ServerEndpoint, verbose=Verbos
     status, response = callServer('put', serverEndpoint, service, open(path, 'rb'),
             {
                 'Accept': responseMimeType,
-                'Content-Disposition': make_content_disposition_header(path)
+                'Content-Disposition': make_content_disposition_header(path.encode('utf-8') if type(path) is unicode_string else path)
             },
             verbose, tikaServerJar, config_path=config_path, requestOptions=requestOptions)
     if csvOutput == 1:

Original file line number	Diff line number	Diff line change
`@@ -475,7 +475,7 @@ def detectType1(option, urlOrPath, serverEndpoint=ServerEndpoint, verbose=Verbos`
`475`	`475`	`status, response = callServer('put', serverEndpoint, service, open(path, 'rb'),`
`476`	`476`	`{`
`477`	`477`	`'Accept': responseMimeType,`
`478`		`- 'Content-Disposition': make_content_disposition_header(path)`
	`478`	`+ 'Content-Disposition': make_content_disposition_header(path.encode('utf-8') if type(path) is unicode_string else path)`
`479`	`479`	`},`
`480`	`480`	`verbose, tikaServerJar, config_path=config_path, requestOptions=requestOptions)`
`481`	`481`	`if csvOutput == 1:`