[UIMA-2097] Treatment of URLs with blanks is incorrect in some places

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 2.3.1SDK
• Fix Version/s: 2.3.1Addons, 2.4.0SDK, 2.4.0AS
• Component/s: Core Java Framework
• Assignee: Marshall Schor
• Reporter: Marshall Schor
• Created: 22/Mar/11 19:25
• Updated: 23/Aug/11 22:04
• Resolved: 06/Apr/11 08:03

Description

User reported Document Analyzer failing to write correct names in the output directory, when the input directory had a blank in it (on Windows). Traced this to failing URL handling.

Proper URL handling seems to need to observe these principles:

1) URL may have "blanks" and other "invalid-in-URI chars"
2) URL may have %20 style encoding of blanks and other need-to-be-escaped characters

Creating files from these: need to use File(a-uri-form).

Creating URIs from URLs - if the URL has unescaped blanks, etc., the form
new URI(aUrl) fails; the aUrl.toURI() fails (i.e., throws an exception due to illegal chars for URI).

To make the URI creation put in the escape chars if they're not there, you need to use the multi-arg form of new URI (see Javadocs).

Fix the patches in UIMA-1879 and UIMA-1748.

Issue Links

• is related to: UIMA-1748 BSFAnnotator fails tests
• is related to: UIMA-1879 xcaswriter and xmicaswriter not handling spaces in file path