Character string 2A sets Espeak volume to 0 and does strange things to pitch and intonation with NVDA

[nvaccessAuto]

Reported by fastfinge on 2009-10-07 01:10
When reading a document in NVDA, coming accross the string "2A" will cause the Espeak volume to be set to 0, and strange things to happen to pitch and intonation. This bug involves oddball characters; I am not sure they will show up in HTML. I have attached a txt file demonstrating the bug.

[nvaccessAuto]

Attachment espeak bug.txt added by fastfinge on 2009-10-07 01:12
Description:
I wasn't sure if the characters required to demonstrate this bug would be filtered out in the HTML or not. This text file definitively demonstrates the problem.

[nvaccessAuto]

Comment 1 by fastfinge on 2009-10-07 01:15
This seems not to happen in firefox. But the bug occurs every time I use wordpad.
Changes:
Changed title from "Character string �2A sets Espeak volume to 0 and does strange things to pitch and intonation with NVDA" to "Character string 2A sets Espeak volume to 0 and does strange things to pitch and intonation with NVDA"

[nvaccessAuto]

Comment 2 by jteh on 2009-10-07 02:06
Ouch. It seems \01 is used by eSpeak for embedded commands. I will commit a fix shortly.
Changes:
Milestone changed from None to 2009.1

[nvaccessAuto]

Comment 3 by jteh on 2009-10-07 02:29
Fixed in r3310. Thanks for reporting.
Changes:
State: closed

[nvaccessAuto]

Comment 4 by pvagner on 2009-10-07 06:25
Are you sure this is NVDA's bug? I have reproducet this with command-line version also.

I think eSpeak should only care about control characters when ssml or whatever markup is chosen.

[nvaccessAuto]

Comment 5 by pvagner on 2009-10-07 06:39
One more thing.
While opening an emails including this string in thunderbird NVDA generates an error.
Here's the output:
ERROR - XMLFormatting.XMLTextParser.parse (08:25:32):
XML: When reading a document in NVDA, coming accross the string "�2A" will

Traceback (most recent call last):
File "XMLFormatting.pyo", line 41, in parse
ExpatError: not well-formed (invalid token): line 1, column 565

And one more...
While opening log viewer with this string inside NVDA gives the following:
ERROR - unhandled exception (08:27:04):
Traceback (most recent call last):
File "gui__init__.pyo", line 229, in onViewLogCommand
File "gui\logViewer.pyo", line 82, in activate
File "gui\logViewer.pyo", line 40, in init
File "gui\logViewer.pyo", line 49, in refresh
File "codecs.pyo", line 666, in read
File "codecs.pyo", line 472, in read
UnicodeDecodeError: 'utf8' codec can't decode bytes in position 76420-76422: invalid data

[nvaccessAuto]

Comment 6 by jteh on 2009-10-07 07:00
Oh, it's certainly an issue in eSpeak. The question is whether it was intentional or not. In any case, we need to work around it for now to prevent annoyance to our users.