[PATCH] Fixes for character encoding problems in atgen.c (0.6.0)

[Hugo Haas]

Summary: the attached patch fixes some character encoding problems for
AT communications.

Hi.

Thinking more about the T610 issues I posted a patch for yesterday, I
realized that it was normal that AT+CSCS="ME" fails with UCS-2
encoding as ME wasn't encoded in UCS-2 in my minicom test.

Looking at atgen.c, I noticed that gnokii did the same mistake, as it
sets the memory type without checking what encoding is used. I also
noted that gnokii wasn't checking the encoding in ReplyReadPhonebook()
when receiving a phone number, and --getphonebook would give me:

  1. Name: Foo Bar/H
  Number: 004500...
  Group id: 0

which would eventually lead to a segmentation fault as numbers aren't
supposed to be that long: GN_PHONEBOOK_NUMBER_MAX_LENGTH equals 49,
and a UCS-2 encoded number could be 50+ byte-long.

The attached patch fixes those problems by:
- adding character encoding code in at_memory_type_set().
- adding character decoding code for the phone number in
  ReplyReadPhonebook().

As I noticed that there were several places where essentially the same
code was used to encode and decode strings, I factorized it in
at_get.c by adding two new functions at_encode() and at_decode().

The total is 52 insertions(+), 51 deletions(-), so the code didn't
get bigger. :-)

This makes my unicode option patch unnecessary. Well, some people
might be interested in using it, I guess.

Regards,

Hugo

-- 
Hugo Haas - http://larve.net/people/hugo/

gnokii-atgen-encoding-fix.diff
Description: Text document

signature.asc
Description: Digital signature