Commit Graph

13 Commits

Author SHA1 Message Date
Luke Street c1635245b8 Use UTF-8 exclusively internally; replace iconv with OSUTF
Filenames are now unconditionally encoded/decoded as
Shift-JIS, which provides a reasonable fallback to
7-bit ASCII.
2021-06-30 13:47:10 -04:00
Minty-Meeo 3c25647b6e The Encoding Update
While Nintendo's own documents claim GameCube and Wii disc file symbol tables only support 7-bit ASCII, this is far from the truth.  Indeed, even some first-party Nintendo games shipped with Shift-JIS encoded file symbol tables.  My guess?  The locale of whatever Windows machine mastered a GameCube or Wii disc influenced how wide character strings (UCS-2) were converted to narrow character strings.  To account for all possibilites, this update adds extensible multi-byte character set options to NOD-Tool.

A rundown of notable changes:
 - "-c XXXXX" option added to set the encoding of the GameCube / Wii ISO(s) being processed.
 - "SystemStringConv" renamed to "DiscLocToSystemConv"
 - "SystemUTF8Conv" renamed to "SystemToDiscLocConv"
 - Help message updated with new info.
 - Bugfix: AddBuildName had a logic error wherein the length of the SystemString was being used instead of length of the disc locale string.  This would corrupt the File Symbol Table if the disc locale string's length was greater than the SystemString's length.
 - Bugfix: recursiveMergeFST was not keeping track of parent indexes at all, meaning nested folders and their contents would be corrupted.  I simply copied the way recursiveBuildFST did things to fix this.
 - Bugfix (Windows): On Windows, for some reason, Sstat was a typedef for _stat (32-bit) instead of _stat64 (64-bit).  This is confounding, because untrimmed Wii ISOs will always be larger than the unsigned 32-bit integer limit (4,699,979,776 bytes vs 4,294,967,295 bytes), meaning the MergeWii errand has never worked for untrimmed ISOs on Windows.  Was this never tested??
 - Bugfix (Windows): Did you know Windows Command Prompt fully supports Unicode?  Stdio streams are now in _O_U16TEXT mode for Windows only.  Previously, attempting to print any character that could not be narrowed to your locale's encoding would either silently fail (std functions), or throw an exception (fmt functions).  As a minor drawback, narrow character print functions can no longer be used when stdio is in _O_U16TEXT mode, necessitating my PR for Logvisor here: (AxioDL/logvisor#7)
 - ExtractionContext::progressCB now uses SystemStringView because widechar printing works correctly on Windows now.
 - progFunc lambda no longer throws exceptions when printing unicode because widechar printing works correctly on Windows now.
 - Top-level constructors and functions with a Codepage_t parameter have also signatures that default to the US-ASCII codepage.
    - DiscGCN constructor
    - DiscBuilderGCN constructor
    - DiscBuilderGCN::CalculateTotalSizeRequired
    - DiscMergerGCN constructor
    - DiscMergerGCN::CalculateTotalSizeRequired
    - DiscWii constructor
    - DiscBuilderWii constructor
    - DiscBuilderWii::CalculateTotalSizeRequired
    - DiscMergerWii constructor
    - DiscMergerWii::CalculateTotalSizeRequired
    - OpenDiscFromImage
 - Conversion between system encoding and disc locale encoding has checks in place to warn the user if string conversion goes awry.
2021-06-27 03:26:20 -05:00
Henrique Gemignani Passos Lima 091262ace1 Use std::optional for CalculateTotalSizeBuild 2019-09-06 22:17:06 +02:00
Lioncash 5935e84dab General: Include headers directly where necessary
Ensures includes are explicitly included where necessary to avoid
indirect inclusions.
2019-08-30 04:34:49 -04:00
Lioncash 2171388b9d General: Make use of override where applicable
Makes it explicit where functions are being overridden in derived
classes/structs.
2019-08-10 01:41:44 -04:00
Jack Andersen be8409681f New code style refactor 2018-12-07 19:21:47 -10:00
Jack Andersen e964a013fe Convert to pragma once 2018-10-06 17:39:24 -10:00
Jack Andersen d5f5db440c Windows build fixes and warning avoidance 2018-05-24 20:38:06 -10:00
Jack Andersen 27a2cb5998 string_view refactor 2017-11-12 20:18:53 -10:00
Jack Andersen 41148a1368 Refactor of extracted directory structure and API simplification 2017-07-01 13:36:16 -10:00
Jack Andersen dd20c1d255 Add floating point progress to builder callbacks 2017-05-21 23:12:48 -04:00
Jack Andersen dc474ad156 Major refactor, better error handling and directory/image merge 2017-02-04 20:19:34 -10:00
Jack Andersen 556a5868bb Humungous refactor 2016-03-04 13:04:30 -10:00