ZeldaMods (Breath of the Wild)

BFEVFL: Difference between revisions

Page Discussion

BFEVFL (view source)

Revision as of 11:55, 16 April 2023

653 bytes added ,  1 year ago
→‎Dictionary entry
VisualWikitext
imported>Leoetlino
Tags: Mobile edit Mobile web edit
 
(13 intermediate revisions by 3 users not shown)
Line 1: Line 1:
<onlyinclude>'''bfevfl''' is a binary file format for [[Event flow|'''ev'''ent '''fl'''ows]].</onlyinclude> In ''Breath of the Wild'' and other Nintendo games such as Splatoon 2, binary event flows are parsed by Nintendo's EventFlow/evfl library.
<onlyinclude>'''BFEVFL''' is a binary file format for [[Event flow|'''ev'''ent '''fl'''ows]].</onlyinclude> In ''Breath of the Wild'' and other Nintendo games such as ''Splatoon 2'' and ''Link's Awakening'', binary event flows are parsed by Nintendo's EventFlow/evfl library.


== Structures ==
== Structures ==
bfevfl is a little endian format, so values are stored in little endian even on Wii U.
BFEVFL is a format that supports both big endian and little endian; however, event flows typically use little endian even on Wii U, as is the case in ''Breath of the Wild''.


bfevfl uses pointers very heavily to refer to strings, to other sections, etc. All pointers are always 64 bit long -- since the same format is used on the Switch which is a 64 bit platform. They must be present in the [[#Relocation table]] if the file is to be loaded correctly by the official bfevfl code.
BFEVFL uses pointers very heavily to refer to strings, to other sections, etc. All pointers are always 64-bit long -- this is because the same format is used on 64-bit platforms like the Switch. Pointers must be present in the [[#Relocation table]] if the file is to be loaded correctly by the official EventFlow code.


As a direct consequence, most sections don't have any fixed order. Therefore, this article will only document the data structures and mention the order Nintendo places sections in. For more details, it is recommended to look at the [https://github.com/leoetlino/evfl <code>evfl</code> library] directly.
As a direct consequence, most sections don't have any fixed order. Therefore, this article will only document the data structures and mention the order Nintendo places sections in. For more details, it is recommended to look at the [https://github.com/leoetlino/evfl <code>evfl</code> library] directly.
Line 16: Line 16:
| 0x0 || char[8] || Magic ("BFEVFL\x00\x00")
| 0x0 || char[8] || Magic ("BFEVFL\x00\x00")
|-
|-
| 0x8 || u16 || Version (0x0300)
| 0x8 || u8 || Version (major)
|-
|-
| 0xa || u8 || Unknown (must be zero)
| 0x9 || u8 || Version (minor)
|-
|-
| 0xb || u8 || Unknown
| 0xa || u8 || Version (patch)
|-
|-
| 0xc || u16 || Byte order mark
| 0xb || u8 || Version (sub-patch)
|-
| 0xc || s16 || Byte order mark
|-
|-
| 0xe || u8 || Alignment (to get the actual value: 1 << raw_value)
| 0xe || u8 || Alignment (to get the actual value: 1 << raw_value)
|-
|-
| 0xf || u8 || Unknown
| 0xf || u8 || Padding
|-
|-
| 0x10 || u32 || File name offset
| 0x10 || int || File name offset
|-
|-
| 0x14 || u16 || Is relocated flag (set in memory)
| 0x14 || u16 || Is relocated flag (set in memory)
Line 34: Line 36:
| 0x16 || u16 || First block offset
| 0x16 || u16 || First block offset
|-
|-
| 0x18 || u32 || Relocation table offset
| 0x18 || int || Relocation table offset
|-
|-
| 0x1c || u32 || File size
| 0x1c || int || File size
|-
|-
| 0x20 || u16 || Number of [[#Flowchart|flowchart]]s
| 0x20 || u16 || Number of [[#Flowchart|flowchart]]s
Line 44: Line 46:
| 0x24 || u32 || Padding
| 0x24 || u32 || Padding
|-
|-
| 0x28 || Flowchart* || Flowchart (nullptr if no flowchart)
| 0x28 || Flowchart** || Flowchart (nullptr if no flowchart)
|-
|-
| 0x30 || [[#Dictionary|Dictionary]]* || Flowchart name dictionary
| 0x30 || [[#Dictionary|Dictionary]]* || Flowchart name dictionary
|-
|-
| 0x38 || Timeline* || Timeline (nullptr if no timeline)
| 0x38 || Timeline** || Timeline (nullptr if no timeline)
|-
|-
| 0x40 || Dictionary* || Timeline name dictionary
| 0x40 || Dictionary* || Timeline name dictionary
Line 63: Line 65:
| 0x0 || char[4] || Magic ("RELT")
| 0x0 || char[4] || Magic ("RELT")
|-
|-
| 0x4 || u32 || Offset to relocation table start
| 0x4 || int || Offset to relocation table start (<code>table_start_offset</code>)
|-
|-
| 0x8 || u32 || Number of sections
| 0x8 || int || Number of sections
|-
|-
| 0xc || u32 || Padding
| 0xc || u32 || Padding
|-
|-
| 0x10 || Section[num_sections] || Sections
| 0x10 || Section[num_sections] || Sections
|-
| - || Entry[...] || Section entries (emitted in the same order as each section)
|}
|}


Note: the number of sections is almost always 1 because a single section can already fit 2^32 - 1 entries. If you need more than 4 billion entries, you are probably doing something ''very wrong''.
Note: the number of sections is almost always 1 because a single section can already fit 2^32 - 1 entries. If you need more than 4 billion entries, you are probably doing something ''very wrong''.
The table base pointer is calculated as follows: <code>reinterpret_cast<char*>(&table) - table_start_offset</code>


==== Relocation table section ====
==== Relocation table section ====
Line 78: Line 84:
! Offset !! Type !! Description
! Offset !! Type !! Description
|-
|-
| 0x0 || void* || Alternative base offset (unused by Nintendo)
| 0x0 || void* || Optional base pointer (<code>ptr</code>)
|-
| 0x8 || u32 || Used to calculate the base pointer if an alternative base offset is used
|-
|-
| 0xc || u32 || Data end offset (before alignment)
| 0x8 || int || Optional base pointer offset (<code>offset</code>)
|-
|-
| 0x10 || u32 || Number of entries to skip
| 0xc || int || Size
|-
|-
| 0x14 || u32 || Number of entries (includes skipped entries)
| 0x10 || int || Index of the first entry in this section
|-
|-
| 0x18 || Entry[num_entries] || Entries
| 0x14 || int || Number of entries
|}
|}
The base pointer (<code>base</code>) is calculated as follows:
* If the optional base pointer field is nullptr, the base pointer is equal to the table base pointer.
* Otherwise, it is equal to <code>ptr - offset</code>.


==== Relocation table entry ====
==== Relocation table entry ====
Line 95: Line 103:
! Offset !! Type !! Description
! Offset !! Type !! Description
|-
|-
| 0x0 || u32 || Offset to pointers to relocate
| 0x0 || u32 || Offset to pointers to relocate, relative to the table base pointer
|-
|-
| 0x4 || u32 || Bit field that determines which pointers need to be relocated (up to 32 contiguous pointers starting from the listed offset)
| 0x4 || u32 || Bit field that determines which pointers need to be relocated (up to 32 contiguous pointers starting from the listed offset)
|}
|}
Offsets are treated as signed 32-bit integers.
Pointers are relocated by overwriting each 64-bit pointer field with <code>base + offset</code>.


=== String pool ===
=== String pool ===
Line 130: Line 142:
! Offset !! Type !! Description
! Offset !! Type !! Description
|-
|-
| 0x0 || u32 || Compact representation of bit index
| 0x0 || u32 || Bit index
|-
|-
| 0x4 || u16 || Next index if bit is 0
| 0x4 || u16 || Index of next node if bit is 0
|-
|-
| 0x6 || u16 || Next index if bit is 1
| 0x6 || u16 || Index of next node if bit is 1
|-
|-
| 0x8 || PascalString* || Name
| 0x8 || PascalString* || Name
|}
|}
The compact representation of the bit index has two parts:
* Bits 3-7: index of the byte that should be checked
* Bits 0-2: index of the bit in that byte
A bit index can be translated to its compact representation using:
<syntaxhighlight lang="python">
def get_compact_representation(bit_idx: int) -> int:
    byte_idx = bit_idx // 8
    return (byte_idx << 3) | (bit_idx - 8*byte_idx)
</syntaxhighlight>


Example: <code>Hello</code> corresponds to 100100001100101011011000110110001101111. Bits 0-3 are 1, bit 4 is 0, etc.
Example: <code>Hello</code> corresponds to 100100001100101011011000110110001101111. Bits 0-3 are 1, bit 4 is 0, etc.


=== Container ===
=== Container ===
(The official name for this structure in the EventFlow library is ore::ResMetadata.)
(The official name for this structure in the EventFlow library is ore::ResMetaData.)


Containers are key-value mappings with keys being strings. A [[#Dictionary]] is used to store keys.
Containers are key-value mappings with keys being strings. A [[#Dictionary]] is used to store keys.
Line 196: Line 197:
| String || String (not in string pool, follows item structure immediately) || 1
| String || String (not in string pool, follows item structure immediately) || 1
|-
|-
| Wide string ("wstring") || (not implemented) || ?
| Wide string ("wstring") || Wide string (not in string pool, follows item structure immediately) || 1
|-
|-
| Int array || Int[n] || n
| Int array || Int[n] || n
Line 206: Line 207:
| String array || String[n] (aligned to 8-byte boundaries this time) || n
| String array || String[n] (aligned to 8-byte boundaries this time) || n
|-
|-
| Wstring array || (not implemented) || ?
| Wstring array || WString[n] (aligned to 8-byte boundaries this time) || n
|-
|-
| Actor identifier || Two strings: actor name + secondary name || 2
| Actor identifier || Two strings: actor name + secondary name || 2
|}
|}
Wide strings use wchar_t which is 32-bit long on Switch.


==== ContainerDataType enum ====
==== ContainerDataType enum ====
Line 268: Line 271:
| 0x34 || u16 || Entry point index for associated entry point (0xffff if none)
| 0x34 || u16 || Entry point index for associated entry point (0xffff if none)
|-
|-
| 0x36 || u8 || Cut number? This is set to 1 for flowcharts. Timeline actors sometimes use a different value here. {{check}}
| 0x36 || u8 || Cut number? This is set to 1 for flowcharts. Timeline actors sometimes use a different value here. {{check}} In BotW, this value is passed as the @MA actor parameter<ref>Switch 1.5.0 0x7100DA8628</ref>.
|-
|-
| 0x37 || u8 || Padding
| 0x37 || u8 || Padding
Line 346: Line 349:
| 0x1a || u16 || Number of variable definitions
| 0x1a || u16 || Number of variable definitions
|-
|-
| 0x1c || u16 || Main event index
| 0x1c || u16 || Main event index. 0xffff if the entry point doesn't point to any event.
|-
|-
| 0x1e || u16 || Padding
| 0x1e || u16 || Padding
Line 497: Line 500:
! Offset !! Type !! Description
! Offset !! Type !! Description
|-
|-
| 0x0 || float || Duration {{check}}
| 0x0 || float || Start time {{check}}
|-
|-
| 0x4 || u32 || ? {{check}}
| 0x4 || u32 || ? {{check}}
Line 532: Line 535:
* Actors
* Actors
* Clips
* Clips
* Oneshots? {{check}}
* Oneshots {{check}}
* Subtimelines
* Subtimelines
* Triggers
* Triggers
* Cuts
* Cuts
* Clip param containers
* Clip param containers
* Oneshot param containers {{check}}
* Cut param containers {{check}}


=== Flowchart ===
=== Flowchart ===
Line 566: Line 571:
* [https://github.com/leoetlino/event-editor EventEditor]: graphical editor for event flows
* [https://github.com/leoetlino/event-editor EventEditor]: graphical editor for event flows


<hr>
== References ==
 
<references/>
<references/>


[[Category:File formats]]
[[Category:File formats]]
Retrieved from "https://zeldamods.org/wiki/Special:MobileDiff/1641...11752"