098ddd3a95
make copy constructor and a assignment operator private to avoid misuse of helper class scope guard
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-11-29 11:56:39 +01:00
9f80a2a2a9
update parser.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-11-29 11:33:49 +01:00
c08a5581dd
ColladaParser::CopyVertex: don't make a local copy of the vertex' indices.
...
Instead, we address the relevant indices in the main vector directly by
adding the input offsets to the per-vertex base offset.
2014-11-20 17:53:06 +01:00
8dead93504
ColladaParser: don't arbitrarily limit number of per-vertex attributes.
2014-11-20 16:42:41 +01:00
ca3ce4703d
Rename ColladaParser::CopyPrimitive to ::CopyVertex.
...
That's a bit closer to what the function actually does.
2014-11-20 13:12:09 +01:00
13a157ea66
For Polylists, manually compute the start offset outside CopyPrimitive.
2014-11-20 13:10:15 +01:00
e568c2df9d
ColladaParser::ReadIndexData: don't rely on the specified number of primitives.
...
Especially TriStrips don't specify triangle count upfront, so one needs to
sum up the count of triangles read from all <p> tags in a tristrip collection.
2014-11-19 14:40:43 +01:00
72f67ea0cd
Whitespace changes
2014-11-19 14:40:42 +01:00
593484829b
Add Function ColladaParser::ReadPrimTriStrips.
...
This does the index mangling for reading tristrips.
2014-11-19 14:40:42 +01:00
0f40dcca7e
Add calculation for nuber of Primitives for TriStrips
2014-11-19 14:40:42 +01:00
7c4bf383d7
Move vertex copy operation into switch statement.
...
We will be adding special handling for Tristrips in an upcoming commit,
which unfortunately doesn't map to a common copy loop as it is.
2014-11-19 14:40:42 +01:00
4ef244f672
Extract function ColladaParser::CopyPrimitive
2014-11-19 14:40:42 +01:00
70d2c8571b
ColladaParser::ReadPrimitives: get rid of "idx" iterator.
...
We will be adding non-sequential access to the indices vector in one of the upcoming commits.
2014-11-19 14:04:49 +01:00
543bd5c0d0
ColladaParser::ReadPrimitives: rename "a" to currentPrimitive
2014-11-19 13:51:22 +01:00
c7b6b5ea2f
Support lowercase 'ns' and 'ni' obj material tokens
...
Previously, the obj material parser would interpret all key tokens
starting with 'n' as 'newmtl' tokens. However, there are other tokens
that can start with 'n', namely 'ns' and 'ni' for material properties.
Make the parser robust to such cases, despite their canonical forms
being uppercase.
Resolves #410 .
2014-11-18 16:56:46 -08:00
dbf9536213
bugfix: add correct handling for metric line end for example file.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-11-02 20:30:45 +01:00
1a5695ff48
update: add parsing of metric nodes.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-11-02 19:01:04 +01:00
6843c42da1
update:
...
- merge master into branch
- fix metrix header parsing.
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-10-28 21:45:09 +01:00
f84589fec4
Merge branch 'master' into opengex_support
2014-10-24 15:24:32 +02:00
cee583db06
bugfix: fix compiler warning: converting double to float.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-10-22 23:05:25 +02:00
b483be3069
Merge pull request #402 from jdduke/disable_bmesh
...
Temporarily disable Blender BMesh conversion
2014-10-10 22:46:25 -07:00
a1a1bf77a8
Temporarily disable Blender BMesh conversion
...
Disable BMesh conversion while associated unresolved issues receive due
attention. Referencing #400 , #373 , #318 #315 and #132 .
2014-10-10 14:16:42 -07:00
487c9449d2
bugfix: search for token checks for the given tokenlist against the lowercased info coming from the asset file. So we also have to used a lowercased token. Thanks to Robinson for the report.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-10-10 15:59:42 +02:00
88f6e8213b
Android JNI asset extraction support (API 9+)
...
- installation fixes
Signed-off-by: Filip Wasil <filip.wasil@tieto.com>
2014-10-07 17:05:14 +02:00
7224476ee1
Android JNI asset extraction support (API 9+)
...
Signed-off-by: Filip Wasil <filip.wasil@tieto.com>
2014-10-07 16:33:43 +02:00
397989e169
Null-terminate string
...
In theory it shouldn't be necessary since the length is specified but there are parts like https://github.com/assimp/assimp/blob/master/code/MaterialSystem.cpp#L74 that depend on it.
2014-09-26 01:29:23 +02:00
a5afbcf29c
refactoring: remove duplicate code.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-23 23:34:05 +02:00
775b26e614
use std:: namespace for most cmath functions:
...
http://en.cppreference.com/w/cpp/header/cmath
2014-09-23 00:43:33 +02:00
b359deb7fd
update:
...
- add assert test in obj-import
- fix material token in Obj.
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-21 20:37:36 +02:00
4dcd316430
Use strict inequality when validating BMesh uvloop access
...
Change the inequality to a strict inequality when validating
BMesh uvloop access during face conversion.
2014-09-20 10:23:21 -07:00
6a8f69ad5d
Merge pull request #386 from jdduke/rebase_fbx_fov
...
Rebase FBX regression db entries
2014-09-16 09:37:37 -07:00
aac40dd5b8
Merge pull request #385 from jdduke/fix_blend_and_rebase_regression_tests
...
Fix non-BMesh-containing .blend import
2014-09-16 01:10:53 -07:00
046b7e5f47
Rebase FBX regression db entries
...
The FBX importer now properly parses field-of-view tokens. Rebase
only the failing FBX entries to accomodate the fix, also removing
the comment noting the (now incorrect) comment about improper import
of the FOV property.
2014-09-15 18:14:38 -07:00
c482458cae
Fix non-BMesh containing .blend import
...
BlenderBMeshConverter was erroneously asserting that it contained valid
mesh upon construction. The assertion should only occur when
triangulation is requested, allowing non-BMesh blender meshes to load
properly.
Note that this fixes a number of .blend regression tests that are
currently baselined as failures.
2014-09-15 17:19:33 -07:00
ab6b5a7b21
Fix gcc warnings. (initialization order)
2014-09-15 23:35:03 +04:00
35c9a33718
Fix crash for Blender poly to face conversion without UVs
...
BlenderBMeshConverter assumes that the BMesh has UVs, but that is not
always the case. Handle the non-UV case by only adding a T face when
UVs are present.
2014-09-13 09:43:00 -07:00
485aa8b39d
Fix null check in Q3BSPFileImporter
...
Reverse the null condition, preventing invalid texture stream
dereferencing. Fixes #380 .
2014-09-12 19:37:10 -07:00
db582ee024
bugfix: replace nullptr with NULL:
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-12 18:12:18 +02:00
49c9786b0a
update: avoid mesh generation on heap, when the mesh will be empty ( obj-loader ).
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-12 15:40:50 +02:00
7dff0c6d52
refactoring: remove not use iosystem.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-12 12:05:42 +02:00
75d3c8e9f2
update: metric parsing ongoing.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-09-12 12:00:03 +02:00
43c82b0b38
Remove std::isnan and std::isinf usage from utFastAtofTest
...
These functions were added in C++11, and should not be used nakedly
in the current code base. Replace them with suitable C++03 constructs.
2014-09-09 16:34:10 -07:00
66e608a393
Replace cppunit with the googletest framework
...
The GTest framework has a more active support base, and natively
supports CMake. Introduce it as an external dependency (using CMake's
ExternalProject_Add), replacing cppunit and porting the associated unit
tests.
2014-09-07 16:52:03 -07:00
c3b35f3933
update: add endline.
2014-09-06 09:19:28 +02:00
185f01963f
LWO: replace legacy reinterpret_cast based code by memcpy to solve Android SIGBUS problems due to unaligned accesses. Fixes #351
2014-09-04 20:49:48 -07:00
1a594b861a
Collada: use a counter instead of clock() to generate unique names. Closes #357
2014-09-04 18:50:37 -07:00
ed9a466c30
update:
...
- add OpenGEX parser sceleton
- add base datatypes for OpenGEX support.
2014-09-04 19:41:14 +02:00
8bf26065de
[FBX] parse upper cased Int/Enum properties
2014-09-01 19:06:59 +03:00
625d2b7c4a
Fix #325 ; Blender UV unwrap issue
...
The BMesh converter might be missing more details, but this should get basic UVs working for now.
2014-08-30 00:50:26 -04:00
201807f733
Merge pull request #350 from Gargaj/ambientlight
...
Enable ambient lights
2014-08-28 09:20:04 -07:00
d333e08343
add ambient light type, allow collada to load it
2014-08-28 16:27:29 +02:00
e2401c8b58
fix va_end call
2014-08-28 15:51:01 +02:00
6f828087f9
add stdarg.h
2014-08-28 15:45:48 +02:00
19de24a442
...and some more assbin leftovers
2014-08-28 15:42:22 +02:00
8f0208e5a8
whoops, still some residue from ASSBIN
2014-08-28 15:41:17 +02:00
c47dc2802e
convert assxml exporter
2014-08-28 15:38:57 +02:00
2e9195819a
start assxml exporter basics
2014-08-28 15:17:53 +02:00
7f45f5fc95
bugfix: fix license.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-08-27 23:01:35 +02:00
8126dce94d
update: add sceleton for OpenGEX importer.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-08-27 23:00:19 +02:00
9d04a45088
Collada : Add missing parsing <image> into <effect><profile> markup
2014-08-27 10:15:13 +02:00
33ffb0003e
Collada Export: escape user-defined strings in XML output.
2014-08-23 15:42:47 -07:00
fddae20cb7
Merge pull request #338 from jdduke/fast_atof_fix_final
...
Gracefully handle NaN/inf values in fast_atoreal_move
2014-08-20 19:29:44 -07:00
ed124ccbab
Gracefully handle NaN/inf values in fast_atoreal_move
...
There are legitimate cases where inf/nan values are embeddded in a mesh.
Such values should not cause loading to fail, and indeed, previous versions of
Assimp supported their existence. Update the new fast_atoreal_move method
to gracefully parse such values, allowing case-insensitive checks for "NAN",
"INF" and "INFINITY" as per the atof guidelines found at
http://en.cppreference.com/w/cpp/string/byte/atof .
Note that the inf/nan text parsing is fairly loose, but is fast for the general
case and should handle most legitimate inf/nan values.
2014-08-20 18:42:37 -07:00
7ff7b783ee
Merge pull request #335 from abma/master
...
fix most warnings with -pedantic
2014-08-17 17:49:21 -07:00
6c5f9ecb06
fix most warnings with -pendantic
2014-08-18 02:09:06 +02:00
6636e596ab
Merge pull request #334 from jdduke/fbx_binary_fix
...
Avoid raw reinterpret_casts in the FBX parser
2014-08-17 17:01:10 -07:00
48a772b20b
fix compiler warnings:
...
- defined but not used
- deprecated conversion from string constant to ‘char*’
2014-08-18 00:46:21 +02:00
d795491c48
fix unittests on linux / enable on travis-ci
2014-08-18 00:34:45 +02:00
56ddb4f4af
fix some compile warnings:
...
- passing NULL to non-pointer
- unused vars
- unused function
- order of init
- parentheses
2014-08-17 23:28:00 +02:00
aafc8d5f3f
CMake: changed cached bools into "option"
2014-08-17 23:10:06 +02:00
69810a2a07
Avoid raw reinterpret_casts in the FBX parser
...
As reinterpret_cast can break strict aliasing rules, causing runtime failure
on Android, replace such usage in FBXParser with memcpy. Also provide a
utility routine for both performing the copy and asserting the validity of
the buffer length relative to the copied region.
2014-08-17 08:41:45 -07:00
5e265610fb
bugfix: add obj-material handling for <color> 0 instead of <color> r g b.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-08-16 11:39:28 +02:00
58b78fa265
comment
2014-08-08 13:10:05 +02:00
fa99aa0d3a
add remaining specializations
...
hopefully i didnt mess this up too bad
2014-08-08 12:56:11 +02:00
b712bf1770
Change to looped writes
...
to avoid struct packing issues
2014-08-08 12:46:29 +02:00
9f795c64d5
Merge pull request #326 from Gargaj/newprogresshandler
...
More elaborate progress handler
2014-08-07 11:45:29 +02:00
63a110468f
Merge pull request #327 from Gargaj/assbinintegrate
...
Integrate import/export for .ASSBIN format
2014-08-07 11:41:35 +02:00
5886154247
tweaks according to the comments of kimkulling
2014-08-07 11:30:44 +02:00
c3977bd132
unify data types
2014-08-07 09:32:00 +02:00
61ecff3474
missed one 32/64 type conversion
2014-08-07 09:30:20 +02:00
b53a425fe4
make templates gcc compatible
2014-08-07 09:28:50 +02:00
2775141da3
#include should be outside
2014-08-07 09:07:09 +02:00
3c172988bb
remove BUG notices (discussed with acgessler)
2014-08-07 08:43:52 +02:00
df4b17d145
code style
2014-08-07 08:26:34 +02:00
aa5c1a1a23
syntax
2014-08-07 08:26:04 +02:00
7925dcadba
avoid NULL, just in case
...
(loader might just probably crash anyway)
2014-08-07 08:25:10 +02:00
523d87bbe7
use 0..fileSize as progress metric (easier for loaders)
2014-08-07 08:22:47 +02:00
061911bdbf
add support for compressed, drop support for shortened
2014-08-07 08:05:07 +02:00
fb546b694e
code style
2014-08-06 15:25:39 +02:00
1aeaef2037
add lights and cameras
2014-08-06 15:01:04 +02:00
97c9ce1d5b
add animations and textures
2014-08-06 14:44:06 +02:00
886a704c17
some notes to self
2014-08-06 14:43:47 +02:00
5048793096
add material import
2014-08-06 14:21:53 +02:00
f38dd8e6e2
add mesh loader
2014-08-06 14:06:08 +02:00
ad298cd84f
fix code style
2014-08-06 13:29:09 +02:00
ac5c1f9df1
add to cmake
2014-08-06 13:27:09 +02:00
87dff6a00e
start importer, load nodes
2014-08-06 13:26:26 +02:00
5b512dd9dd
add info + header check
2014-08-06 00:52:05 +02:00
9cdecc780b
Fix wrong nesting
...
Otherwise doesn't compile if ASSIMP_BUILD_NO_3DS_IMPORTER is defined because the #endif is inside the namespace block
2014-08-06 00:44:30 +02:00
0f822d38cd
add basic framework for importer
2014-08-06 00:39:54 +02:00
cd86818888
- add compression support
...
- remove unneeded code
- change log warning to exception
- document
2014-08-05 23:18:45 +02:00
fc9b4e2adf
follow assimp code style (tabs)
2014-08-05 22:55:20 +02:00
2592ff0796
convert exporter from tools/assimp_cmd
2014-08-05 22:54:45 +02:00
3d5e1b5cbc
add basic framework (files, etc)
2014-08-05 20:54:38 +02:00
6b8cbe5c83
this isn't needed anymore
2014-08-05 00:10:23 +02:00
dbc553343c
Granulate file loading too
2014-08-05 00:05:18 +02:00
dfe1f03e5b
Split off postprocessing progress
2014-08-04 23:57:08 +02:00
203d5d6fcc
Obj: make mtl importer accept kd, ks etc. as well (Kd, Ks being the correct form). This is for keeping up compatibility with assimp's own Obj exporter, which has been inaccurate about this since inception.
2014-08-02 21:18:21 -07:00
f5b9f5cb50
Merge pull request #321 from Gargaj/patch-1
...
Add FieldOfView token type (fixes #245 )
2014-08-02 11:38:23 -07:00
e5ed62581d
Handle both types of element names
...
MAX seems to add an "s" to "Binormal" and "Tangent" when exporting
2014-08-02 14:57:43 +02:00
11af0866dc
Add FieldOfView token type ( fixes #245 )
...
"FieldOfView" has its own token type (basically just one float), if it's missing the camera will keep reverting to default FOV in ASSIMP.
2014-08-02 14:47:59 +02:00
5d861a8250
bugfix:
...
- fix gcc-specific compiler warnings.
2014-07-31 15:09:49 +02:00
9eba139feb
Merge pull request #317 from assimp/export-3ds
...
3DS Exporter
2014-07-23 22:26:58 +02:00
a07cde4a81
3DS Exporter: export scene hierarchy.
2014-07-23 22:08:35 +02:00
026f32733d
3DS Export: fix texture export.
2014-07-23 20:14:32 +02:00
d5db6f39e6
Fix CI build.
2014-07-23 18:54:05 +02:00
6fde07f7e5
Initial implementation of 3DS Exporter. Hierarchy exporting and smoothing groups missing.
2014-07-23 18:27:22 +02:00
0c5605d07d
Don't use isspace() in OBJ file parser.
2014-07-22 21:59:23 -07:00
79c56adea4
Don't use isdigit() in fast_atof.
2014-07-22 21:58:41 -07:00
0166518132
Extend StreamWriter to write to a buffer and to allow re-positioning of the write cursor within the buffer.
2014-07-22 20:36:16 +02:00
3419fedb3c
Add StreamWriter utility (mostly symmetric to StreamReader).
2014-07-22 17:55:37 +02:00
36c82fe5b0
Allow several spaces between numbers in OBJ files.
2014-07-21 23:03:08 -07:00
3e9fab3bfc
Allow numbers starting with decimal in fast_atof.
2014-07-21 22:43:13 -07:00
d655199043
bugfix: fix stupid test introduced by myself.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-07-19 16:22:15 +02:00
46ca36cb6f
update:
...
- remove dead code.
- check for a possible division by zero.
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-07-19 14:45:21 +02:00
2edccb7f34
fast_atof: If a literal ends with a trailing dot (.), only parse further if the next character is a digit. In cases where the dot ends the literal (i.e. "1.") this would cause strtoul10_64 to throw an exception. To preserve as much of the old behaviour, we still consume trailing dots though. This "regression" was introduced (exposed) by #261 , which added the extra check to strtoul10_64 that triggered here. This commit now fixes #304 (IFC file reading broken due to IFC files containing "(1.,1.,1.)" lists.
2014-07-13 15:08:28 +02:00
7e48c1cce9
FBX: fix memory leak, parsed TokenList never released. See #307 .
2014-07-13 14:11:58 +02:00
5ae010028b
FBX: fix memory leak. Connection objects in Document never freed. See #307
2014-07-13 14:07:19 +02:00
a30ea8e92c
Fix overflow in STL header colour reading.
...
When reading the STL header for a "COLOR=rgb" part, the bytes were treated as signed chars, when in fact they can range from 0-255. This meant that any value greater than 127 would overflow, leading to an incorrect colour.
This change fixes the issue by treating the header as unsigned chars.
2014-07-13 00:34:08 +10:00
ce37d3bcad
Fixed possible bug with map_emmissive keyword and removed second check for ambient texture keyword
2014-07-11 08:11:54 +02:00
afa2d3da94
Improved texture map data in mtl exporter (in most cases an upper first character is used)
2014-07-10 08:26:32 +02:00
47e7667071
Added alpha texture map (map_d)
2014-07-09 15:51:09 +02:00
231e4e0cd0
update: commit not saved changes.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-07-04 00:23:02 +02:00
c6b516b68b
bugfix: handling possible nullptr access.
...
update: replace clear be resize( 0 ) at some places to avoid double allocations ( only a micro optimization ).
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-07-04 00:22:13 +02:00
b3b732c12b
update: some micro improvements, replace post-increment operator with pre-increment operator.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-07-03 20:33:52 +02:00
0856ff9659
update: replace some clear / push-back by resize( 0 ) / push_back to avoid redundat allocations.
...
Signed-off-by: Kim Kulling <kim.kulling@googlemail.com>
2014-06-29 21:16:37 +02:00
7abfcd1a4d
Merge pull request #300 from eyethereal/fix-obj-exporter-vertex-references
...
Fixed vertex reference output format in face element for OBJ exporter
2014-06-21 18:55:51 +02:00
1207f84f7b
Merge pull request #298 from Madrich/master
...
Collada fixes, adding XFile Exporter
2014-06-21 18:54:11 +02:00
2c34f24497
Fix XFile define
2014-06-21 17:23:30 +02:00
9fbc9c4fb7
Merge pull request #301 from eyethereal/obj-exporter-no-name-no-group
...
OBJ Exporter: No "g" lines with empty names
2014-06-21 15:21:27 +02:00
77faf04aa3
OBJ Exporter: No "g" lines with empty names
...
In the OBJ spec ( http://www.martinreddy.net/gfx/3d/OBJ.spec ), in the section labeled "Grouping" -> "Syntax", the structure of the "g" group statement is defined. Though this statement allows multiple names on a single line, it is unclear whether there must be at least one name on the line. However, the examples don't show any "g" group statements with no names. So, let's be conservative and not write out a "g" group statement that doesn't have a name. These empty "g" statements were prompting an error message from the three.js OBJ loader code.
2014-06-20 16:06:04 -07:00
d10d20f62f
Update ObjExporter.cpp
...
In the OBJ spec ( http://www.martinreddy.net/gfx/3d/OBJ.spec ), in the section labeled "Referencing groups of vertices", there is no support for allowing a trailing slash with no valid vertex normal reference at the end. This commit modifies the vertex reference output code to prevent this behavior, as it has been causing compatibility issues with other software that parse OBJ files.
2014-06-20 15:08:29 -07:00
1cb01c54a3
Fix IOSFixed
...
Fix Author/AuthoringTool in Collada by MetaData
2014-06-20 00:08:11 +02:00
be5fc42e16
Fix memory corruption in the aiGetExportFormatDescription() API. Clarify Exporter doc.
2014-06-17 16:10:55 +02:00
75fd29ac19
SceneCombiner: implement proper copying of meta data. This entire module will need to be revamped in future as it is highly likely to cause further regressions as assimp data structures are modified.
2014-06-17 15:18:10 +02:00
d044ce7a62
FBXConverter: avoid creating an empty metadata slot (UDP3DSMAX was erased from list _after_ its size was taken)/
2014-06-17 14:31:14 +02:00
244f981f6a
ObjExporter: cleanup code. Fix bug writing normal: normals would not be transformed by world transform.
2014-06-17 14:15:48 +02:00
5efc0207f3
Merge branch 'master' of https://github.com/assimp/assimp
2014-06-13 10:59:16 +02:00
16ae05eeeb
obj loader: added multiline support
...
collada loaded: bug fixed when importing several files, whith one containing animation
2014-06-13 10:56:10 +02:00
edc7a950c4
ReFix the stuff before
...
Fix XFileExporter Normal
Fix Collada (Triangle->Poly)
2014-06-11 00:41:18 +02:00
Kim Kulling
Wolfgang Herget
Johannes Ebersold
Jared Duke
Alexander Gessler
Filip Wasil
Gargaj
abma
Jared Duke
Denis Biryukov
George Papadopoulos
Adam Petrone
unknown
Kim Kulling
Tyson Grant Nottingham
Andrew Short
Robin
Madrich
Nathan Morse
Gregory Jaegy