Write the ACIS payload of 3DSOLID, REGION and BODY to DXF and DWG#1150
Open
ilCosmico wants to merge 15 commits into
Open
Write the ACIS payload of 3DSOLID, REGION and BODY to DXF and DWG#1150ilCosmico wants to merge 15 commits into
ilCosmico wants to merge 15 commits into
Conversation
… and 3 for pre-2013 targets
…try entity layout The section content is byte-identical to the reference produced by the reference writer for the same payloads. A residual issue in the section framing (page count / section id) still prevents third party readers from attaching the payloads; the managed round-trip reads it back correctly.
…entities After the ACIS payload the entity carries a wireframe-data-present flag and, from R2007 on, a trailing int. Without them the readers pick up garbage bits past the payload and run off the object data.
The 350 group is mandatory since the 2007 format: when it misses, AutoCAD discards the whole DXF as invalid input.
The CAD writers emit SAT text as version 1 blocks at every release, with the leading bit set, and reserve version 2 for the raw binary payload with the bit clear. Writing text through version 2 produces entities the modeler rejects.
The writer left every section id at 0 and skipped the empty section that the format places first in the descriptor map. Plain drawings open anyway since applications resolve sections by name, but the AcDs data store keys on the section id, so its records were dropped as out of range. Number the sections from 1 and prepend the empty section, matching the reference.
The reference writers set the has-revision-guid flag to 0 and write no revision fields. Emitting the flag as 1 with placeholder fields makes the Autodesk modeler reject the entity as out of range, even though the managed reader skips the fields and round-trips either way.
Decoded from the raw bits of the reference entities: the R2013+ stream has no ACIS empty bit, it opens with the wireframe block, then the trailing int, the revision guid in its native layout and the end marker. The extra leading bit shifted every following field, so the CAD readers picked up garbage values and dropped the entities as out of range.
Decoded from an Autodesk-written R2013 file of the same scene: the block is four bits (acis empty, no wireframe, acis empty again, no revision guid), not the wireframe-plus-guid layout written before. The stray leading bit and the random revision guid made the modeler reject the solids.
…oCAD fresh-save layout
The last header field of the _data_ segment locates the payload area: its value times 16 is the offset from the segment header to the first payload. The writer hardcoded the value captured from a nine-record reference file, so the CAD readers anchored the payloads at the wrong offset for any other record count and every ACIS payload came back as garbage. Compute the field from the directory size instead; the value now matches the reference files at every record count.
Contributor
Author
|
Rebased on current master (which now includes #1149) and added one commit. Following the discussion in #1152, I audited the AcDs writer field by field against chapter 24 of the .dwg file specification, and that audit found one real defect: in the search segment, the per-handle index stored the record's directory position instead of its rank in the sorted handle order. The two coincide when the directory is already sorted by handle, which is why every file opened fine so far; a file saved by AutoCAD with out-of-order handles shows the correct behavior. The new commit fixes it and adds regression tests built from that file. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This is the write side of #1139 (reading the ACIS payload of 3DSOLID, REGION and BODY), now merged. The DWG and DXF writers used to skip those entities; they now write the ACIS payload back.
The payload is the same SAT or SAB byte stream regardless of the file type. Only the container changes with the version, so the writer puts the bytes where each version expects them:
The R2013 and later modeler geometry entity read path is aligned with the same layout (no ACIS empty bit and no silhouette block before the wireframe), so the reader and writer agree.
Tests cover the SAT text codec, the DXF ACDSDATA section, and the DXF SAT round-trip, next to the existing writer tests.
One dependency worth noting: R2013 and later DWG files also need the small viewport reference-type fix in #1149 to open cleanly in AutoCAD.