[hneogy]

• Tickets addressed: satTle2elem: type 4 sets run as SGP4, and three ways one set fails the whole file #1676
• Review: By commit
• Merge strategy: Merge (no squash)

Description

Fix TLE reader failures involving blank launch metadata, unsupported ephemeris types, numeric satellite names, and malformed records. Preserve the existing fail-fast default while adding explicit recovery through on_error="skip".

Parsing fixes

• Accept blank international designators in columns 10–17 of line 1. launchDate remains None; launchNo and pol retain their defaults of 1 and "A", which are placeholders rather than known launch metadata.
• Accept ephemeris types 0, 2, and a blank field for SGP4 conversion. Reject other types, including 4 (SGP4-XP), instead of producing states with an incompatible propagator.
• Recognize complete record layouts before counting element-line prefixes, preserving names such as 1KUNS-PF, 2026-195B, 1 TEST, and 2 TEST. Support these prefixes in optional 2LE file titles.
• Ignore trailing blank lines.
• Reject recognizable orphaned element lines in satellite-name and file-title positions, including truncated lines, malformed catalog fields, and shortened headers. Intentional element-like titles can be disambiguated with a 0 prefix; satellite names retain that prefix.

Optional error policy

satTle2elem(path)                   # Default: raise on the first record error
satTle2elem(path, on_error="raise")  # Explicitly request the default
satTle2elem(path, on_error="skip")   # Warn and return readable records

on_error is keyword-only. Unsupported values raise ValueError before opening the file.

Skip mode catches ValueError and ArithmeticError during complete-record parsing or orbital conversion, including checksum failures, unsupported ephemeris types, epoch overflow, and division by zero. Each skipped record produces a printed warning. If every attempted record fails, the first original exception is raised.

File access errors, file-level format errors, and unrelated exceptions propagate in either mode. The released default remains supported, so this addition requires no deprecation period.

Documentation

Update the API docstring, release notes, and known-issue entries to describe both policies, supported inputs, metadata defaults, and recovery limitations.

Validation

• 1,296 TLE/OMM tests passed.
• Constellation integration test passed using Matplotlib’s headless backend.
• 8,135 additional adversarial checks passed, covering interacting fixes and shortened orphan lines.
• Sphinx documentation built without warnings.
• git diff --check passed.

The full repository test suite and a clean native rebuild were not rerun for the final revision.

[schaubh]

1. Record identification and satellite-name preservation

   The original prefix check needed additional handling for valid names such as 1 TEST and 2 TEST. The updated reader recognizes complete record layouts before counting element-line prefixes, preserving names and their associated satellites. It also ignores trailing blank lines and rejects recognizable orphaned element lines—including truncated lines, malformed catalog fields, and shortened headers—in both title positions. Intentional element-like titles can use the documented 0 prefix.

2. Malformed-record recovery

   Catching only ValueError left numerical failures capable of aborting recovery, including epoch overflow and division by zero. Skip mode now handles both ValueError and ArithmeticError during record parsing and orbital conversion. Unrelated exceptions still propagate. Regression tests cover failures at the beginning, middle, and end of files, preserve valid records and their orbital elements, and verify that an entirely unreadable file raises its first original error.

3. Backward compatibility through on_error

   Unconditionally skipping failed records would change the released API’s behavior. The new keyword-only on_error option preserves "raise" as the default and makes recovery explicitly opt-in with "skip". Existing callers continue to stop at the first record error without receiving partial results. Invalid policy values are rejected before opening the file, and file-level validation remains active in both modes. Because the existing default remains supported, this addition does not require a deprecation period.

4. Documentation and final validation

   The API documentation, release notes, and known-issue entries now describe the supported ephemeris types, blank-designator placeholders, title disambiguation, both error policies, and recovery limitations. The workarounds also account for the OMM reader being unavailable in affected older releases.

   Final local validation passed: 1,296 TLE/OMM tests, the constellation integration test using a headless plotting backend, and 8,135 additional adversarial checks. Sphinx built without warnings. The full repository suite and a clean native rebuild were not rerun locally. I found no additional actionable defects in the complete updated diff.

[schaubh]

@hneogy , thanks for the support. I made a view review comments you can look over, and I pushed commits to address them. If you agree then please close my comments and let me know. I'll then approve.

[hneogy]

thanks for going through it so carefully! I've looked over your four commits and they look good to me. making the skip opt-in with on_error makes sense for the released API, catching the arithmetic errors too is a good call, and thanks for fixing my known-issue workaround, I'd missed that the OMM reader isn't in those releases. the TLE tests and CelesTrak's analyst file (219 of 219) still check out on the updated branch. resolving your comments now.