First CRAN submission. The version moves to 1.0.0 to mark a stable interface for read_lst_file(), the fetch_*() parsers and run_app().
summary() on an "lst" object now returns a "summary.lst" object (number of lines, estimation method, covariance step) with its own print() method, instead of printing with cat() and returning the input invisibly. Printed at the console, the output is unchanged.
The Shiny app’s ZIP download no longer prints the names of the files it adds to the console.
DESCRIPTION declares Language: en-US, and an inst/WORDLIST records the technical terms used in the documentation.
A synthetic test corpus replaces the external corpus of study listings, which is no longer part of the repository. inst/extdata/lst/ holds 39 small, invented listings written by data-raw/make_synthetic_lst.R, with a manifest and the expected values of every case. The cases cover FO, FOCE, FOCE-I, LAPLACE (with the M3 likelihood), ITS, SAEM, IMP (including evaluation-only steps), IMPMAP and BAYES; chained estimation steps and multiple problems; successful, terminated and evaluation-only runs; completed, aborted and singular covariance steps; eigenvalues that are not printed, negative or zero; fixed, block and SAME parameters; non-finite standard errors; extreme magnitudes; wrapped records; Windows line endings; and truncated, damaged or failed runs. The tests compare every parser with these known values, and the “Getting Started” vignette parses some of the cases.
Examples, vignettes, README, snapshot tests and the Shiny app test use the synthetic listings. The listings from study runs that were bundled in inst/testdata/ and tests/ have been removed, and the snapshots are recorded as readable deparsed values.
fetch_thetas() keeps standard errors aligned with their parameters when a field prints ......... (a fixed THETA) or +Infinity. Previously the following values moved one column to the left and a padding warning was issued. ......... and NaN now give NA, +Infinity gives Inf, in every parameter table.
fetch_etas() and fetch_sigmas() return all random effects when there are ten or more. From the tenth on, NONMEM labels them ET10, EP10 and so on, and these rows were dropped. Matrix rows longer than 12 values, which continue on a record without +, are read in full.
fetch_etas() reads ETA shrinkage records that NONMEM wraps after ten values, instead of returning NA shrinkage with a warning.
fetch_condn() again returns NA without a warning when the covariance step ran with EIGENVLS. PRINTED: NO. Since 0.1.2 this flag was looked for only in the final estimation step, but NONMEM prints it in the problem header.
Method headers for LAPLACE (LAPLACIAN CONDITIONAL ESTIMATION, with or without interaction), ITS (ITERATIVE TWO STAGE) and BAYES as NONMEM prints it (MCMC BAYESIAN ANALYSIS) are recognized.
lstparsR 0.1.2
Preserve scientific exponents and missing matrix diagonals, and read printed ETA shrinkage with or without an equals sign.
Select a consistent final estimation step across parameter and scalar results.
Handle truncated objective output, grouped eigenvalues and singular matrices.
Preserve duplicate uploads, expose incomplete results, and improve table layout and download isolation.
Clarify supported output layouts and first-submission metadata.
lstparsR 0.1.1
Bug fixes
fetch_ofv(): No longer warns “OFV not found” for $SIMULATION-only runs (e.g. VPC simulation outputs). Such runs legitimately produce no objective function value; the function now returns NA_real_ silently in that case. The warning is still emitted when an estimation step ran but its OFV cannot be parsed.
fetch_condn(): No longer warns “Eigenvalue section not found” when the user set EIGENVLS. PRINTED: NO in $COV. In that case the covariance step ran but the eigenvalue block is intentionally absent; the function now returns NA_real_ silently. The warning is still emitted when eigenvalues are expected but cannot be located.
Reported by Christiane on multiple unrelated projects.
Internal
New internal helpers .has_estimation_step() and .eigenvalues_suppressed() in R/utils-internal.R.
lstparsR 0.1.0
Initial development release.
read_lst_file(): reads NONMEM .lst files; returns S3 class "lst".