LoginSign Up
← Back to all sparks
P

parzer

ANALYTICS
Velocity0.0

Parse geographic coordinates. Contribute to ropensci/parzer development by creating an account on GitHub.

github.com/ropensci/parzer ↗

A coordinate parser whose entire job is surviving how badly humans write latitude and longitude.

geospatialstring-parsingcoordinatesropenscimaintenance
◆Current state
parzer converts messy coordinate strings — degrees, minutes, seconds, assorted symbols, arbitrary whitespace — into decimal degrees. Development is slow and sporadic, with three-year gaps between releases, and the work splits between C++ performance in the internal scrub() path and a long tail of parsing bugs. The most recent release, 0.4.4, fixed two genuinely dangerous ones: a leading space could silently drop a negative sign, and an E in a longitude string returned NA while a W parsed fine.
◆Where it's heading
The package has settled its scope — 0.4.1 explicitly rewrote the documentation to say it parses coordinates rather than validates them — and now moves only when someone finds a string it mishandles. Recent work has also been about shedding weight: Rcpp dependence reduced, the C++ requirement dropped from DESCRIPTION, suggested dependencies removed, and the vignette builder moved to Quarto. Maintainership passed to a new maintainer in 2022 and the package has stayed within rOpenSci.
◆Prediction
The next release will most likely be another batch of parsing edge cases reported by users, since that is what every release since 0.2.0 has been. Nothing in these entries points to new functionality.

◆Recent moves

  1. 1y ago

    Fixes dropped negative signs and mis-parsed E longitudes

    Two parsing bugs of the worst kind for this package: a leading space could make a negative sign disappear, and an E in a longitude string returned NA where W parsed correctly. Both produce wrong coordinates rather than errors. The release also trims dependencies and rewrites C++ hot paths to assign directly instead of pushing back.

    View source ↗
  2. 4y ago

    Scope clarified: parsing, not coordinate validation

    Documentation-only, but it settles a recurring misunderstanding by stating that the package parses varied coordinate formats and does not exist to validate them. A new maintainer takes over in the same release.

    View source ↗
  3. 5y ago

    Faster scrub(); works around non-UTF8 locales on Windows

    Speeds up the internal scrubbing function used by nearly every exported function, and routes calls through a locale-adjusted path so Windows users on non-UTF8 MBCS locales get correct results. Locale-dependent parsing failures are hard for users to diagnose, which makes this more valuable than its size suggests.

    View source ↗
  4. 5y ago

    Fixes factor conversion in parse_llstr() on older R

    A single fix for older R versions where stringsAsFactors defaulted to TRUE and broke an internal step. No effect on current R installations.

    View source ↗
  5. 5y ago

    parse_llstr() parses latitude and longitude from one string

    Adds the function for the most common real-world case — a single string containing both coordinates — which previously required splitting by hand. The scrub() character set was widened at the same time.

    View source ↗
  6. 6y ago

    More degree symbols recognised; NA handling fixed in C++

    Widens the set of degree-like symbols the parser accepts and stops NAs from generating warnings on the C++ side. Routine additions to the long tail of formats the package tolerates.

    View source ↗