Module: Emfsvg::FormatHelpers

Defined in:
lib/emfsvg/format_helpers.rb

Overview

C-printf-compatible float formatting. libemf2svg uses C's printf with the platform's default rounding (round-half-to-even on the actual IEEE-754 double bits). Ruby's Float#round and format("%.4f", v) both lose precision for values landing near the 4-decimal midpoint because they operate on the display value rather than the true bits.

Example: 459.04375 stored as 459.04374999… → C: "459.0437", Ruby format: "459.0438". We expose the actual bits via "%.20f" then round via BigDecimal for exact IEEE-754 semantics.

Class Method Summary collapse

Class Method Details

.fmt(n) ⇒ Object

Format a single float as "%.4f" with C printf semantics.



19
20
21
22
 
# File 'lib/emfsvg/format_helpers.rb', line 19

def fmt(n)
  bd = BigDecimal("%.20f" % n.to_f)
  format("%.4f", bd.round(4, half: :even).to_f)
end

.fmt_xy(x, y) ⇒ Object

Format a coordinate pair as "X,Y " (trailing space, matching libemf2svg's point_draw output).



26
27
28
 
# File 'lib/emfsvg/format_helpers.rb', line 26

def fmt_xy(x, y)
  "#{fmt(x)},#{fmt(y)} "
end