klammertext/sks/font/font.py

import re
import klammer_base

class Subsuper(klammer_base.Klammer_base):
    def __init__(self, K, position):
        super().__init__(K)
        self.position = position
        """
        print("Subsuper:", self.args)

        parts = self.args.strip().split()
        if len(parts) != 2:
            raise Exception(f"Two arguments required, but argument is: {self.args.strip()}")
        self.base, self.script = parts
        """

    def html(self):
        return f"{self.base}<{self.position}>{self.script}</sub>"

    def tex(self):
        if self.position == "sub":
            return f"${self.base}_{self.script}$"
        else:
            return f"${self.base}^{self.script}$" 

# The typeface-declaration marker.  A klammer cannot declare a typeface
# itself: in html the @font-face must be in the page's css and the files
# localized into <output>/fonts/, and in tex the \newfontfamily must be in
# the preamble -- both are @document's, and @document runs AFTER the
# klammers it contains (application is post-order).  So @typeface leaves
# this marker where the typeface is needed, and @document collects the
# markers, resolves and localizes each typeface, declares it, and replaces
# the marker with whatever inline switch the target needs (a font-switch
# command for tex; for html the marker sits IN the span's font-family value
# and becomes the family name the store resolved the typeface to, which
# is what the localized css declares -- the name as WRITTEN may be the
# store's directory name, "noto-sans-egyptian-hieroglyphs", and a span
# carrying that matched no @font-face until 2026-09-12).  Same shape as
# the __CAPTION__ markers.
#
# SYNC: typeface_marker_rgx in sks/document/document_class.h -- a C++/Python
# pair, so the constant cannot be single-sourced the way a .k/.py pair can
# (TODO #47); change both or neither.
def typeface_marker(name, scale):
    return f"__TYPEFACE__{name}__{scale}__"

class Typeface(klammer_base.Klammer_base):
    """A temporary typeface change (@typeface).  See sks/font/font.k."""

    def __init__(self, K):
        super().__init__(K)

    def html(self):
        # font-size-adjust is switched OFF for the span: the document
        # normalizes families to a common x-height (font.css), which is
        # meaningless for a script whose glyphs have no x-height at all --
        # hieroglyphs, cuneiform -- and would resize them unpredictably.
        # The family name is quoted -- an unquoted digit-initial word
        # ("Source Sans 3") is invalid CSS -- with SINGLE quotes, because
        # this sits inside a double-quoted html attribute.  What is quoted
        # here is the MARKER: @document replaces it with the resolved
        # family name (see the marker note above).
        # The size is the scale times the body's own enlargement: the body
        # is rendered at "font-size-adjust: ex-height 0.5" (font.css), so
        # the visible text is larger than its nominal size, and a span
        # with the adjust off at a plain "Nem" sat about a fifth smaller
        # against it than the same :scale in pdf.  @document emits the
        # factor as --typeface_scale (document_html.cpp, font_definitions);
        # the fallback 1 is for a fragment rendered without @document.
        # The "*" of calc() survives the @eval re-read (tested 2026-09-18);
        # the "--" of the property name is hidden below with the rest.
        style = "', serif; font-size-adjust: none"
        style += f"; font-size: calc({float(self.scale)}em * var(--typeface_scale, 1))"
        # :middle -- for a script with no baseline relationship to the
        # text.  A browser sizes the span's inline box from the FONT's
        # ascender and descender, not the glyph's ink, and centres it on
        # the content area's midpoint; a hieroglyph font declares an
        # ascender of 1.0 em and no descender (NewGardinerOmni; Noto:
        # 1.324/-0.326), so at :scale 1.75 the box sat half an em too
        # high and every table row grew from 31 to 47 px (Noto: 62).
        # line-height 1 drops the body's leading from the enlarged box and
        # vertical-align middle centres it on the PARENT's x-height, which
        # is why this uses the text's metrics, not the script font's zeros.
        # Measured 2026-09-18: 40 px, the floor the ink allows; line-height
        # 0 alone gave 31 px with the glyphs overlapping the rows above.
        # Off by default: a Latin typeface must keep the shared baseline.
        if self.middle:
            style += "; line-height: 1; vertical-align: middle"
        if self.lift_length():
            # Relative positioning moves the ink and nothing else: the
            # line box is laid out as if the span had not moved, which is
            # what a baseline correction wants (vertical-align would
            # re-lay the line).  Negative top = up, as \raisebox in tex.
            style += f"; position: relative; top: -{self.lift_length()}"
        # The STYLE is markup, so its typographically active characters are
        # hidden: the html target's transforms turn a single quote into a
        # curly one, which broke the css font-family value.  The MARKER is
        # not hidden: a hyphen in the typeface name would become a KTESC
        # sequence, which on the :files path is still there when @document
        # looks for the marker.  The TEXT is writer content and is
        # deliberately left alone -- its quotes and dashes should be
        # transformed.
        hide = klammer_base.hide_typographic
        return (f'<span style="{hide("font-family: " + chr(39))}'
                + typeface_marker(self.name, self.scale) + hide(style) + '">'
                + f'{self.text}</span>')

    def tex(self):
        # The marker becomes the font-switch command; the group is what
        # makes the change temporary (the retired @font emitted a bare
        # \setmainfont, which changed the font for the rest of the
        # document).
        group = "{" + typeface_marker(self.name, self.scale) + " " + self.text + "}"
        if self.lift_length():
            # A \raisebox produced HERE is safe from the generator escaping
            # pass: this klammer is applied after a generator's result has
            # been escaped, so its backslash is never swept -- the trap
            # @nl.tex fell into (2026-09-07) is a klammer applied inside a
            # generator's ARGUMENT, which @typeface shares with @b and @i.
            return "\\raisebox{" + self.lift_length() + "}" + group
        return group

    def lift_length(self):
        # :lift, validated to the units both targets can move by.  The
        # length argtype admits w/h fractions and a quoted string, which
        # mean nothing as a vertical shift; a bare number would be columns
        # or lines and is refused for the same reason.
        lift = self.lift
        if not lift:
            return ""
        if not re.fullmatch(r"-?(\d+\.?\d*|\.\d+)(em|pt|px)", lift):
            raise Exception(f"@typeface :lift takes a length in em, pt or px, not '{lift}'")
        return lift

    def txt(self):
        # A typeface cannot be expressed in plain text.  Declared in the
        # .k description, so this is documented behaviour rather than a
        # silent per-target divergence.
        return self.text

def html_fontify(text, font_symbol, font_size):
    cls = {"r" : "",
           "i" :"ritalic",
           "b" : "bold",
           "s" : "sansserif",
           "t" : "monospace",
           "c" : "code"}[font_symbol]
    result = text
    if cls:
        if font_symbol == "c":
            result = re.sub(" ", "&#160;", result)
        cls = f'class="{cls}"'
    sty = ""
    if font_size != 1:
        sty = f'style="font-size: {float(font_size) * 100}%"'
    attr = (cls + " " + sty).strip()
    if attr:
        result = f'<span {attr}>{result}</span>'
    return result


def tex_fontify(text, font_symbol, font_size):
    command = {"r" : "\\rmfamily",
               "i" :"\\it",
               "b" : "\\bfseries",
               "s" : "\\sffamily",
               "t" : "\\ttfamily",
               "c" : "\\ttfamily"}[font_symbol]
    result = f"{{{command}\\scalefont{{{font_size}}}{{{text}}}}}"
    return result