I am done

rl/Lib/site-packages/fontTools/subset/__main__.py

@@ -0,0 +1,6 @@
+import sys
+from fontTools.subset import main
+
+
+if __name__ == "__main__":
+    sys.exit(main())

rl/Lib/site-packages/fontTools/subset/cff.py

@@ -0,0 +1,184 @@
+from fontTools.misc import psCharStrings
+from fontTools import ttLib
+from fontTools.pens.basePen import NullPen
+from fontTools.misc.roundTools import otRound
+from fontTools.misc.loggingTools import deprecateFunction
+from fontTools.subset.util import _add_method, _uniq_sort
+
+
+class _ClosureGlyphsT2Decompiler(psCharStrings.SimpleT2Decompiler):
+    def __init__(self, components, localSubrs, globalSubrs):
+        psCharStrings.SimpleT2Decompiler.__init__(self, localSubrs, globalSubrs)
+        self.components = components
+
+    def op_endchar(self, index):
+        args = self.popall()
+        if len(args) >= 4:
+            from fontTools.encodings.StandardEncoding import StandardEncoding
+
+            # endchar can do seac accent bulding; The T2 spec says it's deprecated,
+            # but recent software that shall remain nameless does output it.
+            adx, ady, bchar, achar = args[-4:]
+            baseGlyph = StandardEncoding[bchar]
+            accentGlyph = StandardEncoding[achar]
+            self.components.add(baseGlyph)
+            self.components.add(accentGlyph)
+
+
+@_add_method(ttLib.getTableClass("CFF "))
+def closure_glyphs(self, s):
+    cff = self.cff
+    assert len(cff) == 1
+    font = cff[cff.keys()[0]]
+    glyphSet = font.CharStrings
+
+    decompose = s.glyphs
+    while decompose:
+        components = set()
+        for g in decompose:
+            if g not in glyphSet:
+                continue
+            gl = glyphSet[g]
+
+            subrs = getattr(gl.private, "Subrs", [])
+            decompiler = _ClosureGlyphsT2Decompiler(components, subrs, gl.globalSubrs)
+            decompiler.execute(gl)
+        components -= s.glyphs
+        s.glyphs.update(components)
+        decompose = components
+
+
+def _empty_charstring(font, glyphName, isCFF2, ignoreWidth=False):
+    c, fdSelectIndex = font.CharStrings.getItemAndSelector(glyphName)
+    if isCFF2 or ignoreWidth:
+        # CFF2 charstrings have no widths nor 'endchar' operators
+        c.setProgram([] if isCFF2 else ["endchar"])
+    else:
+        if hasattr(font, "FDArray") and font.FDArray is not None:
+            private = font.FDArray[fdSelectIndex].Private
+        else:
+            private = font.Private
+        dfltWdX = private.defaultWidthX
+        nmnlWdX = private.nominalWidthX
+        pen = NullPen()
+        c.draw(pen)  # this will set the charstring's width
+        if c.width != dfltWdX:
+            c.program = [c.width - nmnlWdX, "endchar"]
+        else:
+            c.program = ["endchar"]
+
+
+@_add_method(ttLib.getTableClass("CFF "))
+def prune_pre_subset(self, font, options):
+    cff = self.cff
+    # CFF table must have one font only
+    cff.fontNames = cff.fontNames[:1]
+
+    if options.notdef_glyph and not options.notdef_outline:
+        isCFF2 = cff.major > 1
+        for fontname in cff.keys():
+            font = cff[fontname]
+            _empty_charstring(font, ".notdef", isCFF2=isCFF2)
+
+    # Clear useless Encoding
+    for fontname in cff.keys():
+        font = cff[fontname]
+        # https://github.com/fonttools/fonttools/issues/620
+        font.Encoding = "StandardEncoding"
+
+    return True  # bool(cff.fontNames)
+
+
+@_add_method(ttLib.getTableClass("CFF "))
+def subset_glyphs(self, s):
+    cff = self.cff
+    for fontname in cff.keys():
+        font = cff[fontname]
+        cs = font.CharStrings
+
+        glyphs = s.glyphs.union(s.glyphs_emptied)
+
+        # Load all glyphs
+        for g in font.charset:
+            if g not in glyphs:
+                continue
+            c, _ = cs.getItemAndSelector(g)
+
+        if cs.charStringsAreIndexed:
+            indices = [i for i, g in enumerate(font.charset) if g in glyphs]
+            csi = cs.charStringsIndex
+            csi.items = [csi.items[i] for i in indices]
+            del csi.file, csi.offsets
+            if hasattr(font, "FDSelect"):
+                sel = font.FDSelect
+                sel.format = None
+                sel.gidArray = [sel.gidArray[i] for i in indices]
+            newCharStrings = {}
+            for indicesIdx, charsetIdx in enumerate(indices):
+                g = font.charset[charsetIdx]
+                if g in cs.charStrings:
+                    newCharStrings[g] = indicesIdx
+            cs.charStrings = newCharStrings
+        else:
+            cs.charStrings = {g: v for g, v in cs.charStrings.items() if g in glyphs}
+        font.charset = [g for g in font.charset if g in glyphs]
+        font.numGlyphs = len(font.charset)
+
+        if s.options.retain_gids:
+            isCFF2 = cff.major > 1
+            for g in s.glyphs_emptied:
+                _empty_charstring(font, g, isCFF2=isCFF2, ignoreWidth=True)
+
+    return True  # any(cff[fontname].numGlyphs for fontname in cff.keys())
+
+
+@_add_method(ttLib.getTableClass("CFF "))
+def prune_post_subset(self, ttfFont, options):
+    cff = self.cff
+    for fontname in cff.keys():
+        font = cff[fontname]
+        cs = font.CharStrings
+
+        # Drop unused FontDictionaries
+        if hasattr(font, "FDSelect"):
+            sel = font.FDSelect
+            indices = _uniq_sort(sel.gidArray)
+            sel.gidArray = [indices.index(ss) for ss in sel.gidArray]
+            arr = font.FDArray
+            arr.items = [arr[i] for i in indices]
+            del arr.file, arr.offsets
+
+    # Desubroutinize if asked for
+    if options.desubroutinize:
+        cff.desubroutinize()
+
+    # Drop hints if not needed
+    if not options.hinting:
+        self.remove_hints()
+    elif not options.desubroutinize:
+        self.remove_unused_subroutines()
+    return True
+
+
+@deprecateFunction(
+    "use 'CFFFontSet.desubroutinize()' instead", category=DeprecationWarning
+)
+@_add_method(ttLib.getTableClass("CFF "))
+def desubroutinize(self):
+    self.cff.desubroutinize()
+
+
+@deprecateFunction(
+    "use 'CFFFontSet.remove_hints()' instead", category=DeprecationWarning
+)
+@_add_method(ttLib.getTableClass("CFF "))
+def remove_hints(self):
+    self.cff.remove_hints()
+
+
+@deprecateFunction(
+    "use 'CFFFontSet.remove_unused_subroutines' instead", category=DeprecationWarning
+)
+@_add_method(ttLib.getTableClass("CFF "))
+def remove_unused_subroutines(self):
+    self.cff.remove_unused_subroutines()

rl/Lib/site-packages/fontTools/subset/svg.py

@@ -0,0 +1,253 @@
+from __future__ import annotations
+
+import re
+from functools import lru_cache
+from itertools import chain, count
+from typing import Dict, Iterable, Iterator, List, Optional, Set, Tuple
+
+try:
+    from lxml import etree
+except ImportError:
+    # lxml is required for subsetting SVG, but we prefer to delay the import error
+    # until subset_glyphs() is called (i.e. if font to subset has an 'SVG ' table)
+    etree = None
+
+from fontTools import ttLib
+from fontTools.subset.util import _add_method
+from fontTools.ttLib.tables.S_V_G_ import SVGDocument
+
+
+__all__ = ["subset_glyphs"]
+
+
+GID_RE = re.compile(r"^glyph(\d+)$")
+
+NAMESPACES = {
+    "svg": "http://www.w3.org/2000/svg",
+    "xlink": "http://www.w3.org/1999/xlink",
+}
+XLINK_HREF = f'{{{NAMESPACES["xlink"]}}}href'
+
+
+# TODO(antrotype): Replace with functools.cache once we are 3.9+
+@lru_cache(maxsize=None)
+def xpath(path):
+    # compile XPath upfront, caching result to reuse on multiple elements
+    return etree.XPath(path, namespaces=NAMESPACES)
+
+
+def group_elements_by_id(tree: etree.Element) -> Dict[str, etree.Element]:
+    # select all svg elements with 'id' attribute no matter where they are
+    # including the root element itself:
+    # https://github.com/fonttools/fonttools/issues/2548
+    return {el.attrib["id"]: el for el in xpath("//svg:*[@id]")(tree)}
+
+
+def parse_css_declarations(style_attr: str) -> Dict[str, str]:
+    # https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/style
+    # https://developer.mozilla.org/en-US/docs/Web/CSS/Syntax#css_declarations
+    result = {}
+    for declaration in style_attr.split(";"):
+        if declaration.count(":") == 1:
+            property_name, value = declaration.split(":")
+            property_name = property_name.strip()
+            result[property_name] = value.strip()
+        elif declaration.strip():
+            raise ValueError(f"Invalid CSS declaration syntax: {declaration}")
+    return result
+
+
+def iter_referenced_ids(tree: etree.Element) -> Iterator[str]:
+    # Yield all the ids that can be reached via references from this element tree.
+    # We currently support xlink:href (as used by <use> and gradient templates),
+    # and local url(#...) links found in fill or clip-path attributes
+    # TODO(anthrotype): Check we aren't missing other supported kinds of reference
+    find_svg_elements_with_references = xpath(
+        ".//svg:*[ "
+        "starts-with(@xlink:href, '#') "
+        "or starts-with(@fill, 'url(#') "
+        "or starts-with(@clip-path, 'url(#') "
+        "or contains(@style, ':url(#') "
+        "]",
+    )
+    for el in chain([tree], find_svg_elements_with_references(tree)):
+        ref_id = href_local_target(el)
+        if ref_id is not None:
+            yield ref_id
+
+        attrs = el.attrib
+        if "style" in attrs:
+            attrs = {**dict(attrs), **parse_css_declarations(el.attrib["style"])}
+        for attr in ("fill", "clip-path"):
+            if attr in attrs:
+                value = attrs[attr]
+                if value.startswith("url(#") and value.endswith(")"):
+                    ref_id = value[5:-1]
+                    assert ref_id
+                    yield ref_id
+
+
+def closure_element_ids(
+    elements: Dict[str, etree.Element], element_ids: Set[str]
+) -> None:
+    # Expand the initial subset of element ids to include ids that can be reached
+    # via references from the initial set.
+    unvisited = element_ids
+    while unvisited:
+        referenced: Set[str] = set()
+        for el_id in unvisited:
+            if el_id not in elements:
+                # ignore dangling reference; not our job to validate svg
+                continue
+            referenced.update(iter_referenced_ids(elements[el_id]))
+        referenced -= element_ids
+        element_ids.update(referenced)
+        unvisited = referenced
+
+
+def subset_elements(el: etree.Element, retained_ids: Set[str]) -> bool:
+    # Keep elements if their id is in the subset, or any of their children's id is.
+    # Drop elements whose id is not in the subset, and either have no children,
+    # or all their children are being dropped.
+    if el.attrib.get("id") in retained_ids:
+        # if id is in the set, don't recurse; keep whole subtree
+        return True
+    # recursively subset all the children; we use a list comprehension instead
+    # of a parentheses-less generator expression because we don't want any() to
+    # short-circuit, as our function has a side effect of dropping empty elements.
+    if any([subset_elements(e, retained_ids) for e in el]):
+        return True
+    assert len(el) == 0
+    parent = el.getparent()
+    if parent is not None:
+        parent.remove(el)
+    return False
+
+
+def remap_glyph_ids(
+    svg: etree.Element, glyph_index_map: Dict[int, int]
+) -> Dict[str, str]:
+    # Given {old_gid: new_gid} map, rename all elements containing id="glyph{gid}"
+    # special attributes
+    elements = group_elements_by_id(svg)
+    id_map = {}
+    for el_id, el in elements.items():
+        m = GID_RE.match(el_id)
+        if not m:
+            continue
+        old_index = int(m.group(1))
+        new_index = glyph_index_map.get(old_index)
+        if new_index is not None:
+            if old_index == new_index:
+                continue
+            new_id = f"glyph{new_index}"
+        else:
+            # If the old index is missing, the element correspond to a glyph that was
+            # excluded from the font's subset.
+            # We rename it to avoid clashes with the new GIDs or other element ids.
+            new_id = f".{el_id}"
+            n = count(1)
+            while new_id in elements:
+                new_id = f"{new_id}.{next(n)}"
+
+        id_map[el_id] = new_id
+        el.attrib["id"] = new_id
+
+    return id_map
+
+
+def href_local_target(el: etree.Element) -> Optional[str]:
+    if XLINK_HREF in el.attrib:
+        href = el.attrib[XLINK_HREF]
+        if href.startswith("#") and len(href) > 1:
+            return href[1:]  # drop the leading #
+    return None
+
+
+def update_glyph_href_links(svg: etree.Element, id_map: Dict[str, str]) -> None:
+    # update all xlink:href="#glyph..." attributes to point to the new glyph ids
+    for el in xpath(".//svg:*[starts-with(@xlink:href, '#glyph')]")(svg):
+        old_id = href_local_target(el)
+        assert old_id is not None
+        if old_id in id_map:
+            new_id = id_map[old_id]
+            el.attrib[XLINK_HREF] = f"#{new_id}"
+
+
+def ranges(ints: Iterable[int]) -> Iterator[Tuple[int, int]]:
+    # Yield sorted, non-overlapping (min, max) ranges of consecutive integers
+    sorted_ints = iter(sorted(set(ints)))
+    try:
+        start = end = next(sorted_ints)
+    except StopIteration:
+        return
+    for v in sorted_ints:
+        if v - 1 == end:
+            end = v
+        else:
+            yield (start, end)
+            start = end = v
+    yield (start, end)
+
+
+@_add_method(ttLib.getTableClass("SVG "))
+def subset_glyphs(self, s) -> bool:
+    if etree is None:
+        raise ImportError("No module named 'lxml', required to subset SVG")
+
+    # glyph names (before subsetting)
+    glyph_order: List[str] = s.orig_glyph_order
+    # map from glyph names to original glyph indices
+    rev_orig_glyph_map: Dict[str, int] = s.reverseOrigGlyphMap
+    # map from original to new glyph indices (after subsetting)
+    glyph_index_map: Dict[int, int] = s.glyph_index_map
+
+    new_docs: List[SVGDocument] = []
+    for doc in self.docList:
+        glyphs = {
+            glyph_order[i] for i in range(doc.startGlyphID, doc.endGlyphID + 1)
+        }.intersection(s.glyphs)
+        if not glyphs:
+            # no intersection: we can drop the whole record
+            continue
+
+        svg = etree.fromstring(
+            # encode because fromstring dislikes xml encoding decl if input is str.
+            # SVG xml encoding must be utf-8 as per OT spec.
+            doc.data.encode("utf-8"),
+            parser=etree.XMLParser(
+                # Disable libxml2 security restrictions to support very deep trees.
+                # Without this we would get an error like this:
+                # `lxml.etree.XMLSyntaxError: internal error: Huge input lookup`
+                # when parsing big fonts e.g. noto-emoji-picosvg.ttf.
+                huge_tree=True,
+                # ignore blank text as it's not meaningful in OT-SVG; it also prevents
+                # dangling tail text after removing an element when pretty_print=True
+                remove_blank_text=True,
+                # don't replace entities; we don't expect any in OT-SVG and they may
+                # be abused for XXE attacks
+                resolve_entities=False,
+            ),
+        )
+
+        elements = group_elements_by_id(svg)
+        gids = {rev_orig_glyph_map[g] for g in glyphs}
+        element_ids = {f"glyph{i}" for i in gids}
+        closure_element_ids(elements, element_ids)
+
+        if not subset_elements(svg, element_ids):
+            continue
+
+        if not s.options.retain_gids:
+            id_map = remap_glyph_ids(svg, glyph_index_map)
+            update_glyph_href_links(svg, id_map)
+
+        new_doc = etree.tostring(svg, pretty_print=s.options.pretty_svg).decode("utf-8")
+
+        new_gids = (glyph_index_map[i] for i in gids)
+        for start, end in ranges(new_gids):
+            new_docs.append(SVGDocument(new_doc, start, end, doc.compressed))
+
+    self.docList = new_docs
+
+    return bool(self.docList)

rl/Lib/site-packages/fontTools/subset/util.py

@@ -0,0 +1,25 @@
+"""Private utility methods used by the subset modules"""
+
+
+def _add_method(*clazzes):
+    """Returns a decorator function that adds a new method to one or
+    more classes."""
+
+    def wrapper(method):
+        done = []
+        for clazz in clazzes:
+            if clazz in done:
+                continue  # Support multiple names of a clazz
+            done.append(clazz)
+            assert clazz.__name__ != "DefaultTable", "Oops, table class not found."
+            assert not hasattr(
+                clazz, method.__name__
+            ), "Oops, class '%s' has method '%s'." % (clazz.__name__, method.__name__)
+            setattr(clazz, method.__name__, method)
+        return None
+
+    return wrapper
+
+
+def _uniq_sort(l):
+    return sorted(set(l))