Method repl_fontnames

src/__init__.py:7609–7684

Populate 'font_buffers'. For each font candidate, store its xref and the list of names by which PDF text may refer to it (there may be multiple).

(doc)

            return new_buffer

        def repl_fontnames(doc):
            """Populate 'font_buffers'.

            For each font candidate, store its xref and the list of names
            by which PDF text may refer to it (there may be multiple).
            """

            def norm_name(name):
                """Recreate font name that contains PDF hex codes.

                E.g. #20 -> space, chr(32)
                """
                while "#" in name:
                    p = name.find("#")
                    c = int(name[p + 1 : p + 3], 16)
                    name = name.replace(name[p : p + 3], chr(c))
                return name

            def get_fontnames(doc, item):
                """Return a list of fontnames for an item of page.get_fonts().

                There may be multiple names e.g. for Type0 fonts.
                """
                fontname = item[3]
                names = [fontname]
                fontname = doc.xref_get_key(item[0], "BaseFont")[1][1:]
                fontname = norm_name(fontname)
                if fontname not in names:
                    names.append(fontname)
                descendents = doc.xref_get_key(item[0], "DescendantFonts")
                if descendents[0] != "array":
                    return names
                descendents = descendents[1][1:-1]
                if descendents.endswith(" 0 R"):
                    xref = int(descendents[:-4])
                    descendents = doc.xref_object(xref, compressed=True)
                p1 = descendents.find("/BaseFont")
                if p1 >= 0:
                    p2 = descendents.find("/", p1 + 1)
                    p1 = min(descendents.find("/", p2 + 1), descendents.find(">>", p2 + 1))
                    fontname = descendents[p2 + 1 : p1]
                    fontname = norm_name(fontname)
                    if fontname not in names:
                        names.append(fontname)
                return names

            for i in range(doc.page_count):
                for f in doc.get_page_fonts(i, full=True):
                    font_xref = f[0]  # font xref
                    font_ext = f[1]  # font file extension
                    basename = f[3]  # font basename

                    if font_ext not in (  # skip if not supported by fontTools
                        "otf",
                        "ttf",
                        "woff",
                        "woff2",
                    ):