dotfiles/bin/utf8_nf

#!/bin/env python3

import gzip
import argparse
import re
import sys
import io
import os
from abc import ABC, abstractclassmethod, abstractmethod, abstractstaticmethod
from typing import List, TextIO


def extract_nerdfont_ranges(font_patcher: TextIO):
    range_re = re.compile(r"'((Src|Sym)(Start|End))':\s*([xXA-Fa-f0-9]+|None)")
    name_re = re.compile(r"'Name':\s+\"(.*?)\"")
    for line in font_patcher:
        results = range_re.findall(line)
        if len(results) != 4:
            continue
        props = {r[0]: int(r[3], base=16) if r[3] != "None" else None for r in results}
        r = (
            props["SrcStart"] or props["SymStart"],
            props["SrcEnd"] or props["SymEnd"],
        )
        if r != (0, 0):
            m = name_re.search(line)
            if not m:
                raise Exception("unable to find name: %s" % line)
            yield r[0], r[1], m[1]


class Scope(ABC):
    comment_override = None

    def __init__(self, ranges: List[tuple], output: TextIO):
        self.ranges = ranges
        self.length = len(ranges)
        self.output = output
        self.cursor = 0
        self.escape_char = None
        self.re = re.compile(self.pattern())

    @property
    def empty(self) -> bool:
        return self.cursor >= self.length

    @property
    def current(self) -> tuple:
        return self.ranges[self.cursor]

    def pop(self) -> bool:
        if self.empty:
            return False
        ss, se, comment = self.current
        self.write_line(ss, se, self.comment_override or comment)
        self.cursor += 1

    def read_line(self, line: str) -> bool:
        m = self.re.match(line)
        if not m:
            raise Exception("unkown line: " + line)
        ts = int(m[1], base=16)
        te = int(m[3], base=16) if m[3] else ts
        while not self.empty and self.current[1] < ts:
            self.pop()
        if self.empty:
            return True
        ss, se, _ = self.current
        # output no intersection
        if te < ss:
            return True
        # dont output subset
        if ts >= ss and te <= se:
            return False
        if ss <= te:                       # clip left intersection
            self.write_line(ts, ss - 1, m[5])
        if se <= te:
            self.pop()
        if ts <= se:
            self.write_line(se + 1, te, m[5])
        if se < ts:
            return True
        return False

    def write_line(self, start: int = None, end: int = None, comment: str = None) -> None:
        if end < start:
            return
        self.output.write(self.format(start, end, comment) + '\n')

    @abstractmethod
    def pattern(self) -> str:
        pass

    @abstractmethod
    def format(self, start: int, end: int, arg: any) -> str:
        pass


class CharmapScope(Scope):

    def pattern(self) -> str:
        return r'<U([A-Z0-9]+)>(..<U([A-Z0-9]+)>)?\s+(\S+)\s+(.*?)$'

    def format(self, start: int, end: int, comment: str) -> str:
        return "%s%s %s %s" % (
            "<U%04X>" % start,
            "..<U%04X>" % end if end > start else "",
            "".join(map(lambda x: "%sx%02x" % (self.escape_char, x), chr(start).encode('utf-8'))),
            comment,
        )


class WidthScope(Scope):
    comment_override = '2'

    def pattern(self) -> str:
        return r'<U([A-Z0-9]+)>(...<U([A-Z0-9]+)>)?(\s+)(\d+)$'

    def format(self, start: int, end: int, comment: str) -> str:
        return "%s%s\t%s" % (
            "<U%04X>" % start,
            "...<U%04X>" % end if end > start else "",
            comment
        )


def inject_nerdfont_ranges(ranges: List[tuple], textin: TextIO, textout: TextIO):
    comment_char = None
    keyword_re = re.compile(r'<(\w+)>\s+(\S+)\s*')
    charmap_scope = CharmapScope(ranges, textout)
    width_scope = WidthScope(ranges, textout)
    scope = None

    for line in textin:
        line = line.strip()
        if line.startswith("CHARMAP"):
            scope = charmap_scope
        elif line.startswith("WIDTH"):
            scope = width_scope
        elif line.startswith("END "):
            while not scope.empty:
                scope.pop()
            scope = None
        elif comment_char and line.startswith(comment_char):
            pass
        elif scope and not scope.empty:
            if scope.read_line(line) is False:
                continue
        elif line:
            m = keyword_re.match(line)
            if m:
                if m[1] == "comment_char":
                    comment_char = m[2]
                elif m[1] == 'escape_char':
                    charmap_scope.escape_char = m[2]

        textout.write(line + "\n")


def test():
    TARGET = (
        "<comment_char> %\n"
        "<escape_char> /\n\n"
        "CHARMAP\n"
        "<U0000>                /x00         no\n"
        "<U0001>..<U0005>       /x01         left\n"
        "<U0006>                /x06         subset\n"
        "<U0007>                /x07         subset\n"
        "<U0008>..<U0009>       /x08         right\n"
        "<U4E00>..<U4E99>       /xe4/xb8/x80 superset\n"
        "<U5E00>                /x00         gap\n"
        "END CHARMAP\n"
        "WIDTH\n"
        "<U0000>...<U0004>\t0\n"
        "<U6F00>...<U7FFF>\t1\n"
        "END WIDTH\n"
    )
    ranges = [
        (4, 8, "test"),
        (0x4e03, 0x4e0a, "test2"),
        (0x4F00, 0x4F00, "test3"),
        (0x6F00, 0x6F00, "tail"),
    ]
    EXPECT = (
        "<comment_char> %\n"
        "<escape_char> /\n\n"
        "CHARMAP\n"
        "<U0000>                /x00         no\n"
        "<U0001>..<U0003> /x01 left\n"
        "<U0004>..<U0008> /x04 test\n"
        "<U0009> /x09 right\n"
        "<U4E00>..<U4E02> /xe4/xb8/x80 superset\n"
        "<U4E03>..<U4E0A> /xe4/xb8/x83 test2\n"
        "<U4E0B>..<U4E99> /xe4/xb8/x8b superset\n"
        "<U4F00> /xe4/xbc/x80 test3\n"
        "<U5E00>                /x00         gap\n"
        "<U6F00> /xe6/xbc/x80 tail\n"
        "END CHARMAP\n"
        "WIDTH\n"
        "<U0000>...<U0003>\t0\n"
        "<U0004>...<U0008>\t2\n"
        "<U4E03>...<U4E0A>\t2\n"
        "<U4F00>\t2\n"
        "<U6F00>\t2\n"
        "<U6F01>...<U7FFF>\t1\n"
        "END WIDTH\n"
    )
    target = io.StringIO(TARGET)
    result = io.StringIO()
    inject_nerdfont_ranges(ranges, target, result)
    if result.getvalue() != EXPECT:
        print("\033[42m origin\033[0m")
        print(TARGET)
        print()
        print("\033[42m inject\033[0m")
        for r in ranges:
            print("%04X %04X %s" % r)
        print()
        print("\033[42m expect \033[0m", len(EXPECT))
        print(EXPECT)
        print()
        print("\033[42m result \033[0m", len(result.getvalue()))
        a = EXPECT.split('\n')
        b = result.getvalue().split('\n')
        for i in range(max(len(a), len(b))):
            print("\033[%dm%s\033[0m" % (32 if a[i] == b[i] else 31, b[i]))
    else:
        print("pass")


if __name__ == "__main__":
    if os.environ.get('DEBUGGING') == '1':
        test()
    else:
        parser = argparse.ArgumentParser(
                description="patch charmap to make NerdFont icons double width"
        )
        parser.add_argument(
            "-i", "--in-charmap",
            dest="in_charmap",
            default="/usr/share/i18n/charmaps/UTF-8.gz",
            help="input charmap file path")
        parser.add_argument(
            "-o", "--out",
            dest="out_charmap",
            default="/usr/share/i18n/charmaps/UTF-8NF.gz",
            help="output charmap file path")
        parser.add_argument(
            "-f", "--font-patcher",
            dest="font_patcher",
            required=True,
            help="file path of font_patcher from NerdFont")
        parser.add_argument(
            "--plainout",
            dest="plainout",
            action="store_true",
            help="write to stdout in plain-text")
        args = parser.parse_args()

        font_patcher = open(args.font_patcher, 'r',)
        double_width_ranges = extract_nerdfont_ranges(font_patcher)

        in_charmap = gzip.open(args.in_charmap, 'rt', encoding="ascii")
        if args.plainout:
            out_charmap = sys.stdout
        elif args.out_charmap.endswith('.gz'):
            out_charmap = gzip.open(args.out_charmap, 'wt')
        else:
            out_charmap = open(args.out_charmap, 'wt')
        ranges = sorted(
            filter(lambda x: x[2] in ['Seti-UI + Custom', 'Devicons'], double_width_ranges),
            key=lambda x: x[0]
        )
        for r in ranges:
            print("%04X-%04X %s" % r)
        inject_nerdfont_ranges(ranges, in_charmap, out_charmap)

    # add `en_US.UTF-8NF UTF-8NF` to `/etc/locale.gen`
    # run `locale-gen`
    # update `/etc/locale.conf` to `LANG=en_US.UTF-8NF`
    # restart