#!/usr/bin/env bash
#
# scripts/subset-fonts.sh — Plan 01-12 Wave 1 Task 1.
#
# One-off subsetting recipe for the Mokosh design-integration font bundle:
#
#   Lora variable (normal + italic, axis wght 400-700) — display family per
#   R2 designer reply 2026-05-19 (substitutes Newsreader for Cyrillic
#   coverage; Cyreal foundry, OFL-1.1)
#   IBM Plex Sans (Regular, Medium, SemiBold, Bold) — UI body family
#   IBM Plex Mono (Regular, Medium) — diagnostic / timer family
#
# All faces are subsetted to Latin (U+0020-007E + U+00A0-00FF) + Cyrillic
# basic (U+0400-045F + supplemental code points) per RESEARCH §1 + §6.
# Total bundled artifact target: ~250-300 KB per R2 substitution adjustment.
#
# Upstream sources:
#   Lora-Cyrillic: https://github.com/cyrealtype/Lora-Cyrillic (main branch)
#   IBM Plex:      https://github.com/IBM/plex (master branch)
#
# Output: src/shared/fonts/*.woff2 (7 or 8 files depending on Lora italic
# upstream layout — A5 in RESEARCH Assumptions Log).
#
# Re-run trigger: when the upstream Lora-Cyrillic or IBM/plex releases
# bump a major version with bug fixes; or when the UNICODES range needs
# expansion (Cyrillic supplement, extra punctuation). The committed
# WOFF2 files are the source of truth — running this script overwrites
# them only on intentional regeneration.
#
# References:
#   - fontTools pyftsubset:
#     https://fonttools.readthedocs.io/en/latest/subset/index.html
#   - WOFF2 (RFC 8081): https://www.rfc-editor.org/rfc/rfc8081
#   - OFL-1.1 best practices: https://scripts.sil.org/OFL_web
#
# Google bash style applies; see https://google.github.io/styleguide/shellguide.html

set -euo pipefail

readonly SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
readonly REPO_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"
readonly FONTS_OUT_DIR="${REPO_ROOT}/src/shared/fonts"

# Subset coverage per RESEARCH §1 + §6:
#   Latin core           U+0020-007E
#   Latin supplement     U+00A0-00FF (covers Cyrillic context: nbsp, ©, ®, ™ etc.)
#   Latin extended bits  U+0131 (dotless i), U+0152-0153 (Œœ), U+0301 (combining acute)
#   Cyrillic basic       U+0400-045F (full RU + UA basic alphabet)
#   Cyrillic supplement  U+0490-0491 (Ґґ), U+04B0-04B1 (Ұұ — Kazakh)
#   Numero sign          U+2116 (№ — common Russian punctuation)
readonly UNICODES='U+0020-007E,U+00A0-00FF,U+0131,U+0152-0153,U+0301,U+0400-045F,U+0490-0491,U+04B0-04B1,U+2116'

# Common pyftsubset flags shared across faces.
readonly -a PYFTSUBSET_FLAGS=(
  "--unicodes=${UNICODES}"
  '--flavor=woff2'
  "--layout-features=*"
  '--no-hinting'
  '--desubroutinize'
  '--name-IDs=*'
  '--name-legacy'
  '--name-languages=*'
)

# subset_face <input-ttf> <output-woff2>
subset_face() {
  local -r input_ttf="$1"
  local -r output_woff2="$2"
  if [[ ! -f "${input_ttf}" ]]; then
    echo "ERROR: input TTF not found: ${input_ttf}" >&2
    return 1
  fi
  pyftsubset "${input_ttf}" \
    "${PYFTSUBSET_FLAGS[@]}" \
    "--output-file=${output_woff2}"
  local -r bytes="$(stat -c%s "${output_woff2}")"
  echo "  subsetted: ${output_woff2} (${bytes} bytes)"
}

main() {
  local -r scratch="${1:-}"
  if [[ -z "${scratch}" || ! -d "${scratch}" ]]; then
    cat >&2 <<EOF
Usage: $(basename "$0") <scratch-dir>

  <scratch-dir> must contain the upstream TTFs:
    Lora-VariableFont_wght.ttf
    Lora-Italic-VariableFont_wght.ttf
    IBMPlexSans-Regular.ttf, -Medium.ttf, -SemiBold.ttf, -Bold.ttf
    IBMPlexMono-Regular.ttf, -Medium.ttf

  Fetch recipe:
    SCRATCH=\$(mktemp -d)
    curl -L -o "\${SCRATCH}/Lora-VariableFont_wght.ttf" \\
      "https://raw.githubusercontent.com/cyrealtype/Lora-Cyrillic/main/fonts/variable/Lora%5Bwght%5D.ttf"
    curl -L -o "\${SCRATCH}/Lora-Italic-VariableFont_wght.ttf" \\
      "https://raw.githubusercontent.com/cyrealtype/Lora-Cyrillic/main/fonts/variable/Lora-Italic%5Bwght%5D.ttf"
    for v in Regular Medium SemiBold Bold; do
      curl -L -o "\${SCRATCH}/IBMPlexSans-\${v}.ttf" \\
        "https://raw.githubusercontent.com/IBM/plex/master/packages/plex-sans/fonts/complete/ttf/IBMPlexSans-\${v}.ttf"
    done
    for v in Regular Medium; do
      curl -L -o "\${SCRATCH}/IBMPlexMono-\${v}.ttf" \\
        "https://raw.githubusercontent.com/IBM/plex/master/packages/plex-mono/fonts/complete/ttf/IBMPlexMono-\${v}.ttf"
    done
EOF
    return 2
  fi

  mkdir -p "${FONTS_OUT_DIR}"

  echo "==> Subsetting Lora (variable, Cyreal foundry; R2 substitute for Newsreader)"
  subset_face "${scratch}/Lora-VariableFont_wght.ttf"        "${FONTS_OUT_DIR}/Lora-VariableFont.woff2"
  subset_face "${scratch}/Lora-Italic-VariableFont_wght.ttf" "${FONTS_OUT_DIR}/Lora-Italic-VariableFont.woff2"

  echo "==> Subsetting IBM Plex Sans (4 weights)"
  local variant
  for variant in Regular Medium SemiBold Bold; do
    subset_face "${scratch}/IBMPlexSans-${variant}.ttf" "${FONTS_OUT_DIR}/IBMPlexSans-${variant}.woff2"
  done

  echo "==> Subsetting IBM Plex Mono (2 weights)"
  for variant in Regular Medium; do
    subset_face "${scratch}/IBMPlexMono-${variant}.ttf" "${FONTS_OUT_DIR}/IBMPlexMono-${variant}.woff2"
  done

  echo
  echo "==> Bundle summary:"
  du -ch "${FONTS_OUT_DIR}"/*.woff2 | tail -1
  echo
  echo "==> Done. Files emitted under ${FONTS_OUT_DIR}."
}

main "$@"