OneWheelGeek revised this gist 1 month ago. Go to revision
1 file changed, 28 insertions, 4 deletions
pdfify.sh
| @@ -164,7 +164,8 @@ for arg in "$@"; do | |||
| 164 | 164 | echo "" | |
| 165 | 165 | echo -e "${BOLD}Frontmatter:${RESET}" | |
| 166 | 166 | echo -e " title, subtitle, author, header, footer, toc-level, date," | |
| 167 | - | echo -e " numbersections (true/false), numberfrom (1-4), watermark" | |
| 167 | + | echo -e " numbersections (true/false), numberfrom (1-4), watermark," | |
| 168 | + | echo -e " pagebreak (true/false — page break before each top-level heading)" | |
| 168 | 169 | exit 0 ;; | |
| 169 | 170 | *) POSITIONAL+=("$arg") ;; | |
| 170 | 171 | esac | |
| @@ -226,7 +227,7 @@ RUN npm install -g @mermaid-js/mermaid-cli | |||
| 226 | 227 | ENV PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=true | |
| 227 | 228 | ENV PUPPETEER_EXECUTABLE_PATH=/usr/bin/chromium | |
| 228 | 229 | ENV CHROME_PATH=/usr/bin/chromium | |
| 229 | - | RUN echo '{"maxTextSize": 90000, "flowchart": {"useMaxWidth": true}, "theme": "default"}' > /opt/mermaid-config.json | |
| 230 | + | RUN echo '{"maxTextSize": 90000, "flowchart": {"useMaxWidth": true}, "theme": "base", "themeVariables": {"primaryColor": "#3B82F6", "primaryBorderColor": "#1E40AF", "primaryTextColor": "#1E293B", "lineColor": "#475569", "xyChart": {"backgroundColor": "transparent", "plotColorPalette": "#2563EB,#DC2626,#16A34A,#D97706,#9333EA,#0891B2"}}}' > /opt/mermaid-config.json | |
| 230 | 231 | RUN echo '{"args": ["--no-sandbox", "--disable-setuid-sandbox", "--disable-dev-shm-usage", "--disable-gpu"]}' > /opt/puppeteer-config.json | |
| 231 | 232 | WORKDIR /work | |
| 232 | 233 | ENTRYPOINT ["/bin/bash"] | |
| @@ -324,6 +325,7 @@ convert_file() { | |||
| 324 | 325 | FM_NUMBERSECTIONS=$(extract_fm "numbersections") | |
| 325 | 326 | FM_NUMBERFROM=$(extract_fm "numberfrom") | |
| 326 | 327 | FM_WATERMARK=$(extract_fm "watermark") | |
| 328 | + | FM_PAGEBREAK=$(extract_fm "pagebreak") | |
| 327 | 329 | fi | |
| 328 | 330 | ||
| 329 | 331 | # --- CLI overrides frontmatter --- | |
| @@ -352,6 +354,8 @@ convert_file() { | |||
| 352 | 354 | local FILE_TOC_LEVEL="${FM_TOC_LEVEL:-3}" | |
| 353 | 355 | local FILE_NUMBERS=1 | |
| 354 | 356 | [[ "$FM_NUMBERSECTIONS" == "false" ]] && FILE_NUMBERS=0 | |
| 357 | + | local FILE_PAGEBREAK=1 | |
| 358 | + | [[ "$FM_PAGEBREAK" == "false" ]] && FILE_PAGEBREAK=0 | |
| 355 | 359 | ||
| 356 | 360 | # Auto-determine numberfrom based on structure (if not explicitly set) | |
| 357 | 361 | local FILE_NUMBER_FROM="${FM_NUMBERFROM:-}" | |
| @@ -593,7 +597,7 @@ while IFS= read -r line || [[ -n "$line" ]]; do | |||
| 593 | 597 | NEXT_CHAR="${line:${#BREAK_HASHES}:1}" | |
| 594 | 598 | if [[ "$NEXT_CHAR" != "#" ]]; then | |
| 595 | 599 | H1_COUNT=$((H1_COUNT + 1)) | |
| 596 | - | if [[ $H1_COUNT -gt 1 ]]; then | |
| 600 | + | if [[ $H1_COUNT -gt 1 && $FILE_PAGEBREAK -eq 1 ]]; then | |
| 597 | 601 | echo "" >> "$BREAK_INJECTED" | |
| 598 | 602 | echo '```{=latex}' >> "$BREAK_INJECTED" | |
| 599 | 603 | echo '\newpage' >> "$BREAK_INJECTED" | |
| @@ -663,6 +667,15 @@ while IFS= read -r line || [[ -n "$line" ]]; do | |||
| 663 | 667 | fi | |
| 664 | 668 | done < "$CALLOUT_MD" | |
| 665 | 669 | ||
| 670 | + | # --- Strip YAML frontmatter so pandoc doesn't generate its own title --- | |
| 671 | + | # pdfify already parses frontmatter above; letting pandoc see it causes a | |
| 672 | + | # duplicate title (pandoc's \maketitle + pdfify's custom title banner). | |
| 673 | + | if head -1 "$TEMP_MD" | grep -q '^---'; then | |
| 674 | + | STRIPPED_FM=$(mktemp /tmp/pdfify-nofm-XXXXXX.md) | |
| 675 | + | awk 'NR==1 && /^---/{skip=1; next} skip && /^---/{skip=0; next} !skip' "$TEMP_MD" > "$STRIPPED_FM" | |
| 676 | + | mv "$STRIPPED_FM" "$TEMP_MD" | |
| 677 | + | fi | |
| 678 | + | ||
| 666 | 679 | # --- Lua filter: protect brackets in headings for titlesec --- | |
| 667 | 680 | # Square brackets in headings break titlesec (\SQSPL@scan error) because LaTeX | |
| 668 | 681 | # interprets [ as the start of an optional argument. | |
| @@ -754,7 +767,10 @@ cat > "$PREAMBLE" <<'LATEX' | |||
| 754 | 767 | % Background on code blocks via mdframed | |
| 755 | 768 | \usepackage[framemethod=tikz]{mdframed} | |
| 756 | 769 | ||
| 757 | - | % Override pandoc's Shaded environment | |
| 770 | + | % Override pandoc's Shaded environment (define first if pandoc didn't) | |
| 771 | + | \makeatletter | |
| 772 | + | \@ifundefined{Shaded}{\newenvironment{Shaded}{}{}}{} | |
| 773 | + | \makeatother | |
| 758 | 774 | \renewenvironment{Shaded}{% | |
| 759 | 775 | \begin{mdframed}[ | |
| 760 | 776 | backgroundcolor=codebg, | |
| @@ -915,6 +931,12 @@ cat > "$PREAMBLE" <<'LATEX' | |||
| 915 | 931 | \usepackage{tabularx} | |
| 916 | 932 | \arrayrulecolor{codeborder} | |
| 917 | 933 | ||
| 934 | + | % Alternating row shading | |
| 935 | + | \definecolor{tablerowgray}{HTML}{F3F4F6} | |
| 936 | + | \let\oldlongtable\longtable | |
| 937 | + | \let\endoldlongtable\endlongtable | |
| 938 | + | \renewenvironment{longtable}{\rowcolors{2}{white}{tablerowgray}\oldlongtable}{\endoldlongtable} | |
| 939 | + | ||
| 918 | 940 | % Allow line breaks in table cells and shrink monospace to fit | |
| 919 | 941 | \usepackage{array} | |
| 920 | 942 | \renewcommand{\arraystretch}{1.4} | |
| @@ -966,6 +988,7 @@ FM_DATE_HASH="${FM_DATE_HASH:-}" | |||
| 966 | 988 | FM_DATE_DIRTY="${FM_DATE_DIRTY:-}" | |
| 967 | 989 | FILE_NUMBERS="${FILE_NUMBERS:-1}" | |
| 968 | 990 | FILE_NUMBER_FROM="${FILE_NUMBER_FROM:-2}" | |
| 991 | + | FILE_PAGEBREAK="${FILE_PAGEBREAK:-1}" | |
| 969 | 992 | ||
| 970 | 993 | # Escape LaTeX special characters in text fields (uses sed to avoid | |
| 971 | 994 | # bash parameter substitution brace-parsing issues with } in replacements) | |
| @@ -1191,6 +1214,7 @@ docker run --rm \ | |||
| 1191 | 1214 | -e "FILE_NUMBER_FROM=$FILE_NUMBER_FROM" \ | |
| 1192 | 1215 | -e "HIDE_FIRST_H1=$HIDE_FIRST_H1" \ | |
| 1193 | 1216 | -e "FM_WATERMARK=$FM_WATERMARK" \ | |
| 1217 | + | -e "FILE_PAGEBREAK=$FILE_PAGEBREAK" \ | |
| 1194 | 1218 | --tmpfs /tmp:exec \ | |
| 1195 | 1219 | "$IMAGE_NAME" "/work/$CONVERT_BASENAME" "$INPUT_FILE" "/output/$OUTPUT_FILE" \ | |
| 1196 | 1220 | || { | |
OneWheelGeek revised this gist 1 month ago. Go to revision
1 file changed, 1273 insertions
pdfify.sh(file created)
| @@ -0,0 +1,1273 @@ | |||
| 1 | + | #!/usr/bin/env bash | |
| 2 | + | set -euo pipefail | |
| 3 | + | ||
| 4 | + | # pdfify — Convert Markdown to beautiful PDF via Docker | |
| 5 | + | # Supports: images, mermaid diagrams, tables, code blocks, Obsidian callouts | |
| 6 | + | # Usage: ./pdfify <file.md> [file2.md ...] [options] | |
| 7 | + | ||
| 8 | + | VERSION="1.2.0" | |
| 9 | + | IMAGE_NAME="pdfify" | |
| 10 | + | GIST_ID="23f4514a1f0da1347d3f89926c23b68f" | |
| 11 | + | GIST_RAW="https://gist.githubusercontent.com/jclement/${GIST_ID}/raw/pdfify.sh" | |
| 12 | + | SELF="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/$(basename "${BASH_SOURCE[0]}")" | |
| 13 | + | ||
| 14 | + | # --- Colors --- | |
| 15 | + | RED='\033[0;31m' | |
| 16 | + | GREEN='\033[0;32m' | |
| 17 | + | YELLOW='\033[0;33m' | |
| 18 | + | BLUE='\033[0;34m' | |
| 19 | + | MAGENTA='\033[0;35m' | |
| 20 | + | CYAN='\033[0;36m' | |
| 21 | + | BOLD='\033[1m' | |
| 22 | + | DIM='\033[2m' | |
| 23 | + | RESET='\033[0m' | |
| 24 | + | ||
| 25 | + | # --- Portable SHA-256 (macOS has shasum, Linux often has sha256sum) --- | |
| 26 | + | _sha256() { shasum -a 256 "$@" 2>/dev/null || sha256sum "$@"; } | |
| 27 | + | ||
| 28 | + | info() { echo -e "${BLUE}::${RESET} ${BOLD}$*${RESET}"; } | |
| 29 | + | success() { echo -e "${GREEN}✓${RESET} $*"; } | |
| 30 | + | warn() { echo -e "${YELLOW}⚠${RESET} $*"; } | |
| 31 | + | detail() { echo -e " ${DIM}→${RESET} $*"; } | |
| 32 | + | header() { echo -e "\n${MAGENTA}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}"; echo -e "${MAGENTA} ${BOLD}$*${RESET}"; echo -e "${MAGENTA}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${RESET}\n"; } | |
| 33 | + | ||
| 34 | + | # --- Self-update --- | |
| 35 | + | do_update() { | |
| 36 | + | info "Updating pdfify from gist..." | |
| 37 | + | local raw_url tmp | |
| 38 | + | raw_url=$(curl -fsSL "https://api.github.com/gists/${GIST_ID}" 2>/dev/null | grep '"raw_url"' | head -1 | sed 's/.*"raw_url": *"//;s/".*//') | |
| 39 | + | [[ -z "$raw_url" ]] && raw_url="$GIST_RAW" | |
| 40 | + | tmp=$(mktemp) | |
| 41 | + | if curl -fsSL "$raw_url" -o "$tmp" 2>/dev/null; then | |
| 42 | + | if [[ -s "$tmp" ]] && head -1 "$tmp" | grep -q '^#!/'; then | |
| 43 | + | chmod +x "$tmp" | |
| 44 | + | mv "$tmp" "$SELF" | |
| 45 | + | success "Updated to latest version" | |
| 46 | + | detail "${CYAN}${SELF}${RESET}" | |
| 47 | + | else | |
| 48 | + | rm -f "$tmp" | |
| 49 | + | echo -e "${RED}Error:${RESET} Downloaded file doesn't look like a script" | |
| 50 | + | exit 1 | |
| 51 | + | fi | |
| 52 | + | else | |
| 53 | + | rm -f "$tmp" | |
| 54 | + | echo -e "${RED}Error:${RESET} Failed to download update" | |
| 55 | + | exit 1 | |
| 56 | + | fi | |
| 57 | + | exit 0 | |
| 58 | + | } | |
| 59 | + | ||
| 60 | + | check_for_update() { | |
| 61 | + | local remote_hash local_hash raw_url | |
| 62 | + | raw_url=$(curl -fsSL --connect-timeout 2 --max-time 3 "https://api.github.com/gists/${GIST_ID}" 2>/dev/null | grep '"raw_url"' | head -1 | sed 's/.*"raw_url": *"//;s/".*//') || return 0 | |
| 63 | + | [[ -z "$raw_url" ]] && return 0 | |
| 64 | + | remote_hash=$(curl -fsSL --connect-timeout 2 --max-time 5 "$raw_url" 2>/dev/null | _sha256 | cut -d' ' -f1) || return 0 | |
| 65 | + | local_hash=$(_sha256 < "$SELF" | cut -d' ' -f1) | |
| 66 | + | if [[ -n "$remote_hash" && "$remote_hash" != "$local_hash" ]]; then | |
| 67 | + | echo -e "${YELLOW}⚠${RESET} ${DIM}A newer version of pdfify is available. Run ${CYAN}pdfify --update${DIM} to upgrade.${RESET}" | |
| 68 | + | fi | |
| 69 | + | } | |
| 70 | + | ||
| 71 | + | # --- Args (CLI overrides frontmatter; "" means "use frontmatter default") --- | |
| 72 | + | REBUILD=0 | |
| 73 | + | WATCH=0 | |
| 74 | + | OPEN=0 | |
| 75 | + | PREVIEW=0 | |
| 76 | + | OUT_FILE="" | |
| 77 | + | NEXT_KEY="" | |
| 78 | + | POSITIONAL=() | |
| 79 | + | ||
| 80 | + | # CLI overrides — empty string means "not set, defer to frontmatter" | |
| 81 | + | CLI_TOC_LEVEL="" | |
| 82 | + | CLI_NUMBERS="" | |
| 83 | + | CLI_NUMBER_FROM="" | |
| 84 | + | CLI_TITLE="" | |
| 85 | + | CLI_SUBTITLE="" | |
| 86 | + | CLI_AUTHOR="" | |
| 87 | + | CLI_HEADER="" | |
| 88 | + | CLI_FOOTER="" | |
| 89 | + | CLI_DATE="" | |
| 90 | + | CLI_WATERMARK="" | |
| 91 | + | ||
| 92 | + | for arg in "$@"; do | |
| 93 | + | if [[ -n "$NEXT_KEY" ]]; then | |
| 94 | + | case "$NEXT_KEY" in | |
| 95 | + | toc-level) CLI_TOC_LEVEL="$arg" ;; | |
| 96 | + | number-from) CLI_NUMBER_FROM="$arg" ;; | |
| 97 | + | out) OUT_FILE="$arg" ;; | |
| 98 | + | title) CLI_TITLE="$arg" ;; | |
| 99 | + | subtitle) CLI_SUBTITLE="$arg" ;; | |
| 100 | + | author) CLI_AUTHOR="$arg" ;; | |
| 101 | + | header) CLI_HEADER="$arg" ;; | |
| 102 | + | footer) CLI_FOOTER="$arg" ;; | |
| 103 | + | date) CLI_DATE="$arg" ;; | |
| 104 | + | watermark) CLI_WATERMARK="$arg" ;; | |
| 105 | + | esac | |
| 106 | + | NEXT_KEY="" | |
| 107 | + | continue | |
| 108 | + | fi | |
| 109 | + | case "$arg" in | |
| 110 | + | --rebuild) REBUILD=1 ;; | |
| 111 | + | --update) do_update ;; | |
| 112 | + | --watch) WATCH=1 ;; | |
| 113 | + | --open) OPEN=1 ;; | |
| 114 | + | --preview) PREVIEW=1; OPEN=1 ;; | |
| 115 | + | --no-numbers) CLI_NUMBERS="false" ;; | |
| 116 | + | --numbers) CLI_NUMBERS="true" ;; | |
| 117 | + | --clean) echo -e "${BLUE}::${RESET} ${BOLD}Removing Docker image ${CYAN}${IMAGE_NAME}${RESET}..." | |
| 118 | + | docker rmi "$IMAGE_NAME" >/dev/null 2>&1 && echo -e "${GREEN}✓${RESET} Image removed" || echo -e "${DIM}Image not found${RESET}" | |
| 119 | + | exit 0 ;; | |
| 120 | + | --toc-level) NEXT_KEY="toc-level" ;; | |
| 121 | + | --toc-level=*) CLI_TOC_LEVEL="${arg#*=}" ;; | |
| 122 | + | --number-from) NEXT_KEY="number-from" ;; | |
| 123 | + | --number-from=*) CLI_NUMBER_FROM="${arg#*=}" ;; | |
| 124 | + | --out) NEXT_KEY="out" ;; | |
| 125 | + | --out=*) OUT_FILE="${arg#*=}" ;; | |
| 126 | + | --title) NEXT_KEY="title" ;; | |
| 127 | + | --title=*) CLI_TITLE="${arg#*=}" ;; | |
| 128 | + | --subtitle) NEXT_KEY="subtitle" ;; | |
| 129 | + | --subtitle=*) CLI_SUBTITLE="${arg#*=}" ;; | |
| 130 | + | --author) NEXT_KEY="author" ;; | |
| 131 | + | --author=*) CLI_AUTHOR="${arg#*=}" ;; | |
| 132 | + | --header) NEXT_KEY="header" ;; | |
| 133 | + | --header=*) CLI_HEADER="${arg#*=}" ;; | |
| 134 | + | --footer) NEXT_KEY="footer" ;; | |
| 135 | + | --footer=*) CLI_FOOTER="${arg#*=}" ;; | |
| 136 | + | --date) NEXT_KEY="date" ;; | |
| 137 | + | --date=*) CLI_DATE="${arg#*=}" ;; | |
| 138 | + | --watermark) NEXT_KEY="watermark" ;; | |
| 139 | + | --watermark=*) CLI_WATERMARK="${arg#*=}" ;; | |
| 140 | + | --version) echo "pdfify v${VERSION}"; exit 0 ;; | |
| 141 | + | --help|-h) echo -e "${BOLD}pdfify${RESET} v${VERSION} — Markdown to PDF" | |
| 142 | + | echo "" | |
| 143 | + | echo -e "${BOLD}Usage:${RESET} pdfify ${CYAN}<file.md> [file2.md ...]${RESET} [options]" | |
| 144 | + | echo "" | |
| 145 | + | echo -e "${BOLD}Options:${RESET}" | |
| 146 | + | echo -e " ${DIM}--out FILE${RESET} Output file (single input only)" | |
| 147 | + | echo -e " ${DIM}--toc-level N${RESET} TOC depth: 0=none, 1=H1, 2=H2, 3=H3 (default: 3)" | |
| 148 | + | echo -e " ${DIM}--numbers${RESET} Enable numbered headings (default)" | |
| 149 | + | echo -e " ${DIM}--no-numbers${RESET} Disable numbered headings" | |
| 150 | + | echo -e " ${DIM}--number-from N${RESET} Start numbering at heading level N (default: 2)" | |
| 151 | + | echo -e " ${DIM}--open${RESET} Open PDF after generation" | |
| 152 | + | echo -e " ${DIM}--preview${RESET} Render to /tmp and open (no permanent file)" | |
| 153 | + | echo -e " ${DIM}--watch${RESET} Watch for changes and regenerate" | |
| 154 | + | echo -e " ${DIM}--rebuild${RESET} Force rebuild the Docker image" | |
| 155 | + | echo -e " ${DIM}--clean${RESET} Remove the Docker image" | |
| 156 | + | echo -e " ${DIM}--update${RESET} Update pdfify to latest version from gist" | |
| 157 | + | echo -e " ${DIM}--version${RESET} Show version" | |
| 158 | + | echo "" | |
| 159 | + | echo -e "${BOLD}Overrides${RESET} (CLI trumps frontmatter):" | |
| 160 | + | echo -e " ${DIM}--title TEXT${RESET} ${DIM}--subtitle TEXT${RESET}" | |
| 161 | + | echo -e " ${DIM}--author TEXT${RESET} ${DIM}--header TEXT${RESET}" | |
| 162 | + | echo -e " ${DIM}--footer TEXT${RESET} ${DIM}--date TEXT${RESET}" | |
| 163 | + | echo -e " ${DIM}--watermark TEXT${RESET}" | |
| 164 | + | echo "" | |
| 165 | + | echo -e "${BOLD}Frontmatter:${RESET}" | |
| 166 | + | echo -e " title, subtitle, author, header, footer, toc-level, date," | |
| 167 | + | echo -e " numbersections (true/false), numberfrom (1-4), watermark" | |
| 168 | + | exit 0 ;; | |
| 169 | + | *) POSITIONAL+=("$arg") ;; | |
| 170 | + | esac | |
| 171 | + | done | |
| 172 | + | ||
| 173 | + | if [[ ${#POSITIONAL[@]} -lt 1 ]]; then | |
| 174 | + | echo -e "${BOLD}Usage:${RESET} pdfify ${CYAN}<file.md> [file2.md ...]${RESET} [options]" | |
| 175 | + | echo -e " Run ${CYAN}pdfify --help${RESET} for all options" | |
| 176 | + | exit 1 | |
| 177 | + | fi | |
| 178 | + | ||
| 179 | + | if [[ -n "$OUT_FILE" && ${#POSITIONAL[@]} -gt 1 ]]; then | |
| 180 | + | echo -e "${RED}Error:${RESET} --out cannot be used with multiple input files" | |
| 181 | + | exit 1 | |
| 182 | + | fi | |
| 183 | + | ||
| 184 | + | # --- Open helper --- | |
| 185 | + | open_pdf() { | |
| 186 | + | local pdf="$1" | |
| 187 | + | if command -v open >/dev/null 2>&1; then | |
| 188 | + | open "$pdf" | |
| 189 | + | elif command -v xdg-open >/dev/null 2>&1; then | |
| 190 | + | xdg-open "$pdf" | |
| 191 | + | fi | |
| 192 | + | } | |
| 193 | + | ||
| 194 | + | header "pdfify v${VERSION}" | |
| 195 | + | ||
| 196 | + | # --- Embedded Dockerfile --- | |
| 197 | + | DOCKERFILE=$(cat <<'DOCKERFILE_END' | |
| 198 | + | FROM node:20-slim | |
| 199 | + | ENV DEBIAN_FRONTEND=noninteractive | |
| 200 | + | RUN apt-get update -qq && \ | |
| 201 | + | apt-get install -y --no-install-recommends \ | |
| 202 | + | pandoc \ | |
| 203 | + | texlive-latex-recommended \ | |
| 204 | + | texlive-latex-extra \ | |
| 205 | + | texlive-fonts-recommended \ | |
| 206 | + | texlive-fonts-extra \ | |
| 207 | + | texlive-xetex \ | |
| 208 | + | lmodern \ | |
| 209 | + | librsvg2-bin \ | |
| 210 | + | chromium \ | |
| 211 | + | ca-certificates \ | |
| 212 | + | fonts-liberation \ | |
| 213 | + | fonts-roboto \ | |
| 214 | + | fonts-roboto-unhinted \ | |
| 215 | + | fonts-noto-color-emoji \ | |
| 216 | + | wget \ | |
| 217 | + | fontconfig \ | |
| 218 | + | && rm -rf /var/lib/apt/lists/* | |
| 219 | + | RUN mkdir -p /usr/share/fonts/truetype/roboto-mono && \ | |
| 220 | + | for style in Regular Bold Italic BoldItalic Medium MediumItalic Light LightItalic; do \ | |
| 221 | + | wget -q "https://github.com/googlefonts/RobotoMono/raw/main/fonts/ttf/RobotoMono-${style}.ttf" \ | |
| 222 | + | -O "/usr/share/fonts/truetype/roboto-mono/RobotoMono-${style}.ttf" 2>/dev/null || true; \ | |
| 223 | + | done && \ | |
| 224 | + | fc-cache -f | |
| 225 | + | RUN npm install -g @mermaid-js/mermaid-cli | |
| 226 | + | ENV PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=true | |
| 227 | + | ENV PUPPETEER_EXECUTABLE_PATH=/usr/bin/chromium | |
| 228 | + | ENV CHROME_PATH=/usr/bin/chromium | |
| 229 | + | RUN echo '{"maxTextSize": 90000, "flowchart": {"useMaxWidth": true}, "theme": "default"}' > /opt/mermaid-config.json | |
| 230 | + | RUN echo '{"args": ["--no-sandbox", "--disable-setuid-sandbox", "--disable-dev-shm-usage", "--disable-gpu"]}' > /opt/puppeteer-config.json | |
| 231 | + | WORKDIR /work | |
| 232 | + | ENTRYPOINT ["/bin/bash"] | |
| 233 | + | DOCKERFILE_END | |
| 234 | + | ) | |
| 235 | + | ||
| 236 | + | # --- Build Docker image --- | |
| 237 | + | echo "" | |
| 238 | + | if [[ $REBUILD -eq 1 ]]; then | |
| 239 | + | info "Removing existing Docker image ${CYAN}${IMAGE_NAME}${RESET}..." | |
| 240 | + | docker rmi "$IMAGE_NAME" >/dev/null 2>&1 || true | |
| 241 | + | success "Image removed" | |
| 242 | + | fi | |
| 243 | + | ||
| 244 | + | if docker image inspect "$IMAGE_NAME" >/dev/null 2>&1; then | |
| 245 | + | info "Docker image ${GREEN}${IMAGE_NAME}${RESET} found ${DIM}(cached)${RESET}" | |
| 246 | + | success "Reusing existing image" | |
| 247 | + | else | |
| 248 | + | info "Building Docker image ${CYAN}${IMAGE_NAME}${RESET}..." | |
| 249 | + | detail "Installing: pandoc, XeLaTeX, mermaid-cli, Chromium, fonts" | |
| 250 | + | detail "This takes 2-3 minutes on first run (cached after)" | |
| 251 | + | echo "" | |
| 252 | + | BUILD_CTX=$(mktemp -d) | |
| 253 | + | echo "$DOCKERFILE" | DOCKER_BUILDKIT=0 docker build -t "$IMAGE_NAME" -f - "$BUILD_CTX" 2>&1 | while IFS= read -r line; do | |
| 254 | + | if [[ "$line" =~ ^Step\ ([0-9]+)/([0-9]+) ]]; then | |
| 255 | + | echo -e " ${CYAN}[${BASH_REMATCH[1]}/${BASH_REMATCH[2]}]${RESET} ${DIM}${line#*: }${RESET}" | |
| 256 | + | elif [[ "$line" == *"Successfully tagged"* ]]; then | |
| 257 | + | echo -e " ${GREEN}${line}${RESET}" | |
| 258 | + | elif [[ "$line" == *"ERROR"* || "$line" == *"error"* ]]; then | |
| 259 | + | echo -e " ${RED}${line}${RESET}" | |
| 260 | + | fi | |
| 261 | + | done | |
| 262 | + | rm -rf "$BUILD_CTX" | |
| 263 | + | if ! docker image inspect "$IMAGE_NAME" >/dev/null 2>&1; then | |
| 264 | + | echo -e "\n${RED}Docker build failed. Re-running with full output:${RESET}\n" | |
| 265 | + | BUILD_CTX=$(mktemp -d) | |
| 266 | + | echo "$DOCKERFILE" | DOCKER_BUILDKIT=0 docker build -t "$IMAGE_NAME" -f - "$BUILD_CTX" | |
| 267 | + | rm -rf "$BUILD_CTX" | |
| 268 | + | exit 1 | |
| 269 | + | fi | |
| 270 | + | success "Docker image built" | |
| 271 | + | fi | |
| 272 | + | ||
| 273 | + | # === Per-file conversion === | |
| 274 | + | convert_file() { | |
| 275 | + | local INPUT_PATH="$1" | |
| 276 | + | local OUTPUT_OVERRIDE="$2" | |
| 277 | + | ||
| 278 | + | # --- Resolve paths --- | |
| 279 | + | local INPUT INPUT_DIR INPUT_FILE OUTPUT OUTPUT_DIR OUTPUT_FILE | |
| 280 | + | INPUT="$(cd "$(dirname "$INPUT_PATH")" && pwd)/$(basename "$INPUT_PATH")" | |
| 281 | + | if [[ ! -f "$INPUT" ]]; then | |
| 282 | + | echo -e "${RED}Error:${RESET} $INPUT_PATH not found" | |
| 283 | + | return 1 | |
| 284 | + | fi | |
| 285 | + | INPUT_DIR="$(dirname "$INPUT")" | |
| 286 | + | INPUT_FILE="$(basename "$INPUT")" | |
| 287 | + | OUTPUT="${OUTPUT_OVERRIDE:-${INPUT%.md}.pdf}" | |
| 288 | + | OUTPUT_DIR="$(cd "$(dirname "$OUTPUT")" 2>/dev/null && pwd || (mkdir -p "$(dirname "$OUTPUT")" && cd "$(dirname "$OUTPUT")" && pwd))" | |
| 289 | + | OUTPUT="${OUTPUT_DIR}/$(basename "$OUTPUT")" | |
| 290 | + | OUTPUT_FILE="$(basename "$OUTPUT")" | |
| 291 | + | ||
| 292 | + | # Preview mode: write temp file in input dir (Docker-mountable), move to /tmp after | |
| 293 | + | local PREVIEW_FINAL="" | |
| 294 | + | if [[ $PREVIEW -eq 1 ]]; then | |
| 295 | + | local base="${INPUT_FILE%.md}" | |
| 296 | + | PREVIEW_FINAL="/tmp/pdfify-preview-${base}.pdf" | |
| 297 | + | OUTPUT_FILE=".pdfify-preview-${base}.pdf" | |
| 298 | + | OUTPUT="${OUTPUT_DIR}/${OUTPUT_FILE}" | |
| 299 | + | fi | |
| 300 | + | ||
| 301 | + | info "Input: ${CYAN}${INPUT}${RESET}" | |
| 302 | + | if [[ -n "$PREVIEW_FINAL" ]]; then | |
| 303 | + | info "Output: ${CYAN}${PREVIEW_FINAL}${RESET} ${DIM}(preview)${RESET}" | |
| 304 | + | else | |
| 305 | + | info "Output: ${CYAN}${OUTPUT}${RESET}" | |
| 306 | + | fi | |
| 307 | + | ||
| 308 | + | # --- Parse YAML frontmatter --- | |
| 309 | + | local FM_TITLE="" FM_SUBTITLE="" FM_AUTHOR="" FM_FOOTER="" FM_HEADER="" | |
| 310 | + | local FM_TOC_LEVEL="" FM_DATE="" FM_NUMBERSECTIONS="" FM_NUMBERFROM="" FM_WATERMARK="" | |
| 311 | + | local FM_DATE_HASH="" FM_DATE_DIRTY="" FM_DATE_LABEL="" | |
| 312 | + | ||
| 313 | + | if head -1 "$INPUT" | grep -q '^---'; then | |
| 314 | + | local FM_BLOCK | |
| 315 | + | FM_BLOCK=$(awk 'NR==1 && /^---/{found=1; next} found && /^---/{exit} found{print}' "$INPUT") | |
| 316 | + | extract_fm() { echo "$FM_BLOCK" | sed -n "s/^$1:[[:space:]]*//p" | sed 's/^["'"'"']\(.*\)["'"'"']$/\1/'; } | |
| 317 | + | FM_TITLE=$(extract_fm "title") | |
| 318 | + | FM_AUTHOR=$(extract_fm "author") | |
| 319 | + | FM_SUBTITLE=$(extract_fm "subtitle") | |
| 320 | + | FM_FOOTER=$(extract_fm "footer") | |
| 321 | + | FM_HEADER=$(extract_fm "header") | |
| 322 | + | FM_TOC_LEVEL=$(extract_fm "toc-level") | |
| 323 | + | FM_DATE=$(extract_fm "date") | |
| 324 | + | FM_NUMBERSECTIONS=$(extract_fm "numbersections") | |
| 325 | + | FM_NUMBERFROM=$(extract_fm "numberfrom") | |
| 326 | + | FM_WATERMARK=$(extract_fm "watermark") | |
| 327 | + | fi | |
| 328 | + | ||
| 329 | + | # --- CLI overrides frontmatter --- | |
| 330 | + | [[ -n "$CLI_TITLE" ]] && FM_TITLE="$CLI_TITLE" | |
| 331 | + | [[ -n "$CLI_SUBTITLE" ]] && FM_SUBTITLE="$CLI_SUBTITLE" | |
| 332 | + | [[ -n "$CLI_AUTHOR" ]] && FM_AUTHOR="$CLI_AUTHOR" | |
| 333 | + | [[ -n "$CLI_FOOTER" ]] && FM_FOOTER="$CLI_FOOTER" | |
| 334 | + | [[ -n "$CLI_HEADER" ]] && FM_HEADER="$CLI_HEADER" | |
| 335 | + | [[ -n "$CLI_DATE" ]] && FM_DATE="$CLI_DATE" | |
| 336 | + | [[ -n "$CLI_WATERMARK" ]] && FM_WATERMARK="$CLI_WATERMARK" | |
| 337 | + | [[ -n "$CLI_TOC_LEVEL" ]] && FM_TOC_LEVEL="$CLI_TOC_LEVEL" | |
| 338 | + | [[ -n "$CLI_NUMBER_FROM" ]] && FM_NUMBERFROM="$CLI_NUMBER_FROM" | |
| 339 | + | [[ -n "$CLI_NUMBERS" ]] && FM_NUMBERSECTIONS="$CLI_NUMBERS" | |
| 340 | + | ||
| 341 | + | # --- Auto-detect document structure --- | |
| 342 | + | # Count H1 headings (outside code blocks) | |
| 343 | + | local H1_COUNT=0 IN_CODE_SCAN=0 FIRST_H1_TEXT="" | |
| 344 | + | while IFS= read -r scanline || [[ -n "$scanline" ]]; do | |
| 345 | + | [[ "$scanline" =~ ^\`\`\` ]] && { if [[ $IN_CODE_SCAN -eq 0 ]]; then IN_CODE_SCAN=1; else IN_CODE_SCAN=0; fi; continue; } | |
| 346 | + | if [[ $IN_CODE_SCAN -eq 0 && "$scanline" =~ ^#\ ]]; then | |
| 347 | + | H1_COUNT=$((H1_COUNT + 1)) | |
| 348 | + | [[ $H1_COUNT -eq 1 ]] && FIRST_H1_TEXT="${scanline#\# }" | |
| 349 | + | fi | |
| 350 | + | done < "$INPUT" | |
| 351 | + | ||
| 352 | + | local FILE_TOC_LEVEL="${FM_TOC_LEVEL:-3}" | |
| 353 | + | local FILE_NUMBERS=1 | |
| 354 | + | [[ "$FM_NUMBERSECTIONS" == "false" ]] && FILE_NUMBERS=0 | |
| 355 | + | ||
| 356 | + | # Auto-determine numberfrom based on structure (if not explicitly set) | |
| 357 | + | local FILE_NUMBER_FROM="${FM_NUMBERFROM:-}" | |
| 358 | + | local HIDE_FIRST_H1=0 | |
| 359 | + | if [[ -z "$FILE_NUMBER_FROM" ]]; then | |
| 360 | + | if [[ $H1_COUNT -eq 1 ]]; then | |
| 361 | + | # Single H1 = document title; number from H2, hide H1 in body | |
| 362 | + | FILE_NUMBER_FROM=2 | |
| 363 | + | HIDE_FIRST_H1=1 | |
| 364 | + | # Use H1 text as title if no title set | |
| 365 | + | [[ -z "$FM_TITLE" ]] && FM_TITLE="$FIRST_H1_TEXT" | |
| 366 | + | detail "Auto: ${DIM}single H1 detected → using as title, numbering from H2${RESET}" | |
| 367 | + | else | |
| 368 | + | # Multiple H1s = sections; number from H1 | |
| 369 | + | FILE_NUMBER_FROM=1 | |
| 370 | + | detail "Auto: ${DIM}${H1_COUNT} H1s detected → numbering from H1${RESET}" | |
| 371 | + | fi | |
| 372 | + | fi | |
| 373 | + | # Default date: current date/time | |
| 374 | + | # Set to "none" in frontmatter or --date to suppress | |
| 375 | + | FM_DATE_HASH="${FM_DATE_HASH:-}" | |
| 376 | + | FM_DATE_DIRTY="${FM_DATE_DIRTY:-}" | |
| 377 | + | if [[ "$FM_DATE" == "none" || "$FM_DATE" == "false" ]]; then | |
| 378 | + | FM_DATE="" | |
| 379 | + | elif [[ -z "$FM_DATE" && -z "$CLI_DATE" ]]; then | |
| 380 | + | FM_DATE="$(date +"%Y-%m-%d %H:%M")" | |
| 381 | + | fi | |
| 382 | + | ||
| 383 | + | echo "" | |
| 384 | + | [[ -n "$FM_TITLE" ]] && detail "Title: ${CYAN}${FM_TITLE}${RESET}" | |
| 385 | + | [[ -n "$FM_SUBTITLE" ]] && detail "Subtitle: ${CYAN}${FM_SUBTITLE}${RESET}" | |
| 386 | + | [[ -n "$FM_AUTHOR" ]] && detail "Author: ${CYAN}${FM_AUTHOR}${RESET}" | |
| 387 | + | [[ -n "$FM_HEADER" ]] && detail "Header: ${CYAN}${FM_HEADER}${RESET}" | |
| 388 | + | [[ -n "$FM_FOOTER" ]] && detail "Footer: ${CYAN}${FM_FOOTER}${RESET}" | |
| 389 | + | detail "Date: ${CYAN}${FM_DATE}${RESET}" | |
| 390 | + | detail "TOC: ${CYAN}level ${FILE_TOC_LEVEL}${RESET}" | |
| 391 | + | detail "Numbered: ${CYAN}$([ $FILE_NUMBERS -eq 1 ] && echo "yes (from H${FILE_NUMBER_FROM})" || echo no)${RESET}" | |
| 392 | + | [[ -n "$FM_WATERMARK" ]] && detail "Watermark: ${CYAN}${FM_WATERMARK}${RESET}" | |
| 393 | + | ||
| 394 | + | # --- Git hash for source file --- | |
| 395 | + | local GIT_STAMP="" | |
| 396 | + | ||
| 397 | + | # --- Discover images referenced in the markdown --- | |
| 398 | + | echo "" | |
| 399 | + | info "Scanning ${CYAN}${INPUT_FILE}${RESET} for assets..." | |
| 400 | + | ||
| 401 | + | IMAGES=() | |
| 402 | + | while IFS= read -r img; do | |
| 403 | + | [[ -z "$img" ]] && continue | |
| 404 | + | [[ "$img" =~ ^https?:// ]] && continue | |
| 405 | + | if [[ -f "$INPUT_DIR/$img" ]]; then | |
| 406 | + | IMAGES+=("$img") | |
| 407 | + | success "Image: ${CYAN}${img}${RESET} ${DIM}($(du -h "$INPUT_DIR/$img" | cut -f1 | tr -d ' '))${RESET}" | |
| 408 | + | else | |
| 409 | + | warn "Image: ${YELLOW}${img}${RESET} ${RED}(not found)${RESET}" | |
| 410 | + | fi | |
| 411 | + | done < <(sed -n 's/.*!\[[^]]*\](\([^)]*\)).*/\1/p' "$INPUT"; sed -n 's/.*src="\([^"]*\)".*/\1/p' "$INPUT") | |
| 412 | + | ||
| 413 | + | MERMAID_COUNT=$(grep -c '```mermaid' "$INPUT" || true) | |
| 414 | + | if [[ $MERMAID_COUNT -gt 0 ]]; then | |
| 415 | + | success "Mermaid diagrams: ${CYAN}${MERMAID_COUNT}${RESET}" | |
| 416 | + | fi | |
| 417 | + | ||
| 418 | + | CALLOUT_COUNT=$(grep -c '> \[!' "$INPUT" || true) | |
| 419 | + | if [[ $CALLOUT_COUNT -gt 0 ]]; then | |
| 420 | + | success "Callouts: ${CYAN}${CALLOUT_COUNT}${RESET}" | |
| 421 | + | fi | |
| 422 | + | ||
| 423 | + | TABLE_COUNT=$(grep -c '^|' "$INPUT" || true) | |
| 424 | + | CODE_COUNT=$(grep -c '```' "$INPUT" || true) | |
| 425 | + | CODE_COUNT=$(( (CODE_COUNT - MERMAID_COUNT * 2) / 2 )) | |
| 426 | + | [[ $TABLE_COUNT -gt 0 ]] && detail "Tables: ${TABLE_COUNT} rows" | |
| 427 | + | [[ $CODE_COUNT -gt 0 ]] && detail "Code blocks: ~${CODE_COUNT}" | |
| 428 | + | ||
| 429 | + | echo "" | |
| 430 | + | info "Found ${GREEN}${#IMAGES[@]}${RESET} image(s), ${GREEN}${MERMAID_COUNT}${RESET} mermaid diagram(s), ${GREEN}${CALLOUT_COUNT}${RESET} callout(s)" | |
| 431 | + | ||
| 432 | + | # --- Write the conversion script to a temp file (mounted into Docker) --- | |
| 433 | + | CONVERT_SCRIPT="${INPUT_DIR}/.pdfify-convert-$$.sh" | |
| 434 | + | trap 'rm -f "$CONVERT_SCRIPT"' EXIT | |
| 435 | + | cat > "$CONVERT_SCRIPT" <<'INNER_SCRIPT' | |
| 436 | + | #!/bin/bash | |
| 437 | + | set -euo pipefail | |
| 438 | + | ||
| 439 | + | RED='\033[0;31m' | |
| 440 | + | GREEN='\033[0;32m' | |
| 441 | + | YELLOW='\033[0;33m' | |
| 442 | + | BLUE='\033[0;34m' | |
| 443 | + | CYAN='\033[0;36m' | |
| 444 | + | BOLD='\033[1m' | |
| 445 | + | DIM='\033[2m' | |
| 446 | + | RESET='\033[0m' | |
| 447 | + | ||
| 448 | + | info() { echo -e "${BLUE}::${RESET} ${BOLD}$*${RESET}"; } | |
| 449 | + | success() { echo -e "${GREEN}✓${RESET} $*"; } | |
| 450 | + | detail() { echo -e " ${DIM}→${RESET} $*"; } | |
| 451 | + | ||
| 452 | + | INPUT_FILE="$1" | |
| 453 | + | OUTPUT_FILE="$2" | |
| 454 | + | WORKDIR="/work" | |
| 455 | + | ||
| 456 | + | cd "$WORKDIR" | |
| 457 | + | ||
| 458 | + | # --- Step 0: Strip first H1 if it's being used as document title --- | |
| 459 | + | HIDE_FIRST_H1="${HIDE_FIRST_H1:-0}" | |
| 460 | + | EFFECTIVE_INPUT="$INPUT_FILE" | |
| 461 | + | if [[ "$HIDE_FIRST_H1" == "1" ]]; then | |
| 462 | + | STRIPPED=$(mktemp /tmp/pdfify-stripped-XXXXXX.md) | |
| 463 | + | FOUND_H1=0 | |
| 464 | + | IN_CODE_BLK=0 | |
| 465 | + | IN_FMATTER=0 | |
| 466 | + | while IFS= read -r line || [[ -n "$line" ]]; do | |
| 467 | + | [[ "$line" =~ ^\`\`\` ]] && { if [[ $IN_CODE_BLK -eq 0 ]]; then IN_CODE_BLK=1; else IN_CODE_BLK=0; fi; } | |
| 468 | + | if [[ "$line" == "---" && $IN_CODE_BLK -eq 0 ]]; then | |
| 469 | + | if [[ $IN_FMATTER -eq 0 && $FOUND_H1 -eq 0 ]]; then IN_FMATTER=1; else IN_FMATTER=0; fi | |
| 470 | + | fi | |
| 471 | + | # Skip the first H1 (and any blank line immediately after) | |
| 472 | + | if [[ $FOUND_H1 -eq 0 && $IN_CODE_BLK -eq 0 && $IN_FMATTER -eq 0 && "$line" =~ ^#\ ]]; then | |
| 473 | + | FOUND_H1=1 | |
| 474 | + | continue | |
| 475 | + | fi | |
| 476 | + | # Skip blank line right after removed H1 | |
| 477 | + | if [[ $FOUND_H1 -eq 1 && -z "$line" ]]; then | |
| 478 | + | FOUND_H1=2 | |
| 479 | + | continue | |
| 480 | + | fi | |
| 481 | + | [[ $FOUND_H1 -eq 1 ]] && FOUND_H1=2 | |
| 482 | + | echo "$line" >> "$STRIPPED" | |
| 483 | + | done < "$INPUT_FILE" | |
| 484 | + | EFFECTIVE_INPUT="$(basename "$STRIPPED")" | |
| 485 | + | detail "Stripped first H1 (promoted to title)" | |
| 486 | + | fi | |
| 487 | + | ||
| 488 | + | # --- Step 1: Pre-process Obsidian callouts --- | |
| 489 | + | info "Pre-processing callouts..." | |
| 490 | + | ||
| 491 | + | CALLOUT_MD=$(mktemp /tmp/pdfify-callout-XXXXXX.md) | |
| 492 | + | IN_CALLOUT=0 | |
| 493 | + | CALLOUT_TYPE="" | |
| 494 | + | CALLOUT_TITLE="" | |
| 495 | + | CALLOUT_BUF="" | |
| 496 | + | CALLOUT_COUNT=0 | |
| 497 | + | ||
| 498 | + | flush_callout() { | |
| 499 | + | if [[ $IN_CALLOUT -eq 1 && -n "$CALLOUT_TYPE" ]]; then | |
| 500 | + | CALLOUT_COUNT=$((CALLOUT_COUNT + 1)) | |
| 501 | + | local latex_type | |
| 502 | + | case "${CALLOUT_TYPE,,}" in | |
| 503 | + | info|note) latex_type="calloutinfo" ;; | |
| 504 | + | tip|hint) latex_type="callouttip" ;; | |
| 505 | + | warning|caution) latex_type="calloutwarning" ;; | |
| 506 | + | danger|error|bug) latex_type="calloutdanger" ;; | |
| 507 | + | example) latex_type="calloutexample" ;; | |
| 508 | + | quote|cite) latex_type="calloutquote" ;; | |
| 509 | + | *) latex_type="calloutinfo" ;; | |
| 510 | + | esac | |
| 511 | + | echo "" >> "$CALLOUT_MD" | |
| 512 | + | echo '```{=latex}' >> "$CALLOUT_MD" | |
| 513 | + | echo "\\begin{${latex_type}}{${CALLOUT_TITLE}}" >> "$CALLOUT_MD" | |
| 514 | + | echo '```' >> "$CALLOUT_MD" | |
| 515 | + | echo "" >> "$CALLOUT_MD" | |
| 516 | + | echo "$CALLOUT_BUF" >> "$CALLOUT_MD" | |
| 517 | + | echo "" >> "$CALLOUT_MD" | |
| 518 | + | echo '```{=latex}' >> "$CALLOUT_MD" | |
| 519 | + | echo "\\end{${latex_type}}" >> "$CALLOUT_MD" | |
| 520 | + | echo '```' >> "$CALLOUT_MD" | |
| 521 | + | echo "" >> "$CALLOUT_MD" | |
| 522 | + | fi | |
| 523 | + | IN_CALLOUT=0 | |
| 524 | + | CALLOUT_TYPE="" | |
| 525 | + | CALLOUT_TITLE="" | |
| 526 | + | CALLOUT_BUF="" | |
| 527 | + | } | |
| 528 | + | ||
| 529 | + | while IFS= read -r line || [[ -n "$line" ]]; do | |
| 530 | + | if [[ "$line" =~ ^\>\ *\[!([a-zA-Z]+)\]\ *(.*) ]]; then | |
| 531 | + | flush_callout | |
| 532 | + | IN_CALLOUT=1 | |
| 533 | + | CALLOUT_TYPE="${BASH_REMATCH[1]}" | |
| 534 | + | CALLOUT_TITLE="${BASH_REMATCH[2]:-${BASH_REMATCH[1]^}}" | |
| 535 | + | continue | |
| 536 | + | fi | |
| 537 | + | ||
| 538 | + | if [[ $IN_CALLOUT -eq 1 ]]; then | |
| 539 | + | if [[ "$line" =~ ^\>\ ?(.*) ]]; then | |
| 540 | + | CALLOUT_BUF="${CALLOUT_BUF}${BASH_REMATCH[1]} | |
| 541 | + | " | |
| 542 | + | continue | |
| 543 | + | else | |
| 544 | + | flush_callout | |
| 545 | + | fi | |
| 546 | + | fi | |
| 547 | + | ||
| 548 | + | echo "$line" >> "$CALLOUT_MD" | |
| 549 | + | done < "${STRIPPED:-$INPUT_FILE}" | |
| 550 | + | flush_callout | |
| 551 | + | ||
| 552 | + | if [[ $CALLOUT_COUNT -gt 0 ]]; then | |
| 553 | + | success "Converted $CALLOUT_COUNT callout(s)" | |
| 554 | + | fi | |
| 555 | + | ||
| 556 | + | # --- Step 1b+1c: Inject page breaks (after TOC, before each H1) --- | |
| 557 | + | BREAK_INJECTED=$(mktemp /tmp/pdfify-breaks-XXXXXX.md) | |
| 558 | + | H1_COUNT=0 | |
| 559 | + | IN_FM=0 | |
| 560 | + | IN_CODE=0 | |
| 561 | + | DONE_TOC_BREAK=0 | |
| 562 | + | while IFS= read -r line || [[ -n "$line" ]]; do | |
| 563 | + | # Track code blocks (``` opens/closes) | |
| 564 | + | if [[ "$line" =~ ^\`\`\` ]]; then | |
| 565 | + | if [[ $IN_CODE -eq 0 ]]; then IN_CODE=1; else IN_CODE=0; fi | |
| 566 | + | echo "$line" >> "$BREAK_INJECTED" | |
| 567 | + | continue | |
| 568 | + | fi | |
| 569 | + | ||
| 570 | + | # Track frontmatter (only at start of file) | |
| 571 | + | if [[ "$line" == "---" && $IN_CODE -eq 0 ]]; then | |
| 572 | + | if [[ $IN_FM -eq 0 && $H1_COUNT -eq 0 ]]; then IN_FM=1; else IN_FM=0; fi | |
| 573 | + | echo "$line" >> "$BREAK_INJECTED" | |
| 574 | + | continue | |
| 575 | + | fi | |
| 576 | + | ||
| 577 | + | if [[ $IN_CODE -eq 0 && $IN_FM -eq 0 ]]; then | |
| 578 | + | # Before first content after frontmatter: inject TOC page break | |
| 579 | + | if [[ $DONE_TOC_BREAK -eq 0 && "$TOC_LEVEL" -gt 0 && -n "$line" ]]; then | |
| 580 | + | echo "" >> "$BREAK_INJECTED" | |
| 581 | + | echo '```{=latex}' >> "$BREAK_INJECTED" | |
| 582 | + | echo '\newpage' >> "$BREAK_INJECTED" | |
| 583 | + | echo '```' >> "$BREAK_INJECTED" | |
| 584 | + | echo "" >> "$BREAK_INJECTED" | |
| 585 | + | DONE_TOC_BREAK=1 | |
| 586 | + | fi | |
| 587 | + | ||
| 588 | + | # Page break before each top-level section (except first) | |
| 589 | + | # Build the marker: numberfrom=1 → "# ", numberfrom=2 → "## " | |
| 590 | + | BREAK_HASHES=$(printf '#%.0s' $(seq 1 "$FILE_NUMBER_FROM")) | |
| 591 | + | if [[ "$line" == "${BREAK_HASHES} "* ]]; then | |
| 592 | + | # Make sure it's exactly that level, not deeper | |
| 593 | + | NEXT_CHAR="${line:${#BREAK_HASHES}:1}" | |
| 594 | + | if [[ "$NEXT_CHAR" != "#" ]]; then | |
| 595 | + | H1_COUNT=$((H1_COUNT + 1)) | |
| 596 | + | if [[ $H1_COUNT -gt 1 ]]; then | |
| 597 | + | echo "" >> "$BREAK_INJECTED" | |
| 598 | + | echo '```{=latex}' >> "$BREAK_INJECTED" | |
| 599 | + | echo '\newpage' >> "$BREAK_INJECTED" | |
| 600 | + | echo '```' >> "$BREAK_INJECTED" | |
| 601 | + | echo "" >> "$BREAK_INJECTED" | |
| 602 | + | fi | |
| 603 | + | fi | |
| 604 | + | fi | |
| 605 | + | fi | |
| 606 | + | ||
| 607 | + | echo "$line" >> "$BREAK_INJECTED" | |
| 608 | + | done < "$CALLOUT_MD" | |
| 609 | + | rm -f "$CALLOUT_MD" | |
| 610 | + | CALLOUT_MD="$BREAK_INJECTED" | |
| 611 | + | ||
| 612 | + | # --- Step 2: Pre-render Mermaid blocks to PNG --- | |
| 613 | + | info "Pre-rendering Mermaid diagrams..." | |
| 614 | + | ||
| 615 | + | TEMP_MD=$(mktemp /tmp/pdfify-XXXXXX.md) | |
| 616 | + | MERMAID_COUNT=0 | |
| 617 | + | IN_MERMAID=0 | |
| 618 | + | MERMAID_BUF="" | |
| 619 | + | ||
| 620 | + | while IFS= read -r line || [[ -n "$line" ]]; do | |
| 621 | + | if [[ "$line" =~ ^\`\`\`mermaid ]]; then | |
| 622 | + | IN_MERMAID=1 | |
| 623 | + | MERMAID_BUF="" | |
| 624 | + | continue | |
| 625 | + | fi | |
| 626 | + | ||
| 627 | + | if [[ $IN_MERMAID -eq 1 ]]; then | |
| 628 | + | if [[ "$line" =~ ^\`\`\` ]]; then | |
| 629 | + | IN_MERMAID=0 | |
| 630 | + | MERMAID_COUNT=$((MERMAID_COUNT + 1)) | |
| 631 | + | MERMAID_FILE="/tmp/mermaid-${MERMAID_COUNT}.mmd" | |
| 632 | + | MERMAID_PNG="/tmp/mermaid-${MERMAID_COUNT}.png" | |
| 633 | + | ||
| 634 | + | echo "$MERMAID_BUF" > "$MERMAID_FILE" | |
| 635 | + | ||
| 636 | + | detail "Rendering diagram ${CYAN}#${MERMAID_COUNT}${RESET}..." | |
| 637 | + | mmdc -i "$MERMAID_FILE" \ | |
| 638 | + | -o "$MERMAID_PNG" \ | |
| 639 | + | -w 1600 \ | |
| 640 | + | -b transparent \ | |
| 641 | + | -c /opt/mermaid-config.json \ | |
| 642 | + | -p /opt/puppeteer-config.json \ | |
| 643 | + | 2>/dev/null || { | |
| 644 | + | echo -e " ${YELLOW}⚠${RESET} Diagram $MERMAID_COUNT failed — inserting as code block" | |
| 645 | + | echo '```' >> "$TEMP_MD" | |
| 646 | + | echo "$MERMAID_BUF" >> "$TEMP_MD" | |
| 647 | + | echo '```' >> "$TEMP_MD" | |
| 648 | + | continue | |
| 649 | + | } | |
| 650 | + | ||
| 651 | + | SIZE=$(du -h "$MERMAID_PNG" 2>/dev/null | cut -f1 | tr -d ' ') | |
| 652 | + | success "Diagram #${MERMAID_COUNT} rendered ${DIM}(${SIZE})${RESET}" | |
| 653 | + | ||
| 654 | + | echo "" >> "$TEMP_MD" | |
| 655 | + | echo "\\" >> "$TEMP_MD" | |
| 656 | + | echo "" >> "$TEMP_MD" | |
| 657 | + | else | |
| 658 | + | MERMAID_BUF="${MERMAID_BUF}${line} | |
| 659 | + | " | |
| 660 | + | fi | |
| 661 | + | else | |
| 662 | + | echo "$line" >> "$TEMP_MD" | |
| 663 | + | fi | |
| 664 | + | done < "$CALLOUT_MD" | |
| 665 | + | ||
| 666 | + | # --- Lua filter: protect brackets in headings for titlesec --- | |
| 667 | + | # Square brackets in headings break titlesec (\SQSPL@scan error) because LaTeX | |
| 668 | + | # interprets [ as the start of an optional argument. | |
| 669 | + | BRACKET_FILTER=$(mktemp /tmp/pdfify-bracket-filter-XXXXXX.lua) | |
| 670 | + | cat > "$BRACKET_FILTER" <<'LUAFILTER' | |
| 671 | + | -- Protect square brackets in headings to prevent titlesec \SQSPL@scan errors. | |
| 672 | + | -- Brackets in headings make titlesec think they are optional arguments. | |
| 673 | + | -- We replace [ and ] with \lbrack/\rbrack in all inline types. | |
| 674 | + | ||
| 675 | + | function Header(el) | |
| 676 | + | if FORMAT ~= "latex" and FORMAT ~= "pdf" then return nil end | |
| 677 | + | ||
| 678 | + | el = el:walk { | |
| 679 | + | Str = function(s) | |
| 680 | + | if s.text:find("[%[%]]") then | |
| 681 | + | local t = s.text:gsub("%[", "\\lbrack{}"):gsub("%]", "\\rbrack{}") | |
| 682 | + | return pandoc.RawInline("latex", t) | |
| 683 | + | end | |
| 684 | + | end, | |
| 685 | + | Code = function(c) | |
| 686 | + | -- All code in headings must use \oldtexttt to bypass seqsplit | |
| 687 | + | -- (seqsplit in titlesec moving arguments causes \SQSPL@scan errors) | |
| 688 | + | local t = c.text | |
| 689 | + | t = t:gsub("\\", "\\textbackslash ") | |
| 690 | + | t = t:gsub("%%", "\\%%") | |
| 691 | + | t = t:gsub("%#", "\\#") | |
| 692 | + | t = t:gsub("%$", "\\$") | |
| 693 | + | t = t:gsub("%&", "\\&") | |
| 694 | + | t = t:gsub("_", "\\_") | |
| 695 | + | t = t:gsub("%{", "\\{") | |
| 696 | + | t = t:gsub("%}", "\\}") | |
| 697 | + | t = t:gsub("~", "\\textasciitilde{}") | |
| 698 | + | t = t:gsub("%^", "\\textasciicircum{}") | |
| 699 | + | t = t:gsub("%[", "\\lbrack{}"):gsub("%]", "\\rbrack{}") | |
| 700 | + | return pandoc.RawInline("latex", "\\oldtexttt{" .. t .. "}") | |
| 701 | + | end | |
| 702 | + | } | |
| 703 | + | return el | |
| 704 | + | end | |
| 705 | + | LUAFILTER | |
| 706 | + | ||
| 707 | + | echo "" | |
| 708 | + | info "Generating PDF with Pandoc + XeLaTeX..." | |
| 709 | + | detail "Engine: xelatex" | |
| 710 | + | detail "Font: Roboto / Roboto Mono" | |
| 711 | + | detail "Margins: 0.5in, Font size: 10pt" | |
| 712 | + | echo "" | |
| 713 | + | ||
| 714 | + | # Write LaTeX preamble for modern styling | |
| 715 | + | PREAMBLE=$(mktemp /tmp/pdfify-preamble-XXXXXX.tex) | |
| 716 | + | cat > "$PREAMBLE" <<'LATEX' | |
| 717 | + | % --- Modern color scheme --- | |
| 718 | + | \usepackage{xcolor} | |
| 719 | + | \definecolor{accent}{HTML}{374151} | |
| 720 | + | \definecolor{accentdark}{HTML}{111827} | |
| 721 | + | \definecolor{codebg}{HTML}{F8F9FA} | |
| 722 | + | \definecolor{codeborder}{HTML}{E2E8F0} | |
| 723 | + | \definecolor{headrulecolor}{HTML}{E2E8F0} | |
| 724 | + | ||
| 725 | + | % --- Callout colors --- | |
| 726 | + | \definecolor{infobg}{HTML}{EFF6FF} | |
| 727 | + | \definecolor{infobar}{HTML}{3B82F6} | |
| 728 | + | \definecolor{infofg}{HTML}{1E40AF} | |
| 729 | + | \definecolor{tipbg}{HTML}{F0FDF4} | |
| 730 | + | \definecolor{tipbar}{HTML}{22C55E} | |
| 731 | + | \definecolor{tipfg}{HTML}{166534} | |
| 732 | + | \definecolor{warningbg}{HTML}{FFFBEB} | |
| 733 | + | \definecolor{warningbar}{HTML}{F59E0B} | |
| 734 | + | \definecolor{warningfg}{HTML}{92400E} | |
| 735 | + | \definecolor{dangerbg}{HTML}{FEF2F2} | |
| 736 | + | \definecolor{dangerbar}{HTML}{EF4444} | |
| 737 | + | \definecolor{dangerfg}{HTML}{991B1B} | |
| 738 | + | \definecolor{examplebg}{HTML}{F5F3FF} | |
| 739 | + | \definecolor{examplebar}{HTML}{8B5CF6} | |
| 740 | + | \definecolor{examplefg}{HTML}{5B21B6} | |
| 741 | + | \definecolor{quotecallbg}{HTML}{F8F9FA} | |
| 742 | + | \definecolor{quotecallbar}{HTML}{6B7280} | |
| 743 | + | \definecolor{quotecallfg}{HTML}{374151} | |
| 744 | + | ||
| 745 | + | % --- Code block wrapping and styling --- | |
| 746 | + | \usepackage{fvextra} | |
| 747 | + | \DefineVerbatimEnvironment{Highlighting}{Verbatim}{ | |
| 748 | + | breaklines, | |
| 749 | + | breakanywhere, | |
| 750 | + | commandchars=\\\{\}, | |
| 751 | + | fontsize=\small | |
| 752 | + | } | |
| 753 | + | ||
| 754 | + | % Background on code blocks via mdframed | |
| 755 | + | \usepackage[framemethod=tikz]{mdframed} | |
| 756 | + | ||
| 757 | + | % Override pandoc's Shaded environment | |
| 758 | + | \renewenvironment{Shaded}{% | |
| 759 | + | \begin{mdframed}[ | |
| 760 | + | backgroundcolor=codebg, | |
| 761 | + | hidealllines=true, | |
| 762 | + | roundcorner=4pt, | |
| 763 | + | innertopmargin=8pt, | |
| 764 | + | innerbottommargin=8pt, | |
| 765 | + | innerleftmargin=10pt, | |
| 766 | + | innerrightmargin=10pt, | |
| 767 | + | skipabove=10pt, | |
| 768 | + | skipbelow=10pt | |
| 769 | + | ] | |
| 770 | + | }{% | |
| 771 | + | \end{mdframed} | |
| 772 | + | } | |
| 773 | + | ||
| 774 | + | % --- Callout environments --- | |
| 775 | + | \newenvironment{calloutbase}[3]{% | |
| 776 | + | \begin{mdframed}[ | |
| 777 | + | backgroundcolor=#1, | |
| 778 | + | linecolor=#2, | |
| 779 | + | linewidth=3pt, | |
| 780 | + | topline=false, | |
| 781 | + | bottomline=false, | |
| 782 | + | rightline=false, | |
| 783 | + | innertopmargin=12pt, | |
| 784 | + | innerbottommargin=12pt, | |
| 785 | + | innerleftmargin=12pt, | |
| 786 | + | innerrightmargin=12pt, | |
| 787 | + | skipabove=12pt, | |
| 788 | + | skipbelow=12pt, | |
| 789 | + | roundcorner=0pt | |
| 790 | + | ] | |
| 791 | + | \textbf{\color{#2}#3}\par\smallskip\setlength{\parindent}{0pt} | |
| 792 | + | }{% | |
| 793 | + | \end{mdframed} | |
| 794 | + | } | |
| 795 | + | ||
| 796 | + | \newenvironment{calloutinfo}[1]{\begin{calloutbase}{infobg}{infobar}{#1}}{\end{calloutbase}} | |
| 797 | + | \newenvironment{callouttip}[1]{\begin{calloutbase}{tipbg}{tipbar}{#1}}{\end{calloutbase}} | |
| 798 | + | \newenvironment{calloutwarning}[1]{\begin{calloutbase}{warningbg}{warningbar}{#1}}{\end{calloutbase}} | |
| 799 | + | \newenvironment{calloutdanger}[1]{\begin{calloutbase}{dangerbg}{dangerbar}{#1}}{\end{calloutbase}} | |
| 800 | + | \newenvironment{calloutexample}[1]{\begin{calloutbase}{examplebg}{examplebar}{#1}}{\end{calloutbase}} | |
| 801 | + | \newenvironment{calloutquote}[1]{\begin{calloutbase}{quotecallbg}{quotecallbar}{#1}}{\end{calloutbase}} | |
| 802 | + | ||
| 803 | + | % --- PDF bookmarks (sidebar navigation in PDF viewers) --- | |
| 804 | + | \usepackage{bookmark} | |
| 805 | + | \bookmarksetup{ | |
| 806 | + | numbered=false, | |
| 807 | + | open, | |
| 808 | + | openlevel=2 | |
| 809 | + | } | |
| 810 | + | ||
| 811 | + | % --- Title banner --- | |
| 812 | + | \definecolor{titlebg}{HTML}{E5E7EB} | |
| 813 | + | ||
| 814 | + | % --- Page break after TOC --- | |
| 815 | + | \let\oldtableofcontents\tableofcontents | |
| 816 | + | \renewcommand{\tableofcontents}{\oldtableofcontents\clearpage} | |
| 817 | + | ||
| 818 | + | % --- TOC styling --- | |
| 819 | + | \usepackage{tocloft} | |
| 820 | + | \setlength{\cftbeforetoctitleskip}{0.5em} | |
| 821 | + | \renewcommand{\cfttoctitlefont}{\LARGE\bfseries\color{accentdark}\scshape} | |
| 822 | + | \renewcommand{\cftaftertoctitle}{\par\vspace{2pt}{\color{headrulecolor}\hrule height 1pt}\vspace{10pt}} | |
| 823 | + | \renewcommand{\cftsecfont}{\bfseries\color{accentdark}} | |
| 824 | + | \renewcommand{\cftsecpagefont}{\bfseries\color{accentdark}} | |
| 825 | + | \renewcommand{\cftsubsecfont}{\color{accent}} | |
| 826 | + | \renewcommand{\cftsubsecpagefont}{\color{accent}} | |
| 827 | + | \renewcommand{\cftsubsubsecfont}{\small\color{accent}} | |
| 828 | + | \renewcommand{\cftsubsubsecpagefont}{\small\color{accent}} | |
| 829 | + | \renewcommand{\cftsecleader}{\cftdotfill{\cftsecdotsep}} | |
| 830 | + | \renewcommand{\cftsecdotsep}{\cftdotsep} | |
| 831 | + | \setlength{\cftbeforesecskip}{6pt} | |
| 832 | + | \setlength{\cftbeforesubsecskip}{2pt} | |
| 833 | + | ||
| 834 | + | % --- Heading font --- | |
| 835 | + | \newfontfamily\headingfont{Roboto}[BoldFont={Roboto Bold}] | |
| 836 | + | ||
| 837 | + | % --- Symbol fallback (arrows, etc.) --- | |
| 838 | + | \usepackage{newunicodechar} | |
| 839 | + | \newfontfamily\fallbackfont{Liberation Sans}[Scale=MatchLowercase] | |
| 840 | + | \newunicodechar{→}{{\fallbackfont →}} | |
| 841 | + | \newunicodechar{←}{{\fallbackfont ←}} | |
| 842 | + | \newunicodechar{↔}{{\fallbackfont ↔}} | |
| 843 | + | \newunicodechar{⇒}{{\fallbackfont ⇒}} | |
| 844 | + | \newunicodechar{⇐}{{\fallbackfont ⇐}} | |
| 845 | + | \newunicodechar{✓}{{\fallbackfont ✓}} | |
| 846 | + | \newunicodechar{✗}{{\fallbackfont ✗}} | |
| 847 | + | ||
| 848 | + | % --- Modern section headings (tight, bold, dark) --- | |
| 849 | + | \usepackage{titlesec} | |
| 850 | + | ||
| 851 | + | % H1: # headings — large, small caps, dark, with rule | |
| 852 | + | \titleformat{\section} | |
| 853 | + | {\LARGE\headingfont\bfseries\color{accentdark}\addfontfeatures{LetterSpace=5}\scshape} | |
| 854 | + | {\thesection}{0.5em}{}[\vspace{2pt}{\color{headrulecolor}\titlerule[1pt]}] | |
| 855 | + | \titlespacing*{\section}{0pt}{20pt}{10pt} | |
| 856 | + | ||
| 857 | + | % H2: ## headings | |
| 858 | + | \titleformat{\subsection} | |
| 859 | + | {\Large\headingfont\bfseries\color{accentdark}\addfontfeatures{LetterSpace=-1}} | |
| 860 | + | {\thesubsection}{0.5em}{} | |
| 861 | + | \titlespacing*{\subsection}{0pt}{16pt}{8pt} | |
| 862 | + | ||
| 863 | + | % H3: ### headings | |
| 864 | + | \titleformat{\subsubsection} | |
| 865 | + | {\large\bfseries\color{accent}} | |
| 866 | + | {\thesubsubsection}{0.5em}{} | |
| 867 | + | \titlespacing*{\subsubsection}{0pt}{12pt}{6pt} | |
| 868 | + | ||
| 869 | + | % H4: #### headings | |
| 870 | + | \titleformat{\paragraph}[hang] | |
| 871 | + | {\normalsize\bfseries\color{accent}} | |
| 872 | + | {\theparagraph}{0.5em}{} | |
| 873 | + | \titlespacing*{\paragraph}{0pt}{10pt}{4pt} | |
| 874 | + | ||
| 875 | + | %%SECNUMDEPTH_PLACEHOLDER%% | |
| 876 | + | ||
| 877 | + | % --- Page style (header/footer injected by pdfify) --- | |
| 878 | + | \usepackage{fancyhdr} | |
| 879 | + | \pagestyle{fancy} | |
| 880 | + | \fancyhf{} | |
| 881 | + | \renewcommand{\headrulewidth}{0pt} | |
| 882 | + | \renewcommand{\footrulewidth}{0pt} | |
| 883 | + | \setlength{\headheight}{14pt} | |
| 884 | + | %%HEADER_PLACEHOLDER%% | |
| 885 | + | %%FOOTER_PLACEHOLDER%% | |
| 886 | + | % Make plain style identical to fancy (so title/TOC pages get the same footer) | |
| 887 | + | \fancypagestyle{plain}{\fancyhf{}\renewcommand{\headrulewidth}{0pt}\renewcommand{\footrulewidth}{0pt}%%FOOTER_PLAIN%%} | |
| 888 | + | ||
| 889 | + | % --- Blockquote styling (plain > quotes, not callouts) --- | |
| 890 | + | \usepackage{etoolbox} | |
| 891 | + | \renewenvironment{quote}{% | |
| 892 | + | \begin{mdframed}[ | |
| 893 | + | backgroundcolor=infobg, | |
| 894 | + | linecolor=infobar, | |
| 895 | + | linewidth=3pt, | |
| 896 | + | topline=false, | |
| 897 | + | bottomline=false, | |
| 898 | + | rightline=false, | |
| 899 | + | innertopmargin=12pt, | |
| 900 | + | innerbottommargin=12pt, | |
| 901 | + | innerleftmargin=12pt, | |
| 902 | + | innerrightmargin=12pt, | |
| 903 | + | skipabove=10pt, | |
| 904 | + | skipbelow=10pt, | |
| 905 | + | roundcorner=0pt | |
| 906 | + | ]% | |
| 907 | + | }{% | |
| 908 | + | \end{mdframed}% | |
| 909 | + | } | |
| 910 | + | ||
| 911 | + | % --- Table styling --- | |
| 912 | + | \usepackage{booktabs} | |
| 913 | + | \usepackage{colortbl} | |
| 914 | + | \usepackage{longtable} | |
| 915 | + | \usepackage{tabularx} | |
| 916 | + | \arrayrulecolor{codeborder} | |
| 917 | + | ||
| 918 | + | % Allow line breaks in table cells and shrink monospace to fit | |
| 919 | + | \usepackage{array} | |
| 920 | + | \renewcommand{\arraystretch}{1.4} | |
| 921 | + | \let\oldtexttt\texttt | |
| 922 | + | \renewcommand{\texttt}[1]{{\small\oldtexttt{\seqsplit{#1}}}} | |
| 923 | + | \usepackage{seqsplit} | |
| 924 | + | \setlength{\tabcolsep}{4pt} | |
| 925 | + | ||
| 926 | + | % --- Images constrained to page --- | |
| 927 | + | \usepackage{grffile} | |
| 928 | + | \usepackage[export]{adjustbox} | |
| 929 | + | \let\oldincludegraphics\includegraphics | |
| 930 | + | \renewcommand{\includegraphics}[2][]{% | |
| 931 | + | \oldincludegraphics[max width=\textwidth,max height=0.45\textheight,keepaspectratio,#1]{#2}% | |
| 932 | + | } | |
| 933 | + | ||
| 934 | + | % --- Figures don't float --- | |
| 935 | + | \usepackage{float} | |
| 936 | + | \floatplacement{figure}{H} | |
| 937 | + | ||
| 938 | + | % --- Caption styling --- | |
| 939 | + | \usepackage{caption} | |
| 940 | + | \captionsetup{labelformat=empty,font={small,color=gray},skip=4pt} | |
| 941 | + | ||
| 942 | + | % --- Tighter lists --- | |
| 943 | + | \usepackage{enumitem} | |
| 944 | + | \setlist{nosep,leftmargin=1.5em} | |
| 945 | + | ||
| 946 | + | % --- Links --- | |
| 947 | + | \usepackage{hyperref} | |
| 948 | + | \hypersetup{ | |
| 949 | + | colorlinks=true, | |
| 950 | + | linkcolor=accent, | |
| 951 | + | urlcolor=accent, | |
| 952 | + | citecolor=accent | |
| 953 | + | } | |
| 954 | + | ||
| 955 | + | % --- Horizontal rules --- | |
| 956 | + | \renewcommand{\rule}[2]{\textcolor{headrulecolor}{\vrule width \textwidth height 0.5pt}} | |
| 957 | + | LATEX | |
| 958 | + | ||
| 959 | + | TOC_LEVEL="${TOC_LEVEL:-3}" | |
| 960 | + | FM_FOOTER="${FM_FOOTER:-}" | |
| 961 | + | FM_HEADER="${FM_HEADER:-}" | |
| 962 | + | FM_AUTHOR="${FM_AUTHOR:-}" | |
| 963 | + | FM_DATE="${FM_DATE:-}" | |
| 964 | + | FM_DATE_LABEL="${FM_DATE_LABEL:-}" | |
| 965 | + | FM_DATE_HASH="${FM_DATE_HASH:-}" | |
| 966 | + | FM_DATE_DIRTY="${FM_DATE_DIRTY:-}" | |
| 967 | + | FILE_NUMBERS="${FILE_NUMBERS:-1}" | |
| 968 | + | FILE_NUMBER_FROM="${FILE_NUMBER_FROM:-2}" | |
| 969 | + | ||
| 970 | + | # Escape LaTeX special characters in text fields (uses sed to avoid | |
| 971 | + | # bash parameter substitution brace-parsing issues with } in replacements) | |
| 972 | + | latex_escape() { | |
| 973 | + | printf '%s' "$1" | sed \ | |
| 974 | + | -e 's/\\/@@BSLASH@@/g' \ | |
| 975 | + | -e 's/&/\\&/g' \ | |
| 976 | + | -e 's/%/\\%/g' \ | |
| 977 | + | -e 's/\$/\\$/g' \ | |
| 978 | + | -e 's/#/\\#/g' \ | |
| 979 | + | -e 's/_/\\_/g' \ | |
| 980 | + | -e 's/{/\\{/g' \ | |
| 981 | + | -e 's/}/\\}/g' \ | |
| 982 | + | -e 's/~/\\textasciitilde{}/g' \ | |
| 983 | + | -e 's/\^/\\textasciicircum{}/g' \ | |
| 984 | + | -e 's/@@BSLASH@@/\\textbackslash{}/g' | |
| 985 | + | } | |
| 986 | + | ||
| 987 | + | # Inject title banner into preamble | |
| 988 | + | FM_TITLE="${FM_TITLE:-}" | |
| 989 | + | FM_TITLE_TEX="$(latex_escape "$FM_TITLE")" | |
| 990 | + | FM_SUBTITLE_TEX="$(latex_escape "${FM_SUBTITLE:-}")" | |
| 991 | + | FM_AUTHOR_TEX="$(latex_escape "${FM_AUTHOR:-}")" | |
| 992 | + | ||
| 993 | + | { | |
| 994 | + | if [[ -n "$FM_TITLE" ]]; then | |
| 995 | + | cat <<'TITLE_STATIC' | |
| 996 | + | \makeatletter | |
| 997 | + | \renewcommand{\maketitle}{% | |
| 998 | + | \thispagestyle{fancy}% | |
| 999 | + | \vspace*{-\topskip}% | |
| 1000 | + | \vspace*{-\headsep}% | |
| 1001 | + | \vspace*{-\headheight}% | |
| 1002 | + | \vspace*{-0.55in}% | |
| 1003 | + | \noindent\hspace*{-0.5in}% | |
| 1004 | + | \fcolorbox{titlebg}{titlebg}{% | |
| 1005 | + | \parbox{\dimexpr\paperwidth-2\fboxsep-2\fboxrule}{% | |
| 1006 | + | \hspace*{0.3in}\begin{minipage}{\dimexpr\textwidth}% | |
| 1007 | + | \vspace{20pt}% | |
| 1008 | + | TITLE_STATIC | |
| 1009 | + | ||
| 1010 | + | echo " {\\fontsize{28}{34}\\selectfont\\bfseries\\color{black}${FM_TITLE_TEX}}\\\\[6pt]%" | |
| 1011 | + | ||
| 1012 | + | FM_SUBTITLE="${FM_SUBTITLE:-}" | |
| 1013 | + | if [[ -n "$FM_SUBTITLE" ]]; then | |
| 1014 | + | echo " {\\fontsize{14}{18}\\selectfont\\color{black}${FM_SUBTITLE_TEX}}\\\\[8pt]%" | |
| 1015 | + | fi | |
| 1016 | + | ||
| 1017 | + | if [[ -n "$FM_AUTHOR" ]]; then | |
| 1018 | + | echo " {\\fontsize{11}{14}\\selectfont\\color{black}${FM_AUTHOR_TEX}}\\\\[6pt]%" | |
| 1019 | + | fi | |
| 1020 | + | ||
| 1021 | + | if [[ -n "$FM_DATE" ]]; then | |
| 1022 | + | DATE_VAL="" | |
| 1023 | + | if [[ -n "$FM_DATE_HASH" ]]; then | |
| 1024 | + | DATE_VAL="${FM_DATE% · *} · {\\texttt{${FM_DATE_HASH}}}" | |
| 1025 | + | else | |
| 1026 | + | DATE_VAL="${FM_DATE}" | |
| 1027 | + | fi | |
| 1028 | + | DIRTY_PART="" | |
| 1029 | + | if [[ -n "${FM_DATE_DIRTY:-}" ]]; then | |
| 1030 | + | DIRTY_PART=" {\\color{gray}\\itshape (dirty)}" | |
| 1031 | + | fi | |
| 1032 | + | if [[ -n "$FM_DATE_LABEL" ]]; then | |
| 1033 | + | echo " {\\fontsize{10}{12}\\selectfont\\color{black}${DATE_VAL} {\\color{gray}--- ${FM_DATE_LABEL}}${DIRTY_PART}}\\\\[4pt]%" | |
| 1034 | + | else | |
| 1035 | + | echo " {\\fontsize{10}{12}\\selectfont\\color{black}${DATE_VAL}${DIRTY_PART}}\\\\[4pt]%" | |
| 1036 | + | fi | |
| 1037 | + | fi | |
| 1038 | + | ||
| 1039 | + | cat <<'TITLE_END' | |
| 1040 | + | \vspace{6pt}% | |
| 1041 | + | \end{minipage}% | |
| 1042 | + | }% | |
| 1043 | + | }% | |
| 1044 | + | \par\vspace{20pt}% | |
| 1045 | + | } | |
| 1046 | + | \makeatother | |
| 1047 | + | TITLE_END | |
| 1048 | + | echo '\AtBeginDocument{\maketitle}' | |
| 1049 | + | else | |
| 1050 | + | echo '\renewcommand{\maketitle}{}' | |
| 1051 | + | fi | |
| 1052 | + | } >> "$PREAMBLE" | |
| 1053 | + | ||
| 1054 | + | # Inject header/footer into preamble | |
| 1055 | + | GIT_STAMP="${GIT_STAMP:-}" | |
| 1056 | + | ||
| 1057 | + | FOOTER_L="" | |
| 1058 | + | FOOTER_C="" | |
| 1059 | + | FOOTER_R="\\\\fancyfoot[R]{\\\\color{gray}\\\\small Page \\\\thepage\\\\ of \\\\pageref*{LastPage}}" | |
| 1060 | + | ||
| 1061 | + | [[ -n "$FM_FOOTER" ]] && FOOTER_L="\\\\fancyfoot[L]{\\\\color{gray}\\\\small ${FM_FOOTER}}" | |
| 1062 | + | ||
| 1063 | + | sed -i "s|%%FOOTER_PLACEHOLDER%%|\\\\usepackage{lastpage}${FOOTER_L}${FOOTER_C}${FOOTER_R}|" "$PREAMBLE" | |
| 1064 | + | sed -i "s|%%FOOTER_PLAIN%%|${FOOTER_L}${FOOTER_C}${FOOTER_R}|" "$PREAMBLE" | |
| 1065 | + | ||
| 1066 | + | if [[ -n "$FM_HEADER" ]]; then | |
| 1067 | + | sed -i "s|%%HEADER_PLACEHOLDER%%|\\\\fancyhead[C]{\\\\color{gray}\\\\small ${FM_HEADER}}|" "$PREAMBLE" | |
| 1068 | + | else | |
| 1069 | + | sed -i "s|%%HEADER_PLACEHOLDER%%||" "$PREAMBLE" | |
| 1070 | + | fi | |
| 1071 | + | ||
| 1072 | + | # Inject watermark if set | |
| 1073 | + | FM_WATERMARK="${FM_WATERMARK:-}" | |
| 1074 | + | if [[ -n "$FM_WATERMARK" ]]; then | |
| 1075 | + | cat >> "$PREAMBLE" <<WATERMARK | |
| 1076 | + | \\usepackage{eso-pic} | |
| 1077 | + | \\usepackage{tikz} | |
| 1078 | + | \\AddToShipoutPictureFG{% | |
| 1079 | + | \\begin{tikzpicture}[remember picture,overlay] | |
| 1080 | + | \\node[rotate=45,opacity=0.12,scale=10,text=red] at (current page.center) {\\textsf{\\textbf{\\MakeUppercase{${FM_WATERMARK}}}}}; | |
| 1081 | + | \\end{tikzpicture}% | |
| 1082 | + | } | |
| 1083 | + | WATERMARK | |
| 1084 | + | detail "Watermark: ${CYAN}${FM_WATERMARK}${RESET}" | |
| 1085 | + | fi | |
| 1086 | + | ||
| 1087 | + | # Build TOC flags | |
| 1088 | + | # When numbering is on, headings shift by -1, so TOC depth needs +1 to compensate | |
| 1089 | + | TOC_FLAGS=() | |
| 1090 | + | if [[ "$TOC_LEVEL" -gt 0 ]]; then | |
| 1091 | + | TOC_FLAGS+=(--toc --toc-depth="$TOC_LEVEL") | |
| 1092 | + | detail "TOC depth: ${CYAN}${TOC_LEVEL}${RESET}" | |
| 1093 | + | else | |
| 1094 | + | detail "TOC: ${DIM}disabled${RESET}" | |
| 1095 | + | fi | |
| 1096 | + | ||
| 1097 | + | AUTHOR_FLAGS=() | |
| 1098 | + | if [[ -n "$FM_AUTHOR" ]]; then | |
| 1099 | + | AUTHOR_FLAGS+=(-M "author=$FM_AUTHOR") | |
| 1100 | + | fi | |
| 1101 | + | ||
| 1102 | + | # Numbered sections | |
| 1103 | + | NUMBER_FLAGS=() | |
| 1104 | + | if [[ "$FILE_NUMBERS" == "1" ]]; then | |
| 1105 | + | NUMBER_FLAGS+=(--number-sections) | |
| 1106 | + | ||
| 1107 | + | # numberfrom controls which heading level starts getting numbers | |
| 1108 | + | # pandoc: section=1, subsection=2, subsubsection=3 | |
| 1109 | + | cat >> "$PREAMBLE" <<SECNUM | |
| 1110 | + | \\setcounter{secnumdepth}{4} | |
| 1111 | + | SECNUM | |
| 1112 | + | ||
| 1113 | + | if [[ "$FILE_NUMBER_FROM" -ge 2 ]]; then | |
| 1114 | + | # H1 (\section) unnumbered, H2 numbered as 1, 2, 3 | |
| 1115 | + | cat >> "$PREAMBLE" <<'SECNUM2' | |
| 1116 | + | \makeatletter | |
| 1117 | + | \renewcommand{\thesection}{} | |
| 1118 | + | \renewcommand{\thesubsection}{\arabic{subsection}} | |
| 1119 | + | \renewcommand{\thesubsubsection}{\thesubsection.\arabic{subsubsection}} | |
| 1120 | + | % Remove section number from titleformat without changing style | |
| 1121 | + | \titleformat{\section} | |
| 1122 | + | {\LARGE\headingfont\bfseries\color{accentdark}\addfontfeatures{LetterSpace=5}\scshape} | |
| 1123 | + | {}{0em}{}[\vspace{2pt}{\color{headrulecolor}\titlerule[1pt]}] | |
| 1124 | + | \makeatother | |
| 1125 | + | SECNUM2 | |
| 1126 | + | fi | |
| 1127 | + | ||
| 1128 | + | if [[ "$FILE_NUMBER_FROM" -ge 3 ]]; then | |
| 1129 | + | cat >> "$PREAMBLE" <<'SECNUM3' | |
| 1130 | + | \renewcommand{\thesubsection}{} | |
| 1131 | + | \renewcommand{\thesubsubsection}{\arabic{subsubsection}} | |
| 1132 | + | \titleformat{\subsection} | |
| 1133 | + | {\Large\headingfont\bfseries\color{accentdark}\addfontfeatures{LetterSpace=-1}} | |
| 1134 | + | {}{0em}{} | |
| 1135 | + | SECNUM3 | |
| 1136 | + | fi | |
| 1137 | + | fi | |
| 1138 | + | ||
| 1139 | + | # Remove placeholder | |
| 1140 | + | sed -i 's|%%SECNUMDEPTH_PLACEHOLDER%%||' "$PREAMBLE" | |
| 1141 | + | ||
| 1142 | + | pandoc "$TEMP_MD" \ | |
| 1143 | + | -o "$OUTPUT_FILE" \ | |
| 1144 | + | --pdf-engine=xelatex \ | |
| 1145 | + | --lua-filter="$BRACKET_FILTER" \ | |
| 1146 | + | --resource-path=".:$WORKDIR" \ | |
| 1147 | + | --columns=72 \ | |
| 1148 | + | -V geometry:"margin=0.5in,includehead,includefoot" \ | |
| 1149 | + | -V fontsize=10pt \ | |
| 1150 | + | -V mainfont="Roboto" \ | |
| 1151 | + | -V monofont="Roboto Mono" \ | |
| 1152 | + | "${TOC_FLAGS[@]}" \ | |
| 1153 | + | "${AUTHOR_FLAGS[@]}" \ | |
| 1154 | + | "${NUMBER_FLAGS[@]}" \ | |
| 1155 | + | --highlight-style=tango \ | |
| 1156 | + | -H "$PREAMBLE" \ | |
| 1157 | + | --standalone | |
| 1158 | + | ||
| 1159 | + | rm -f "$TEMP_MD" "$CALLOUT_MD" "$PREAMBLE" "${BRACKET_FILTER:-}" "${STRIPPED:-}" /tmp/mermaid-*.mmd /tmp/mermaid-*.png | |
| 1160 | + | ||
| 1161 | + | PAGES=$(strings "$OUTPUT_FILE" 2>/dev/null | grep -c '/Type /Page' || echo "?") | |
| 1162 | + | SIZE=$(du -h "$OUTPUT_FILE" | cut -f1 | tr -d ' ') | |
| 1163 | + | success "PDF generated: ${CYAN}${SIZE}${RESET}, ~${CYAN}${PAGES}${RESET} pages" | |
| 1164 | + | INNER_SCRIPT | |
| 1165 | + | ||
| 1166 | + | chmod +x "$CONVERT_SCRIPT" | |
| 1167 | + | ||
| 1168 | + | # --- Run Docker --- | |
| 1169 | + | echo "" | |
| 1170 | + | info "Launching Docker container..." | |
| 1171 | + | detail "Mounting: ${CYAN}${INPUT_DIR}${RESET} → /work ${DIM}(read-only)${RESET}" | |
| 1172 | + | detail "Output: ${CYAN}${OUTPUT_DIR}${RESET} → /output" | |
| 1173 | + | echo "" | |
| 1174 | + | ||
| 1175 | + | CONVERT_BASENAME="$(basename "$CONVERT_SCRIPT")" | |
| 1176 | + | docker run --rm \ | |
| 1177 | + | -v "$INPUT_DIR:/work:ro" \ | |
| 1178 | + | -v "$OUTPUT_DIR:/output" \ | |
| 1179 | + | -e "TOC_LEVEL=$FILE_TOC_LEVEL" \ | |
| 1180 | + | -e "FM_FOOTER=$FM_FOOTER" \ | |
| 1181 | + | -e "FM_HEADER=$FM_HEADER" \ | |
| 1182 | + | -e "FM_AUTHOR=$FM_AUTHOR" \ | |
| 1183 | + | -e "FM_TITLE=$FM_TITLE" \ | |
| 1184 | + | -e "FM_SUBTITLE=$FM_SUBTITLE" \ | |
| 1185 | + | -e "FM_DATE=$FM_DATE" \ | |
| 1186 | + | -e "FM_DATE_LABEL=${FM_DATE_LABEL:-}" \ | |
| 1187 | + | -e "FM_DATE_HASH=${FM_DATE_HASH:-}" \ | |
| 1188 | + | -e "FM_DATE_DIRTY=${FM_DATE_DIRTY:-}" \ | |
| 1189 | + | -e "GIT_STAMP=${GIT_STAMP:-}" \ | |
| 1190 | + | -e "FILE_NUMBERS=$FILE_NUMBERS" \ | |
| 1191 | + | -e "FILE_NUMBER_FROM=$FILE_NUMBER_FROM" \ | |
| 1192 | + | -e "HIDE_FIRST_H1=$HIDE_FIRST_H1" \ | |
| 1193 | + | -e "FM_WATERMARK=$FM_WATERMARK" \ | |
| 1194 | + | --tmpfs /tmp:exec \ | |
| 1195 | + | "$IMAGE_NAME" "/work/$CONVERT_BASENAME" "$INPUT_FILE" "/output/$OUTPUT_FILE" \ | |
| 1196 | + | || { | |
| 1197 | + | echo "" | |
| 1198 | + | echo -e " ${RED}${BOLD}Error producing PDF.${RESET} Docker/pandoc exited with a non-zero status." | |
| 1199 | + | echo "" | |
| 1200 | + | return 1 | |
| 1201 | + | } | |
| 1202 | + | ||
| 1203 | + | # Move preview file to /tmp and clean up | |
| 1204 | + | if [[ -n "$PREVIEW_FINAL" ]]; then | |
| 1205 | + | mv "$OUTPUT" "$PREVIEW_FINAL" | |
| 1206 | + | OUTPUT="$PREVIEW_FINAL" | |
| 1207 | + | fi | |
| 1208 | + | ||
| 1209 | + | echo "" | |
| 1210 | + | echo -e " ${GREEN}${BOLD}PDF created:${RESET} ${CYAN}${OUTPUT}${RESET}" | |
| 1211 | + | echo "" | |
| 1212 | + | ||
| 1213 | + | # Open if requested | |
| 1214 | + | if [[ $OPEN -eq 1 ]]; then | |
| 1215 | + | open_pdf "$OUTPUT" | |
| 1216 | + | fi | |
| 1217 | + | } | |
| 1218 | + | ||
| 1219 | + | # --- Process each input file --- | |
| 1220 | + | run_all() { | |
| 1221 | + | local FAILED=0 | |
| 1222 | + | for input_file in "${POSITIONAL[@]}"; do | |
| 1223 | + | convert_file "$input_file" "$OUT_FILE" || FAILED=$((FAILED + 1)) | |
| 1224 | + | done | |
| 1225 | + | ||
| 1226 | + | if [[ $FAILED -eq 0 ]]; then | |
| 1227 | + | header "Complete! (${#POSITIONAL[@]} file(s))" | |
| 1228 | + | else | |
| 1229 | + | header "${FAILED} of ${#POSITIONAL[@]} file(s) failed" | |
| 1230 | + | fi | |
| 1231 | + | } | |
| 1232 | + | ||
| 1233 | + | run_all | |
| 1234 | + | ||
| 1235 | + | # --- Watch mode --- | |
| 1236 | + | if [[ $WATCH -eq 1 ]]; then | |
| 1237 | + | info "Watching for changes... ${DIM}(Ctrl+C to stop)${RESET}" | |
| 1238 | + | echo "" | |
| 1239 | + | ||
| 1240 | + | # Get initial checksums (using a temp file instead of associative array for bash 3 compat) | |
| 1241 | + | CHECKSUM_FILE=$(mktemp) | |
| 1242 | + | trap 'rm -f "$CHECKSUM_FILE"' EXIT | |
| 1243 | + | for f in "${POSITIONAL[@]}"; do | |
| 1244 | + | fpath="$(cd "$(dirname "$f")" && pwd)/$(basename "$f")" | |
| 1245 | + | echo "$(_sha256 < "$fpath" | cut -d' ' -f1) $fpath" >> "$CHECKSUM_FILE" | |
| 1246 | + | done | |
| 1247 | + | ||
| 1248 | + | while true; do | |
| 1249 | + | sleep 2 | |
| 1250 | + | CHANGED=0 | |
| 1251 | + | for f in "${POSITIONAL[@]}"; do | |
| 1252 | + | fpath="$(cd "$(dirname "$f")" && pwd)/$(basename "$f")" | |
| 1253 | + | NEW_HASH=$(_sha256 < "$fpath" | cut -d' ' -f1) | |
| 1254 | + | OLD_HASH=$(grep " $fpath\$" "$CHECKSUM_FILE" | cut -d' ' -f1) | |
| 1255 | + | if [[ "$NEW_HASH" != "$OLD_HASH" ]]; then | |
| 1256 | + | CHANGED=1 | |
| 1257 | + | # Update stored checksum | |
| 1258 | + | grep -v " $fpath\$" "$CHECKSUM_FILE" > "${CHECKSUM_FILE}.tmp" || true | |
| 1259 | + | echo "$NEW_HASH $fpath" >> "${CHECKSUM_FILE}.tmp" | |
| 1260 | + | mv "${CHECKSUM_FILE}.tmp" "$CHECKSUM_FILE" | |
| 1261 | + | fi | |
| 1262 | + | done | |
| 1263 | + | if [[ $CHANGED -eq 1 ]]; then | |
| 1264 | + | echo "" | |
| 1265 | + | info "Change detected — rebuilding..." | |
| 1266 | + | echo "" | |
| 1267 | + | run_all | |
| 1268 | + | fi | |
| 1269 | + | done | |
| 1270 | + | fi | |
| 1271 | + | ||
| 1272 | + | # Check for updates (runs after success, fast timeout) | |
| 1273 | + | check_for_update | |
Newer
Older