Revision of pdfify.sh

167

+

echo -e " numbersections (true/false), numberfrom (1-4), watermark,"

168

+

echo -e " pagebreak (true/false — page break before each top-level heading)"

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

328

+

FM_PAGEBREAK=$(extract_fm "pagebreak")

357

+

local FILE_PAGEBREAK=1

358

+

[[ "$FM_PAGEBREAK" == "false" ]] && FILE_PAGEBREAK=0

600

+

if [[ $H1_COUNT -gt 1 && $FILE_PAGEBREAK -eq 1 ]]; then

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

+

770

+

% Override pandoc's Shaded environment (define first if pandoc didn't)

771

+

\makeatletter

772

+

\@ifundefined{Shaded}{\newenvironment{Shaded}{}{}}{}

773

+

\makeatother

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

+

991

+

FILE_PAGEBREAK="${FILE_PAGEBREAK:-1}"

1217

+

-e "FILE_PAGEBREAK=$FILE_PAGEBREAK" \

			@@ -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 / pdfify.sh

OneWheelGeek revised this gist 5 months ago · 24f66a0

OneWheelGeek revised this gist 5 months ago · fd1d9c1