///| Inline serialization (text, emphasis, links, images, code spans, ...).
///|
///| Split from serializer.mbt to keep that file focused on block-level
///| serialization. Table-cell variants and wikilink escaping live here
///| because they are part of the same inline-rendering surface.
///|
/// Write literal text, escaping the characters that would otherwise be read
/// back as markup. Inline text in the CST is already unescaped, so this is
/// what keeps `serialize(parse(x))` stable.
fn write_escaped_markdown_literal(
content : String,
buf : StringBuilder,
) -> Unit {
let len = content.length()
let mut i = 0
for c in content {
match c {
'\\' =>
// Only a backslash that would start an escape sequence needs doubling.
if i + 1 < len &&
is_punctuation(content.unsafe_get(i + 1).unsafe_to_char()) {
buf.write_string("\\\\")
} else {
buf.write_char('\\')
}
'*' | '_' | '[' | '`' | '<' | '&' => {
buf.write_char('\\')
buf.write_char(c)
}
'~' => {
if (i > 0 && content.unsafe_get(i - 1) == '~') ||
(i + 1 < len && content.unsafe_get(i + 1) == '~') {
buf.write_char('\\')
}
buf.write_char('~')
}
'@' => {
let looks_email_like = i > 0 &&
i + 1 < len &&
is_email_local_char(content.unsafe_get(i - 1)) &&
is_ascii_alnum_unit(content.unsafe_get(i + 1))
if looks_email_like {
buf.write_char('\\')
}
buf.write_char('@')
}
_ => buf.write_char(c)
}
i = i + (if c.to_int() > 0xFFFF { 2 } else { 1 })
}
}
///| Normalize GFM bare autolinks to explicit Markdown syntax. This shares the
///|
/// renderer's boundary/delimiter scanner so HTML and Markdown agree.
fn write_escaped_markdown_text(content : String, buf : StringBuilder) -> Unit {
let len = content.length()
let mut pos = 0
while pos < len {
match find_next_autolink(content, pos) {
None => {
write_escaped_markdown_literal(
content.unsafe_substring(start=pos, end=len),
buf,
)
pos = len
}
Some(found) => {
write_escaped_markdown_literal(
content.unsafe_substring(start=pos, end=found.start),
buf,
)
let display = content.unsafe_substring(start=found.start, end=found.end)
match found.kind {
BareAutolinkKind::Url if display.has_prefix("ftp://") => {
// remark-gfm does not normalize bare FTP URLs as links.
buf.write_string("ftp\\://")
buf.write_string(
display.unsafe_substring(start=6, end=display.length()),
)
pos = found.end
}
BareAutolinkKind::Url | BareAutolinkKind::Email => {
buf.write_char('<')
buf.write_string(display)
buf.write_char('>')
pos = found.end
}
BareAutolinkKind::Www => {
buf.write_char('[')
for c in display {
if c == '&' || c == '[' || c == ']' || c == '\\' {
buf.write_char('\\')
}
buf.write_char(c)
}
buf.write_string("](")
write_link_destination("http://" + display, buf)
buf.write_char(')')
pos = found.end
}
}
}
}
}
}
///|
/// Write a link/image title inside double quotes, escaping what would end it.
/// Titles are stored unescaped in the CST, so this is what keeps them valid.
fn write_link_title(title : String, buf : StringBuilder) -> Unit {
buf.write_string(" \"")
for c in title {
if c == '"' || c == '\\' {
buf.write_char('\\')
}
buf.write_char(c)
}
buf.write_char('"')
}
///|
/// Write a link/image destination, escaping parentheses and falling back to
/// the `<...>` form when the destination contains whitespace.
fn write_link_destination(url : String, buf : StringBuilder) -> Unit {
let mut needs_angle = false
for c in url {
if c == ' ' || c == '\t' || c == '\n' || c == '<' || c == '>' {
needs_angle = true
break
}
}
let len = url.length()
let mut i = 0
let escape_backslash = fn(i : Int) -> Bool {
i + 1 < len && is_punctuation(url.unsafe_get(i + 1).unsafe_to_char())
}
if needs_angle {
buf.write_char('<')
for c in url {
match c {
'<' | '>' => {
buf.write_char('\\')
buf.write_char(c)
}
'\\' => {
if escape_backslash(i) {
buf.write_char('\\')
}
buf.write_char(c)
}
_ => buf.write_char(c)
}
i = i + (if c.to_int() > 0xFFFF { 2 } else { 1 })
}
buf.write_char('>')
return
}
for c in url {
match c {
'(' | ')' | '&' => {
buf.write_char('\\')
buf.write_char(c)
}
'\\' => {
if escape_backslash(i) {
buf.write_char('\\')
}
buf.write_char(c)
}
_ => buf.write_char(c)
}
i = i + (if c.to_int() > 0xFFFF { 2 } else { 1 })
}
}
///|
/// Serialize inline content
fn serialize_inlines(inlines : Array[Inline], buf : StringBuilder) -> Unit {
serialize_inlines_with_autolink(inlines, buf, true)
}
///|
/// Serialize children while preventing nested links inside link labels.
fn serialize_inlines_with_autolink(
inlines : Array[Inline],
buf : StringBuilder,
autolink : Bool,
) -> Unit {
for i = 0; i < inlines.length(); i = i + 1 {
let inline = inlines[i]
// A literal `!` immediately followed by a link node would otherwise be
// reparsed as an image. Inside one text node escaping `[` is sufficient,
// so this must be handled at the inline-node boundary.
match inline {
Inline::Text(content~, ..) if content.has_suffix("!") &&
i + 1 < inlines.length() &&
inline_starts_link(inlines[i + 1]) => {
let prefix = content.unsafe_substring(start=0, end=content.length() - 1)
if autolink {
write_escaped_markdown_text(prefix, buf)
} else {
write_escaped_markdown_literal(prefix, buf)
}
buf.write_string("\\!")
}
_ => serialize_inline(inline, buf, autolink~)
}
}
}
///|
/// Whether serializing this inline starts with a Markdown link opener.
fn inline_starts_link(inline : Inline) -> Bool {
match inline {
Inline::Link(..) | Inline::RefLink(..) => true
_ => false
}
}
///|
/// Serialize table cell inline content (escapes pipes)
///
/// Every pipe a cell's content produces has to leave the cell as `\|`, and
/// that includes the ones nested inside code spans, emphasis and links: GFM
/// strips those escapes before the inline parser runs, so an unescaped pipe
/// would split the row instead of staying in the cell. The inlines are
/// rendered into a scratch buffer first so the rule applies uniformly to
/// every inline kind rather than only to top-level text.
fn serialize_table_cell_inlines(
inlines : Array[Inline],
buf : StringBuilder,
) -> Unit {
for inline in inlines {
let rendered = StringBuilder()
serialize_inline(inline, rendered)
write_table_escaped_pipes(rendered.to_string(), buf)
}
}
///|
/// Escape every pipe that is not already protected by an odd backslash run.
fn write_table_escaped_pipes(source : String, buf : StringBuilder) -> Unit {
let mut backslashes = 0
for c in source {
if c == '|' && backslashes % 2 == 0 {
buf.write_char('\\')
}
buf.write_char(c)
if c == '\\' {
backslashes = backslashes + 1
} else {
backslashes = 0
}
}
}
///|
/// Escape characters that would terminate or split a wiki link.
fn serialize_wikilink_part(value : String, buf : StringBuilder) -> Unit {
for c in value {
match c {
'\\' | '|' | ']' => {
buf.write_char('\\')
buf.write_char(c)
}
_ => buf.write_char(c)
}
}
}
///|
fn serialize_wikilink_destination(target : String, fragment : String) -> String {
if fragment.is_empty() {
target
} else {
target + "#" + fragment
}
}
///| Image labels serialize as plain text: nested emphasis, links, and images
///|
/// contribute their text but not their Markdown markers.
fn normalized_image_alt(alt : String) -> String {
let buf = StringBuilder()
write_plain_text(parse_inlines(alt), buf)
buf.to_string()
}
///|
/// Escape the delimiters that would terminate or nest an image label.
fn write_image_alt(alt : String, buf : StringBuilder) -> Unit {
for c in alt {
match c {
'\\' | '[' | ']' => {
buf.write_char('\\')
buf.write_char(c)
}
_ => buf.write_char(c)
}
}
}
///|
/// Serialize a single inline element
fn serialize_inline(
inline : Inline,
buf : StringBuilder,
autolink? : Bool = true,
) -> Unit {
match inline {
Inline::Text(content~, ..) =>
if autolink {
write_escaped_markdown_text(content, buf)
} else {
write_escaped_markdown_literal(content, buf)
}
Inline::SoftBreak(..) => buf.write_char('\n')
Inline::HardBreak(..) =>
// remark uses backslash style by default
buf.write_string("\\\n")
Inline::Emphasis(children~, ..) => {
// Always use * for GFM compatibility (remark default)
buf.write_char('*')
serialize_inlines_with_autolink(children, buf, autolink)
buf.write_char('*')
}
Inline::Strong(children~, ..) => {
// Always use ** for GFM compatibility (remark default)
buf.write_string("**")
serialize_inlines_with_autolink(children, buf, autolink)
buf.write_string("**")
}
Inline::Strikethrough(children~, ..) => {
buf.write_string("~~")
serialize_inlines_with_autolink(children, buf, autolink)
buf.write_string("~~")
}
Inline::Code(content=raw, ..) => {
// The CST keeps the source text, padding included; drop it and let the
// padding rules below re-add whatever this content needs.
let content = strip_code_span_padding(raw)
// Calculate minimum backticks needed (must not match any run in content)
let backticks = calc_code_span_backticks(content)
write_chars(buf, '`', backticks)
// Add padding space if content starts/ends with backtick
// (Spaces at both ends need padding only if NOT all spaces, to prevent trimming)
let needs_padding = content.length() > 0 &&
({
let first = content.get_char(0)
let last = content.get_char(content.length() - 1)
first == Some('`') || last == Some('`')
})
if needs_padding {
buf.write_char(' ')
}
buf.write_string(content)
if needs_padding {
buf.write_char(' ')
}
write_chars(buf, '`', backticks)
}
Inline::Directive(name~, label~, attributes~, ..) => {
buf.write_char(':')
buf.write_string(name)
buf.write_char('[')
for c in label {
if c == ']' || c == '\\' {
buf.write_char('\\')
}
buf.write_char(c)
}
buf.write_char(']')
if !attributes.is_empty() {
serialize_attribute_list(attributes, buf)
}
}
Inline::WikiLink(target~, label~, fragment~, ..) => {
let destination = serialize_wikilink_destination(target, fragment)
buf.write_string("[[")
serialize_wikilink_part(destination, buf)
if !label.is_empty() {
buf.write_char('|')
serialize_wikilink_part(label, buf)
}
buf.write_string("]]")
}
Inline::Link(children~, url~, title~, ..) => {
buf.write_char('[')
serialize_inlines_with_autolink(children, buf, false)
buf.write_string("](")
write_link_destination(url, buf)
if !title.is_empty() {
write_link_title(title, buf)
}
buf.write_char(')')
}
Inline::RefLink(children~, label~, style~, ..) => {
buf.write_char('[')
serialize_inlines_with_autolink(children, buf, false)
match style {
ReferenceStyle::Full => {
buf.write_string("][")
buf.write_string(label)
buf.write_char(']')
}
ReferenceStyle::Collapsed => buf.write_string("][]")
ReferenceStyle::Shortcut => buf.write_char(']')
}
}
Inline::Autolink(url~, ..) => {
buf.write_char('<')
buf.write_string(url)
buf.write_char('>')
}
Inline::Image(alt~, url~, title~, ..) => {
buf.write_string("
write_link_destination(url, buf)
if !title.is_empty() {
write_link_title(title, buf)
}
buf.write_char(')')
}
Inline::RefImage(alt~, label~, style~, ..) => {
let display = normalized_image_alt(alt)
buf.write_string("![")
write_image_alt(display, buf)
let output_style = if style != ReferenceStyle::Full && display != label {
ReferenceStyle::Full
} else {
style
}
match output_style {
ReferenceStyle::Full => {
buf.write_string("][")
buf.write_string(label)
buf.write_char(']')
}
ReferenceStyle::Collapsed => buf.write_string("][]")
ReferenceStyle::Shortcut => buf.write_char(']')
}
}
Inline::HtmlInline(html~, ..) => buf.write_string(html)
Inline::FootnoteReference(label~, ..) => {
buf.write_string("[^")
buf.write_string(label)
buf.write_char(']')
}
}
}