@bendyline/squisq 2.0.1 → 2.2.0

package/src/markdown/attrTokens.ts

@@ -1,243 +0,0 @@
-/**
- * Shared key=value attribute machinery for the two heading annotation forms:
- * the squisq-native `{[templateName key=value …]}` template annotation and
- * the Pandoc-style `{#id .class key=value}` attribute block.
- *
- * Both forms share one interior grammar — whitespace-separated tokens whose
- * values may be quoted (double or single) to include spaces, brackets, or
- * quote characters. This module is the single implementation of that
- * grammar: one tokenizer, one value un/escaper, one quoting rule. The
- * trailing-annotation recognizers live here too so the editor bridge
- * (packages/editor-react/src/tiptapBridge.ts) stays in sync by importing
- * them rather than copying regexes.
- *
- * Value syntax:
- * - Unquoted: no whitespace, quotes, or `[ ] { }` (those would collide with
- *   the surrounding annotation delimiters), e.g. `columns=3`, `name=O'Brien`
- *   (a quote *inside* an unquoted value is literal).
- * - Quoted: `key="value with spaces"` or `key='…'`. Inside a quoted run,
- *   `\` escapes the active quote character and itself; any other `\` is
- *   literal. Quoted values may contain `]` and `}`.
- * - Serialization emits double quotes, switching to single quotes when the
- *   value contains a double quote (markdown character-escape resolution
- *   makes backslash-escaped quotes fragile in `.md` source, so picking the
- *   other quote character is preferred over escaping). A value containing
- *   BOTH quote characters falls back to double quotes with `\"` escapes —
- *   that survives the API path but is best avoided in markdown source.
- */
-
-// ============================================
-// Trailing-annotation recognizers
-// ============================================
-
-/** Regex source for a double-quoted run with `\"` / `\\` escapes. */
-const DQ_RUN = `"(?:[^"\\\\]|\\\\.)*"`;
-/** Regex source for a single-quoted run with `\'` / `\\` escapes. */
-const SQ_RUN = `'(?:[^'\\\\]|\\\\.)*'`;
-
-/**
- * Quote-aware match for a trailing `{[…]}` annotation: the interior is any
- * mix of quoted runs and characters other than `]` and quotes, so quoted
- * values may contain `]`. An empty interior is valid so `{[]}` can remain
- * lossless metadata instead of visible heading text. The trailing
- * `[\s\]}]*` tolerates an accidental doubled `]}` that users sometimes type
- * when learning the syntax.
- */
-const TEMPLATE_STRICT_RE = new RegExp(
-  `\\s*\\{\\[((?:[^\\]"']|${DQ_RUN}|${SQ_RUN})*)\\]\\}[\\s\\]}]*$`,
-);
-
-/**
- * Legacy fallback matching the pre-quote-aware grammar (`[^\]]*` interior).
- * Tried after the strict form so inputs with unbalanced quotes keep parsing
- * exactly as they did before quote support landed.
- */
-const TEMPLATE_LEGACY_RE = /\s*\{\[([^\]]*)\]\}[\s\]}]*$/;
-
-/**
- * Quote-aware match for a trailing Pandoc `{…}` block. The negative
- * lookahead `(?!\[)` keeps it from colliding with the `{[…]}` template
- * form. Quoted values may contain `}`.
- */
-const PANDOC_STRICT_RE = new RegExp(`\\s*\\{(?!\\[)((?:[^}"']|${DQ_RUN}|${SQ_RUN})*)\\}\\s*$`);
-
-/** Legacy fallback for the Pandoc block (`[^}]*` interior). */
-const PANDOC_LEGACY_RE = /\s*\{(?!\[)([^}]*)\}\s*$/;
-
-/** Result of matching a trailing annotation against a text run. */
-export interface TrailingAnnotationMatch {
-  /** Interior of the brace block (between `{[`/`{` and `]}`/`}`), untrimmed. */
-  inner: string;
-  /** Offset of the match start (including leading whitespace) in the input. */
-  index: number;
-}
-
-function matchTrailing(
-  text: string,
-  strict: RegExp,
-  legacy: RegExp,
-): TrailingAnnotationMatch | null {
-  const m = strict.exec(text) ?? legacy.exec(text);
-  return m ? { inner: m[1], index: m.index } : null;
-}
-
-/**
- * Match a trailing `{[templateName key=value …]}` annotation. Quote-aware
- * (values may contain `]` when quoted), with a legacy fallback so inputs
- * the old `[^\]]+` grammar accepted still match.
- */
-export function matchTrailingTemplateAnnotation(text: string): TrailingAnnotationMatch | null {
-  return matchTrailing(text, TEMPLATE_STRICT_RE, TEMPLATE_LEGACY_RE);
-}
-
-/**
- * Match a trailing Pandoc `{#id .class key=value}` block. Quote-aware
- * (values may contain `}` when quoted), with a legacy fallback.
- */
-export function matchTrailingPandocAttr(text: string): TrailingAnnotationMatch | null {
-  return matchTrailing(text, PANDOC_STRICT_RE, PANDOC_LEGACY_RE);
-}
-
-// ============================================
-// Tokenizer
-// ============================================
-
-const WHITESPACE_RE = /\s/;
-
-/**
- * Quote-aware whitespace tokenizer for the interior of either annotation
- * form.
- *
- * A quote run starts only at the beginning of a token or immediately after
- * `=` — a quote elsewhere (e.g. `name=O'Brien`) is a literal character.
- * Inside a run, `\` escapes the active quote character and itself; the
- * escape pair is preserved in the returned token (callers strip and
- * unescape via {@link splitKeyValueToken} / {@link unquoteAttrValue}).
- * An unbalanced opening quote is tolerated: the run extends to the end of
- * the input.
- */
-export function tokenizeAttrTokens(input: string): string[] {
-  const tokens: string[] = [];
-  let i = 0;
-  const n = input.length;
-  while (i < n) {
-    while (i < n && WHITESPACE_RE.test(input[i])) i++;
-    if (i >= n) break;
-
-    let token = '';
-    while (i < n && !WHITESPACE_RE.test(input[i])) {
-      const ch = input[i];
-      const atValueStart = token === '' || token.endsWith('=');
-      if ((ch === '"' || ch === "'") && atValueStart) {
-        token += ch;
-        i++;
-        while (i < n && input[i] !== ch) {
-          if (input[i] === '\\' && i + 1 < n && (input[i + 1] === ch || input[i + 1] === '\\')) {
-            token += input[i] + input[i + 1];
-            i += 2;
-          } else {
-            token += input[i];
-            i++;
-          }
-        }
-        if (i < n && input[i] === ch) {
-          token += ch;
-          i++;
-        }
-      } else {
-        token += ch;
-        i++;
-      }
-    }
-    if (token) tokens.push(token);
-  }
-  return tokens;
-}
-
-// ============================================
-// Value un/escaping
-// ============================================
-
-const DQ_UNESCAPE_RE = /\\([\\"])/g;
-const SQ_UNESCAPE_RE = /\\([\\'])/g;
-
-/**
- * Strip surrounding quotes from a raw token value and unescape the active
- * quote character and backslash. Values not wrapped in matching quotes are
- * returned verbatim (including any interior quotes or backslashes).
- */
-export function unquoteAttrValue(raw: string): string {
-  if (raw.length >= 2) {
-    const q = raw[0];
-    if ((q === '"' || q === "'") && raw.endsWith(q)) {
-      const body = raw.slice(1, -1);
-      return body.replace(q === '"' ? DQ_UNESCAPE_RE : SQ_UNESCAPE_RE, '$1');
-    }
-  }
-  return raw;
-}
-
-/**
- * Split a `key=value` token into its parts, unquoting/unescaping the value.
- * Returns null for tokens with no `=` past the first character (bare words,
- * `#id`, `.class`, or a leading `=`).
- */
-export function splitKeyValueToken(token: string): { key: string; value: string } | null {
-  const eqIdx = token.indexOf('=');
-  if (eqIdx <= 0) return null;
-  return { key: token.slice(0, eqIdx), value: unquoteAttrValue(token.slice(eqIdx + 1)) };
-}
-
-// ============================================
-// Serialization
-// ============================================
-
-const NEEDS_QUOTING_RE = /[\s"'[\]{}]/;
-const DQ_ESCAPE_RE = /[\\"]/g;
-const SQ_ESCAPE_RE = /[\\']/g;
-
-/**
- * Whether a value must be quoted to survive tokenization and the outer
- * annotation regexes: any whitespace, quote character, bracket, or brace —
- * or an empty value, which would otherwise serialize as a bare `key=`.
- */
-export function needsQuoting(value: string): boolean {
-  return value === '' || NEEDS_QUOTING_RE.test(value);
-}
-
-/**
- * Serialize an attribute value, quoting when {@link needsQuoting} says the
- * bare form would not round-trip.
- *
- * Double quotes are canonical; a value containing a double quote (but no
- * single quote) is single-quoted instead, because markdown resolves
- * backslash escapes in text runs before our parser sees them — switching
- * the quote character round-trips through `.md` source where `\"` would
- * not. Backslashes and the active quote character are escaped.
- */
-export function quoteAttrValue(value: string): string {
-  if (!needsQuoting(value)) return value;
-  if (value.includes('"') && !value.includes("'")) {
-    return `'${value.replace(SQ_ESCAPE_RE, (ch) => `\\${ch}`)}'`;
-  }
-  return `"${value.replace(DQ_ESCAPE_RE, (ch) => `\\${ch}`)}"`;
-}
-
-/**
- * Serialize a `{[templateName key=value …]}` template annotation. The inverse
- * of {@link parseStandaloneAnnotation} / the tokenizer: emits the template
- * name (when present) followed by each `key=value` pair in insertion order,
- * quoting values via {@link quoteAttrValue}. Used to re-emit heading-less
- * standalone annotation blocks in `docToMarkdown`.
- */
-export function serializeAnnotation(
-  template: string | undefined,
-  params?: Record<string, string>,
-): string {
-  const parts: string[] = [];
-  if (template) parts.push(template);
-  for (const [key, value] of Object.entries(params ?? {})) {
-    parts.push(`${key}=${quoteAttrValue(value)}`);
-  }
-  return `{[${parts.join(' ')}]}`;
-}