feat: support multiple attributes and `as` directive in inline blocks

Update the inline block parser to support arbitrary key-value pairs.
If an `as` attribute is provided, the block is replaced with a
directive (e.g., `:directive[path]{attrs}`) instead of being
stripped. If no `as` attribute is present, the block is removed.
Additionally, filenames are now automatically generated using a
content hash if the `file` attribute is missing.
This commit is contained in:
hypercross 2026-07-08 16:02:19 +08:00
parent 2a9eee6410
commit 749b1a6a4b
1 changed files with 64 additions and 32 deletions

View File

@ -1,68 +1,100 @@
import { posix } from "path"; import { posix } from "path";
import { createHash } from "crypto";
/** /**
* Regex to match a fenced code block with a `file="..."` attribute, e.g.: * Regex to match a fenced code block with attributes, e.g.:
* *
* ```csv file="card-data.csv" * ```csv file="card-data.csv"
* name, cost, effect * name, cost, effect
* Fireball, 1, Blast em * Fireball, 1, Blast em
* ``` * ```
* *
* ```yaml file="monsters.yaml" * ```csv as="md-table" roll=true
* goblin: * name, cost, effect
* hp: 7 * Fireball, 1, Blast em
* ``` * ```
* *
* The language identifier is optional only `file="..."` matters. * Group 1: language identifier (e.g. "csv")
* * Group 2: attribute string (e.g. `file="card-data.csv" as="md-table" roll=true`)
* Group 1: the info string after the opening fence (e.g. `csv file="card-data.csv"`) * Group 3: the code block body
* Group 2: the code block body
*/ */
const INLINE_FILE_BLOCK_RE = /^```\w*\s+file="([^"]+)"(?:\s+\w+="[^"]*")*\s*\n([\s\S]*?)```\s*$/gm; const INLINE_FILE_BLOCK_RE = /^```(\w*)\s+(.+?)\s*\n([\s\S]*?)```\s*$/gm;
/** /**
* Process fenced code blocks that contain a `file="..."` attribute. * Parse key="value" and key=value pairs from an attribute string.
*/
function parseAttrs(info: string): Record<string, string> {
const attrs: Record<string, string> = {};
const re = /(\w+)\s*=\s*("[^"]*"|\S+)/g;
let m: RegExpExecArray | null;
while ((m = re.exec(info)) !== null) {
attrs[m[1]] = m[2].replace(/^"|"$/g, "");
}
return attrs;
}
/**
* Short content hash for stable generated filenames.
*/
function contentHash(body: string): string {
return createHash("md5").update(body).digest("hex").slice(0, 8);
}
/**
* Process fenced code blocks that contain attributes.
* *
* For each block like: * For each block like:
* ```lang file="card-data.csv" * ```lang file="card-data.csv" as="md-table" roll=true
* content * content
* ``` * ```
* *
* 1. Strips the block from the returned text (so it doesn't render) * 1. Strips the block from the returned text
* 2. Injects a synthetic entry into `index` at the resolved file path * 2. Injects a synthetic entry into `index` at the resolved file path
* 3. If `as` is present, replaces the block with a directive like
* `:md-table[./file.csv]{roll=true}`
* *
* @param content - Raw file content * @param content - Raw file content
* @param fileRelativePath - Normalized path of the containing file (e.g. "/rules/combat.md") * @param fileRelativePath - Normalized path of the containing file (e.g. "/rules/combat.md")
* @param index - The content index to inject synthetic entries into * @param index - The content index to inject synthetic entries into
* @returns Content with inline blocks removed * @returns Content with inline blocks processed
*/ */
export function extractInlineBlocks( export function extractInlineBlocks(
content: string, content: string,
fileRelativePath: string, fileRelativePath: string,
index: Record<string, string>, index: Record<string, string>,
): string { ): string {
const matches: { filename: string; body: string }[] = []; const fileDir = posix.dirname(fileRelativePath);
let match: RegExpExecArray | null;
INLINE_FILE_BLOCK_RE.lastIndex = 0; return content.replace(
INLINE_FILE_BLOCK_RE,
(_match: string, lang: string, infoString: string, body: string) => {
const attrs = parseAttrs(infoString);
while ((match = INLINE_FILE_BLOCK_RE.exec(content)) !== null) { // Determine filename: explicit `file` or content-hash generated
matches.push({ filename: match[1], body: match[2] }); const filename =
} attrs.file || `_inline_${contentHash(body)}.${lang || "txt"}`;
const resolvedPath = posix.join(fileDir, filename);
if (matches.length === 0) return content; // Inject the body as a synthetic file entry
index[resolvedPath] = body;
console.log(
`[inline] Extracted ${resolvedPath} from ${fileRelativePath}`,
);
const stripped = content.replace(INLINE_FILE_BLOCK_RE, ""); // If `as` is set, emit a directive instead of stripping
if (attrs.as) {
const extra = { ...attrs };
delete extra.file;
delete extra.as;
const extraStr = Object.keys(extra).length
? `{${Object.entries(extra)
.map(([k, v]) => `${k}=${v}`)
.join(" ")}}`
: "";
return `:${attrs.as}[./${filename}]${extraStr}`;
}
for (const { filename, body } of matches) { return ""; // no `as` — strip the block entirely
const fileDir = posix.dirname(fileRelativePath); },
const resolvedPath = posix.join(fileDir, filename); );
index[resolvedPath] = body;
console.log(
`[inline] Extracted ${resolvedPath} from ${fileRelativePath}`,
);
}
return stripped;
} }