| 1 | import type { Root, Text } from "mdast"; |
| 2 | import type { InlineMath } from "mdast-util-math"; |
| 3 | import { visit } from "unist-util-visit"; |
| 4 | import { classifyInlineMath } from "./mathClassify"; |
| 5 | import { latexNormalizeForKatex } from "./latexNormalize"; |
| 6 | import { |
| 7 | resolveProtectedInlineMathSource, |
| 8 | restoreProtectedInlineMathSource, |
| 9 | } from "./mathNormalize"; |
| 10 | |
| 11 | type HastLikeNode = { |
| 12 | type: string; |
| 13 | value?: string; |
| 14 | children?: HastLikeNode[]; |
| 15 | }; |
| 16 | |
| 17 | type MathNodeData = { |
| 18 | hChildren?: HastLikeNode[]; |
| 19 | hProperties?: Record<string, unknown>; |
| 20 | }; |
| 21 | |
| 22 | type ParentWithChildren = { |
| 23 | children: Array<{ type: string; value?: unknown; position?: unknown }>; |
| 24 | }; |
| 25 | |
| 26 | type VFileLike = { |
| 27 | value: unknown; |
| 28 | }; |
| 29 | |
| 30 | function siblingText(parent: ParentWithChildren, index: number, offset: -1 | 1): string { |
| 31 | const sibling = parent.children[index + offset]; |
| 32 | if (sibling?.type !== "text" || typeof sibling.value !== "string") return ""; |
| 33 | return offset < 0 ? sibling.value.slice(-120) : sibling.value.slice(0, 120); |
| 34 | } |
| 35 | |
| 36 | function stripAdjacentMarkdown(text: string, side: -1 | 1): string { |
| 37 | let current = text; |
| 38 | while (true) { |
| 39 | const next = side < 0 |
| 40 | ? current.replace(/(?:[*_~]{1,3}|\[)\s*$/, "") |
| 41 | : current.replace(/^\s*(?:[*_~]{1,3}|\])/, ""); |
| 42 | if (next === current) return current; |
| 43 | current = next; |
| 44 | } |
| 45 | } |
| 46 | |
| 47 | function inlineMathContext( |
| 48 | node: InlineMath, |
| 49 | file: VFileLike, |
| 50 | parent: ParentWithChildren, |
| 51 | index: number, |
| 52 | ): { before: string; after: string } { |
| 53 | const source = String(file.value ?? ""); |
| 54 | const start = node.position?.start.offset; |
| 55 | const end = node.position?.end.offset; |
| 56 | if (typeof start === "number" && typeof end === "number") { |
| 57 | return { |
| 58 | before: stripAdjacentMarkdown(source.slice(Math.max(0, start - 120), start), -1), |
| 59 | after: stripAdjacentMarkdown(source.slice(end, end + 120), 1), |
| 60 | }; |
| 61 | } |
| 62 | return { |
| 63 | before: siblingText(parent, index, -1), |
| 64 | after: siblingText(parent, index, 1), |
| 65 | }; |
| 66 | } |
| 67 | |
| 68 | function originalSource(node: InlineMath, file: VFileLike): string { |
| 69 | const source = String(file.value ?? ""); |
| 70 | const start = node.position?.start.offset; |
| 71 | const end = node.position?.end.offset; |
| 72 | if (typeof start === "number" && typeof end === "number") { |
| 73 | const span = source.slice(start, end); |
| 74 | if (span.startsWith("$") && span.endsWith("$")) { |
| 75 | return `$${restoreProtectedInlineMathSource(span.slice(1, -1))}$`; |
| 76 | } |
| 77 | return restoreProtectedInlineMathSource(span); |
| 78 | } |
| 79 | return `$${restoreProtectedInlineMathSource(node.value)}$`; |
| 80 | } |
| 81 | |
| 82 | function inlineMathSources(node: InlineMath, file: VFileLike): { |
| 83 | source: string; |
| 84 | rendered: string; |
| 85 | } { |
| 86 | const parsed = resolveProtectedInlineMathSource(node.value); |
| 87 | const fileSource = String(file.value ?? ""); |
| 88 | const start = node.position?.start.offset; |
| 89 | const end = node.position?.end.offset; |
| 90 | if (typeof start !== "number" || typeof end !== "number") return parsed; |
| 91 | |
| 92 | const span = fileSource.slice(start, end); |
| 93 | if (!span.startsWith("$") || !span.endsWith("$")) return parsed; |
| 94 | return { |
| 95 | source: resolveProtectedInlineMathSource(span.slice(1, -1)).source, |
| 96 | rendered: parsed.rendered, |
| 97 | }; |
| 98 | } |
| 99 | |
| 100 | function setMathValue( |
| 101 | node: { value: string; data?: unknown }, |
| 102 | source: string, |
| 103 | value: string, |
| 104 | ): void { |
| 105 | node.value = value; |
| 106 | |
| 107 | // mdast-util-math caches hast children while parsing. Keep that rendering |
| 108 | // payload in sync with node.value when a later remark plugin normalises it, |
| 109 | // while carrying the unnormalised TeX through to the rehype stage. |
| 110 | const data = node.data as MathNodeData | undefined; |
| 111 | if (data) { |
| 112 | data.hProperties ??= {}; |
| 113 | data.hProperties.dataLatexSource = source; |
| 114 | } |
| 115 | const hChildren = data?.hChildren; |
| 116 | const updateFirstText = (children: HastLikeNode[] | undefined): boolean => { |
| 117 | if (!children) return false; |
| 118 | for (const child of children) { |
| 119 | if (child.type === "text") { |
| 120 | child.value = value; |
| 121 | return true; |
| 122 | } |
| 123 | if (updateFirstText(child.children)) return true; |
| 124 | } |
| 125 | return false; |
| 126 | }; |
| 127 | updateFirstText(hChildren); |
| 128 | } |
| 129 | |
| 130 | /** |
| 131 | * Semantic policy layered after remark-math has parsed Markdown boundaries. |
| 132 | * Code spans/fences never become math nodes, so this plugin only decides how |
| 133 | * already-parsed inline math should render. |
| 134 | */ |
| 135 | export function remarkMathPolicy() { |
| 136 | return (tree: Root, file: VFileLike) => { |
| 137 | visit(tree, "inlineMath", (node, index, parent) => { |
| 138 | if (typeof index !== "number" || !parent) return; |
| 139 | |
| 140 | const typedNode = node as InlineMath; |
| 141 | const typedParent = parent as ParentWithChildren; |
| 142 | const { source, rendered } = inlineMathSources(typedNode, file); |
| 143 | const classificationSource = source.trim(); |
| 144 | const decision = classifyInlineMath( |
| 145 | classificationSource, |
| 146 | inlineMathContext(typedNode, file, typedParent, index), |
| 147 | ); |
| 148 | |
| 149 | if (decision === "math") { |
| 150 | setMathValue(typedNode, source, latexNormalizeForKatex(rendered)); |
| 151 | return; |
| 152 | } |
| 153 | |
| 154 | const value = decision === "currency" |
| 155 | ? `$${classificationSource}` |
| 156 | : originalSource(typedNode, file); |
| 157 | typedParent.children[index] = { type: "text", value } satisfies Text; |
| 158 | }); |
| 159 | |
| 160 | visit(tree, "math", (node) => { |
| 161 | const { source, rendered } = resolveProtectedInlineMathSource(node.value); |
| 162 | setMathValue(node, source, latexNormalizeForKatex(rendered)); |
| 163 | }); |
| 164 | }; |
| 165 | } |
| 166 |