@oh-my-pi/pi-coding-agent 14.0.3 → 14.0.5

package/src/utils/lang-from-path.ts

@@ -0,0 +1,239 @@
+import * as path from "node:path";
+
+/**
+ * Extension segment → [highlight language id, LSP language id].
+ * Highlight ids match tree-sitter / native highlighter; LSP ids match Language Server Protocol.
+ */
+const EXTENSION_LANG: Record<string, readonly [string, string]> = {
+	// TypeScript / JavaScript
+	ts: ["typescript", "typescript"],
+	cts: ["typescript", "typescript"],
+	mts: ["typescript", "typescript"],
+	tsx: ["tsx", "typescriptreact"],
+	js: ["javascript", "javascript"],
+	jsx: ["javascript", "javascriptreact"],
+	mjs: ["javascript", "javascript"],
+	cjs: ["javascript", "javascript"],
+
+	// Systems
+	rs: ["rust", "rust"],
+	go: ["go", "go"],
+	c: ["c", "c"],
+	h: ["c", "c"],
+	cpp: ["cpp", "cpp"],
+	cc: ["cpp", "cpp"],
+	cxx: ["cpp", "cpp"],
+	hh: ["cpp", "cpp"],
+	hpp: ["cpp", "cpp"],
+	hxx: ["cpp", "cpp"],
+	cu: ["cpp", "cpp"],
+	ino: ["cpp", "cpp"],
+	zig: ["zig", "zig"],
+
+	// Scripting
+	py: ["python", "python"],
+	pyi: ["python", "python"],
+	rb: ["ruby", "ruby"],
+	rbw: ["ruby", "ruby"],
+	gemspec: ["ruby", "ruby"],
+	lua: ["lua", "lua"],
+	sh: ["bash", "shellscript"],
+	bash: ["bash", "shellscript"],
+	zsh: ["bash", "shellscript"],
+	ksh: ["bash", "shellscript"],
+	bats: ["bash", "shellscript"],
+	tmux: ["bash", "shellscript"],
+	cgi: ["bash", "shellscript"],
+	fcgi: ["bash", "shellscript"],
+	command: ["bash", "shellscript"],
+	tool: ["bash", "shellscript"],
+	fish: ["fish", "fish"],
+	pl: ["perl", "perl"],
+	pm: ["perl", "perl"],
+	perl: ["perl", "perl"],
+	php: ["php", "php"],
+
+	// JVM
+	java: ["java", "java"],
+	kt: ["kotlin", "kotlin"],
+	ktm: ["kotlin", "kotlin"],
+	kts: ["kotlin", "kotlin"],
+	scala: ["scala", "scala"],
+	sc: ["scala", "scala"],
+	sbt: ["scala", "scala"],
+	groovy: ["groovy", "groovy"],
+	clj: ["clojure", "clojure"],
+	cljc: ["clojure", "clojure"],
+	cljs: ["clojure", "clojure"],
+	edn: ["clojure", "clojure"],
+
+	// .NET
+	cs: ["csharp", "csharp"],
+	fs: ["fsharp", "fsharp"],
+	vb: ["vb", "vb"],
+
+	// Web
+	html: ["html", "html"],
+	htm: ["html", "html"],
+	xhtml: ["html", "html"],
+	css: ["css", "css"],
+	scss: ["scss", "scss"],
+	sass: ["sass", "sass"],
+	less: ["less", "less"],
+	vue: ["vue", "vue"],
+	svelte: ["svelte", "svelte"],
+	astro: ["astro", "astro"],
+
+	// Data
+	json: ["json", "json"],
+	jsonc: ["jsonc", "jsonc"],
+	yaml: ["yaml", "yaml"],
+	yml: ["yaml", "yaml"],
+	toml: ["toml", "toml"],
+	xml: ["xml", "xml"],
+	xsl: ["xml", "xml"],
+	xslt: ["xml", "xml"],
+	svg: ["xml", "xml"],
+	plist: ["xml", "xml"],
+	ini: ["ini", "ini"],
+
+	// Docs
+	md: ["markdown", "markdown"],
+	markdown: ["markdown", "markdown"],
+	mdx: ["markdown", "markdown"],
+	rst: ["restructuredtext", "restructuredtext"],
+	adoc: ["asciidoc", "asciidoc"],
+	tex: ["latex", "latex"],
+
+	// Other languages
+	sql: ["sql", "sql"],
+	graphql: ["graphql", "graphql"],
+	gql: ["graphql", "graphql"],
+	proto: ["protobuf", "protobuf"],
+	dockerfile: ["dockerfile", "dockerfile"],
+	containerfile: ["dockerfile", "dockerfile"],
+	tf: ["hcl", "terraform"],
+	hcl: ["hcl", "hcl"],
+	tfvars: ["hcl", "hcl"],
+	nix: ["nix", "nix"],
+	ex: ["elixir", "elixir"],
+	exs: ["elixir", "elixir"],
+	erl: ["erlang", "erlang"],
+	hrl: ["erlang", "erlang"],
+	hs: ["haskell", "haskell"],
+	ml: ["ocaml", "ocaml"],
+	mli: ["ocaml", "ocaml"],
+	swift: ["swift", "swift"],
+	r: ["r", "r"],
+	jl: ["julia", "julia"],
+	dart: ["dart", "dart"],
+	elm: ["elm", "elm"],
+	v: ["verilog", "v"],
+	nim: ["nim", "nim"],
+	cr: ["crystal", "crystal"],
+	d: ["d", "d"],
+	pas: ["pascal", "pascal"],
+	pp: ["pascal", "pascal"],
+	lisp: ["lisp", "lisp"],
+	lsp: ["lisp", "lisp"],
+	rkt: ["racket", "racket"],
+	scm: ["scheme", "scheme"],
+	ps1: ["powershell", "powershell"],
+	psm1: ["powershell", "powershell"],
+	bat: ["bat", "bat"],
+	cmd: ["bat", "bat"],
+	tla: ["tlaplus", "tlaplus"],
+	tlaplus: ["tlaplus", "tlaplus"],
+	m: ["objc", "plaintext"],
+	mm: ["objc", "plaintext"],
+	sol: ["solidity", "plaintext"],
+	odin: ["odin", "plaintext"],
+	star: ["starlark", "plaintext"],
+	bzl: ["starlark", "plaintext"],
+	sv: ["verilog", "plaintext"],
+	svh: ["verilog", "plaintext"],
+	vh: ["verilog", "plaintext"],
+	vim: ["vim", "plaintext"],
+	ipynb: ["ipynb", "plaintext"],
+	hbs: ["handlebars", "plaintext"],
+	hsb: ["handlebars", "plaintext"],
+	handlebars: ["handlebars", "plaintext"],
+	diff: ["diff", "plaintext"],
+	patch: ["diff", "plaintext"],
+	makefile: ["make", "plaintext"],
+	mk: ["make", "plaintext"],
+	mak: ["make", "plaintext"],
+	cmake: ["cmake", "cmake"],
+	justfile: ["just", "plaintext"],
+	txt: ["text", "plaintext"],
+	text: ["text", "plaintext"],
+	log: ["log", "plaintext"],
+	csv: ["csv", "plaintext"],
+	tsv: ["tsv", "plaintext"],
+	cfg: ["conf", "plaintext"],
+	conf: ["conf", "plaintext"],
+	config: ["conf", "plaintext"],
+	properties: ["conf", "plaintext"],
+	env: ["env", "plaintext"],
+	gitignore: ["conf", "plaintext"],
+	gitattributes: ["conf", "plaintext"],
+	gitmodules: ["conf", "plaintext"],
+	editorconfig: ["conf", "plaintext"],
+	npmrc: ["conf", "plaintext"],
+	prettierrc: ["conf", "plaintext"],
+	eslintrc: ["conf", "plaintext"],
+	prettierignore: ["conf", "plaintext"],
+	eslintignore: ["conf", "plaintext"],
+};
+
+/** Final segment after the last `.` in the full path (prior theme behavior). */
+function themeExtensionKey(filePath: string): string {
+	const extBeg = filePath.lastIndexOf(".");
+	return extBeg !== -1 ? filePath.slice(extBeg + 1).toLowerCase() : filePath.toLowerCase();
+}
+
+function lspExtensionKey(filePath: string): string {
+	const ext = path.extname(filePath).toLowerCase();
+	return ext.startsWith(".") ? ext.slice(1) : "";
+}
+
+/**
+ * Language id for syntax highlighting and UI (icons, read tool), or undefined if unknown.
+ */
+export function getLanguageFromPath(filePath: string): string | undefined {
+	const pair = EXTENSION_LANG[themeExtensionKey(filePath)];
+	if (pair) return pair[0];
+
+	const baseName = path.basename(filePath).toLowerCase();
+	if (baseName.startsWith(".env.")) return "env";
+	if (baseName === "dockerfile" || baseName.startsWith("dockerfile.") || baseName === "containerfile") {
+		return "dockerfile";
+	}
+	if (baseName === "justfile") return "just";
+	if (baseName === "cmakelists.txt") return "cmake";
+
+	return undefined;
+}
+
+/**
+ * LSP language identifier; falls back to `plaintext`.
+ */
+export function detectLanguageId(filePath: string): string {
+	const baseName = path.basename(filePath).toLowerCase();
+	if (baseName === "dockerfile" || baseName.startsWith("dockerfile.") || baseName === "containerfile") {
+		return "dockerfile";
+	}
+	if (baseName === "makefile" || baseName === "gnumakefile") {
+		return "makefile";
+	}
+	if (baseName === "justfile") {
+		return "just";
+	}
+
+	const lspExt = lspExtensionKey(filePath);
+	if (baseName === "cmakelists.txt" || lspExt === "cmake") {
+		return "cmake";
+	}
+
+	return EXTENSION_LANG[lspExt]?.[1] ?? "plaintext";
+}

package/src/utils/sixel.ts

@@ -1,4 +1,4 @@
-import { $env } from "@oh-my-pi/pi-utils";
+import { $env, $flag } from "@oh-my-pi/pi-utils";
 
 const SIXEL_START_REGEX = /\x1bP(?:[0-9;]*)q/u;
 const SIXEL_END_SEQUENCE = "\x1b\\";
@@ -15,7 +15,7 @@ const SIXEL_PLACEHOLDER_PREFIX = "__OMP_SIXEL_SEQUENCE_";
  */
 export function isSixelPassthroughEnabled(): boolean {
 	const forcedProtocol = $env.PI_FORCE_IMAGE_PROTOCOL?.trim().toLowerCase();
-	return forcedProtocol === "sixel" && $env.PI_ALLOW_SIXEL_PASSTHROUGH === "1";
+	return forcedProtocol === "sixel" && $flag("PI_ALLOW_SIXEL_PASSTHROUGH");
 }
 /** Returns true when the text contains a SIXEL start sequence. */
 export function containsSixelSequence(text: string): boolean {

package/src/web/scrapers/types.ts

@@ -2,6 +2,8 @@
  * Shared types and utilities for web-fetch handlers
  */
 import { ptree } from "@oh-my-pi/pi-utils";
+import TurndownService from "turndown";
+import { gfm } from "turndown-plugin-gfm";
 import { ToolAbortError } from "../../tools/tool-errors";
 
 export { formatNumber } from "@oh-my-pi/pi-utils";
@@ -153,41 +155,57 @@ export async function loadPage(url: string, options: LoadPageOptions = {}): Prom
 	return { content: "", contentType: "", finalUrl: url, ok: false };
 }
 
+/** Module-level Turndown instance — matches markit-ai's configuration. */
+const turndown = new TurndownService({
+	headingStyle: "atx",
+	codeBlockStyle: "fenced",
+	bulletListMarker: "-",
+});
+turndown.use(gfm);
+turndown.addRule("strikethrough", {
+	filter: ["del", "s", "strike"],
+	replacement(content) {
+		return `~~${content}~~`;
+	},
+});
+turndown.addRule("heading", {
+	filter: ["h1", "h2", "h3", "h4", "h5", "h6"],
+	replacement(content, node) {
+		const level = Number(node.nodeName.charAt(1));
+		const prefix = "#".repeat(level);
+		const cleaned = content.replace(/\\([.])/g, "$1").trim();
+		return `\n\n${prefix} ${cleaned}\n\n`;
+	},
+});
+
+type TurndownListParent = {
+	nodeName: string;
+	getAttribute(name: string): string | null;
+	children: ArrayLike<unknown>;
+};
+
+turndown.addRule("listItem", {
+	filter: "li",
+	replacement(content, node, options) {
+		content = content.replace(/^\n+/, "").replace(/\n+$/, "\n").replace(/\n/gm, "\n  ");
+		const parent = node.parentNode as unknown as TurndownListParent | null;
+		let prefix = `${options.bulletListMarker} `;
+		if (parent?.nodeName === "OL") {
+			const start = parent.getAttribute("start");
+			const index = Array.prototype.indexOf.call(parent.children, node);
+			prefix = `${(start ? Number(start) : 1) + index}. `;
+		}
+		return prefix + content + (node.nextSibling ? "\n" : "");
+	},
+});
+
 /**
- * Convert basic HTML to markdown
+ * Convert HTML to markdown using Turndown with GFM support.
+ * Strips script/style tags before conversion.
  */
 export function htmlToBasicMarkdown(html: string): string {
-	const stripped = html
-		.replace(/<pre[^>]*><code[^>]*>/g, "\n```\n")
-		.replace(/<\/code><\/pre>/g, "\n```\n")
-		.replace(/<code[^>]*>/g, "`")
-		.replace(/<\/code>/g, "`")
-		.replace(/<strong[^>]*>/g, "**")
-		.replace(/<\/strong>/g, "**")
-		.replace(/<b[^>]*>/g, "**")
-		.replace(/<\/b>/g, "**")
-		.replace(/<em[^>]*>/g, "*")
-		.replace(/<\/em>/g, "*")
-		.replace(/<i[^>]*>/g, "*")
-		.replace(/<\/i>/g, "*")
-		.replace(
-			/<a[^>]*href="([^"]+)"[^>]*>([\s\S]*?)<\/a>/g,
-			(_, href, text) => `[${text.replace(/<[^>]+>/g, "").trim()}](${href})`,
-		)
-		.replace(/<p[^>]*>/g, "\n\n")
-		.replace(/<\/p>/g, "")
-		.replace(/<br\s*\/?>/g, "\n")
-		.replace(/<li[^>]*>/g, "- ")
-		.replace(/<\/li>/g, "\n")
-		.replace(/<\/?[uo]l[^>]*>/g, "\n")
-		.replace(/<h(\d)[^>]*>/g, (_, n) => `\n${"#".repeat(parseInt(n, 10))} `)
-		.replace(/<\/h\d>/g, "\n")
-		.replace(/<blockquote[^>]*>/g, "\n> ")
-		.replace(/<\/blockquote>/g, "\n")
-		.replace(/<[^>]+>/g, "")
-		.replace(/\n{3,}/g, "\n\n")
-		.trim();
-	return decodeHtmlEntities(stripped);
+	const cleaned = html.replace(/<script[\s\S]*?<\/script>/gi, "").replace(/<style[\s\S]*?<\/style>/gi, "");
+	return turndown.turndown(cleaned).trim();
 }
 
 /**

package/src/web/search/providers/codex.ts

@@ -90,6 +90,10 @@ interface CodexResponse {
 	usage?: CodexUsage;
 }
 
+function isImagePlaceholderAnswer(text: string): boolean {
+	return text.trim().toLowerCase() === "(see attached image)";
+}
+
 /**
  * Decodes a JWT token and extracts the payload.
  * @param token - JWT token string
@@ -232,6 +236,7 @@ async function callCodexSearch(
 
 	// Parse SSE stream
 	const answerParts: string[] = [];
+	const streamedAnswerParts: string[] = [];
 	const sources: SearchSource[] = [];
 	let model = requestedModel;
 	let requestId = "";
@@ -241,7 +246,12 @@ async function callCodexSearch(
 		const eventType = typeof rawEvent.type === "string" ? rawEvent.type : "";
 		if (!eventType) continue;
 
-		if (eventType === "response.output_item.done") {
+		if (eventType === "response.output_text.delta") {
+			const delta = typeof rawEvent.delta === "string" ? rawEvent.delta : "";
+			if (delta) {
+				streamedAnswerParts.push(delta);
+			}
+		} else if (eventType === "response.output_item.done") {
 			const item = rawEvent.item as CodexResponseItem | undefined;
 			if (!item) continue;
 
@@ -302,8 +312,17 @@ async function callCodexSearch(
 		}
 	}
 
+	const finalAnswer = answerParts.join("\n\n").trim();
+	const streamedAnswer = streamedAnswerParts.join("").trim();
+	const answer =
+		finalAnswer.length > 0 && !isImagePlaceholderAnswer(finalAnswer)
+			? finalAnswer
+			: streamedAnswer.length > 0
+				? streamedAnswer
+				: finalAnswer;
+
 	return {
-		answer: answerParts.join("\n\n"),
+		answer,
 		sources,
 		model,
 		requestId,