@kolisachint/hoocode-agent 0.4.92 → 0.4.94

package/dist/core/tools/docgrep.js

@@ -0,0 +1,106 @@
+import { Text } from "@kolisachint/hoocode-tui";
+import { Type } from "typebox";
+import { keyHint } from "../../modes/interactive/components/keybinding-hints.js";
+import { theme as appTheme } from "../../modes/interactive/theme/theme.js";
+import { grepDocument, truncateRenderToTokenBudget } from "./filetools-shared.js";
+import { resolveReadPath } from "./path-utils.js";
+import { getTextOutput, invalidArgText, shortenPath, str } from "./render-utils.js";
+import { wrapToolDefinition } from "./tool-definition-wrapper.js";
+const docGrepSchema = Type.Object({
+    path: Type.String({
+        description: "Path to the document to search (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.",
+    }),
+    pattern: Type.String({
+        description: "Literal substring to match per line (NOT a regex).",
+    }),
+    ignoreCase: Type.Optional(Type.Boolean({ description: "Case-insensitive matching. Default false." })),
+    limit: Type.Optional(Type.Number({ description: "Stop after N matches. Default: all matches." })),
+});
+/**
+ * Render grep matches as `#block_id:line :: "snippet"` lines. Each block id
+ * feeds straight back into DocPeek (to hydrate) or a DocEdit patch.
+ */
+export function renderGrepView(view) {
+    const header = `grep ${JSON.stringify(view.pattern)} — ${view.returned} match${view.returned === 1 ? "" : "es"}`;
+    if (view.matches.length === 0)
+        return header;
+    const lines = [header, ""];
+    for (const m of view.matches) {
+        const flag = m.writable ? "" : " (read-only)";
+        lines.push(`#${m.block_id}:${m.line}${flag} :: ${JSON.stringify(m.snippet)}`);
+    }
+    const { text, droppedLines } = truncateRenderToTokenBudget(lines);
+    if (droppedLines === 0)
+        return text;
+    return (`${text}\n\n[Truncated: ${droppedLines} more match line${droppedLines === 1 ? "" : "s"} omitted; ` +
+        `re-run with a smaller limit or a more specific pattern.]`);
+}
+function formatDocGrepCall(args) {
+    const path = str(args?.path);
+    const pathDisplay = path === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg("toolOutput", "...");
+    let text = appTheme.fg("toolTitle", appTheme.bold("DocGrep "));
+    const pattern = str(args?.pattern);
+    if (pattern)
+        text += `${appTheme.fg("accent", JSON.stringify(pattern))} ${appTheme.fg("muted", "in")} `;
+    text += appTheme.fg("accent", pathDisplay);
+    if (args?.ignoreCase)
+        text += appTheme.fg("muted", " (i)");
+    return text;
+}
+function formatDocGrepResult(result, options, showImages) {
+    const output = getTextOutput(result, showImages).trim();
+    if (!output)
+        return "";
+    const lines = output.split("\n");
+    const maxLines = options.expanded ? lines.length : 15;
+    const displayLines = lines.slice(0, maxLines);
+    const remaining = lines.length - maxLines;
+    let text = `\n${displayLines.map((line) => appTheme.fg("toolOutput", line)).join("\n")}`;
+    if (remaining > 0) {
+        text += `${appTheme.fg("muted", `\n... (${remaining} more lines,`)} ${keyHint("app.tools.expand", "to expand")})`;
+    }
+    return text;
+}
+export function createDocGrepToolDefinition(cwd, options) {
+    return {
+        name: "DocGrep",
+        label: "DocGrep",
+        description: "Locate blocks in a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) by literal text, WITHOUT hydrating the whole document. Returns each match as an editable el_ node #id, line number, and snippet. Those el_ #ids target a DocEdit/DocWrite patch directly — DocGrep is the fast path from 'find text' to 'edit it' (far cheaper than a full DocRead). Off by default; enabled with --enable-filetools.",
+        promptSnippet: "Find blocks in a structured/binary document by literal text, without hydrating it",
+        promptGuidelines: [
+            "Use DocGrep to find where something is in a large structured/binary document instead of reading it all: it returns editable el_ node #ids + snippets for a literal substring match. Patch those el_ #ids straight with DocEdit/DocWrite.",
+            "DocGrep's el_ #ids are the edit id-space; they are NOT DocPeek's structural-path ids. To read more context around a match, DocScan/DocPeek by structural path instead.",
+            "pattern is a literal substring (not a regex); pass ignoreCase for case-insensitive matching.",
+            "Works across all supported formats including spreadsheet cell values (xlsx) and slide text (pptx); patch a matched el_ #id with DocEdit, or read more context with DocPeek.",
+        ],
+        parameters: docGrepSchema,
+        async execute(_toolCallId, { path, pattern, ignoreCase, limit }, signal) {
+            if (signal?.aborted)
+                throw new Error("Operation aborted");
+            const absolutePath = resolveReadPath(path, cwd);
+            const view = await grepDocument(absolutePath, pattern, cwd, signal, {
+                ignoreCase,
+                limit,
+                timeoutSecs: options?.timeoutSecs,
+            });
+            return {
+                content: [{ type: "text", text: renderGrepView(view) }],
+                details: { pattern: view.pattern, matches: view.returned },
+            };
+        },
+        renderCall(args, _theme, context) {
+            const text = context.lastComponent ?? new Text("", 0, 0);
+            text.setText(formatDocGrepCall(args));
+            return text;
+        },
+        renderResult(result, options, _theme, context) {
+            const text = context.lastComponent ?? new Text("", 0, 0);
+            text.setText(formatDocGrepResult(result, options, context.showImages));
+            return text;
+        },
+    };
+}
+export function createDocGrepTool(cwd, options) {
+    return wrapToolDefinition(createDocGrepToolDefinition(cwd, options));
+}
+//# sourceMappingURL=docgrep.js.map

package/dist/core/tools/docgrep.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"docgrep.js","sourceRoot":"","sources":["../../../src/core/tools/docgrep.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AAChD,OAAO,EAAe,IAAI,EAAE,MAAM,SAAS,CAAC;AAC5C,OAAO,EAAE,OAAO,EAAE,MAAM,wDAAwD,CAAC;AACjF,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,wCAAwC,CAAC;AAE3E,OAAO,EAAiB,YAAY,EAAE,2BAA2B,EAAE,MAAM,uBAAuB,CAAC;AACjG,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,WAAW,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AACpF,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAElE,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC;IACjC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC;QACjB,WAAW,EACV,qGAAqG;KACtG,CAAC;IACF,OAAO,EAAE,IAAI,CAAC,MAAM,CAAC;QACpB,WAAW,EAAE,oDAAoD;KACjE,CAAC;IACF,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,WAAW,EAAE,2CAA2C,EAAE,CAAC,CAAC;IACrG,KAAK,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,WAAW,EAAE,6CAA6C,EAAE,CAAC,CAAC;CACjG,CAAC,CAAC;AAcH;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,IAAc,EAAU;IACtD,MAAM,MAAM,GAAG,QAAQ,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,QAAM,IAAI,CAAC,QAAQ,SAAS,IAAI,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IACjH,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,MAAM,CAAC;IAC7C,MAAM,KAAK,GAAa,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;IACrC,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;QAC9B,MAAM,IAAI,GAAG,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC;QAC9C,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,IAAI,GAAG,IAAI,OAAO,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;IAC/E,CAAC;IACD,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,2BAA2B,CAAC,KAAK,CAAC,CAAC;IAClE,IAAI,YAAY,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IACpC,OAAO,CACN,GAAG,IAAI,mBAAmB,YAAY,mBAAmB,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,YAAY;QAClG,0DAA0D,CAC1D,CAAC;AAAA,CACF;AAED,SAAS,iBAAiB,CAAC,IAA2E,EAAU;IAC/G,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC7B,MAAM,WAAW,GAChB,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;IACxG,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,WAAW,EAAE,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC;IAC/D,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACnC,IAAI,OAAO;QAAE,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC;IACxG,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;IAC3C,IAAI,IAAI,EAAE,UAAU;QAAE,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;IAC3D,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,SAAS,mBAAmB,CAC3B,MAA2D,EAC3D,OAAgC,EAChC,UAAmB,EACV;IACT,MAAM,MAAM,GAAG,aAAa,CAAC,MAAa,EAAE,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC;IAC/D,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,CAAC;IACvB,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACjC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;IACtD,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,GAAG,QAAQ,CAAC;IAC1C,IAAI,IAAI,GAAG,KAAK,YAAY,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACzF,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QACnB,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,SAAS,cAAc,CAAC,IAAI,OAAO,CAAC,kBAAkB,EAAE,WAAW,CAAC,GAAG,CAAC;IACnH,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,MAAM,UAAU,2BAA2B,CAC1C,GAAW,EACX,OAA4B,EAC2C;IACvE,OAAO;QACN,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;QAChB,WAAW,EACV,0ZAAwZ;QACzZ,aAAa,EAAE,mFAAmF;QAClG,gBAAgB,EAAE;YACjB,0OAA0O;YAC1O,wKAAwK;YACxK,8FAA8F;YAC9F,6KAA6K;SAC7K;QACD,UAAU,EAAE,aAAa;QACzB,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAoB,EAAE,MAAoB,EAAE;YACxG,IAAI,MAAM,EAAE,OAAO;gBAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;YAC1D,MAAM,YAAY,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAChD,MAAM,IAAI,GAAG,MAAM,YAAY,CAAC,YAAY,EAAE,OAAO,EAAE,GAAG,EAAE,MAAM,EAAE;gBACnE,UAAU;gBACV,KAAK;gBACL,WAAW,EAAE,OAAO,EAAE,WAAW;aACjC,CAAC,CAAC;YACH,OAAO;gBACN,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAe,EAAE,IAAI,EAAE,cAAc,CAAC,IAAI,CAAC,EAAE,CAAC;gBAChE,OAAO,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,QAAQ,EAAE;aAC1D,CAAC;QAAA,CACF;QACD,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE;YACjC,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;YACtC,OAAO,IAAI,CAAC;QAAA,CACZ;QACD,YAAY,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE;YAC9C,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,mBAAmB,CAAC,MAAa,EAAE,OAAO,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;YAC9E,OAAO,IAAI,CAAC;QAAA,CACZ;KACD,CAAC;AAAA,CACF;AAED,MAAM,UAAU,iBAAiB,CAAC,GAAW,EAAE,OAA4B,EAAmC;IAC7G,OAAO,kBAAkB,CAAC,2BAA2B,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC;AAAA,CACrE","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport { type GrepView, grepDocument, truncateRenderToTokenBudget } from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docGrepSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription:\n\t\t\t\"Path to the document to search (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\tpattern: Type.String({\n\t\tdescription: \"Literal substring to match per line (NOT a regex).\",\n\t}),\n\tignoreCase: Type.Optional(Type.Boolean({ description: \"Case-insensitive matching. Default false.\" })),\n\tlimit: Type.Optional(Type.Number({ description: \"Stop after N matches. Default: all matches.\" })),\n});\n\nexport type DocGrepToolInput = Static<typeof docGrepSchema>;\n\nexport interface DocGrepToolDetails {\n\tpattern?: string;\n\tmatches?: number;\n}\n\nexport interface DocGrepToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\n/**\n * Render grep matches as `#block_id:line :: \"snippet\"` lines. Each block id\n * feeds straight back into DocPeek (to hydrate) or a DocEdit patch.\n */\nexport function renderGrepView(view: GrepView): string {\n\tconst header = `grep ${JSON.stringify(view.pattern)} — ${view.returned} match${view.returned === 1 ? \"\" : \"es\"}`;\n\tif (view.matches.length === 0) return header;\n\tconst lines: string[] = [header, \"\"];\n\tfor (const m of view.matches) {\n\t\tconst flag = m.writable ? \"\" : \" (read-only)\";\n\t\tlines.push(`#${m.block_id}:${m.line}${flag} :: ${JSON.stringify(m.snippet)}`);\n\t}\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tif (droppedLines === 0) return text;\n\treturn (\n\t\t`${text}\\n\\n[Truncated: ${droppedLines} more match line${droppedLines === 1 ? \"\" : \"s\"} omitted; ` +\n\t\t`re-run with a smaller limit or a more specific pattern.]`\n\t);\n}\n\nfunction formatDocGrepCall(args: { path?: string; pattern?: string; ignoreCase?: boolean } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocGrep \"));\n\tconst pattern = str(args?.pattern);\n\tif (pattern) text += `${appTheme.fg(\"accent\", JSON.stringify(pattern))} ${appTheme.fg(\"muted\", \"in\")} `;\n\ttext += appTheme.fg(\"accent\", pathDisplay);\n\tif (args?.ignoreCase) text += appTheme.fg(\"muted\", \" (i)\");\n\treturn text;\n}\n\nfunction formatDocGrepResult(\n\tresult: { content: Array<{ type: string; text?: string }> },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocGrepToolDefinition(\n\tcwd: string,\n\toptions?: DocGrepToolOptions,\n): ToolDefinition<typeof docGrepSchema, DocGrepToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocGrep\",\n\t\tlabel: \"DocGrep\",\n\t\tdescription:\n\t\t\t\"Locate blocks in a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) by literal text, WITHOUT hydrating the whole document. Returns each match as an editable el_ node #id, line number, and snippet. Those el_ #ids target a DocEdit/DocWrite patch directly — DocGrep is the fast path from 'find text' to 'edit it' (far cheaper than a full DocRead). Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Find blocks in a structured/binary document by literal text, without hydrating it\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocGrep to find where something is in a large structured/binary document instead of reading it all: it returns editable el_ node #ids + snippets for a literal substring match. Patch those el_ #ids straight with DocEdit/DocWrite.\",\n\t\t\t\"DocGrep's el_ #ids are the edit id-space; they are NOT DocPeek's structural-path ids. To read more context around a match, DocScan/DocPeek by structural path instead.\",\n\t\t\t\"pattern is a literal substring (not a regex); pass ignoreCase for case-insensitive matching.\",\n\t\t\t\"Works across all supported formats including spreadsheet cell values (xlsx) and slide text (pptx); patch a matched el_ #id with DocEdit, or read more context with DocPeek.\",\n\t\t],\n\t\tparameters: docGrepSchema,\n\t\tasync execute(_toolCallId, { path, pattern, ignoreCase, limit }: DocGrepToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst view = await grepDocument(absolutePath, pattern, cwd, signal, {\n\t\t\t\tignoreCase,\n\t\t\t\tlimit,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderGrepView(view) }],\n\t\t\t\tdetails: { pattern: view.pattern, matches: view.returned },\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocGrepCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocGrepResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocGrepTool(cwd: string, options?: DocGrepToolOptions): AgentTool<typeof docGrepSchema> {\n\treturn wrapToolDefinition(createDocGrepToolDefinition(cwd, options));\n}\n"]}

package/dist/core/tools/docpeek.d.ts

@@ -0,0 +1,25 @@
+import type { AgentTool } from "@kolisachint/hoocode-agent-core";
+import { type Static, Type } from "typebox";
+import type { ToolDefinition } from "../extensions/types.js";
+import { type ReadView } from "./filetools-shared.js";
+declare const docPeekSchema: Type.TObject<{
+    path: Type.TString;
+    id: Type.TOptional<Type.TArray<Type.TString>>;
+    offset: Type.TOptional<Type.TNumber>;
+    limit: Type.TOptional<Type.TNumber>;
+}>;
+export type DocPeekToolInput = Static<typeof docPeekSchema>;
+export interface DocPeekToolDetails {
+    returned?: number;
+    total?: number;
+}
+export interface DocPeekToolOptions {
+    /** Timeout (seconds) for the filetools invocation. */
+    timeoutSecs?: number;
+}
+/** Render hydrated read blocks using the same id-addressed node dialect as DocRead. */
+export declare function renderReadView(view: ReadView): string;
+export declare function createDocPeekToolDefinition(cwd: string, options?: DocPeekToolOptions): ToolDefinition<typeof docPeekSchema, DocPeekToolDetails | undefined>;
+export declare function createDocPeekTool(cwd: string, options?: DocPeekToolOptions): AgentTool<typeof docPeekSchema>;
+export {};
+//# sourceMappingURL=docpeek.d.ts.map

package/dist/core/tools/docpeek.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"docpeek.d.ts","sourceRoot":"","sources":["../../../src/core/tools/docpeek.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,iCAAiC,CAAC;AAEjE,OAAO,EAAE,KAAK,MAAM,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAG5C,OAAO,KAAK,EAAE,cAAc,EAA2B,MAAM,wBAAwB,CAAC;AACtF,OAAO,EACN,KAAK,QAAQ,EAIb,MAAM,uBAAuB,CAAC;AAK/B,QAAA,MAAM,aAAa;;;;;EAYjB,CAAC;AAEH,MAAM,MAAM,gBAAgB,GAAG,MAAM,CAAC,OAAO,aAAa,CAAC,CAAC;AAE5D,MAAM,WAAW,kBAAkB;IAClC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,KAAK,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,kBAAkB;IAClC,sDAAsD;IACtD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,uFAAuF;AACvF,wBAAgB,cAAc,CAAC,IAAI,EAAE,QAAQ,GAAG,MAAM,CAmBrD;AAuCD,wBAAgB,2BAA2B,CAC1C,GAAG,EAAE,MAAM,EACX,OAAO,CAAC,EAAE,kBAAkB,GAC1B,cAAc,CAAC,OAAO,aAAa,EAAE,kBAAkB,GAAG,SAAS,CAAC,CAuCtE;AAED,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,SAAS,CAAC,OAAO,aAAa,CAAC,CAE5G","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\ttype ReadView,\n\treadDocumentBlocks,\n\trenderDocNodeLines,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docPeekSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription: \"Path to the document to read (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\tid: Type.Optional(\n\t\tType.Array(Type.String(), {\n\t\t\tdescription:\n\t\t\t\t\"Block id(s) to hydrate — use the EXACT ids returned by a prior DocScan (e.g. 'node[title:0]', 'paragraph[3]', 'page[1]'); do NOT hand-construct sub-ranges (an invented 'rows[0-2]' returns nothing — pass the exact 'rows[0-99]' DocScan emitted). These are DocScan's path ids, not DocGrep's el_ node ids. Omit to read a paginated slice of all blocks.\",\n\t\t}),\n\t),\n\toffset: Type.Optional(Type.Number({ description: \"Pagination start when no ids are given. Default 0.\" })),\n\tlimit: Type.Optional(Type.Number({ description: \"Pagination size when no ids are given. Default: all remaining.\" })),\n});\n\nexport type DocPeekToolInput = Static<typeof docPeekSchema>;\n\nexport interface DocPeekToolDetails {\n\treturned?: number;\n\ttotal?: number;\n}\n\nexport interface DocPeekToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\n/** Render hydrated read blocks using the same id-addressed node dialect as DocRead. */\nexport function renderReadView(view: ReadView): string {\n\tconst header = `document — ${view.returned}/${view.total} blocks (offset ${view.offset})`;\n\tconst lines: string[] = [header, \"\", ...renderDocNodeLines(view.nodes)];\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tconst seen = view.offset + view.returned;\n\tconst remaining = view.total - seen;\n\tconst parts = [text];\n\tif (droppedLines > 0) {\n\t\tparts.push(\n\t\t\t`\\n[Truncated: ${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted; ` +\n\t\t\t\t`request fewer ids or a smaller limit.]`,\n\t\t);\n\t}\n\tif (remaining > 0) {\n\t\tparts.push(\n\t\t\t`\\n[${remaining} more block${remaining === 1 ? \"\" : \"s\"} not shown — re-run with offset:${seen} to continue.]`,\n\t\t);\n\t}\n\treturn parts.join(\"\\n\");\n}\n\nfunction formatDocPeekCall(\n\targs: { path?: string; id?: string[]; offset?: number; limit?: number } | undefined,\n): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocPeek \")) + appTheme.fg(\"accent\", pathDisplay);\n\tconst ids = Array.isArray(args?.id) ? args.id.length : 0;\n\tif (ids > 0) {\n\t\ttext += appTheme.fg(\"muted\", ` (${ids} id${ids === 1 ? \"\" : \"s\"})`);\n\t} else {\n\t\tconst range: string[] = [];\n\t\tif (args?.offset !== undefined) range.push(`offset ${args.offset}`);\n\t\tif (args?.limit !== undefined) range.push(`limit ${args.limit}`);\n\t\tif (range.length > 0) text += appTheme.fg(\"muted\", ` (${range.join(\", \")})`);\n\t}\n\treturn text;\n}\n\nfunction formatDocPeekResult(\n\tresult: { content: Array<{ type: string; text?: string }> },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocPeekToolDefinition(\n\tcwd: string,\n\toptions?: DocPeekToolOptions,\n): ToolDefinition<typeof docPeekSchema, DocPeekToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocPeek\",\n\t\tlabel: \"DocPeek\",\n\t\tdescription:\n\t\t\t\"Hydrate specific blocks of a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) by structural-path id — the read half of the token-sensitive loop. Pass the path ids from a prior DocScan (e.g. node[title:0]) to read only those blocks (or omit ids to page through all with offset/limit), instead of dumping the whole document with DocRead. The hydrated nodes carry the editable el_ #ids, which you then patch with DocEdit/DocWrite. Read-only. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Read only specific blocks of a structured/binary document by id\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocPeek to read just the blocks you need (by structural-path id from DocScan, e.g. node[title:0]) instead of a full DocRead — it is much cheaper in tokens. Omit ids and use offset/limit to page through a large document.\",\n\t\t\t\"Pass the EXACT ids DocScan returned; DocPeek does not accept invented sub-ranges (e.g. rows[0-2] when DocScan emitted rows[0-99]).\",\n\t\t\t\"DocPeek takes DocScan's path ids, NOT DocGrep's el_ node ids. The nodes it returns carry the editable el_ #ids — patch those with DocEdit/DocWrite (which auto-extract). If you already have an el_ id from DocGrep, go straight to DocEdit; you do not need DocPeek.\",\n\t\t\t\"Works across all supported formats including spreadsheet cells (hydrating a sheet's rows[a-b] block returns the cell values) and slide text (pptx).\",\n\t\t],\n\t\tparameters: docPeekSchema,\n\t\tasync execute(_toolCallId, { path, id, offset, limit }: DocPeekToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst view = await readDocumentBlocks(absolutePath, cwd, signal, {\n\t\t\t\tids: id,\n\t\t\t\toffset,\n\t\t\t\tlimit,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderReadView(view) }],\n\t\t\t\tdetails: { returned: view.returned, total: view.total },\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocPeekCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocPeekResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocPeekTool(cwd: string, options?: DocPeekToolOptions): AgentTool<typeof docPeekSchema> {\n\treturn wrapToolDefinition(createDocPeekToolDefinition(cwd, options));\n}\n"]}

package/dist/core/tools/docpeek.js

@@ -0,0 +1,112 @@
+import { Text } from "@kolisachint/hoocode-tui";
+import { Type } from "typebox";
+import { keyHint } from "../../modes/interactive/components/keybinding-hints.js";
+import { theme as appTheme } from "../../modes/interactive/theme/theme.js";
+import { readDocumentBlocks, renderDocNodeLines, truncateRenderToTokenBudget, } from "./filetools-shared.js";
+import { resolveReadPath } from "./path-utils.js";
+import { getTextOutput, invalidArgText, shortenPath, str } from "./render-utils.js";
+import { wrapToolDefinition } from "./tool-definition-wrapper.js";
+const docPeekSchema = Type.Object({
+    path: Type.String({
+        description: "Path to the document to read (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.",
+    }),
+    id: Type.Optional(Type.Array(Type.String(), {
+        description: "Block id(s) to hydrate — use the EXACT ids returned by a prior DocScan (e.g. 'node[title:0]', 'paragraph[3]', 'page[1]'); do NOT hand-construct sub-ranges (an invented 'rows[0-2]' returns nothing — pass the exact 'rows[0-99]' DocScan emitted). These are DocScan's path ids, not DocGrep's el_ node ids. Omit to read a paginated slice of all blocks.",
+    })),
+    offset: Type.Optional(Type.Number({ description: "Pagination start when no ids are given. Default 0." })),
+    limit: Type.Optional(Type.Number({ description: "Pagination size when no ids are given. Default: all remaining." })),
+});
+/** Render hydrated read blocks using the same id-addressed node dialect as DocRead. */
+export function renderReadView(view) {
+    const header = `document — ${view.returned}/${view.total} blocks (offset ${view.offset})`;
+    const lines = [header, "", ...renderDocNodeLines(view.nodes)];
+    const { text, droppedLines } = truncateRenderToTokenBudget(lines);
+    const seen = view.offset + view.returned;
+    const remaining = view.total - seen;
+    const parts = [text];
+    if (droppedLines > 0) {
+        parts.push(`\n[Truncated: ${droppedLines} more node line${droppedLines === 1 ? "" : "s"} omitted; ` +
+            `request fewer ids or a smaller limit.]`);
+    }
+    if (remaining > 0) {
+        parts.push(`\n[${remaining} more block${remaining === 1 ? "" : "s"} not shown — re-run with offset:${seen} to continue.]`);
+    }
+    return parts.join("\n");
+}
+function formatDocPeekCall(args) {
+    const path = str(args?.path);
+    const pathDisplay = path === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg("toolOutput", "...");
+    let text = appTheme.fg("toolTitle", appTheme.bold("DocPeek ")) + appTheme.fg("accent", pathDisplay);
+    const ids = Array.isArray(args?.id) ? args.id.length : 0;
+    if (ids > 0) {
+        text += appTheme.fg("muted", ` (${ids} id${ids === 1 ? "" : "s"})`);
+    }
+    else {
+        const range = [];
+        if (args?.offset !== undefined)
+            range.push(`offset ${args.offset}`);
+        if (args?.limit !== undefined)
+            range.push(`limit ${args.limit}`);
+        if (range.length > 0)
+            text += appTheme.fg("muted", ` (${range.join(", ")})`);
+    }
+    return text;
+}
+function formatDocPeekResult(result, options, showImages) {
+    const output = getTextOutput(result, showImages).trim();
+    if (!output)
+        return "";
+    const lines = output.split("\n");
+    const maxLines = options.expanded ? lines.length : 15;
+    const displayLines = lines.slice(0, maxLines);
+    const remaining = lines.length - maxLines;
+    let text = `\n${displayLines.map((line) => appTheme.fg("toolOutput", line)).join("\n")}`;
+    if (remaining > 0) {
+        text += `${appTheme.fg("muted", `\n... (${remaining} more lines,`)} ${keyHint("app.tools.expand", "to expand")})`;
+    }
+    return text;
+}
+export function createDocPeekToolDefinition(cwd, options) {
+    return {
+        name: "DocPeek",
+        label: "DocPeek",
+        description: "Hydrate specific blocks of a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) by structural-path id — the read half of the token-sensitive loop. Pass the path ids from a prior DocScan (e.g. node[title:0]) to read only those blocks (or omit ids to page through all with offset/limit), instead of dumping the whole document with DocRead. The hydrated nodes carry the editable el_ #ids, which you then patch with DocEdit/DocWrite. Read-only. Off by default; enabled with --enable-filetools.",
+        promptSnippet: "Read only specific blocks of a structured/binary document by id",
+        promptGuidelines: [
+            "Use DocPeek to read just the blocks you need (by structural-path id from DocScan, e.g. node[title:0]) instead of a full DocRead — it is much cheaper in tokens. Omit ids and use offset/limit to page through a large document.",
+            "Pass the EXACT ids DocScan returned; DocPeek does not accept invented sub-ranges (e.g. rows[0-2] when DocScan emitted rows[0-99]).",
+            "DocPeek takes DocScan's path ids, NOT DocGrep's el_ node ids. The nodes it returns carry the editable el_ #ids — patch those with DocEdit/DocWrite (which auto-extract). If you already have an el_ id from DocGrep, go straight to DocEdit; you do not need DocPeek.",
+            "Works across all supported formats including spreadsheet cells (hydrating a sheet's rows[a-b] block returns the cell values) and slide text (pptx).",
+        ],
+        parameters: docPeekSchema,
+        async execute(_toolCallId, { path, id, offset, limit }, signal) {
+            if (signal?.aborted)
+                throw new Error("Operation aborted");
+            const absolutePath = resolveReadPath(path, cwd);
+            const view = await readDocumentBlocks(absolutePath, cwd, signal, {
+                ids: id,
+                offset,
+                limit,
+                timeoutSecs: options?.timeoutSecs,
+            });
+            return {
+                content: [{ type: "text", text: renderReadView(view) }],
+                details: { returned: view.returned, total: view.total },
+            };
+        },
+        renderCall(args, _theme, context) {
+            const text = context.lastComponent ?? new Text("", 0, 0);
+            text.setText(formatDocPeekCall(args));
+            return text;
+        },
+        renderResult(result, options, _theme, context) {
+            const text = context.lastComponent ?? new Text("", 0, 0);
+            text.setText(formatDocPeekResult(result, options, context.showImages));
+            return text;
+        },
+    };
+}
+export function createDocPeekTool(cwd, options) {
+    return wrapToolDefinition(createDocPeekToolDefinition(cwd, options));
+}
+//# sourceMappingURL=docpeek.js.map

package/dist/core/tools/docpeek.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"docpeek.js","sourceRoot":"","sources":["../../../src/core/tools/docpeek.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AAChD,OAAO,EAAe,IAAI,EAAE,MAAM,SAAS,CAAC;AAC5C,OAAO,EAAE,OAAO,EAAE,MAAM,wDAAwD,CAAC;AACjF,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,wCAAwC,CAAC;AAE3E,OAAO,EAEN,kBAAkB,EAClB,kBAAkB,EAClB,2BAA2B,GAC3B,MAAM,uBAAuB,CAAC;AAC/B,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,WAAW,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AACpF,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAElE,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC;IACjC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC;QACjB,WAAW,EAAE,mGAAmG;KAChH,CAAC;IACF,EAAE,EAAE,IAAI,CAAC,QAAQ,CAChB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE;QACzB,WAAW,EACV,iWAA6V;KAC9V,CAAC,CACF;IACD,MAAM,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,WAAW,EAAE,oDAAoD,EAAE,CAAC,CAAC;IACzG,KAAK,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,WAAW,EAAE,gEAAgE,EAAE,CAAC,CAAC;CACpH,CAAC,CAAC;AAcH,uFAAuF;AACvF,MAAM,UAAU,cAAc,CAAC,IAAc,EAAU;IACtD,MAAM,MAAM,GAAG,gBAAc,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,KAAK,mBAAmB,IAAI,CAAC,MAAM,GAAG,CAAC;IAC1F,MAAM,KAAK,GAAa,CAAC,MAAM,EAAE,EAAE,EAAE,GAAG,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;IACxE,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,2BAA2B,CAAC,KAAK,CAAC,CAAC;IAClE,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,QAAQ,CAAC;IACzC,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;IACpC,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC;IACrB,IAAI,YAAY,GAAG,CAAC,EAAE,CAAC;QACtB,KAAK,CAAC,IAAI,CACT,iBAAiB,YAAY,kBAAkB,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,YAAY;YACvF,wCAAwC,CACzC,CAAC;IACH,CAAC;IACD,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QACnB,KAAK,CAAC,IAAI,CACT,MAAM,SAAS,cAAc,SAAS,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,qCAAmC,IAAI,gBAAgB,CAC9G,CAAC;IACH,CAAC;IACD,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAAA,CACxB;AAED,SAAS,iBAAiB,CACzB,IAAmF,EAC1E;IACT,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC7B,MAAM,WAAW,GAChB,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;IACxG,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,WAAW,EAAE,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,GAAG,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;IACpG,MAAM,GAAG,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IACzD,IAAI,GAAG,GAAG,CAAC,EAAE,CAAC;QACb,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,KAAK,GAAG,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC;IACrE,CAAC;SAAM,CAAC;QACP,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,IAAI,EAAE,MAAM,KAAK,SAAS;YAAE,KAAK,CAAC,IAAI,CAAC,UAAU,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;QACpE,IAAI,IAAI,EAAE,KAAK,KAAK,SAAS;YAAE,KAAK,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC;QACjE,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,KAAK,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC9E,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,SAAS,mBAAmB,CAC3B,MAA2D,EAC3D,OAAgC,EAChC,UAAmB,EACV;IACT,MAAM,MAAM,GAAG,aAAa,CAAC,MAAa,EAAE,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC;IAC/D,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,CAAC;IACvB,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACjC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;IACtD,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,GAAG,QAAQ,CAAC;IAC1C,IAAI,IAAI,GAAG,KAAK,YAAY,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACzF,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QACnB,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,SAAS,cAAc,CAAC,IAAI,OAAO,CAAC,kBAAkB,EAAE,WAAW,CAAC,GAAG,CAAC;IACnH,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,MAAM,UAAU,2BAA2B,CAC1C,GAAW,EACX,OAA4B,EAC2C;IACvE,OAAO;QACN,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;QAChB,WAAW,EACV,wfAAsf;QACvf,aAAa,EAAE,iEAAiE;QAChF,gBAAgB,EAAE;YACjB,mOAAiO;YACjO,oIAAoI;YACpI,yQAAuQ;YACvQ,qJAAqJ;SACrJ;QACD,UAAU,EAAE,aAAa;QACzB,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAoB,EAAE,MAAoB,EAAE;YAC/F,IAAI,MAAM,EAAE,OAAO;gBAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;YAC1D,MAAM,YAAY,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAChD,MAAM,IAAI,GAAG,MAAM,kBAAkB,CAAC,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE;gBAChE,GAAG,EAAE,EAAE;gBACP,MAAM;gBACN,KAAK;gBACL,WAAW,EAAE,OAAO,EAAE,WAAW;aACjC,CAAC,CAAC;YACH,OAAO;gBACN,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAe,EAAE,IAAI,EAAE,cAAc,CAAC,IAAI,CAAC,EAAE,CAAC;gBAChE,OAAO,EAAE,EAAE,QAAQ,EAAE,IAAI,CAAC,QAAQ,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE;aACvD,CAAC;QAAA,CACF;QACD,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE;YACjC,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;YACtC,OAAO,IAAI,CAAC;QAAA,CACZ;QACD,YAAY,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE;YAC9C,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,mBAAmB,CAAC,MAAa,EAAE,OAAO,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;YAC9E,OAAO,IAAI,CAAC;QAAA,CACZ;KACD,CAAC;AAAA,CACF;AAED,MAAM,UAAU,iBAAiB,CAAC,GAAW,EAAE,OAA4B,EAAmC;IAC7G,OAAO,kBAAkB,CAAC,2BAA2B,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC;AAAA,CACrE","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\ttype ReadView,\n\treadDocumentBlocks,\n\trenderDocNodeLines,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docPeekSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription: \"Path to the document to read (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\tid: Type.Optional(\n\t\tType.Array(Type.String(), {\n\t\t\tdescription:\n\t\t\t\t\"Block id(s) to hydrate — use the EXACT ids returned by a prior DocScan (e.g. 'node[title:0]', 'paragraph[3]', 'page[1]'); do NOT hand-construct sub-ranges (an invented 'rows[0-2]' returns nothing — pass the exact 'rows[0-99]' DocScan emitted). These are DocScan's path ids, not DocGrep's el_ node ids. Omit to read a paginated slice of all blocks.\",\n\t\t}),\n\t),\n\toffset: Type.Optional(Type.Number({ description: \"Pagination start when no ids are given. Default 0.\" })),\n\tlimit: Type.Optional(Type.Number({ description: \"Pagination size when no ids are given. Default: all remaining.\" })),\n});\n\nexport type DocPeekToolInput = Static<typeof docPeekSchema>;\n\nexport interface DocPeekToolDetails {\n\treturned?: number;\n\ttotal?: number;\n}\n\nexport interface DocPeekToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\n/** Render hydrated read blocks using the same id-addressed node dialect as DocRead. */\nexport function renderReadView(view: ReadView): string {\n\tconst header = `document — ${view.returned}/${view.total} blocks (offset ${view.offset})`;\n\tconst lines: string[] = [header, \"\", ...renderDocNodeLines(view.nodes)];\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tconst seen = view.offset + view.returned;\n\tconst remaining = view.total - seen;\n\tconst parts = [text];\n\tif (droppedLines > 0) {\n\t\tparts.push(\n\t\t\t`\\n[Truncated: ${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted; ` +\n\t\t\t\t`request fewer ids or a smaller limit.]`,\n\t\t);\n\t}\n\tif (remaining > 0) {\n\t\tparts.push(\n\t\t\t`\\n[${remaining} more block${remaining === 1 ? \"\" : \"s\"} not shown — re-run with offset:${seen} to continue.]`,\n\t\t);\n\t}\n\treturn parts.join(\"\\n\");\n}\n\nfunction formatDocPeekCall(\n\targs: { path?: string; id?: string[]; offset?: number; limit?: number } | undefined,\n): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocPeek \")) + appTheme.fg(\"accent\", pathDisplay);\n\tconst ids = Array.isArray(args?.id) ? args.id.length : 0;\n\tif (ids > 0) {\n\t\ttext += appTheme.fg(\"muted\", ` (${ids} id${ids === 1 ? \"\" : \"s\"})`);\n\t} else {\n\t\tconst range: string[] = [];\n\t\tif (args?.offset !== undefined) range.push(`offset ${args.offset}`);\n\t\tif (args?.limit !== undefined) range.push(`limit ${args.limit}`);\n\t\tif (range.length > 0) text += appTheme.fg(\"muted\", ` (${range.join(\", \")})`);\n\t}\n\treturn text;\n}\n\nfunction formatDocPeekResult(\n\tresult: { content: Array<{ type: string; text?: string }> },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocPeekToolDefinition(\n\tcwd: string,\n\toptions?: DocPeekToolOptions,\n): ToolDefinition<typeof docPeekSchema, DocPeekToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocPeek\",\n\t\tlabel: \"DocPeek\",\n\t\tdescription:\n\t\t\t\"Hydrate specific blocks of a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) by structural-path id — the read half of the token-sensitive loop. Pass the path ids from a prior DocScan (e.g. node[title:0]) to read only those blocks (or omit ids to page through all with offset/limit), instead of dumping the whole document with DocRead. The hydrated nodes carry the editable el_ #ids, which you then patch with DocEdit/DocWrite. Read-only. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Read only specific blocks of a structured/binary document by id\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocPeek to read just the blocks you need (by structural-path id from DocScan, e.g. node[title:0]) instead of a full DocRead — it is much cheaper in tokens. Omit ids and use offset/limit to page through a large document.\",\n\t\t\t\"Pass the EXACT ids DocScan returned; DocPeek does not accept invented sub-ranges (e.g. rows[0-2] when DocScan emitted rows[0-99]).\",\n\t\t\t\"DocPeek takes DocScan's path ids, NOT DocGrep's el_ node ids. The nodes it returns carry the editable el_ #ids — patch those with DocEdit/DocWrite (which auto-extract). If you already have an el_ id from DocGrep, go straight to DocEdit; you do not need DocPeek.\",\n\t\t\t\"Works across all supported formats including spreadsheet cells (hydrating a sheet's rows[a-b] block returns the cell values) and slide text (pptx).\",\n\t\t],\n\t\tparameters: docPeekSchema,\n\t\tasync execute(_toolCallId, { path, id, offset, limit }: DocPeekToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst view = await readDocumentBlocks(absolutePath, cwd, signal, {\n\t\t\t\tids: id,\n\t\t\t\toffset,\n\t\t\t\tlimit,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderReadView(view) }],\n\t\t\t\tdetails: { returned: view.returned, total: view.total },\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocPeekCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocPeekResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocPeekTool(cwd: string, options?: DocPeekToolOptions): AgentTool<typeof docPeekSchema> {\n\treturn wrapToolDefinition(createDocPeekToolDefinition(cwd, options));\n}\n"]}

package/dist/core/tools/docread.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"docread.d.ts","sourceRoot":"","sources":["../../../src/core/tools/docread.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,iCAAiC,CAAC;AAEjE,OAAO,EAAE,KAAK,MAAM,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAG5C,OAAO,KAAK,EAAE,cAAc,EAA2B,MAAM,wBAAwB,CAAC;AACtF,OAAO,EAGN,KAAK,QAAQ,EAGb,MAAM,uBAAuB,CAAC;AAK/B,QAAA,MAAM,aAAa;;;EAWjB,CAAC;AAEH,MAAM,MAAM,gBAAgB,GAAG,MAAM,CAAC,OAAO,aAAa,CAAC,CAAC;AAE5D,MAAM,WAAW,kBAAkB;IAClC,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,SAAS,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,WAAW,kBAAkB;IAClC,sDAAsD;IACtD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAUD;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CAAC,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,GAAG,SAAS,GAAG,MAAM,CA6B5F;AA6BD,wBAAgB,2BAA2B,CAC1C,GAAG,EAAE,MAAM,EACX,OAAO,CAAC,EAAE,kBAAkB,GAC1B,cAAc,CAAC,OAAO,aAAa,EAAE,kBAAkB,GAAG,SAAS,CAAC,CAwCtE;AAED,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,SAAS,CAAC,OAAO,aAAa,CAAC,CAE5G","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\tDOCREAD_MAX_RENDER_TOKENS,\n\ttype DocNode,\n\ttype Envelope,\n\textractDocument,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docReadSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription:\n\t\t\t\"Path to the document to extract (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\treadonly: Type.Optional(\n\t\tType.Boolean({\n\t\t\tdescription:\n\t\t\t\t\"Analysis-only projection: strips node ids for a smaller view that CANNOT be edited. Omit (default false) when you intend to DocEdit/DocWrite afterwards.\",\n\t\t}),\n\t),\n});\n\nexport type DocReadToolInput = Static<typeof docReadSchema>;\n\nexport interface DocReadToolDetails {\n\ttype?: string;\n\tfidelity?: string;\n\twritable?: boolean;\n\tnodeCount?: number;\n}\n\nexport interface DocReadToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\nfunction countNodes(nodes: DocNode[]): number {\n\tlet total = 0;\n\tfor (const node of nodes) {\n\t\ttotal += 1 + (node.children ? countNodes(node.children) : 0);\n\t}\n\treturn total;\n}\n\n/**\n * Render the envelope as compact, id-addressed text the model edits against.\n * Each line carries the node id so DocEdit/DocWrite patches can target it.\n *\n * The filetools binary has no pagination, so a dense document can project into\n * a huge dump. We truncate the rendered view to a token budget and append an\n * actionable notice instead of flooding the model context.\n */\nexport function renderEnvelopeText(envelope: Envelope, readonly: boolean | undefined): string {\n\tconst header =\n\t\t`document ${envelope.source.path} [${envelope.source.type}, ${envelope.fidelity}, ` +\n\t\t`${envelope.writable ? \"writable\" : \"read-only\"}]`;\n\tconst lines: string[] = [header, \"\"];\n\n\tconst walk = (nodes: DocNode[], depth: number): void => {\n\t\tfor (const node of nodes) {\n\t\t\tconst indent = \"  \".repeat(depth);\n\t\t\tconst idPart = node.id ? `#${node.id} ` : \"\";\n\t\t\tconst attrs = node.attrs?.length ? ` ${node.attrs.map((a) => `${a.name}=\"${a.value}\"`).join(\" \")}` : \"\";\n\t\t\tconst text = node.text !== undefined ? ` :: ${JSON.stringify(node.text)}` : \"\";\n\t\t\tlines.push(`${indent}${idPart}<${node.tag}${attrs}>${text}`);\n\t\t\tif (node.children?.length) walk(node.children, depth + 1);\n\t\t}\n\t};\n\twalk(envelope.structure, 0);\n\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tif (droppedLines === 0) return text;\n\n\tconst hint = readonly\n\t\t? \"narrow to a smaller or more targeted file\"\n\t\t: \"re-run with readonly:true for a smaller analysis-only view, or target a smaller file\";\n\treturn (\n\t\t`${text}\\n\\n[Truncated: document exceeds the ~${DOCREAD_MAX_RENDER_TOKENS} token render budget; ` +\n\t\t`${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted. The ids shown are still valid for ` +\n\t\t`DocEdit/DocWrite — ${hint}.]`\n\t);\n}\n\nfunction formatDocReadCall(args: { path?: string; readonly?: boolean } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocRead \")) + appTheme.fg(\"accent\", pathDisplay);\n\tif (args?.readonly) text += appTheme.fg(\"muted\", \" (readonly)\");\n\treturn text;\n}\n\nfunction formatDocReadResult(\n\tresult: { content: Array<{ type: string; text?: string }>; details?: DocReadToolDetails },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocReadToolDefinition(\n\tcwd: string,\n\toptions?: DocReadToolOptions,\n): ToolDefinition<typeof docReadSchema, DocReadToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocRead\",\n\t\tlabel: \"DocRead\",\n\t\tdescription:\n\t\t\t\"Extract a structured or binary document (XML, drawio, docx/xlsx/pptx, PDF) into editable, id-addressed JSON the agent can patch losslessly. Each node has a stable #id; edit with DocEdit (in place) or DocWrite (to a new path), passing a patch that targets those ids. This is the canonical way to read and edit these formats: never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or rewrite them — that loses the lossless id-map and corrupts formatting. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Extract a structured/binary document into editable, id-addressed JSON\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocRead to open structured/binary documents (XML, drawio, docx/xlsx/pptx, PDF) instead of read; it returns id-addressed nodes you can patch with DocEdit/DocWrite. Never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or edit these formats — that loses the lossless id-map and corrupts the file.\",\n\t\t\t\"DocEdit/DocWrite require a prior DocRead of the same file (the id-map is established by the extract).\",\n\t\t],\n\t\tparameters: docReadSchema,\n\t\tasync execute(_toolCallId, { path, readonly }: DocReadToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst envelope = await extractDocument(absolutePath, cwd, signal, {\n\t\t\t\treadonly,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderEnvelopeText(envelope, readonly) }],\n\t\t\t\tdetails: {\n\t\t\t\t\ttype: envelope.source.type,\n\t\t\t\t\tfidelity: envelope.fidelity,\n\t\t\t\t\twritable: envelope.writable,\n\t\t\t\t\tnodeCount: countNodes(envelope.structure),\n\t\t\t\t},\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocReadTool(cwd: string, options?: DocReadToolOptions): AgentTool<typeof docReadSchema> {\n\treturn wrapToolDefinition(createDocReadToolDefinition(cwd, options));\n}\n"]}
+{"version":3,"file":"docread.d.ts","sourceRoot":"","sources":["../../../src/core/tools/docread.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,iCAAiC,CAAC;AAEjE,OAAO,EAAE,KAAK,MAAM,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAG5C,OAAO,KAAK,EAAE,cAAc,EAA2B,MAAM,wBAAwB,CAAC;AACtF,OAAO,EAGN,KAAK,QAAQ,EAIb,MAAM,uBAAuB,CAAC;AAK/B,QAAA,MAAM,aAAa;;;EAWjB,CAAC;AAEH,MAAM,MAAM,gBAAgB,GAAG,MAAM,CAAC,OAAO,aAAa,CAAC,CAAC;AAE5D,MAAM,WAAW,kBAAkB;IAClC,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,OAAO,CAAC;IACnB,SAAS,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,WAAW,kBAAkB;IAClC,sDAAsD;IACtD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAUD;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CAAC,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,GAAG,SAAS,GAAG,MAAM,CAiB5F;AA6BD,wBAAgB,2BAA2B,CAC1C,GAAG,EAAE,MAAM,EACX,OAAO,CAAC,EAAE,kBAAkB,GAC1B,cAAc,CAAC,OAAO,aAAa,EAAE,kBAAkB,GAAG,SAAS,CAAC,CAyCtE;AAED,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,SAAS,CAAC,OAAO,aAAa,CAAC,CAE5G","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\tDOCREAD_MAX_RENDER_TOKENS,\n\ttype DocNode,\n\ttype Envelope,\n\textractDocument,\n\trenderDocNodeLines,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docReadSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription:\n\t\t\t\"Path to the document to extract (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\treadonly: Type.Optional(\n\t\tType.Boolean({\n\t\t\tdescription:\n\t\t\t\t\"Analysis-only projection: strips node ids for a smaller view that CANNOT be edited. Omit (default false) when you intend to DocEdit/DocWrite afterwards.\",\n\t\t}),\n\t),\n});\n\nexport type DocReadToolInput = Static<typeof docReadSchema>;\n\nexport interface DocReadToolDetails {\n\ttype?: string;\n\tfidelity?: string;\n\twritable?: boolean;\n\tnodeCount?: number;\n}\n\nexport interface DocReadToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\nfunction countNodes(nodes: DocNode[]): number {\n\tlet total = 0;\n\tfor (const node of nodes) {\n\t\ttotal += 1 + (node.children ? countNodes(node.children) : 0);\n\t}\n\treturn total;\n}\n\n/**\n * Render the envelope as compact, id-addressed text the model edits against.\n * Each line carries the node id so DocEdit/DocWrite patches can target it.\n *\n * The filetools binary has no pagination, so a dense document can project into\n * a huge dump. We truncate the rendered view to a token budget and append an\n * actionable notice instead of flooding the model context.\n */\nexport function renderEnvelopeText(envelope: Envelope, readonly: boolean | undefined): string {\n\tconst header =\n\t\t`document ${envelope.source.path} [${envelope.source.type}, ${envelope.fidelity}, ` +\n\t\t`${envelope.writable ? \"writable\" : \"read-only\"}]`;\n\tconst lines: string[] = [header, \"\", ...renderDocNodeLines(envelope.structure)];\n\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tif (droppedLines === 0) return text;\n\n\tconst hint = readonly\n\t\t? \"narrow to a smaller or more targeted file\"\n\t\t: \"re-run with readonly:true for a smaller analysis-only view, or target a smaller file\";\n\treturn (\n\t\t`${text}\\n\\n[Truncated: document exceeds the ~${DOCREAD_MAX_RENDER_TOKENS} token render budget; ` +\n\t\t`${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted. The ids shown are still valid for ` +\n\t\t`DocEdit/DocWrite — ${hint}.]`\n\t);\n}\n\nfunction formatDocReadCall(args: { path?: string; readonly?: boolean } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocRead \")) + appTheme.fg(\"accent\", pathDisplay);\n\tif (args?.readonly) text += appTheme.fg(\"muted\", \" (readonly)\");\n\treturn text;\n}\n\nfunction formatDocReadResult(\n\tresult: { content: Array<{ type: string; text?: string }>; details?: DocReadToolDetails },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocReadToolDefinition(\n\tcwd: string,\n\toptions?: DocReadToolOptions,\n): ToolDefinition<typeof docReadSchema, DocReadToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocRead\",\n\t\tlabel: \"DocRead\",\n\t\tdescription:\n\t\t\t\"Extract a structured or binary document (XML, drawio, docx/xlsx/pptx, PDF) into editable, id-addressed JSON the agent can patch losslessly. Each node has a stable #id; edit with DocEdit (in place) or DocWrite (to a new path), passing a patch that targets those ids. This is the canonical way to read and edit these formats: never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or rewrite them — that loses the lossless id-map and corrupts formatting. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Extract a structured/binary document into editable, id-addressed JSON\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocRead to open structured/binary documents (XML, drawio, docx/xlsx/pptx, PDF) instead of read; it returns id-addressed nodes you can patch with DocEdit/DocWrite. Never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or edit these formats — that loses the lossless id-map and corrupts the file.\",\n\t\t\t\"Flow: scan first, edit second. To understand a document, start with DocRead readonly:true — an analysis-only projection that strips node ids and is much cheaper in tokens. Only do a full (writable) DocRead when you actually intend to edit, since that view carries the whole id-map and is token-heavy.\",\n\t\t\t\"Read once, then edit. A writable DocRead establishes the id-map; DocEdit/DocWrite re-extract on their own, so do not re-run DocRead between edits. If a patch targets stale ids the edit fails and returns the current structure with fresh ids to re-issue against.\",\n\t\t],\n\t\tparameters: docReadSchema,\n\t\tasync execute(_toolCallId, { path, readonly }: DocReadToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst envelope = await extractDocument(absolutePath, cwd, signal, {\n\t\t\t\treadonly,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderEnvelopeText(envelope, readonly) }],\n\t\t\t\tdetails: {\n\t\t\t\t\ttype: envelope.source.type,\n\t\t\t\t\tfidelity: envelope.fidelity,\n\t\t\t\t\twritable: envelope.writable,\n\t\t\t\t\tnodeCount: countNodes(envelope.structure),\n\t\t\t\t},\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocReadTool(cwd: string, options?: DocReadToolOptions): AgentTool<typeof docReadSchema> {\n\treturn wrapToolDefinition(createDocReadToolDefinition(cwd, options));\n}\n"]}

package/dist/core/tools/docread.js

@@ -2,7 +2,7 @@ import { Text } from "@kolisachint/hoocode-tui";
 import { Type } from "typebox";
 import { keyHint } from "../../modes/interactive/components/keybinding-hints.js";
 import { theme as appTheme } from "../../modes/interactive/theme/theme.js";
-import { DOCREAD_MAX_RENDER_TOKENS, extractDocument, truncateRenderToTokenBudget, } from "./filetools-shared.js";
+import { DOCREAD_MAX_RENDER_TOKENS, extractDocument, renderDocNodeLines, truncateRenderToTokenBudget, } from "./filetools-shared.js";
 import { resolveReadPath } from "./path-utils.js";
 import { getTextOutput, invalidArgText, shortenPath, str } from "./render-utils.js";
 import { wrapToolDefinition } from "./tool-definition-wrapper.js";
@@ -32,19 +32,7 @@ function countNodes(nodes) {
 export function renderEnvelopeText(envelope, readonly) {
     const header = `document ${envelope.source.path} [${envelope.source.type}, ${envelope.fidelity}, ` +
         `${envelope.writable ? "writable" : "read-only"}]`;
-    const lines = [header, ""];
-    const walk = (nodes, depth) => {
-        for (const node of nodes) {
-            const indent = "  ".repeat(depth);
-            const idPart = node.id ? `#${node.id} ` : "";
-            const attrs = node.attrs?.length ? ` ${node.attrs.map((a) => `${a.name}="${a.value}"`).join(" ")}` : "";
-            const text = node.text !== undefined ? ` :: ${JSON.stringify(node.text)}` : "";
-            lines.push(`${indent}${idPart}<${node.tag}${attrs}>${text}`);
-            if (node.children?.length)
-                walk(node.children, depth + 1);
-        }
-    };
-    walk(envelope.structure, 0);
+    const lines = [header, "", ...renderDocNodeLines(envelope.structure)];
     const { text, droppedLines } = truncateRenderToTokenBudget(lines);
     if (droppedLines === 0)
         return text;
@@ -85,7 +73,8 @@ export function createDocReadToolDefinition(cwd, options) {
         promptSnippet: "Extract a structured/binary document into editable, id-addressed JSON",
         promptGuidelines: [
             "Use DocRead to open structured/binary documents (XML, drawio, docx/xlsx/pptx, PDF) instead of read; it returns id-addressed nodes you can patch with DocEdit/DocWrite. Never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or edit these formats — that loses the lossless id-map and corrupts the file.",
-            "DocEdit/DocWrite require a prior DocRead of the same file (the id-map is established by the extract).",
+            "Flow: scan first, edit second. To understand a document, start with DocRead readonly:true — an analysis-only projection that strips node ids and is much cheaper in tokens. Only do a full (writable) DocRead when you actually intend to edit, since that view carries the whole id-map and is token-heavy.",
+            "Read once, then edit. A writable DocRead establishes the id-map; DocEdit/DocWrite re-extract on their own, so do not re-run DocRead between edits. If a patch targets stale ids the edit fails and returns the current structure with fresh ids to re-issue against.",
         ],
         parameters: docReadSchema,
         async execute(_toolCallId, { path, readonly }, signal) {

package/dist/core/tools/docread.js.map

@@ -1 +1 @@
-{"version":3,"file":"docread.js","sourceRoot":"","sources":["../../../src/core/tools/docread.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AAChD,OAAO,EAAe,IAAI,EAAE,MAAM,SAAS,CAAC;AAC5C,OAAO,EAAE,OAAO,EAAE,MAAM,wDAAwD,CAAC;AACjF,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,wCAAwC,CAAC;AAE3E,OAAO,EACN,yBAAyB,EAGzB,eAAe,EACf,2BAA2B,GAC3B,MAAM,uBAAuB,CAAC;AAC/B,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,WAAW,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AACpF,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAElE,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC;IACjC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC;QACjB,WAAW,EACV,sGAAsG;KACvG,CAAC;IACF,QAAQ,EAAE,IAAI,CAAC,QAAQ,CACtB,IAAI,CAAC,OAAO,CAAC;QACZ,WAAW,EACV,0JAA0J;KAC3J,CAAC,CACF;CACD,CAAC,CAAC;AAgBH,SAAS,UAAU,CAAC,KAAgB,EAAU;IAC7C,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,KAAK,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9D,CAAC;IACD,OAAO,KAAK,CAAC;AAAA,CACb;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,kBAAkB,CAAC,QAAkB,EAAE,QAA6B,EAAU;IAC7F,MAAM,MAAM,GACX,YAAY,QAAQ,CAAC,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,QAAQ,IAAI;QACnF,GAAG,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,GAAG,CAAC;IACpD,MAAM,KAAK,GAAa,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;IAErC,MAAM,IAAI,GAAG,CAAC,KAAgB,EAAE,KAAa,EAAQ,EAAE,CAAC;QACvD,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC1B,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;YAClC,MAAM,MAAM,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;YAC7C,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACxG,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,OAAO,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAC/E,KAAK,CAAC,IAAI,CAAC,GAAG,MAAM,GAAG,MAAM,IAAI,IAAI,CAAC,GAAG,GAAG,KAAK,IAAI,IAAI,EAAE,CAAC,CAAC;YAC7D,IAAI,IAAI,CAAC,QAAQ,EAAE,MAAM;gBAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;QAC3D,CAAC;IAAA,CACD,CAAC;IACF,IAAI,CAAC,QAAQ,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC;IAE5B,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,2BAA2B,CAAC,KAAK,CAAC,CAAC;IAClE,IAAI,YAAY,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAEpC,MAAM,IAAI,GAAG,QAAQ;QACpB,CAAC,CAAC,2CAA2C;QAC7C,CAAC,CAAC,sFAAsF,CAAC;IAC1F,OAAO,CACN,GAAG,IAAI,yCAAyC,yBAAyB,wBAAwB;QACjG,GAAG,YAAY,kBAAkB,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,8CAA8C;QAC5G,wBAAsB,IAAI,IAAI,CAC9B,CAAC;AAAA,CACF;AAED,SAAS,iBAAiB,CAAC,IAAuD,EAAU;IAC3F,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC7B,MAAM,WAAW,GAChB,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;IACxG,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,WAAW,EAAE,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,GAAG,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;IACpG,IAAI,IAAI,EAAE,QAAQ;QAAE,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,aAAa,CAAC,CAAC;IAChE,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,SAAS,mBAAmB,CAC3B,MAAyF,EACzF,OAAgC,EAChC,UAAmB,EACV;IACT,MAAM,MAAM,GAAG,aAAa,CAAC,MAAa,EAAE,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC;IAC/D,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,CAAC;IACvB,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACjC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;IACtD,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,GAAG,QAAQ,CAAC;IAC1C,IAAI,IAAI,GAAG,KAAK,YAAY,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACzF,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QACnB,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,SAAS,cAAc,CAAC,IAAI,OAAO,CAAC,kBAAkB,EAAE,WAAW,CAAC,GAAG,CAAC;IACnH,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,MAAM,UAAU,2BAA2B,CAC1C,GAAW,EACX,OAA4B,EAC2C;IACvE,OAAO;QACN,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;QAChB,WAAW,EACV,yhBAAuhB;QACxhB,aAAa,EAAE,uEAAuE;QACtF,gBAAgB,EAAE;YACjB,+UAA6U;YAC7U,uGAAuG;SACvG;QACD,UAAU,EAAE,aAAa;QACzB,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAoB,EAAE,MAAoB,EAAE;YACtF,IAAI,MAAM,EAAE,OAAO;gBAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;YAC1D,MAAM,YAAY,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAChD,MAAM,QAAQ,GAAG,MAAM,eAAe,CAAC,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE;gBACjE,QAAQ;gBACR,WAAW,EAAE,OAAO,EAAE,WAAW;aACjC,CAAC,CAAC;YACH,OAAO;gBACN,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAe,EAAE,IAAI,EAAE,kBAAkB,CAAC,QAAQ,EAAE,QAAQ,CAAC,EAAE,CAAC;gBAClF,OAAO,EAAE;oBACR,IAAI,EAAE,QAAQ,CAAC,MAAM,CAAC,IAAI;oBAC1B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;oBAC3B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;oBAC3B,SAAS,EAAE,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC;iBACzC;aACD,CAAC;QAAA,CACF;QACD,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE;YACjC,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;YACtC,OAAO,IAAI,CAAC;QAAA,CACZ;QACD,YAAY,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE;YAC9C,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,mBAAmB,CAAC,MAAa,EAAE,OAAO,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;YAC9E,OAAO,IAAI,CAAC;QAAA,CACZ;KACD,CAAC;AAAA,CACF;AAED,MAAM,UAAU,iBAAiB,CAAC,GAAW,EAAE,OAA4B,EAAmC;IAC7G,OAAO,kBAAkB,CAAC,2BAA2B,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC;AAAA,CACrE","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\tDOCREAD_MAX_RENDER_TOKENS,\n\ttype DocNode,\n\ttype Envelope,\n\textractDocument,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docReadSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription:\n\t\t\t\"Path to the document to extract (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\treadonly: Type.Optional(\n\t\tType.Boolean({\n\t\t\tdescription:\n\t\t\t\t\"Analysis-only projection: strips node ids for a smaller view that CANNOT be edited. Omit (default false) when you intend to DocEdit/DocWrite afterwards.\",\n\t\t}),\n\t),\n});\n\nexport type DocReadToolInput = Static<typeof docReadSchema>;\n\nexport interface DocReadToolDetails {\n\ttype?: string;\n\tfidelity?: string;\n\twritable?: boolean;\n\tnodeCount?: number;\n}\n\nexport interface DocReadToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\nfunction countNodes(nodes: DocNode[]): number {\n\tlet total = 0;\n\tfor (const node of nodes) {\n\t\ttotal += 1 + (node.children ? countNodes(node.children) : 0);\n\t}\n\treturn total;\n}\n\n/**\n * Render the envelope as compact, id-addressed text the model edits against.\n * Each line carries the node id so DocEdit/DocWrite patches can target it.\n *\n * The filetools binary has no pagination, so a dense document can project into\n * a huge dump. We truncate the rendered view to a token budget and append an\n * actionable notice instead of flooding the model context.\n */\nexport function renderEnvelopeText(envelope: Envelope, readonly: boolean | undefined): string {\n\tconst header =\n\t\t`document ${envelope.source.path} [${envelope.source.type}, ${envelope.fidelity}, ` +\n\t\t`${envelope.writable ? \"writable\" : \"read-only\"}]`;\n\tconst lines: string[] = [header, \"\"];\n\n\tconst walk = (nodes: DocNode[], depth: number): void => {\n\t\tfor (const node of nodes) {\n\t\t\tconst indent = \"  \".repeat(depth);\n\t\t\tconst idPart = node.id ? `#${node.id} ` : \"\";\n\t\t\tconst attrs = node.attrs?.length ? ` ${node.attrs.map((a) => `${a.name}=\"${a.value}\"`).join(\" \")}` : \"\";\n\t\t\tconst text = node.text !== undefined ? ` :: ${JSON.stringify(node.text)}` : \"\";\n\t\t\tlines.push(`${indent}${idPart}<${node.tag}${attrs}>${text}`);\n\t\t\tif (node.children?.length) walk(node.children, depth + 1);\n\t\t}\n\t};\n\twalk(envelope.structure, 0);\n\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tif (droppedLines === 0) return text;\n\n\tconst hint = readonly\n\t\t? \"narrow to a smaller or more targeted file\"\n\t\t: \"re-run with readonly:true for a smaller analysis-only view, or target a smaller file\";\n\treturn (\n\t\t`${text}\\n\\n[Truncated: document exceeds the ~${DOCREAD_MAX_RENDER_TOKENS} token render budget; ` +\n\t\t`${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted. The ids shown are still valid for ` +\n\t\t`DocEdit/DocWrite — ${hint}.]`\n\t);\n}\n\nfunction formatDocReadCall(args: { path?: string; readonly?: boolean } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocRead \")) + appTheme.fg(\"accent\", pathDisplay);\n\tif (args?.readonly) text += appTheme.fg(\"muted\", \" (readonly)\");\n\treturn text;\n}\n\nfunction formatDocReadResult(\n\tresult: { content: Array<{ type: string; text?: string }>; details?: DocReadToolDetails },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocReadToolDefinition(\n\tcwd: string,\n\toptions?: DocReadToolOptions,\n): ToolDefinition<typeof docReadSchema, DocReadToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocRead\",\n\t\tlabel: \"DocRead\",\n\t\tdescription:\n\t\t\t\"Extract a structured or binary document (XML, drawio, docx/xlsx/pptx, PDF) into editable, id-addressed JSON the agent can patch losslessly. Each node has a stable #id; edit with DocEdit (in place) or DocWrite (to a new path), passing a patch that targets those ids. This is the canonical way to read and edit these formats: never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or rewrite them — that loses the lossless id-map and corrupts formatting. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Extract a structured/binary document into editable, id-addressed JSON\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocRead to open structured/binary documents (XML, drawio, docx/xlsx/pptx, PDF) instead of read; it returns id-addressed nodes you can patch with DocEdit/DocWrite. Never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or edit these formats — that loses the lossless id-map and corrupts the file.\",\n\t\t\t\"DocEdit/DocWrite require a prior DocRead of the same file (the id-map is established by the extract).\",\n\t\t],\n\t\tparameters: docReadSchema,\n\t\tasync execute(_toolCallId, { path, readonly }: DocReadToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst envelope = await extractDocument(absolutePath, cwd, signal, {\n\t\t\t\treadonly,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderEnvelopeText(envelope, readonly) }],\n\t\t\t\tdetails: {\n\t\t\t\t\ttype: envelope.source.type,\n\t\t\t\t\tfidelity: envelope.fidelity,\n\t\t\t\t\twritable: envelope.writable,\n\t\t\t\t\tnodeCount: countNodes(envelope.structure),\n\t\t\t\t},\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocReadTool(cwd: string, options?: DocReadToolOptions): AgentTool<typeof docReadSchema> {\n\treturn wrapToolDefinition(createDocReadToolDefinition(cwd, options));\n}\n"]}
+{"version":3,"file":"docread.js","sourceRoot":"","sources":["../../../src/core/tools/docread.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AAChD,OAAO,EAAe,IAAI,EAAE,MAAM,SAAS,CAAC;AAC5C,OAAO,EAAE,OAAO,EAAE,MAAM,wDAAwD,CAAC;AACjF,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,wCAAwC,CAAC;AAE3E,OAAO,EACN,yBAAyB,EAGzB,eAAe,EACf,kBAAkB,EAClB,2BAA2B,GAC3B,MAAM,uBAAuB,CAAC;AAC/B,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,WAAW,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AACpF,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AAElE,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC;IACjC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC;QACjB,WAAW,EACV,sGAAsG;KACvG,CAAC;IACF,QAAQ,EAAE,IAAI,CAAC,QAAQ,CACtB,IAAI,CAAC,OAAO,CAAC;QACZ,WAAW,EACV,0JAA0J;KAC3J,CAAC,CACF;CACD,CAAC,CAAC;AAgBH,SAAS,UAAU,CAAC,KAAgB,EAAU;IAC7C,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,KAAK,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC9D,CAAC;IACD,OAAO,KAAK,CAAC;AAAA,CACb;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,kBAAkB,CAAC,QAAkB,EAAE,QAA6B,EAAU;IAC7F,MAAM,MAAM,GACX,YAAY,QAAQ,CAAC,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,MAAM,CAAC,IAAI,KAAK,QAAQ,CAAC,QAAQ,IAAI;QACnF,GAAG,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,GAAG,CAAC;IACpD,MAAM,KAAK,GAAa,CAAC,MAAM,EAAE,EAAE,EAAE,GAAG,kBAAkB,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC;IAEhF,MAAM,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,2BAA2B,CAAC,KAAK,CAAC,CAAC;IAClE,IAAI,YAAY,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAEpC,MAAM,IAAI,GAAG,QAAQ;QACpB,CAAC,CAAC,2CAA2C;QAC7C,CAAC,CAAC,sFAAsF,CAAC;IAC1F,OAAO,CACN,GAAG,IAAI,yCAAyC,yBAAyB,wBAAwB;QACjG,GAAG,YAAY,kBAAkB,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,8CAA8C;QAC5G,wBAAsB,IAAI,IAAI,CAC9B,CAAC;AAAA,CACF;AAED,SAAS,iBAAiB,CAAC,IAAuD,EAAU;IAC3F,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC7B,MAAM,WAAW,GAChB,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;IACxG,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,WAAW,EAAE,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,GAAG,QAAQ,CAAC,EAAE,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;IACpG,IAAI,IAAI,EAAE,QAAQ;QAAE,IAAI,IAAI,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,aAAa,CAAC,CAAC;IAChE,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,SAAS,mBAAmB,CAC3B,MAAyF,EACzF,OAAgC,EAChC,UAAmB,EACV;IACT,MAAM,MAAM,GAAG,aAAa,CAAC,MAAa,EAAE,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC;IAC/D,IAAI,CAAC,MAAM;QAAE,OAAO,EAAE,CAAC;IACvB,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACjC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;IACtD,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;IAC9C,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,GAAG,QAAQ,CAAC;IAC1C,IAAI,IAAI,GAAG,KAAK,YAAY,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACzF,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;QACnB,IAAI,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,SAAS,cAAc,CAAC,IAAI,OAAO,CAAC,kBAAkB,EAAE,WAAW,CAAC,GAAG,CAAC;IACnH,CAAC;IACD,OAAO,IAAI,CAAC;AAAA,CACZ;AAED,MAAM,UAAU,2BAA2B,CAC1C,GAAW,EACX,OAA4B,EAC2C;IACvE,OAAO;QACN,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;QAChB,WAAW,EACV,yhBAAuhB;QACxhB,aAAa,EAAE,uEAAuE;QACtF,gBAAgB,EAAE;YACjB,+UAA6U;YAC7U,gTAA8S;YAC9S,sQAAsQ;SACtQ;QACD,UAAU,EAAE,aAAa;QACzB,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAoB,EAAE,MAAoB,EAAE;YACtF,IAAI,MAAM,EAAE,OAAO;gBAAE,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;YAC1D,MAAM,YAAY,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAChD,MAAM,QAAQ,GAAG,MAAM,eAAe,CAAC,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE;gBACjE,QAAQ;gBACR,WAAW,EAAE,OAAO,EAAE,WAAW;aACjC,CAAC,CAAC;YACH,OAAO;gBACN,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAe,EAAE,IAAI,EAAE,kBAAkB,CAAC,QAAQ,EAAE,QAAQ,CAAC,EAAE,CAAC;gBAClF,OAAO,EAAE;oBACR,IAAI,EAAE,QAAQ,CAAC,MAAM,CAAC,IAAI;oBAC1B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;oBAC3B,QAAQ,EAAE,QAAQ,CAAC,QAAQ;oBAC3B,SAAS,EAAE,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC;iBACzC;aACD,CAAC;QAAA,CACF;QACD,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE;YACjC,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC;YACtC,OAAO,IAAI,CAAC;QAAA,CACZ;QACD,YAAY,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE;YAC9C,MAAM,IAAI,GAAI,OAAO,CAAC,aAAkC,IAAI,IAAI,IAAI,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAC/E,IAAI,CAAC,OAAO,CAAC,mBAAmB,CAAC,MAAa,EAAE,OAAO,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;YAC9E,OAAO,IAAI,CAAC;QAAA,CACZ;KACD,CAAC;AAAA,CACF;AAED,MAAM,UAAU,iBAAiB,CAAC,GAAW,EAAE,OAA4B,EAAmC;IAC7G,OAAO,kBAAkB,CAAC,2BAA2B,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC;AAAA,CACrE","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport {\n\tDOCREAD_MAX_RENDER_TOKENS,\n\ttype DocNode,\n\ttype Envelope,\n\textractDocument,\n\trenderDocNodeLines,\n\ttruncateRenderToTokenBudget,\n} from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docReadSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription:\n\t\t\t\"Path to the document to extract (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\treadonly: Type.Optional(\n\t\tType.Boolean({\n\t\t\tdescription:\n\t\t\t\t\"Analysis-only projection: strips node ids for a smaller view that CANNOT be edited. Omit (default false) when you intend to DocEdit/DocWrite afterwards.\",\n\t\t}),\n\t),\n});\n\nexport type DocReadToolInput = Static<typeof docReadSchema>;\n\nexport interface DocReadToolDetails {\n\ttype?: string;\n\tfidelity?: string;\n\twritable?: boolean;\n\tnodeCount?: number;\n}\n\nexport interface DocReadToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\nfunction countNodes(nodes: DocNode[]): number {\n\tlet total = 0;\n\tfor (const node of nodes) {\n\t\ttotal += 1 + (node.children ? countNodes(node.children) : 0);\n\t}\n\treturn total;\n}\n\n/**\n * Render the envelope as compact, id-addressed text the model edits against.\n * Each line carries the node id so DocEdit/DocWrite patches can target it.\n *\n * The filetools binary has no pagination, so a dense document can project into\n * a huge dump. We truncate the rendered view to a token budget and append an\n * actionable notice instead of flooding the model context.\n */\nexport function renderEnvelopeText(envelope: Envelope, readonly: boolean | undefined): string {\n\tconst header =\n\t\t`document ${envelope.source.path} [${envelope.source.type}, ${envelope.fidelity}, ` +\n\t\t`${envelope.writable ? \"writable\" : \"read-only\"}]`;\n\tconst lines: string[] = [header, \"\", ...renderDocNodeLines(envelope.structure)];\n\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tif (droppedLines === 0) return text;\n\n\tconst hint = readonly\n\t\t? \"narrow to a smaller or more targeted file\"\n\t\t: \"re-run with readonly:true for a smaller analysis-only view, or target a smaller file\";\n\treturn (\n\t\t`${text}\\n\\n[Truncated: document exceeds the ~${DOCREAD_MAX_RENDER_TOKENS} token render budget; ` +\n\t\t`${droppedLines} more node line${droppedLines === 1 ? \"\" : \"s\"} omitted. The ids shown are still valid for ` +\n\t\t`DocEdit/DocWrite — ${hint}.]`\n\t);\n}\n\nfunction formatDocReadCall(args: { path?: string; readonly?: boolean } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocRead \")) + appTheme.fg(\"accent\", pathDisplay);\n\tif (args?.readonly) text += appTheme.fg(\"muted\", \" (readonly)\");\n\treturn text;\n}\n\nfunction formatDocReadResult(\n\tresult: { content: Array<{ type: string; text?: string }>; details?: DocReadToolDetails },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocReadToolDefinition(\n\tcwd: string,\n\toptions?: DocReadToolOptions,\n): ToolDefinition<typeof docReadSchema, DocReadToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocRead\",\n\t\tlabel: \"DocRead\",\n\t\tdescription:\n\t\t\t\"Extract a structured or binary document (XML, drawio, docx/xlsx/pptx, PDF) into editable, id-addressed JSON the agent can patch losslessly. Each node has a stable #id; edit with DocEdit (in place) or DocWrite (to a new path), passing a patch that targets those ids. This is the canonical way to read and edit these formats: never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or rewrite them — that loses the lossless id-map and corrupts formatting. Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Extract a structured/binary document into editable, id-addressed JSON\",\n\t\tpromptGuidelines: [\n\t\t\t\"Use DocRead to open structured/binary documents (XML, drawio, docx/xlsx/pptx, PDF) instead of read; it returns id-addressed nodes you can patch with DocEdit/DocWrite. Never fall back to ad-hoc scripts (python/openpyxl, docx, PyPDF2, unzip, sed) to parse or edit these formats — that loses the lossless id-map and corrupts the file.\",\n\t\t\t\"Flow: scan first, edit second. To understand a document, start with DocRead readonly:true — an analysis-only projection that strips node ids and is much cheaper in tokens. Only do a full (writable) DocRead when you actually intend to edit, since that view carries the whole id-map and is token-heavy.\",\n\t\t\t\"Read once, then edit. A writable DocRead establishes the id-map; DocEdit/DocWrite re-extract on their own, so do not re-run DocRead between edits. If a patch targets stale ids the edit fails and returns the current structure with fresh ids to re-issue against.\",\n\t\t],\n\t\tparameters: docReadSchema,\n\t\tasync execute(_toolCallId, { path, readonly }: DocReadToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst envelope = await extractDocument(absolutePath, cwd, signal, {\n\t\t\t\treadonly,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderEnvelopeText(envelope, readonly) }],\n\t\t\t\tdetails: {\n\t\t\t\t\ttype: envelope.source.type,\n\t\t\t\t\tfidelity: envelope.fidelity,\n\t\t\t\t\twritable: envelope.writable,\n\t\t\t\t\tnodeCount: countNodes(envelope.structure),\n\t\t\t\t},\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocReadResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocReadTool(cwd: string, options?: DocReadToolOptions): AgentTool<typeof docReadSchema> {\n\treturn wrapToolDefinition(createDocReadToolDefinition(cwd, options));\n}\n"]}

package/dist/core/tools/docscan.d.ts

@@ -0,0 +1,29 @@
+import type { AgentTool } from "@kolisachint/hoocode-agent-core";
+import { type Static, Type } from "typebox";
+import type { ToolDefinition } from "../extensions/types.js";
+import { type ScanView } from "./filetools-shared.js";
+declare const docScanSchema: Type.TObject<{
+    path: Type.TString;
+    offset: Type.TOptional<Type.TNumber>;
+    limit: Type.TOptional<Type.TNumber>;
+}>;
+export type DocScanToolInput = Static<typeof docScanSchema>;
+export interface DocScanToolDetails {
+    fileType?: string;
+    returned?: number;
+    total?: number;
+}
+export interface DocScanToolOptions {
+    /** Timeout (seconds) for the filetools invocation. */
+    timeoutSecs?: number;
+}
+/**
+ * Render a scan manifest as compact, id-addressed preview lines. Each line
+ * carries the block id (feeds DocPeek/DocGrep) plus a section label, token
+ * estimate, and a short preview — structure without hydrating full content.
+ */
+export declare function renderScanView(view: ScanView): string;
+export declare function createDocScanToolDefinition(cwd: string, options?: DocScanToolOptions): ToolDefinition<typeof docScanSchema, DocScanToolDetails | undefined>;
+export declare function createDocScanTool(cwd: string, options?: DocScanToolOptions): AgentTool<typeof docScanSchema>;
+export {};
+//# sourceMappingURL=docscan.d.ts.map

package/dist/core/tools/docscan.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"docscan.d.ts","sourceRoot":"","sources":["../../../src/core/tools/docscan.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,iCAAiC,CAAC;AAEjE,OAAO,EAAE,KAAK,MAAM,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAG5C,OAAO,KAAK,EAAE,cAAc,EAA2B,MAAM,wBAAwB,CAAC;AACtF,OAAO,EAAE,KAAK,QAAQ,EAA6C,MAAM,uBAAuB,CAAC;AAKjG,QAAA,MAAM,aAAa;;;;EAMjB,CAAC;AAEH,MAAM,MAAM,gBAAgB,GAAG,MAAM,CAAC,OAAO,aAAa,CAAC,CAAC;AAE5D,MAAM,WAAW,kBAAkB;IAClC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,KAAK,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,kBAAkB;IAClC,sDAAsD;IACtD,WAAW,CAAC,EAAE,MAAM,CAAC;CACrB;AAED;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,QAAQ,GAAG,MAAM,CA2BrD;AAgCD,wBAAgB,2BAA2B,CAC1C,GAAG,EAAE,MAAM,EACX,OAAO,CAAC,EAAE,kBAAkB,GAC1B,cAAc,CAAC,OAAO,aAAa,EAAE,kBAAkB,GAAG,SAAS,CAAC,CAoCtE;AAED,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,kBAAkB,GAAG,SAAS,CAAC,OAAO,aAAa,CAAC,CAE5G","sourcesContent":["import type { AgentTool } from \"@kolisachint/hoocode-agent-core\";\nimport { Text } from \"@kolisachint/hoocode-tui\";\nimport { type Static, Type } from \"typebox\";\nimport { keyHint } from \"../../modes/interactive/components/keybinding-hints.js\";\nimport { theme as appTheme } from \"../../modes/interactive/theme/theme.js\";\nimport type { ToolDefinition, ToolRenderResultOptions } from \"../extensions/types.js\";\nimport { type ScanView, scanDocument, truncateRenderToTokenBudget } from \"./filetools-shared.js\";\nimport { resolveReadPath } from \"./path-utils.js\";\nimport { getTextOutput, invalidArgText, shortenPath, str } from \"./render-utils.js\";\nimport { wrapToolDefinition } from \"./tool-definition-wrapper.js\";\n\nconst docScanSchema = Type.Object({\n\tpath: Type.String({\n\t\tdescription: \"Path to the document to scan (relative or absolute). XML, drawio, OOXML (docx/xlsx/pptx), or PDF.\",\n\t}),\n\toffset: Type.Optional(Type.Number({ description: \"Skip the first N blocks (pagination). Default 0.\" })),\n\tlimit: Type.Optional(Type.Number({ description: \"Maximum number of blocks to return. Default: all remaining.\" })),\n});\n\nexport type DocScanToolInput = Static<typeof docScanSchema>;\n\nexport interface DocScanToolDetails {\n\tfileType?: string;\n\treturned?: number;\n\ttotal?: number;\n}\n\nexport interface DocScanToolOptions {\n\t/** Timeout (seconds) for the filetools invocation. */\n\ttimeoutSecs?: number;\n}\n\n/**\n * Render a scan manifest as compact, id-addressed preview lines. Each line\n * carries the block id (feeds DocPeek/DocGrep) plus a section label, token\n * estimate, and a short preview — structure without hydrating full content.\n */\nexport function renderScanView(view: ScanView): string {\n\tconst header =\n\t\t`document ${view.file_type} — ${view.returned}/${view.total} blocks ` +\n\t\t`(offset ${view.offset}, ~${view.total_tokens} tok total)`;\n\tconst lines: string[] = [header, \"\"];\n\tfor (const block of view.blocks) {\n\t\tconst section = block.section_name ? ` ${block.section_name}#${block.section_number}` : \"\";\n\t\tconst preview = block.preview ? ` :: ${JSON.stringify(block.preview)}` : \"\";\n\t\tlines.push(`#${block.id} [${block.block_type}]${section} ~${block.token_estimate}tok${preview}`);\n\t}\n\n\tconst { text, droppedLines } = truncateRenderToTokenBudget(lines);\n\tconst seen = view.offset + view.returned;\n\tconst remaining = view.total - seen;\n\tconst parts = [text];\n\tif (droppedLines > 0) {\n\t\tparts.push(\n\t\t\t`\\n[Truncated: ${droppedLines} more block line${droppedLines === 1 ? \"\" : \"s\"} omitted from this render. ` +\n\t\t\t\t`Re-run with a tighter limit, or DocGrep/DocPeek to target.]`,\n\t\t);\n\t}\n\tif (remaining > 0) {\n\t\tparts.push(\n\t\t\t`\\n[${remaining} more block${remaining === 1 ? \"\" : \"s\"} not shown — re-run with offset:${seen} to continue.]`,\n\t\t);\n\t}\n\treturn parts.join(\"\\n\");\n}\n\nfunction formatDocScanCall(args: { path?: string; offset?: number; limit?: number } | undefined): string {\n\tconst path = str(args?.path);\n\tconst pathDisplay =\n\t\tpath === null ? invalidArgText(appTheme) : path ? shortenPath(path) : appTheme.fg(\"toolOutput\", \"...\");\n\tlet text = appTheme.fg(\"toolTitle\", appTheme.bold(\"DocScan \")) + appTheme.fg(\"accent\", pathDisplay);\n\tconst range: string[] = [];\n\tif (args?.offset !== undefined) range.push(`offset ${args.offset}`);\n\tif (args?.limit !== undefined) range.push(`limit ${args.limit}`);\n\tif (range.length > 0) text += appTheme.fg(\"muted\", ` (${range.join(\", \")})`);\n\treturn text;\n}\n\nfunction formatDocScanResult(\n\tresult: { content: Array<{ type: string; text?: string }> },\n\toptions: ToolRenderResultOptions,\n\tshowImages: boolean,\n): string {\n\tconst output = getTextOutput(result as any, showImages).trim();\n\tif (!output) return \"\";\n\tconst lines = output.split(\"\\n\");\n\tconst maxLines = options.expanded ? lines.length : 15;\n\tconst displayLines = lines.slice(0, maxLines);\n\tconst remaining = lines.length - maxLines;\n\tlet text = `\\n${displayLines.map((line) => appTheme.fg(\"toolOutput\", line)).join(\"\\n\")}`;\n\tif (remaining > 0) {\n\t\ttext += `${appTheme.fg(\"muted\", `\\n... (${remaining} more lines,`)} ${keyHint(\"app.tools.expand\", \"to expand\")})`;\n\t}\n\treturn text;\n}\n\nexport function createDocScanToolDefinition(\n\tcwd: string,\n\toptions?: DocScanToolOptions,\n): ToolDefinition<typeof docScanSchema, DocScanToolDetails | undefined> {\n\treturn {\n\t\tname: \"DocScan\",\n\t\tlabel: \"DocScan\",\n\t\tdescription:\n\t\t\t\"Cheaply outline a structured/binary document (XML, drawio, docx/xlsx/pptx, PDF) without hydrating it: returns a paginated manifest of blocks, each with a structural-path id, type, section label, token estimate, and a short preview. This is the cheap first step of the token-sensitive loop — far smaller than a full DocRead. Pass the path ids it returns to DocPeek to hydrate just those blocks (the hydrated nodes carry the editable el_ #ids for DocEdit). Off by default; enabled with --enable-filetools.\",\n\t\tpromptSnippet: \"Outline a structured/binary document into a cheap, paginated block manifest\",\n\t\tpromptGuidelines: [\n\t\t\t\"Start here for large structured/binary documents instead of a full DocRead: DocScan returns a paginated outline (structural-path block ids + previews) that is much cheaper in tokens. Page through it with offset/limit.\",\n\t\t\t\"Flow: DocScan to see structure → DocPeek the path ids you want (hydrates them and reveals their editable el_ #ids) → DocEdit. Or jump straight in with DocGrep, which finds blocks by text and returns editable el_ #ids directly.\",\n\t\t],\n\t\tparameters: docScanSchema,\n\t\tasync execute(_toolCallId, { path, offset, limit }: DocScanToolInput, signal?: AbortSignal) {\n\t\t\tif (signal?.aborted) throw new Error(\"Operation aborted\");\n\t\t\tconst absolutePath = resolveReadPath(path, cwd);\n\t\t\tconst view = await scanDocument(absolutePath, cwd, signal, {\n\t\t\t\toffset,\n\t\t\t\tlimit,\n\t\t\t\ttimeoutSecs: options?.timeoutSecs,\n\t\t\t});\n\t\t\treturn {\n\t\t\t\tcontent: [{ type: \"text\" as const, text: renderScanView(view) }],\n\t\t\t\tdetails: { fileType: view.file_type, returned: view.returned, total: view.total },\n\t\t\t};\n\t\t},\n\t\trenderCall(args, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocScanCall(args));\n\t\t\treturn text;\n\t\t},\n\t\trenderResult(result, options, _theme, context) {\n\t\t\tconst text = (context.lastComponent as Text | undefined) ?? new Text(\"\", 0, 0);\n\t\t\ttext.setText(formatDocScanResult(result as any, options, context.showImages));\n\t\t\treturn text;\n\t\t},\n\t};\n}\n\nexport function createDocScanTool(cwd: string, options?: DocScanToolOptions): AgentTool<typeof docScanSchema> {\n\treturn wrapToolDefinition(createDocScanToolDefinition(cwd, options));\n}\n"]}