fix: truncate webfetch content to fit context

Plugins/ZShell/Llm/filetool.cpp

@@ -0,0 +1,151 @@
+#include "filetool.hpp"
+
+#include <QFile>
+#include <QFileInfo>
+#include <QJsonArray>
+
+namespace ZShell::llm {
+
+namespace {
+
+QJsonObject makeOutput(const QString& text) {
+	QJsonObject obj;
+	obj[QStringLiteral("output")] = text;
+	return obj;
+}
+
+QJsonObject makeError(const QString& message) {
+	QJsonObject obj;
+	obj[QStringLiteral("error")] = message;
+	return obj;
+}
+
+} // namespace
+
+FileReadTool::FileReadTool(QObject* parent) : LlmTool(parent) {}
+
+QString FileReadTool::name() const {
+	return QStringLiteral("readfile");
+}
+
+QString FileReadTool::description() const {
+	return QStringLiteral(
+		"Read a text file from the local filesystem and return part of "
+		"its content. At most a limited number of characters is returned "
+		"per call; use offset and limit to page through larger files. "
+		"Binary files are not supported. Full output of webfetch calls "
+		"is stored under /tmp/zshell-llm/webfetch/ and can be read this "
+		"way. This tool is read-only.");
+}
+
+QJsonObject FileReadTool::parameters() const {
+	QJsonObject path;
+	path[QStringLiteral("type")] = QStringLiteral("string");
+	path[QStringLiteral("description")] =
+		QStringLiteral("Path of the file to read");
+
+	QJsonObject offset;
+	offset[QStringLiteral("type")] = QStringLiteral("integer");
+	offset[QStringLiteral("minimum")] = 0;
+	offset[QStringLiteral("description")] =
+		QStringLiteral("Byte offset to start reading from. Defaults to 0.");
+
+	QJsonObject limit;
+	limit[QStringLiteral("type")] = QStringLiteral("integer");
+	limit[QStringLiteral("minimum")] = 1;
+	limit[QStringLiteral("description")] =
+		QStringLiteral("Maximum number of characters to return. Defaults "
+					   "to a value that fits the model's context window.");
+
+	QJsonObject properties;
+	properties[QStringLiteral("path")] = path;
+	properties[QStringLiteral("offset")] = offset;
+	properties[QStringLiteral("limit")] = limit;
+
+	QJsonObject schema;
+	schema[QStringLiteral("type")] = QStringLiteral("object");
+	schema[QStringLiteral("properties")] = properties;
+	QJsonArray required;
+	required.append(QStringLiteral("path"));
+	schema[QStringLiteral("required")] = required;
+	return schema;
+}
+
+void FileReadTool::execute(
+	const QString& toolCallId,
+	const QJsonObject& args,
+	std::function<void(const QJsonObject&)> done) {
+	Q_UNUSED(toolCallId);
+	const QString path = args[QStringLiteral("path")].toString().trimmed();
+	if (path.isEmpty()) {
+		done(makeError(QStringLiteral("Missing required argument 'path'")));
+		return;
+	}
+	const int offset = qMax(0, args[QStringLiteral("offset")].toInt(0));
+	const int budget = inlineBudgetChars();
+	const int limit =
+		qBound(1, args[QStringLiteral("limit")].toInt(budget), budget);
+
+	QFile file(path);
+	if (!file.open(QIODevice::ReadOnly)) {
+		const QFileInfo info(path);
+		if (info.exists() && info.isDir()) {
+			done(makeError(QStringLiteral("Path is a directory, not a "
+										   "file: %1")
+							   .arg(path)));
+		} else {
+			done(makeError(QStringLiteral("Cannot open file: %1").arg(path)));
+		}
+		return;
+	}
+	const qint64 fileSize = file.size();
+	if (fileSize == 0) {
+		file.close();
+		done(makeOutput(QString()));
+		return;
+	}
+	if (offset >= fileSize) {
+		file.close();
+		done(makeError(QStringLiteral("Offset %1 is beyond the end of the "
+									   "file (%2 bytes)")
+						   .arg(offset)
+						   .arg(fileSize)));
+		return;
+	}
+	if (!file.seek(offset)) {
+		file.close();
+		done(makeError(QStringLiteral("Cannot seek in file: %1").arg(path)));
+		return;
+	}
+	// UTF-8 uses at most 4 bytes per character, so this many bytes always
+	// cover `limit` characters.
+	const qint64 wanted = qint64(limit) * 4 + 16;
+	const QByteArray chunk = file.read(qMin(wanted, fileSize - offset));
+	file.close();
+
+	if (chunk.contains('\0')) {
+		done(makeError(QStringLiteral("Binary files are not supported: %1")
+						   .arg(path)));
+		return;
+	}
+	const QString content = QString::fromUtf8(chunk);
+	if (content.size() <= limit) {
+		done(makeOutput(content));
+		return;
+	}
+	const QString head = content.left(limit);
+	const int nextOffset = offset + head.toUtf8().size();
+	const QString output =
+		head +
+		QStringLiteral(
+			"\n\n[... truncated: showing %1 characters (bytes %2-%3 of a "
+			"%4 byte file). Continue with offset=%5.]")
+			.arg(limit)
+			.arg(offset)
+			.arg(nextOffset)
+			.arg(fileSize)
+			.arg(nextOffset);
+	done(makeOutput(output));
+}
+
+} // namespace ZShell::llm