matterviz 0.4.1 → 0.4.3

package/dist/trajectory/parse/h5-utils.js

@@ -0,0 +1,109 @@
+// Shared h5wasm plumbing for HDF5 trajectory parsers: type guards plus the
+// write-to-in-memory-FS/open/cleanup lifecycle both torch-sim and vaspout
+// parsing need (the FS write is the expensive part, so parsers share one open
+// file handle instead of re-opening per format probe).
+import { is_elem_symbol } from '../../element/helpers';
+import * as h5wasm from 'h5wasm';
+export const is_hdf5_dataset = (entity) => entity !== null && `to_array` in entity && entity instanceof h5wasm.Dataset;
+export const is_hdf5_group = (entity) => entity !== null && `keys` in entity && entity instanceof h5wasm.Group;
+// Datasets in interrupted files can be torn mid-chunk, making to_array throw —
+// treat unreadable datasets like missing ones so callers keep what parsed so far.
+export const read_dataset = (h5_file, path) => {
+    try {
+        const entity = h5_file.get(path);
+        return is_hdf5_dataset(entity) ? entity.to_array() : null;
+    }
+    catch {
+        return null;
+    }
+};
+// h5wasm returns JS strings for variable-length UTF-8 but may hand back byte
+// arrays for fixed-length string datasets (like VASP's |S2 ion_types).
+export const to_string_array = (data) => {
+    if (!Array.isArray(data))
+        return null;
+    const decoder = new TextDecoder();
+    const strings = [];
+    for (const item of data) {
+        if (typeof item === `string`)
+            strings.push(item.trim());
+        else if (item instanceof Uint8Array)
+            strings.push(decoder.decode(item).trim());
+        else
+            return null;
+    }
+    return strings;
+};
+// Integer datasets (like int64 number_ion_types) can surface as BigInt
+// values or typed arrays depending on dtype; normalize to plain numbers.
+export const to_number_array = (data) => {
+    const values = Array.isArray(data)
+        ? data
+        : ArrayBuffer.isView(data)
+            ? Array.from(data)
+            : null;
+    if (!values)
+        return null;
+    const numbers = values.map((item) => (typeof item === `bigint` ? Number(item) : item));
+    return numbers.every((item) => typeof item === `number` && Number.isFinite(item))
+        ? numbers
+        : null;
+};
+// Scalar datasets surface as plain numbers, 1-element arrays, or BigInt
+// depending on dtype; normalize to a finite number or null.
+export const to_scalar_number = (data) => {
+    const value = Array.isArray(data) ? data[0] : data;
+    if (typeof value === `bigint`)
+        return Number(value);
+    return typeof value === `number` && Number.isFinite(value) ? value : null;
+};
+// Apply VASP's POSCAR-style universal scaling factor to lattice vectors
+export const scale_matrix = (matrix, scale) => scale === 1 ? matrix : matrix.map((row) => row.map((val) => val * scale));
+// Expand VASP's (ion_types, number_ion_types) pair into per-atom symbols,
+// e.g. ([Ga, Sb], [1, 1]) -> [Ga, Sb]. Throws on unknown symbols and
+// malformed counts so torn/corrupt HDF5 input fails loudly, not by dropping atoms.
+export const expand_ion_types = (ion_types, ion_counts) => {
+    if (ion_types.length !== ion_counts.length) {
+        throw new Error(`ion_types (${ion_types.length}) and ion_counts (${ion_counts.length}) length mismatch`);
+    }
+    const elements = [];
+    for (const [type_idx, symbol] of ion_types.entries()) {
+        if (!is_elem_symbol(symbol)) {
+            throw new Error(`Unknown element symbol in ion_types: ${symbol}`);
+        }
+        const ion_count = ion_counts[type_idx];
+        if (!Number.isFinite(ion_count) || !Number.isInteger(ion_count) || ion_count < 0) {
+            throw new Error(`Invalid ion count for ${symbol}: ${ion_count}`);
+        }
+        for (let count = 0; count < ion_count; count++)
+            elements.push(symbol);
+    }
+    return elements;
+};
+// Writes the buffer to h5wasm's in-memory FS under a unique temp name, opens
+// it, runs the callback (awaiting async ones so cleanup can't race reads),
+// and always closes + unlinks afterwards.
+export async function with_h5_file(buffer, filename, callback) {
+    const { FS } = await h5wasm.ready;
+    const file_basename = filename
+        ?.split(`/`)
+        .at(-1)
+        ?.replaceAll(/[^\w.-]/g, `_`) ?? `temp`;
+    const unique_suffix = `${Date.now()}-${Math.random().toString(36).slice(2)}`;
+    const temp_filename = `${file_basename}-${unique_suffix}.h5`;
+    FS.writeFile(temp_filename, new Uint8Array(buffer));
+    let h5_file = null;
+    try {
+        h5_file = new h5wasm.File(temp_filename, `r`);
+        return await callback(h5_file);
+    }
+    finally {
+        h5_file?.close();
+        try {
+            FS.unlink(temp_filename);
+        }
+        catch {
+            /* temp file cleanup is best-effort */
+        }
+    }
+}

package/dist/trajectory/parse/hdf5.d.ts

@@ -1,2 +1,2 @@
 import type { TrajectoryType } from '../index';
-export declare function parse_torch_sim_hdf5(buffer: ArrayBuffer, filename?: string): Promise<TrajectoryType>;
+export declare function parse_hdf5_trajectory(buffer: ArrayBuffer, filename?: string): Promise<TrajectoryType>;

package/dist/trajectory/parse/hdf5.js

@@ -1,78 +1,96 @@
-// HDF5 trajectory parsing (torch-sim / generic format)
+// HDF5 trajectory parsing: dispatches between VASP vaspout.h5 and the
+// torch-sim / generic dataset-alias layout after a single file open.
 import { calc_lattice_params, transpose_3x3_matrix } from '../../math';
-import * as h5wasm from 'h5wasm';
-import { convert_atomic_numbers, create_trajectory_frame, validate_3x3_matrix, } from '../helpers';
-const is_hdf5_dataset = (entity) => entity !== null && `to_array` in entity && entity instanceof h5wasm.Dataset;
-const is_hdf5_group = (entity) => entity !== null && `keys` in entity && entity instanceof h5wasm.Group;
-export async function parse_torch_sim_hdf5(buffer, filename) {
-    const { FS } = await h5wasm.ready;
-    const file_basename = filename
-        ?.split(`/`)
-        .at(-1)
-        ?.replaceAll(/[^\w.-]/g, `_`) ?? `temp`;
-    const unique_suffix = `${Date.now()}-${Math.random().toString(36).slice(2)}`;
-    const temp_filename = `${file_basename}-${unique_suffix}.h5`;
-    FS.writeFile(temp_filename, new Uint8Array(buffer));
-    let h5_file = null;
-    try {
-        h5_file = new h5wasm.File(temp_filename, `r`);
-        const found_paths = {};
-        let total_groups_found = 0;
-        const find_dataset = (names) => {
-            const discover = (parent, path = ``) => {
-                total_groups_found++;
-                for (const name of parent.keys()) {
-                    const item = parent.get(name);
-                    const full_path = path ? `${path}/${name}` : `/${name}`;
-                    if (names.includes(name) && is_hdf5_dataset(item)) {
-                        const found_name = names.find((dataset_name) => dataset_name === name);
-                        if (found_name)
-                            found_paths[found_name] = full_path;
-                        return item;
-                    }
-                    if (is_hdf5_group(item)) {
-                        const result = discover(item, full_path);
-                        if (result)
-                            return result;
-                    }
+import { convert_atomic_numbers, count_elements, create_trajectory_frame, validate_3x3_matrix, } from '../helpers';
+import { is_hdf5_dataset, is_hdf5_group, read_dataset, with_h5_file } from './h5-utils';
+import { is_vaspout_h5_file, parse_vaspout_h5_file } from './vaspout-h5';
+// Routes an opened HDF5 file to the right parser: vaspout.h5 has the VASP 6.x
+// root-group layout, everything else falls through to the torch-sim recursive
+// dataset-name search. One FS write + open, no parse-and-fallback.
+export async function parse_hdf5_trajectory(buffer, filename) {
+    return with_h5_file(buffer, filename, (h5_file) => is_vaspout_h5_file(h5_file)
+        ? parse_vaspout_h5_file(h5_file)
+        : parse_torch_sim_h5_file(h5_file));
+}
+// Dataset-name aliases searched (in order) across the whole group tree
+const POSITION_ALIASES = [`positions`, `coords`, `coordinates`];
+const ATOMIC_NUMBER_ALIASES = [`atomic_numbers`, `numbers`, `Z`, `species`];
+const CELL_ALIASES = [`cell`, `cells`, `lattice`];
+const ENERGY_ALIASES = [`potential_energy`, `energy`];
+function parse_torch_sim_h5_file(h5_file) {
+    const alias_groups = [POSITION_ALIASES, ATOMIC_NUMBER_ALIASES, CELL_ALIASES, ENERGY_ALIASES];
+    const all_aliases = new Set(alias_groups.flat());
+    const found_paths = {};
+    const found_alias_groups = new Set();
+    let total_groups_found = 0;
+    const discover = (parent, path = ``) => {
+        total_groups_found++;
+        for (const name of parent.keys()) {
+            const item = parent.get(name);
+            const full_path = path ? `${path}/${name}` : `/${name}`;
+            if (is_hdf5_dataset(item) && all_aliases.has(name)) {
+                found_paths[name] ??= full_path;
+                for (const [group_idx, aliases] of alias_groups.entries()) {
+                    if (aliases.includes(name))
+                        found_alias_groups.add(group_idx);
                 }
-                return null;
-            };
-            return discover(h5_file);
-        };
-        const positions_data = find_dataset([`positions`, `coords`, `coordinates`])?.to_array();
-        const atomic_numbers_data = find_dataset([
-            `atomic_numbers`,
-            `numbers`,
-            `Z`,
-            `species`,
-        ])?.to_array();
-        const cells_data = find_dataset([`cell`, `cells`, `lattice`])?.to_array();
-        const energies_data = find_dataset([`potential_energy`, `energy`])?.to_array();
-        if (!positions_data || !atomic_numbers_data) {
-            const missing_datasets = [];
-            if (!positions_data) {
-                missing_datasets.push(`positions (tried: positions, coords, coordinates)`);
+                if (found_alias_groups.size === alias_groups.length)
+                    return;
             }
-            if (!atomic_numbers_data) {
-                missing_datasets.push(`atomic numbers (tried: atomic_numbers, numbers, Z, species)`);
+            else if (is_hdf5_group(item)) {
+                discover(item, full_path);
+                if (found_alias_groups.size === alias_groups.length)
+                    return;
             }
-            const missing_str = missing_datasets.join(`, `);
-            const available_str = Array.from(h5_file.keys()).join(`, `);
-            throw new Error(`Missing required dataset(s) in HDF5 file: ${missing_str}. Available datasets: ${available_str}`);
         }
-        const positions_are_frames = positions_data.length > 0 &&
-            positions_data.every((entry) => Array.isArray(entry) && entry.every((coord) => Array.isArray(coord)));
-        const positions = positions_are_frames ? positions_data : [positions_data];
-        const atomic_numbers_are_frames = atomic_numbers_data.length > 0 &&
-            atomic_numbers_data.every((entry) => Array.isArray(entry));
-        const atomic_numbers = atomic_numbers_are_frames
-            ? atomic_numbers_data
-            : [atomic_numbers_data];
-        const frames = positions.map((frame_pos, idx) => {
+    };
+    discover(h5_file);
+    const first_path = (names) => names.map((name) => found_paths[name]).find(Boolean);
+    const find_dataset = (names) => {
+        const path = first_path(names);
+        return path ? read_dataset(h5_file, path) : null;
+    };
+    const positions_data = find_dataset(POSITION_ALIASES);
+    const atomic_numbers_data = find_dataset(ATOMIC_NUMBER_ALIASES);
+    const cells_data = find_dataset(CELL_ALIASES);
+    const energies_data = find_dataset(ENERGY_ALIASES);
+    if (!positions_data || !atomic_numbers_data) {
+        const missing_datasets = [];
+        if (!positions_data) {
+            missing_datasets.push(`positions (tried: ${POSITION_ALIASES.join(`, `)})`);
+        }
+        if (!atomic_numbers_data) {
+            missing_datasets.push(`atomic numbers (tried: ${ATOMIC_NUMBER_ALIASES.join(`, `)})`);
+        }
+        const missing_str = missing_datasets.join(`, `);
+        const available_str = [...h5_file.keys()].join(`, `);
+        throw new Error(`Missing required dataset(s) in HDF5 file: ${missing_str}. Available datasets: ${available_str}`);
+    }
+    const positions_are_frames = positions_data.length > 0 &&
+        positions_data.every((entry) => Array.isArray(entry) && entry.every((coord) => Array.isArray(coord)));
+    const positions = positions_are_frames ? positions_data : [positions_data];
+    const atomic_numbers_are_frames = atomic_numbers_data.length > 0 &&
+        atomic_numbers_data.every((entry) => Array.isArray(entry));
+    const atomic_numbers = atomic_numbers_are_frames
+        ? atomic_numbers_data
+        : [atomic_numbers_data];
+    // A constant cell may be stored as a single 3x3 matrix; normalize to a
+    // per-frame array so cells[idx] always yields a full lattice (indexing a bare
+    // 3x3 with a frame index would hand a single row to validate_3x3_matrix)
+    const cells_are_frames = cells_data?.every((entry) => Array.isArray(entry) && entry.every((row) => Array.isArray(row)));
+    const cells = cells_data
+        ? cells_are_frames
+            ? cells_data
+            : [cells_data]
+        : null;
+    const frames = [];
+    let dropped_steps = 0;
+    for (const [idx, frame_pos] of positions.entries()) {
+        try {
             const frame_atomic_numbers = atomic_numbers[idx] || atomic_numbers[0];
             const frame_elements = convert_atomic_numbers(frame_atomic_numbers);
-            const cell = cells_data?.[idx];
+            // constant-cell files reuse the single normalized cell for every frame
+            const cell = cells_are_frames ? cells?.[idx] : cells?.[0];
             const lattice_mat = cell ? transpose_3x3_matrix(validate_3x3_matrix(cell)) : undefined;
             const energy_entry = energies_data?.[idx];
             const energy = Array.isArray(energy_entry) ? energy_entry[0] : energy_entry;
@@ -83,45 +101,37 @@ export async function parse_torch_sim_hdf5(buffer, filename) {
                 metadata.volume = calc_lattice_params(lattice_mat).volume;
             }
             const pbc = lattice_mat ? [true, true, true] : [false, false, false];
-            return create_trajectory_frame(frame_pos, frame_elements, lattice_mat, pbc, idx, metadata);
-        });
-        const first_frame_elements = frames[0]?.structure.sites.map((site) => site.species[0].element) ?? [];
-        return {
-            frames,
-            metadata: {
-                source_format: `hdf5_trajectory`,
-                frame_count: frames.length,
-                num_atoms: first_frame_elements.length,
-                periodic_boundary_conditions: cells_data ? [true, true, true] : [false, false, false],
-                element_counts: first_frame_elements.reduce((counts, element) => {
-                    counts[element] = (counts[element] || 0) + 1;
-                    return counts;
-                }, {}),
-                discovered_datasets: {
-                    positions: found_paths.positions ||
-                        found_paths.coords ||
-                        found_paths.coordinates ||
-                        `unknown`,
-                    atomic_numbers: found_paths.atomic_numbers ||
-                        found_paths.numbers ||
-                        found_paths.Z ||
-                        found_paths.species ||
-                        `unknown`,
-                    cells: found_paths.cell || found_paths.cells || found_paths.lattice,
-                    energies: found_paths.potential_energy || found_paths.energy,
-                },
-                total_groups_found,
-                has_cell_info: Boolean(cells_data),
-            },
-        };
-    }
-    finally {
-        h5_file?.close();
-        try {
-            FS.unlink(temp_filename);
+            frames.push(create_trajectory_frame(frame_pos, frame_elements, lattice_mat, pbc, idx, metadata));
         }
-        catch {
-            /* temp file cleanup is best-effort */
+        catch (err) {
+            // Same torn-tail resiliency as the vaspout parser: interrupted writers
+            // zero-fill trailing chunks (atomic number 0, degenerate cells), which
+            // fails validation — keep the frames parsed so far and report the rest
+            // as dropped. A file whose very first frame is unparsable still throws.
+            if (frames.length === 0)
+                throw err;
+            dropped_steps = positions.length - idx;
+            break;
         }
     }
+    const first_frame_elements = frames[0]?.structure.sites.map((site) => site.species[0].element) ?? [];
+    return {
+        frames,
+        metadata: {
+            source_format: `hdf5_trajectory`,
+            frame_count: frames.length,
+            num_atoms: first_frame_elements.length,
+            periodic_boundary_conditions: cells_data ? [true, true, true] : [false, false, false],
+            element_counts: count_elements(first_frame_elements),
+            discovered_datasets: {
+                positions: first_path(POSITION_ALIASES) ?? `unknown`,
+                atomic_numbers: first_path(ATOMIC_NUMBER_ALIASES) ?? `unknown`,
+                cells: first_path(CELL_ALIASES),
+                energies: first_path(ENERGY_ALIASES),
+            },
+            total_groups_found,
+            has_cell_info: Boolean(cells_data),
+            ...(dropped_steps > 0 ? { dropped_steps } : {}),
+        },
+    };
 }

package/dist/trajectory/parse/index.d.ts

@@ -1,12 +1,12 @@
-import { is_trajectory_file } from '../format-detect';
-import { TrajFrameReader } from '../frame-reader';
-import type { FrameLoader, ParseProgress, TrajectoryType } from '../index';
+import type { ParseProgress, TrajectoryType } from '../index';
 import type { AtomTypeMapping, LoadingOptions } from '../types';
-export { LARGE_FILE_THRESHOLD, MAX_BIN_FILE_SIZE, MAX_TEXT_FILE_SIZE, } from '../constants';
+export declare const LARGE_FILE_THRESHOLD: number;
+export declare const MAX_BIN_FILE_SIZE: number;
+export declare const MAX_TEXT_FILE_SIZE: number;
 export type { AtomTypeMapping, LoadingOptions } from '../types';
-export { is_trajectory_file, TrajFrameReader };
+export { is_trajectory_file } from '../format-detect';
+export { TrajFrameReader } from '../frame-reader';
 export declare function parse_trajectory_data(data: unknown, filename?: string, atom_type_mapping?: AtomTypeMapping): Promise<TrajectoryType>;
 export declare function get_unsupported_format_message(filename: string, content: string): string | null;
 export declare function parse_trajectory_async(data: ArrayBuffer | string, filename: string, on_progress?: (progress: ParseProgress) => void, options?: LoadingOptions): Promise<TrajectoryType>;
-export declare function create_frame_loader(filename: string): FrameLoader;
 export declare function load_binary_traj(resp: Response, type: string, fallback?: boolean): Promise<ArrayBuffer | string>;

package/dist/trajectory/parse/index.js

@@ -1,15 +1,14 @@
 // Parsing functions for trajectory data from various formats
 import { is_binary } from '../../io/is-binary';
 import { is_plain_object } from '../../utils';
+import { DEFAULTS } from '../../settings';
 import { is_parsed_structure, parse_xyz } from '../../structure/parse';
-import { INDEX_SAMPLE_RATE, LARGE_FILE_THRESHOLD } from '../constants';
-import { strip_compression_extensions } from '../../io';
-import { ext_hint, FORMAT_PATTERNS, is_trajectory_file } from '../format-detect';
+import { ext_hint, FORMAT_PATTERNS, indexed_trajectory_format, is_indexable_trajectory_filename, } from '../format-detect';
 import { TrajFrameReader } from '../frame-reader';
 import { count_xyz_frames } from '../helpers';
 import { parse_ase_trajectory } from './ase';
 import { get_traj_parse_warnings, reset_traj_parse_warnings, traj_warn } from './diagnostics';
-import { parse_torch_sim_hdf5 } from './hdf5';
+import { parse_hdf5_trajectory } from './hdf5';
 import { parse_lammps_trajectory } from './lammps';
 import { parse_pymatgen_trajectory } from './pymatgen';
 import { parse_vasp_xdatcar } from './vasp';
@@ -21,16 +20,23 @@ const assert_frame_structure = (structure, label) => {
         throw new Error(`Invalid structure in ${context}: expected non-empty 'sites' array with species and coordinates`);
     }
 };
-// Re-export constants and types for consumers
-export { LARGE_FILE_THRESHOLD, MAX_BIN_FILE_SIZE, MAX_TEXT_FILE_SIZE, } from '../constants';
-export { is_trajectory_file, TrajFrameReader };
+// Constants for trajectory parsing and large file handling
+export const LARGE_FILE_THRESHOLD = 400 * 1024 * 1024; // 400MB
+const INDEX_SAMPLE_RATE = 100; // Default sample rate for frame indexing
+// Fallback thresholds for component usage without loading_options, derived from the
+// settings schema so settings-driven contexts (e.g. the VSCode extension) and direct
+// component use agree on when large-file/indexed loading kicks in.
+export const MAX_BIN_FILE_SIZE = DEFAULTS.trajectory.bin_file_threshold; // 50MB
+export const MAX_TEXT_FILE_SIZE = DEFAULTS.trajectory.text_file_threshold; // 25MB
+export { is_trajectory_file } from '../format-detect';
+export { TrajFrameReader } from '../frame-reader';
 export async function parse_trajectory_data(data, filename, atom_type_mapping) {
     reset_traj_parse_warnings();
     if (data instanceof ArrayBuffer) {
         if (FORMAT_PATTERNS.ase(data, filename))
             return parse_ase_trajectory(data, filename);
         if (FORMAT_PATTERNS.hdf5(data, filename))
-            return parse_torch_sim_hdf5(data, filename);
+            return parse_hdf5_trajectory(data, filename);
         throw new Error(`Unsupported binary format${filename ? `: ${filename}` : ``}`);
     }
     if (typeof data === `string`) {
@@ -45,19 +51,15 @@ export async function parse_trajectory_data(data, filename, atom_type_mapping) {
         }
         // Single XYZ fallback (content-sniffed when the filename gives no format hint,
         // e.g. blob: object URLs whose basenames are UUIDs)
-        const xyz_hint = ext_hint(filename, /\.(xyz|extxyz)$/);
+        const xyz_hint = ext_hint(filename, /\.(?:xyz|extxyz)$/);
         if (xyz_hint || (xyz_hint === null && count_xyz_frames(content) === 1)) {
-            try {
-                const structure = parse_xyz(content);
-                if (structure) {
-                    return {
-                        frames: [{ structure, step: 0, metadata: {} }],
-                        metadata: { source_format: `single_xyz`, frame_count: 1 },
-                    };
-                }
-            }
-            catch {
-                // Single-frame XYZ parsing failed, continue to JSON parsing.
+            // parse_xyz never throws (it returns null); fall through to JSON parsing.
+            const structure = parse_xyz(content);
+            if (structure) {
+                return {
+                    frames: [{ structure, step: 0, metadata: {} }],
+                    metadata: { source_format: `single_xyz`, frame_count: 1 },
+                };
             }
         }
         try {
@@ -157,10 +159,9 @@ export async function parse_trajectory_async(data, filename, on_progress, option
         // Use indexed loading for supported large files (including compressed names).
         // When the filename gives no format hint (e.g. blob: URLs), sniff a content
         // prefix for XYZ frames so large extensionless files still get indexed.
-        const base_filename = strip_compression_extensions(filename);
-        const can_index = /\.(xyz|extxyz|traj)$/.test(base_filename) ||
+        const can_index = is_indexable_trajectory_filename(filename) ||
             (typeof data === `string` &&
-                ext_hint(filename, /\.(xyz|extxyz)$/) === null &&
+                ext_hint(filename, /\.(?:xyz|extxyz)$/) === null &&
                 count_xyz_frames(data.slice(0, 2 ** 20)) >= 1);
         if (should_use_indexing && can_index) {
             return attach_parse_warnings(await parse_with_unified_loader(data, filename, { index_sample_rate, extract_plot_metadata }, on_progress));
@@ -216,9 +217,7 @@ async function parse_with_unified_loader(data, filename, options, on_progress) {
     }
     const stage = `Ready: ${total_frames} frames indexed`;
     on_progress?.({ current: 100, total: 100, stage });
-    const source_format = filename.toLowerCase().endsWith(`.traj`)
-        ? `ase_trajectory`
-        : `xyz_trajectory`;
+    const source_format = `${indexed_trajectory_format(filename)}_trajectory`;
     return {
         frames,
         metadata: { source_format, frame_count: total_frames },
@@ -229,13 +228,6 @@ async function parse_with_unified_loader(data, filename, options, on_progress) {
         frame_loader: loader,
     };
 }
-// Factory function for frame loader (simplified)
-export function create_frame_loader(filename) {
-    if (!/\.(xyz|extxyz|traj)$/.exec(filename.toLowerCase())) {
-        throw new Error(`Unsupported format for frame loading: ${filename}`);
-    }
-    return new TrajFrameReader(filename);
-}
 export async function load_binary_traj(resp, type, fallback = false) {
     try {
         // Read binary from a clone so the original can be used for text fallback

package/dist/trajectory/parse/lammps.js

@@ -1,7 +1,7 @@
 import { ELEM_SYMBOLS } from '../../labels';
 import * as math from '../../math';
-import { coerce_elem_symbol } from '../../element';
-import { create_trajectory_frame } from '../helpers';
+import { coerce_elem_symbol } from '../../element/helpers';
+import { count_elements, create_trajectory_frame } from '../helpers';
 import { traj_warn } from './diagnostics';
 const is_periodic = (token) => token.toLowerCase().startsWith(`p`);
 // Parse LAMMPS box bounds → lattice matrix. Handles orthogonal and triclinic boxes.
@@ -60,11 +60,11 @@ export function parse_lammps_trajectory(content, filename, atom_type_mapping) {
         if (!skip_to(`ITEM: TIMESTEP`))
             break;
         idx++;
-        const timestep = parseInt(read_line(), 10) || 0;
+        const timestep = Math.trunc(Number(read_line())) || 0;
         if (!skip_to(`ITEM: NUMBER OF ATOMS`))
             break;
         idx++;
-        const num_atoms = parseInt(read_line(), 10);
+        const num_atoms = Math.trunc(Number(read_line()));
         if (!num_atoms || num_atoms <= 0)
             continue;
         // BOX BOUNDS: orthogonal="pp pp pp", triclinic="xy xz yz pp pp pp"
@@ -111,7 +111,7 @@ export function parse_lammps_trajectory(content, filename, atom_type_mapping) {
         const frac_to_cart = use_scaled ? math.create_frac_to_cart(lattice_matrix) : null;
         for (let atom = 0; atom < num_atoms && idx < lines.length; atom++) {
             const parts = read_line().split(/\s+/);
-            const coords = pos_cols.map((col_idx) => parseFloat(parts[col_idx]));
+            const coords = pos_cols.map((col_idx) => Number(parts[col_idx]));
             if (coords.some(isNaN) || parts.length <= max_col_idx)
                 continue;
             // Convert scaled coordinates to Cartesian if needed
@@ -119,7 +119,7 @@ export function parse_lammps_trajectory(content, filename, atom_type_mapping) {
             let element_symbol;
             if (type_col !== undefined) {
                 // Map atom type to element using custom mapping or default (type 1 -> H, etc.)
-                const atom_type = parseInt(parts[type_col], 10) || 1;
+                const atom_type = Math.trunc(Number(parts[type_col])) || 1;
                 atom_types_found.add(atom_type);
                 element_symbol = get_element(atom_type);
             }
@@ -134,7 +134,7 @@ export function parse_lammps_trajectory(content, filename, atom_type_mapping) {
                 }
             }
             else if (id_col !== undefined) {
-                const atom_id = parseInt(parts[id_col], 10) || 1;
+                const atom_id = Math.trunc(Number(parts[id_col])) || 1;
                 atom_types_found.add(atom_id);
                 if (!id_fallback_warning_emitted) {
                     traj_warn(`LAMMPS parser fallback: mapping atom IDs to elements from ID column; this may be incorrect for large or sequential IDs. Prefer a TYPE column when available.`);
@@ -159,11 +159,7 @@ export function parse_lammps_trajectory(content, filename, atom_type_mapping) {
         throw new Error(`No valid frames found in LAMMPS trajectory`);
     }
     const first_frame = frames[0];
-    const element_counts = first_frame.structure.sites.reduce((counts, site) => {
-        const elem = site.species[0].element;
-        counts[elem] = (counts[elem] || 0) + 1;
-        return counts;
-    }, {});
+    const element_counts = count_elements(first_frame.structure.sites.map((site) => site.species[0].element));
     return {
         frames,
         metadata: {

package/dist/trajectory/parse/vasp.js

@@ -1,14 +1,15 @@
 import * as math from '../../math';
-import { is_elem_symbol } from '../../element';
+import { is_elem_symbol } from '../../element/helpers';
 import { create_trajectory_frame, validate_3x3_matrix } from '../helpers';
+import { parse_leading_num } from '../../utils';
 // Parse the 7-line XDATCAR header at lines[start]: title, scale factor, 3 lattice rows
 // (multiplied by scale), element names, element counts
 function parse_xdatcar_header(lines, start) {
-    const scale = parseFloat(lines[start + 1]);
+    const scale = parse_leading_num(lines[start + 1]);
     const rows = lines.slice(start + 2, start + 5).map((line) => line
         .trim()
         .split(/\s+/)
-        .map((val) => parseFloat(val) * scale));
+        .map((val) => Number(val) * scale));
     const names = lines[start + 5].trim().split(/\s+/);
     const counts = lines[start + 6].trim().split(/\s+/).map(Number);
     return { scale, rows, names, counts };
@@ -31,7 +32,9 @@ export function parse_vasp_xdatcar(content, filename) {
     const bad_element = element_names.find((name) => !is_elem_symbol(name));
     if (bad_element)
         throw new Error(`Invalid element symbol in XDATCAR: ${bad_element}`);
-    let elements = element_names.flatMap((name, idx) => Array(element_counts[idx]).fill(name));
+    // "Na Cl" + [2, 2] -> [Na, Na, Cl, Cl]
+    const expand_element_counts = (names, counts) => names.flatMap((name, idx) => Array(counts[idx]).fill(name));
+    let elements = expand_element_counts(element_names, element_counts);
     const frames = [];
     let line_idx = 7;
     let frac_to_cart = math.create_frac_to_cart(lattice_matrix);
@@ -49,14 +52,14 @@ export function parse_vasp_xdatcar(content, filename) {
                 if (hdr.names.length === hdr.counts.length &&
                     hdr.names.every(is_elem_symbol) &&
                     hdr.counts.every((count) => Number.isInteger(count) && count > 0)) {
-                    elements = hdr.names.flatMap((name, idx) => Array(hdr.counts[idx]).fill(name));
+                    elements = expand_element_counts(hdr.names, hdr.counts);
                 }
             }
         }
         const config_line = lines[config_idx];
         line_idx = config_idx + 1;
-        const step_match = /configuration=\s*(\d+)/.exec(config_line);
-        const step = step_match ? parseInt(step_match[1], 10) : frames.length + 1;
+        const step_match = /configuration=\s*(?<step>\d+)/.exec(config_line);
+        const step = step_match ? Math.trunc(Number(step_match[1])) : frames.length + 1;
         const positions = [];
         for (let idx = 0; idx < elements.length && line_idx < lines.length; idx++) {
             const coords = lines[line_idx].trim().split(/\s+/).slice(0, 3).map(Number);

package/dist/trajectory/parse/vaspout-electronic.d.ts

@@ -0,0 +1,13 @@
+import type { ElectronicBandStructure, ElectronicDos } from '../../spectral';
+import type * as h5wasm from 'h5wasm';
+export interface VaspoutElectronicData {
+    dos: ElectronicDos | null;
+    bands: ElectronicBandStructure | null;
+}
+export declare const read_vaspout_dos: (h5_file: h5wasm.File) => ElectronicDos | null;
+export declare const line_mode_labels: (labels: string[] | null, per_segment_dataset: number | null, n_kpoints: number) => {
+    labels: (string | null)[];
+    per_segment: number;
+} | null;
+export declare const read_vaspout_bands: (h5_file: h5wasm.File, dos?: ElectronicDos | null) => ElectronicBandStructure | null;
+export declare const read_vaspout_electronic: (h5_file: h5wasm.File) => VaspoutElectronicData | null;