matterviz 0.3.5 → 0.3.6

package/dist/src/lib/trajectory/TrajectoryInfoPane.svelte

@@ -0,0 +1,313 @@
+<script lang="ts">
+  import type { InfoItem } from '$lib/layout'
+  import InfoPaneCards from '$lib/overlays/InfoPaneCards.svelte'
+  import { format_num } from '$lib/labels'
+  import DraggablePane from '$lib/overlays/DraggablePane.svelte'
+  import { get_electro_neg_formula } from '$lib/composition'
+  import { SETTINGS_CONFIG } from '$lib/settings'
+  import type { AnyStructure } from '$lib/structure'
+  import type { ComponentProps } from 'svelte'
+  import type { TrajectoryType } from './index'
+
+  let {
+    trajectory,
+    current_step_idx,
+    current_filename,
+    current_file_path,
+    file_size,
+    file_object,
+    pane_open = $bindable(false),
+    toggle_props,
+    pane_props,
+    ...rest
+  }: Omit<ComponentProps<typeof DraggablePane>, `children`> & {
+    trajectory: TrajectoryType
+    current_step_idx: number
+    current_filename?: string | null
+    current_file_path?: string | null
+    file_size?: number | null
+    file_object?: File | null
+    pane_open?: boolean
+    toggle_props?: ComponentProps<typeof DraggablePane>[`toggle_props`]
+    pane_props?: ComponentProps<typeof DraggablePane>[`pane_props`]
+  } = $props()
+
+  // Helper functions
+  const format_size = (bytes: number) =>
+    bytes > 1024 * 1024
+      ? `${format_num(bytes / (1024 * 1024), `.2~f`)} MB`
+      : `${format_num(bytes / 1024, `.2~f`)} KB`
+
+  const is_valid_number = (val: unknown): val is number =>
+    typeof val === `number` && isFinite(val)
+
+  const extract_numeric_array = (frames: typeof trajectory.frames, prop: string) =>
+    frames.map((frame) => frame.metadata?.[prop]).filter(is_valid_number)
+
+  const format_range = (values: number[], unit = ``, decimals = `.2~f`) => {
+    if (!values.length) return null
+    if (values.length === 1) {
+      return `${format_num(values[0], decimals)} ${unit}`.trim()
+    }
+    const [min, max] = [Math.min(...values), Math.max(...values)]
+    return `${format_num(min, decimals)} - ${format_num(max, decimals)} ${unit}`
+      .trim()
+  }
+
+  const safe_item = (
+    label: string,
+    value: string | null,
+    key?: string,
+    tooltip?: string,
+  ): InfoItem | null => value ? { label, value, key, tooltip } : null
+
+  const is_info_item = (item: unknown): item is InfoItem => Boolean(item)
+
+  const safe_formula = (structure: AnyStructure) => {
+    try {
+      return get_electro_neg_formula(structure)
+    } catch {
+      return null
+    }
+  }
+
+  // Get trajectory info organized by sections
+  let info_pane_data = $derived.by(() => {
+    if (
+      (!trajectory?.frames?.length && !trajectory?.total_frames) ||
+      current_step_idx < 0 ||
+      current_step_idx >= (trajectory.total_frames ?? trajectory.frames?.length ?? 0)
+    ) return []
+
+    // For indexed trajectories, we might not have the current frame loaded
+    const current_frame = trajectory.frames?.[current_step_idx]
+    const total_frames = trajectory.total_frames ?? trajectory.frames?.length ?? 0
+
+    const sections: { title: string; items: InfoItem[] }[] = []
+
+    // File info section
+    const file_items = [
+      current_filename &&
+      safe_item(
+        `Name`,
+        current_filename,
+        `file-name`,
+        current_file_path || undefined,
+      ),
+      file_size && file_size > 0 &&
+      safe_item(`File Size`, format_size(file_size), `file-size`),
+      file_object?.lastModified &&
+      safe_item(
+        `Modified`,
+        new Date(file_object.lastModified).toLocaleString(),
+        `file-modified`,
+      ),
+      trajectory.metadata?.source_format &&
+      safe_item(`Format`, String(trajectory.metadata.source_format), `file-format`),
+    ].filter(is_info_item)
+
+    if (file_items.length > 0) {
+      sections.push({ title: `File`, items: file_items })
+    }
+
+    // Trajectory info section (always show this)
+    const traj_items = [
+      safe_item(
+        `Total Frames`,
+        `${format_num(total_frames, `.3~s`)} (current: ${
+          format_num(current_step_idx + 1, `.3~s`)
+        })`,
+        `total-frames`,
+      ),
+      trajectory.is_indexed &&
+      safe_item(
+        `Indexed`,
+        `Yes`,
+        `indexed-mode`,
+        SETTINGS_CONFIG.trajectory.use_indexing.description,
+      ),
+      trajectory.indexed_frames &&
+      safe_item(
+        `Index Points`,
+        `${trajectory.indexed_frames.length}`,
+        `index-points`,
+        `Number of frames indexed for fast seeking`,
+      ),
+      trajectory.plot_metadata &&
+      safe_item(
+        `Plot Metadata`,
+        `${trajectory.plot_metadata.length} frames`,
+        `plot-metadata`,
+        `Pre-extracted metadata for plotting`,
+      ),
+    ].filter(is_info_item)
+
+    if (traj_items.length > 0) {
+      sections.push({ title: `Trajectory`, items: traj_items })
+    }
+
+    // Structure info section (only if we have the current frame)
+    if (current_frame?.structure?.sites) {
+      const { structure } = current_frame
+      const lattice = `lattice` in structure ? structure.lattice : null
+      const { volume, a, b, c, alpha, beta, gamma } = lattice || {}
+      const formula = safe_formula(structure)
+
+      const structure_items = [
+        safe_item(`Atoms`, `${structure.sites.length}`, `atoms`),
+        formula && safe_item(`Formula`, String(formula), `formula`),
+        is_valid_number(volume) && volume > 0 &&
+        safe_item(`Volume`, `${format_num(volume, `.3~s`)} ų`, `volume`),
+        is_valid_number(volume) && volume > 0 && structure.sites.length > 0 &&
+        safe_item(
+          `Density`,
+          `${format_num(structure.sites.length / volume, `.4~s`)} atoms/ų`,
+          `density`,
+        ),
+        [a, b, c].every(is_valid_number) &&
+        safe_item(
+          `Cell Lengths`,
+          `${format_num(a as number, `.3~f`)}, ${format_num(b as number, `.3~f`)}, ${
+            format_num(c as number, `.3~f`)
+          } Å`,
+          `cell-lengths`,
+        ),
+        [alpha, beta, gamma].every(is_valid_number) &&
+        safe_item(
+          `Cell Angles`,
+          `${format_num(alpha as number, `.2~f`)}°, ${
+            format_num(beta as number, `.2~f`)
+          }°, ${format_num(gamma as number, `.2~f`)}°`,
+          `cell-angles`,
+        ),
+      ].filter(is_info_item)
+
+      if (structure_items.length > 0) {
+        sections.push({ title: `Structure`, items: structure_items })
+      }
+    } else if (trajectory.is_indexed) {
+      // For indexed trajectories, show a note that frame data is loaded on demand
+      const structure_items = [
+        safe_item(
+          `Frame Loading`,
+          `On-demand`,
+          `frame-loading`,
+          `Structure data loaded when frame is accessed`,
+        ),
+      ].filter(is_info_item)
+
+      if (structure_items.length > 0) {
+        sections.push({ title: `Structure`, items: structure_items })
+      }
+    }
+
+    // Energy section (only for regular trajectories with multiple frames)
+    if (!trajectory.is_indexed && trajectory.frames.length > 1) {
+      const energies = extract_numeric_array(trajectory.frames, `energy`)
+      if (energies.length > 1) {
+        const current_energy = current_frame?.metadata?.energy
+        const energy_range = format_range(energies, `eV`, `.3~s`)
+
+        const energy_items = [
+          is_valid_number(current_energy) &&
+          safe_item(
+            `Current Energy`,
+            `${format_num(current_energy, `.3~s`)} eV`,
+            `energy-current`,
+          ),
+          energy_range && safe_item(`Energy Range`, energy_range, `energy-range`),
+        ].filter(is_info_item)
+
+        if (energy_items.length > 0) {
+          sections.push({ title: `Energy`, items: energy_items })
+        }
+      }
+    }
+
+    // Forces section (only for regular trajectories with multiple frames)
+    if (!trajectory.is_indexed && trajectory.frames.length > 1) {
+      const forces = extract_numeric_array(trajectory.frames, `force_max`)
+      if (forces.length > 1) {
+        const current_force = current_frame?.metadata?.force_max
+        const force_range = format_range(forces, `eV/Å`, `.3~s`)
+
+        const force_items = [
+          is_valid_number(current_force) &&
+          safe_item(
+            `Max Force`,
+            `${format_num(current_force, `.3~s`)} eV/Å`,
+            `force-current`,
+          ),
+          force_range && safe_item(`Force Range`, force_range, `force-range`),
+        ].filter(is_info_item)
+
+        if (force_items.length > 0) {
+          sections.push({ title: `Forces`, items: force_items })
+        }
+      }
+    }
+
+    // Volume change section (only for regular trajectories)
+    if (
+      !trajectory.is_indexed && current_frame?.structure &&
+      trajectory.frames.length > 1
+    ) {
+      const lattice = `lattice` in current_frame.structure
+        ? current_frame.structure.lattice
+        : null
+      if (lattice) {
+        const volumes = trajectory.frames.map((
+          { structure },
+        ) => (`lattice` in structure && structure.lattice?.volume))
+          .filter(is_valid_number)
+          .filter((volume) => volume > 0)
+
+        if (volumes.length > 1) {
+          const vol_change = (Math.max(...volumes) - Math.min(...volumes)) /
+            Math.min(...volumes)
+          if (Math.abs(vol_change) > 0.1 && is_valid_number(vol_change)) {
+            const vol_items = [safe_item(
+              `Volume Change`,
+              `${format_num(vol_change, `.2~%`)}`,
+              `vol-change`,
+            )].filter(is_info_item)
+
+            if (vol_items.length > 0) {
+              sections.push({ title: `Volume`, items: vol_items })
+            }
+          }
+        }
+      }
+    }
+    return sections
+  })
+
+  let info_cards = $derived(
+    info_pane_data.map(({ title, items }) => ({ title, rows: items })),
+  )
+  let n_info_items = $derived(
+    info_pane_data.reduce((count, { items }) => count + items.length, 0),
+  )
+</script>
+
+<DraggablePane
+  bind:show={pane_open}
+  max_width="24em"
+  toggle_props={{
+    title: pane_open ? `` : `Trajectory info`,
+    ...toggle_props,
+    class: `trajectory-info-toggle ${toggle_props?.class ?? ``}`,
+  }}
+  open_icon="Cross"
+  closed_icon="Info"
+  pane_props={{ ...pane_props, class: `trajectory-info-pane ${pane_props?.class ?? ``}` }}
+  {...rest}
+>
+  <h4 style="margin-top: 0">Trajectory Info</h4>
+  <InfoPaneCards
+    cards={info_cards}
+    filter_placeholder="Filter trajectory info"
+    empty_label="trajectory info"
+    show_filter={n_info_items > 5}
+  />
+</DraggablePane>

package/dist/src/lib/trajectory/constants.ts

@@ -0,0 +1,7 @@
+// Constants for trajectory parsing and large file handling
+export const MAX_SAFE_STRING_LENGTH = 0x1fffffe8 * 0.5 // 50% of JS max string length as safety
+export const MAX_METADATA_SIZE = 50 * 1024 * 1024 // 50MB limit for metadata
+export const LARGE_FILE_THRESHOLD = 400 * 1024 * 1024 // 400MB
+export const INDEX_SAMPLE_RATE = 100 // Default sample rate for frame indexing
+export const MAX_BIN_FILE_SIZE = 100 * 1024 * 1024 // 100MB default for ArrayBuffer files
+export const MAX_TEXT_FILE_SIZE = 50 * 1024 * 1024 // 50MB default for string files

package/dist/src/lib/trajectory/extract.ts

@@ -0,0 +1,196 @@
+// Data extraction functions for trajectory analysis and plotting
+import { get_density } from '$lib/structure/index'
+import type { TrajectoryDataExtractor, TrajectoryFrame, TrajectoryType } from './index'
+
+// Common data extractor that extracts energy and structural properties
+export const energy_data_extractor: TrajectoryDataExtractor = (
+  frame: TrajectoryFrame,
+): Record<string, number> => {
+  const data: Record<string, number> = {
+    Step: frame.step,
+  }
+
+  if (frame.metadata) {
+    // Extract energy-related properties
+    const energy_fields = [
+      `energy`,
+      `energy_per_atom`,
+      `potential_energy`,
+      `kinetic_energy`,
+      `total_energy`,
+    ]
+
+    for (const field of energy_fields) {
+      if (field in frame.metadata && typeof frame.metadata[field] === `number`) {
+        data[field] = frame.metadata[field]
+      }
+    }
+  }
+
+  return data
+}
+
+// Data extractor for forces and stresses
+export const force_stress_data_extractor: TrajectoryDataExtractor = (
+  frame: TrajectoryFrame,
+): Record<string, number> => {
+  const data: Record<string, number> = {
+    Step: frame.step,
+  }
+
+  if (frame.metadata) {
+    // Calculate force properties from forces array if available (preferred)
+    if (frame.metadata.forces && Array.isArray(frame.metadata.forces)) {
+      const forces = frame.metadata.forces as number[][]
+      if (forces.length > 0) {
+        const force_magnitudes = forces.map((force) => Math.hypot(...force))
+        data.force_max = Math.max(...force_magnitudes)
+        // Calculate RMS (root mean square) of force magnitudes
+        data.force_norm = Math.sqrt(
+          force_magnitudes.reduce((sum, f) => sum + f ** 2, 0) / force_magnitudes.length,
+        )
+      }
+    } else {
+      // Fallback to metadata values if forces array not available
+      if (frame.metadata.force_max && typeof frame.metadata.force_max === `number`) {
+        data.force_max = frame.metadata.force_max
+      }
+      // Prefer force_norm if available, fall back to force_rms
+      if (frame.metadata.force_norm && typeof frame.metadata.force_norm === `number`) {
+        data.force_norm = frame.metadata.force_norm
+      } else if (frame.metadata.force_rms && typeof frame.metadata.force_rms === `number`) {
+        data.force_norm = frame.metadata.force_rms // Use force_rms as fallback
+      }
+    }
+
+    // Extract other stress and pressure properties (no duplicates expected)
+    const other_stress_fields = [`stress_max`, `stress_frobenius`, `stress_trace`, `pressure`]
+    for (const field of other_stress_fields) {
+      if (field in frame.metadata && typeof frame.metadata[field] === `number`) {
+        data[field] = frame.metadata[field]
+      }
+    }
+  }
+
+  return data
+}
+
+// Data extractor for structural properties
+export const structural_data_extractor: TrajectoryDataExtractor = (
+  frame: TrajectoryFrame,
+): Record<string, number> => {
+  const data: Record<string, number> = {
+    Step: frame.step,
+  }
+
+  // Extract lattice properties (preferred source for volume)
+  if (`lattice` in frame.structure) {
+    const lattice = frame.structure.lattice
+    data.volume = lattice.volume // Use consistent lowercase naming
+    data.a = lattice.a
+    data.b = lattice.b
+    data.c = lattice.c
+    data.alpha = lattice.alpha
+    data.beta = lattice.beta
+    data.gamma = lattice.gamma
+  }
+
+  if (frame.metadata) {
+    // Extract other structural properties, avoiding volume duplicate
+    const structural_fields = [`temperature`]
+
+    for (const field of structural_fields) {
+      if (field in frame.metadata && typeof frame.metadata[field] === `number`)
+        data[field] = frame.metadata[field]
+    }
+
+    // Handle density separately - prefer metadata, but calculate if not available
+    if (frame.metadata.density && typeof frame.metadata.density === `number`) {
+      data.density = frame.metadata.density
+    } else if (`lattice` in frame.structure) {
+      try {
+        data.density = get_density(frame.structure)
+      } catch (error) {
+        console.warn(`Failed to calculate density for frame ${frame.step}:`, error)
+      }
+    }
+
+    // Only use metadata volume if lattice volume is not available
+    if (!data.volume && frame.metadata.volume && typeof frame.metadata.volume === `number`) {
+      data.volume = frame.metadata.volume
+    }
+
+    // Note: pressure is handled by force_stress_data_extractor to avoid duplication
+  } else if (`lattice` in frame.structure) {
+    // Calculate density even when no metadata is available
+    try {
+      data.density = get_density(frame.structure)
+    } catch (error) {
+      console.warn(`Failed to calculate density for frame ${frame.step}:`, error)
+    }
+  }
+
+  return data
+}
+
+// Helper function to check if a property varies across trajectory frames
+function property_varies(
+  trajectory: TrajectoryType,
+  property_key: string,
+  tolerance = 1e-10,
+): boolean {
+  if (trajectory.frames.length <= 1) return false
+
+  const values: number[] = []
+  for (const frame of trajectory.frames) {
+    // Check both direct structure properties and metadata
+    let value: number | undefined
+
+    if (`lattice` in frame.structure) {
+      const lattice_value = (frame.structure.lattice as Record<string, unknown>)[property_key]
+      if (typeof lattice_value === `number`) value = lattice_value
+    }
+
+    if (value === undefined && frame.metadata && property_key in frame.metadata) {
+      const metadata_value = frame.metadata[property_key]
+      if (typeof metadata_value === `number`) {
+        value = metadata_value
+      }
+    }
+
+    if (value !== undefined) {
+      values.push(value)
+    }
+  }
+
+  if (values.length <= 1) return false
+
+  const first_value = values[0]
+  return values.some((value) => Math.abs(value - first_value) > tolerance)
+}
+
+// Combined data extractor that extracts all common properties
+export const full_data_extractor: TrajectoryDataExtractor = (
+  frame: TrajectoryFrame,
+  trajectory: TrajectoryType,
+): Record<string, number> => {
+  const base_data = {
+    ...energy_data_extractor(frame, trajectory),
+    ...force_stress_data_extractor(frame, trajectory),
+    ...structural_data_extractor(frame, trajectory),
+  }
+
+  // Check which lattice parameters vary
+  const lattice_params = [`a`, `b`, `c`, `alpha`, `beta`, `gamma`]
+  const result = { ...base_data }
+
+  // Add metadata to specify which properties don't vary
+  for (const param of lattice_params) {
+    if (!property_varies(trajectory, param)) {
+      // Mark individual lattice parameters as constant
+      result[`constant_${param}`] = 1
+    }
+  }
+
+  return result
+}

package/dist/src/lib/trajectory/format-detect.ts

@@ -0,0 +1,96 @@
+// Format detection for trajectory files
+import {
+  COMPRESSION_EXTENSIONS_REGEX,
+  CONFIG_DIRS_REGEX,
+  MD_SIM_EXCLUDE_REGEX,
+  TRAJ_EXTENSIONS_REGEX,
+  TRAJ_FALLBACK_EXTENSIONS_REGEX,
+  TRAJ_KEYWORDS_SIMPLE_REGEX,
+  XDATCAR_REGEX,
+} from '$lib/constants'
+import { count_xyz_frames } from './helpers'
+
+export function strip_compression_extensions(filename: string): string {
+  let base_name = filename.toLowerCase()
+  while (COMPRESSION_EXTENSIONS_REGEX.test(base_name)) {
+    base_name = base_name.replace(COMPRESSION_EXTENSIONS_REGEX, ``)
+  }
+  return base_name
+}
+
+// Unified format detection
+export const FORMAT_PATTERNS = {
+  ase: (data: unknown, filename?: string) => {
+    const base_name = filename ? strip_compression_extensions(filename) : undefined
+    if (!base_name?.endsWith(`.traj`) || !(data instanceof ArrayBuffer)) {
+      return false
+    }
+    const view = new Uint8Array(data.slice(0, 24))
+    return [0x2d, 0x20, 0x6f, 0x66, 0x20, 0x55, 0x6c, 0x6d].every(
+      (byte, idx) => view[idx] === byte,
+    )
+  },
+
+  hdf5: (data: unknown, filename?: string) => {
+    const base_name = filename ? strip_compression_extensions(filename) : undefined
+    const has_ext = base_name?.match(/\.(h5|hdf5)$/)
+    if (!has_ext || !(data instanceof ArrayBuffer) || data.byteLength < 8) return false
+    const signature = new Uint8Array(data.slice(0, 8))
+    return [0x89, 0x48, 0x44, 0x46, 0x0d, 0x0a, 0x1a, 0x0a].every(
+      (b, idx) => signature[idx] === b,
+    )
+  },
+
+  vasp: (data: string, filename?: string) => {
+    const basename = filename?.toLowerCase().split(`/`).pop() || ``
+    if (basename === `xdatcar` || basename.startsWith(`xdatcar`)) return true
+    const lines = data.trim().split(/\r?\n/)
+    return (
+      lines.length >= 10 &&
+      lines.some((line) => line.includes(`Direct configuration=`)) &&
+      !isNaN(parseFloat(lines[1])) &&
+      lines.slice(2, 5).every((line) => line.trim().split(/\s+/).length === 3)
+    )
+  },
+
+  xyz_multi: (data: string, filename?: string) => {
+    const base = filename ? strip_compression_extensions(filename) : ``
+    if (!/\.(xyz|extxyz)$/.test(base)) return false
+    return count_xyz_frames(data) >= 2
+  },
+
+  lammpstrj: (data: string, filename?: string) => {
+    const base = filename ? strip_compression_extensions(filename) : ``
+    if (!base.endsWith(`.lammpstrj`)) return false
+    return data.includes(`ITEM: TIMESTEP`) && data.includes(`ITEM: ATOMS`)
+  },
+} as const
+
+// Check if file is a trajectory (supports both filename-only and content-based detection)
+export function is_trajectory_file(filename: string, content?: string): boolean {
+  if (CONFIG_DIRS_REGEX.test(filename)) return false
+  const base_name = strip_compression_extensions(filename)
+
+  // For xyz/extxyz files, use content-based detection if available
+  if (/\.(xyz|extxyz)$/i.test(base_name)) {
+    if (content) return count_xyz_frames(content) >= 2
+    return TRAJ_KEYWORDS_SIMPLE_REGEX.test(base_name)
+  }
+
+  // Always detect these specific trajectory formats
+  if (TRAJ_EXTENSIONS_REGEX.test(base_name) || XDATCAR_REGEX.test(base_name)) return true
+
+  // Special exclusion for generic md_simulation pattern with certain extensions
+  if (MD_SIM_EXCLUDE_REGEX.test(base_name)) return false
+
+  // For .h5/.hdf5 files, require trajectory keywords
+  if (/\.(h5|hdf5)$/i.test(base_name)) {
+    return TRAJ_KEYWORDS_SIMPLE_REGEX.test(base_name)
+  }
+
+  // For other extensions, require both keywords and specific extensions
+  return (
+    TRAJ_KEYWORDS_SIMPLE_REGEX.test(base_name) &&
+    TRAJ_FALLBACK_EXTENSIONS_REGEX.test(base_name)
+  )
+}