Add custom words to the `en_US` dictionary (#24108)

GitOrigin-RevId: 44587567340b4a6ff357f6a321055554a800d799

services/web/frontend/js/features/source-editor/hunspell/HunspellManager.ts

@@ -149,7 +149,7 @@ export class HunspellManager {
       this.hunspellWorker.postMessage({
         type: 'init',
         lang: this.language,
-        learnedWords: this.learnedWords, // TODO: add words
+        learnedWords: this.learnedWords,
         baseAssetPath: this.baseAssetPath,
         dictionariesRoot: this.dictionariesRoot,
       })

services/web/frontend/js/features/source-editor/hunspell/hunspell.worker.ts

@@ -1,5 +1,6 @@
 import Hunspell from './wasm/hunspell'
 import hunspellWasmPath from './wasm/hunspell.wasm'
+import { buildAdditionalDictionary } from './wordlists/dictionary-additions'
 
 type SpellChecker = {
   spell(words: string[]): { index: number }[]
@@ -52,6 +53,7 @@ const createSpellChecker = async ({
     'number',
     'number',
   ])
+  const addDic = cwrap('Hunspell_add_dic', 'number', ['number', 'number'])
   const addWord = cwrap('Hunspell_add', 'number', ['number', 'number'])
   const removeWord = cwrap('Hunspell_remove', 'number', ['number', 'number'])
   const freeList = cwrap('Hunspell_free_list', 'number', [
@@ -81,11 +83,13 @@ const createSpellChecker = async ({
   const affPtr = stringToNewUTF8('/dictionaries/index.aff')
   const spellPtr = create(affPtr, dicPtr)
 
-  for (const word of learnedWords) {
-    const wordPtr = stringToNewUTF8(word)
-    addWord(spellPtr, wordPtr)
-    _free(wordPtr)
-  }
+  FS.writeFile(
+    '/dictionaries/extra.dic',
+    await buildAdditionalDictionary(lang, learnedWords)
+  )
+  const extraDicPtr = stringToNewUTF8('/dictionaries/extra.dic')
+  addDic(spellPtr, extraDicPtr)
+  _free(extraDicPtr)
 
   const spellChecker: SpellChecker = {
     spell(words) {

services/web/frontend/js/features/source-editor/hunspell/wasm/hunspell.d.ts

@@ -25,6 +25,12 @@ declare class Hunspell {
     input: string[]
   ): (spellPtr: number, suggestionListPtr: number, wordPtr: number) => number
 
+  cwrap(
+    method: 'Hunspell_add_dic',
+    output: string,
+    input: string[]
+  ): (spellPtr: number, wordPtr: number) => number
+
   cwrap(
     method: 'Hunspell_add',
     output: string,

services/web/frontend/js/features/source-editor/hunspell/wordlists/dictionary-additions.ts

@@ -0,0 +1,18 @@
+const dictionaryAdditions = new Set(['en_US'])
+
+export const buildAdditionalDictionary = async (
+  lang: string,
+  learnedWords: string[]
+) => {
+  const words = [...learnedWords]
+
+  if (dictionaryAdditions.has(lang)) {
+    const wordList = await import(`./${lang}.txt`).then(m => m.default)
+    words.push(...wordList.split('\n').filter(Boolean))
+  }
+
+  // the first line contains the approximate word count
+  words.unshift(String(words.length))
+
+  return new TextEncoder().encode(words.join('\n'))
+}

services/web/frontend/js/features/source-editor/hunspell/wordlists/en_US.txt

@@ -0,0 +1,535 @@
+Abaqus
+abelian
+ACF
+ACS
+AdaBoost
+AFM
+AGN
+AIC
+AlexNet
+AMR
+anharmonicity
+ANNs
+ANOVA
+ANSYS
+apriori
+arcsec
+ARIMA
+ARMA
+arXiv
+ASIC
+ASR
+asymptotics
+AUC
+autoencoder
+autoencoders
+AWGN
+axion
+backend
+backhaul
+backpropagation
+Balmer
+beamformer
+beamforming
+beamline
+BER
+BFGS
+BiLSTM
+binarization
+binarized
+Biot
+bistability
+BLEU
+BMI
+Bogoliubov
+Bonferroni
+Boussinesq
+Cartan
+cartesian
+casted
+CDF
+CDM
+centerline
+CFD
+CFL
+Chebyshev
+Cholesky
+CIFAR
+CMB
+CNC
+CNF
+Colab
+collisionless
+comorbidities
+compactification
+COMSOL
+confounders
+CoV
+CRF
+crosslinking
+CSI
+CubeSat
+CUDA
+cumulant
+cumulants
+CVD
+DAC
+DAQ
+datacenter
+datasheet
+DBSCAN
+DCT
+debiasing
+Debye
+deformability
+delocalization
+denoise
+denoised
+denoising
+DenseNet
+densification
+dephasing
+depthwise
+detrended
+deviatoric
+DevOps
+DFS
+DFT
+dichalcogenides
+dimerization
+discretizations
+discriminability
+dispatchable
+distinguishability
+distributionally
+DNN
+DNNs
+DOF
+DOFs
+DQN
+DRL
+Drude
+DSC
+DTU
+dynamicity
+ECM
+eco
+EDA
+EfficientNet
+eigen
+eigenmode
+eigenmodes
+EKF
+electrolyzer
+EMG
+endogeneity
+ensembling
+equivariant
+ERP
+ESS
+Eulerian
+exceedance
+explainability
+Fabry
+FCN
+FDM
+FEA
+feedforward
+FFT
+finetune
+finetuned
+finetuning
+Floquet
+fluorophore
+fluorophores
+Fock
+forcings
+FOV
+FPGA
+FPGAs
+freestream
+Frobenius
+FSM
+FTIR
+functionalization
+functionalized
+functionals
+FWHM
+GaAs
+Galerkin
+GANs
+gapless
+GARCH
+gaussian
+Gaussianity
+GCN
+GDPR
+GHG
+Gini
+GIS
+GLM
+GMM
+GNN
+GNNs
+GNSS
+GPP
+GPT
+grayscale
+GRU
+Gurobi
+hadronization
+Hankel
+Hartree
+HCI
+HCl
+heatmap
+heatmaps
+hemodynamic
+heteroskedasticity
+HMD
+holonomic
+homodyne
+homophily
+Hopfield
+HPC
+HVAC
+hypergraph
+hypergraphs
+hyperparameter
+hyperspectral
+hypersurface
+hypersurfaces
+hysteretic
+ICP
+ICT
+IDF
+iff
+IID
+ILP
+ImageJ
+ImageNet
+IMU
+incentivizes
+incentivizing
+injectivity
+inpainting
+interferogram
+interparticle
+intertemporal
+intraday
+invasively
+IoU
+IQR
+Ising
+iso
+Jaccard
+Jupyter
+JWST
+Kaggle
+Kalman
+Keras
+ket
+keypoint
+keypoints
+Kinect
+KNN
+kpc
+KPI
+KPIs
+Krylov
+Kubernetes
+Kutta
+LabVIEW
+Langevin
+Larmor
+LCA
+LDA
+Lennard
+Levenberg
+LHC
+LHS
+LIGO
+Likert
+Lindblad
+linewidth
+LLM
+LLMs
+logit
+logits
+lognormal
+LoRa
+LQR
+LSTM
+LSTMs
+LTE
+LTI
+macroscale
+Majorana
+makespan
+MAPE
+MapReduce
+Marquardt
+MCMC
+MCU
+MDP
+measurability
+medoids
+mesoscale
+metaheuristic
+metaheuristics
+metamodel
+meV
+microenvironment
+microfluidic
+microfluidics
+microgrid
+microgrids
+microscale
+microservice
+microservices
+MILP
+MIMO
+MIP
+misclassifications
+missingness
+MLE
+MLP
+MLPs
+mmWave
+MNIST
+MobileNet
+monodisperse
+MOSFET
+MOSFETs
+MPC
+MPI
+MQTT
+MSE
+multiclass
+multiobjective
+multiphysics
+multipoles
+multiscale
+mV
+NaOH
+NER
+neuro
+neuromorphic
+NIR
+NIST
+NLP
+NN
+NNs
+Noether
+nonconvex
+nondegenerate
+nonlinearities
+nonlinearly
+nonlocal
+nonstationary
+Nusselt
+ODEs
+odometry
+OFDM
+OLS
+omics
+OpenCV
+OpenFOAM
+OpenMP
+OpenStreetMap
+operationalization
+overdamped
+overfit
+overfitting
+overpotential
+PageRank
+parameterization
+parameterizations
+parameterizing
+paraxial
+passivation
+PCA
+PCC
+PCR
+PDEs
+PDMS
+Peltier
+perceptron
+perceptrons
+perturbatively
+photovoltaics
+piezo
+PIV
+PLA
+planform
+PMMA
+PMT
+polarimetric
+polarizers
+polytropic
+posedness
+poset
+postprocessing
+PPO
+Prandtl
+preconditioner
+premia
+prespecified
+pretrained
+pretraining
+prosumers
+proto
+PSD
+PSNR
+PSO
+PV
+PWM
+QFT
+quadcopter
+quadrotor
+quintiles
+Raman
+RANSAC
+RBF
+RDF
+recyclability
+regularizer
+ReLU
+reparameterization
+ResNet
+reweighted
+reweighting
+RHS
+Ricci
+Riesz
+RL
+RLC
+RMSE
+RNN
+RNNs
+RoBERTa
+ROS
+RPC
+RSSI
+RTT
+Runge
+SaaS
+SAR
+scalings
+scatterplot
+Schwarz
+scikit
+SCM
+Scopus
+SDE
+SDG
+SDN
+SDP
+SDR
+SDSS
+SED
+serverless
+setpoint
+setpoints
+SFR
+sharding
+sigmoidal
+Simulink
+SiO
+SLA
+SMA
+SMC
+Smirnov
+SMT
+SNE
+Sobel
+Sobolev
+sociodemographic
+softmax
+SolidWorks
+soliton
+solitons
+SOTA
+spacelike
+spanwise
+sparsification
+SPI
+spintronic
+spintronics
+SSIM
+Stackelberg
+Stata
+STFT
+STL
+STM
+stochasticity
+streamwise
+Strouhal
+subcarrier
+subcarriers
+subgraph
+subgraphs
+subgrid
+sublinear
+submodular
+subproblems
+superlinear
+supremum
+SVD
+SVM
+SVMs
+TEM
+TeV
+thermalization
+thermo
+thermomechanical
+thermophysical
+thresholding
+Tikhonov
+timelike
+timeseries
+timestep
+Toeplitz
+TOF
+TP
+tradeoff
+tradeoffs
+transcriptomic
+translationally
+tri
+Uber
+UMAP
+UML
+undeformed
+underactuated
+underexplored
+underfitting
+undoped
+UNet
+unmodeled
+unpolarized
+VAE
+VAEs
+VGG
+virial
+viscoelasticity
+ViT
+Voigt
+Voronoi
+voxel
+voxels
+Waals
+walkthrough
+Wasserstein
+wavefunction
+wavefunctions
+wavenumber
+wavenumbers
+wettability
+Wilcoxon
+XAI
+XGBoost
+Xilinx
+XRD
+Yukawa
+Zehnder

services/web/types/assets.d.ts

@@ -22,3 +22,8 @@ declare module '*.wasm' {
   const src: string
   export default src
 }
+
+declare module '*.txt' {
+  const src: string
+  export default src
+}

services/web/webpack.config.js

@@ -158,6 +158,13 @@ module.exports = {
           filename: 'js/[name]-[contenthash][ext]',
         },
       },
+      {
+        test: /\.txt$/,
+        type: 'asset/source',
+        generator: {
+          filename: 'js/[name]-[contenthash][ext]',
+        },
+      },
       {
         // Pass Less files through less-loader/css-loader/mini-css-extract-
         // plugin (note: run in reverse order)