feat: expose nmr-correlation package as a new nmr-cli command

app/scripts/nmr-cli/package.json

@@ -25,6 +25,7 @@
     "lodash.merge": "^4.6.2",
     "mf-parser": "^3.9.2",
     "ml-spectra-processing": "^14.34.0",
+    "nmr-correlation": "^3.0.2",
     "nmr-processing": "^22.23.6",
     "openchemlib": "^9.25.0",
     "playwright": "1.62.1",

app/scripts/nmr-cli/src/correlation.ts

@@ -0,0 +1,76 @@
+import { buildCorrelationData } from 'nmr-correlation'
+import type { Options as CorrelationOptions, Spectra } from 'nmr-correlation'
+import { FifoLogger } from 'fifo-logger'
+import {
+  buildWebSource,
+  core,
+  parsingOptions,
+  processSpectra,
+} from './parse/prase-spectra'
+
+// Default tolerances
+const DEFAULT_TOLERANCE_H = 0.02
+const DEFAULT_TOLERANCE_C = 0.25
+
+export interface CorrelationInput {
+  url: string
+  mf: string
+  toleranceH?: number
+  toleranceC?: number
+}
+
+function resolveTolerance(value: number | undefined, fallback: number): number {
+  return value === undefined || Number.isNaN(value) ? fallback : value
+}
+
+export async function generateCorrelationData(input: CorrelationInput) {
+  const { url, mf, toleranceH, toleranceC } = input
+  const logger = new FifoLogger()
+
+  const source = buildWebSource(url)
+
+  const { state } = await core.readFromWebSource(source, {
+    ...parsingOptions,
+    logger,
+  })
+
+  const spectraBeforeProcessing = state.data ? [...state.data.spectra] : []
+
+  if (state.data) {
+    processSpectra(
+      state.data,
+      { autoProcessing: true, autoDetection: true },
+      logger
+    )
+  }
+
+  // processSpectra replaces a spectrum's array slot with a new object only
+  // when it successfully parses it; on failure it leaves the original raw
+  // object in place (see its catch block) instead of removing it. Compare
+  // by reference against the pre-processing snapshot to filter those out,
+  // so buildCorrelationData never sees a spectrum it can't actually read.
+  // Note: a pre-existing bug (see https://github.com/NFDI4Chem/nmrkit/issues/139)
+  // currently makes every spectrum fail this step, so real cross-spectrum correlation links are untested here.
+  const spectra = (state.data?.spectra ?? []).filter(
+    (spectrum, index) => spectrum !== spectraBeforeProcessing[index]
+  )
+
+  const options: CorrelationOptions = {
+    mf,
+    tolerance: {
+      H: resolveTolerance(toleranceH, DEFAULT_TOLERANCE_H),
+      C: resolveTolerance(toleranceC, DEFAULT_TOLERANCE_C),
+    },
+  }
+
+  let correlationData
+  try {
+    correlationData = buildCorrelationData(spectra as Spectra, options)
+  } catch (error) {
+    throw new Error(
+      `Failed to build correlation data: ${error instanceof Error ? error.message : String(error)}`
+    )
+  }
+
+  return { ...correlationData, logs: logger.getLogs() }
+}

app/scripts/nmr-cli/src/index.ts

@@ -4,6 +4,7 @@ import { parseSpectra } from './parse/prase-spectra'
 import { generateSpectrumFromPublicationString } from './publication-string'
 import { generateNMRiumFromPeaks } from './peaks-to-nmrium'
 import type { PeaksToNMRiumInput } from './peaks-to-nmrium'
+import { generateCorrelationData } from './correlation'
 import { hideBin } from 'yargs/helpers'
 import { parsePredictionCommand } from './prediction'
 import { readFileSync } from 'fs'
@@ -15,8 +16,15 @@ Usage: nmr-cli  <command> [options]
 Commands:
   parse-spectra                Parse a spectra file to NMRium file
   parse-publication-string     resurrect spectrum from the publication string 
-  predict                      Predict spectrum from Mol 
+  predict                      Predict spectrum from Mol
   peaks-to-nmrium              Convert a peak list to NMRium object
+  correlation                  Build correlation data from NMR spectra fetched from a URL
+
+Options for 'correlation' command:
+  -u, --url                Spectra ZIP file URL
+      --mf                Molecular formula
+      --tolerance-h       H tolerance override (default: 0.02)
+      --tolerance-c       C tolerance override (default: 0.25)
 
 Options for 'parse-spectra' command:
   -u, --url                File URL  
@@ -250,12 +258,61 @@ const peaksToNMRiumCommand: CommandModule = {
   },
 }
 
+// Define the correlation command
+const correlationCommand: CommandModule = {
+  command: ['correlation', 'corr'],
+  describe: 'Build correlation data from NMR spectra fetched from a URL',
+  builder: yargs => {
+    return yargs.options({
+      u: {
+        alias: 'url',
+        describe: 'Spectra ZIP file URL',
+        type: 'string',
+        demandOption: true,
+        nargs: 1,
+      },
+      mf: {
+        describe: 'Molecular formula',
+        type: 'string',
+        demandOption: true,
+        nargs: 1,
+      },
+      'tolerance-h': {
+        describe: 'H tolerance override (default: 0.02)',
+        type: 'number',
+      },
+      'tolerance-c': {
+        describe: 'C tolerance override (default: 0.25)',
+        type: 'number',
+      },
+    })
+  },
+  handler: async argv => {
+    try {
+      const result = await generateCorrelationData({
+        url: argv.u as string,
+        mf: argv.mf as string,
+        toleranceH: argv['tolerance-h'] as number | undefined,
+        toleranceC: argv['tolerance-c'] as number | undefined,
+      })
+      console.log(JSON.stringify(result))
+    } catch (error) {
+      console.error(
+        'Error:',
+        error instanceof Error ? error.message : String(error),
+      )
+      process.exit(1)
+    }
+  },
+}
+
 yargs(hideBin(process.argv))
   .usage(usageMessage)
   .command(parseFileCommand)
   .command(parsePublicationCommand)
   .command(parsePredictionCommand)
   .command(peaksToNMRiumCommand)
+  .command(correlationCommand)
   .showHelpOnFail(true)
   .help()
   .parse()

app/scripts/nmr-cli/src/parse/prase-spectra.ts

@@ -196,19 +196,22 @@ async function processAndSerialize(
   outputResult({ nmriumState: { data, version }, images, logs }, o);
 }
 
-async function loadSpectrumFromURL(options: RequiredKey<FileOptionsArgs, 'u'>, logger: FifoLogger) {
-  const { u: url, include, exclude } = options;
-
+function buildWebSource(url: string) {
   const { pathname: relativePath, origin: baseURL } = new URL(url)
-  const source = {
+  return {
     entries: [
       {
         relativePath,
       },
     ],
     baseURL,
   }
+}
+
+async function loadSpectrumFromURL(options: RequiredKey<FileOptionsArgs, 'u'>, logger: FifoLogger) {
+  const { u: url, include, exclude } = options;
 
+  const source = buildWebSource(url)
 
   const { state } = await core.readFromWebSource(source, { ...parsingOptions, fileFilter: { include, exclude }, logger });
 
@@ -257,4 +260,4 @@ function parseSpectra(argv: yargs.ArgumentsCamelCase<FileOptionsArgs>
 
 
 
-export { loadSpectrumFromFilePath, loadSpectrumFromURL, parseSpectra }
+export { loadSpectrumFromFilePath, loadSpectrumFromURL, parseSpectra, processSpectra, parsingOptions, core, buildWebSource }