gpml2pvjson/bin/gpml2pvjson

#!/usr/bin/env node

var _ = require("lodash");
var crypto = require("crypto");
var fs = require("fs");
var hl = require("highland");
var npmPackage = require("../package.json");
var ndjson = require("ndjson");
var program = require("commander");
var VError = require("verror");

var gpml2pvjson = require("../es5/2013a/toPvjson").toPvjson;

program
  .version(npmPackage.version)
  .description("Converts GPML (XML) to pvjson (JSON)")
  .option(
    "--id [string]",
    'Specify unique ID of this pathway, e.g., "http://identifiers.org/wikipathways/WP4"'
  )
  .option(
    "--pathway-version [string]",
    'Specify version of this pathway, e.g., "90358"'
  );

// NOTE: this is in addition to the automatically generated help text
program.on("--help", function() {
  console.log("  Examples:");
  console.log();
  console.log("    Display pvjson in command line:");
  console.log(
    `    $ gpml2pvjson --id http://identifiers.org/wikipathways/WP554 < ./test/input/WP554_77712.gpml`
  );

  console.log("    Save pvjson to new file:");
  console.log(
    "    $ gpml2pvjson < ./test/input/WP554_77712.gpml > ./WP554_77712.json"
  );

  console.log("    Download from WikiPathways and convert:");
  console.log(
    `    $ curl "http://webservice.wikipathways.org/getPathwayAs?fileType=xml&pwId=WP554&revision=77712&format=xml" | xpath "*/ns1:data/text()" | base64 --decode | gpml2pvjson --id http://identifiers.org/wikipathways/WP554 --pathway-version=77712`
  );

  console.log("    Get w/ WikiPathways API and convert:");
  console.log(
    `    $ wikipathways-api-client get-pathway WP4 | gpml2pvjson --id http://identifiers.org/wikipathways/WP4`
  );
});

program.parse(process.argv);

var id = program.id;
var pathwayVersion = program.pathwayVersion;
// NOTE If an id is not provided, the CLI generates a hash of the input to use as the id. See
// https://bentrask.com/?q=hash://sha256/98493caa8b37eaa26343bbf73f232597a3ccda20498563327a4c3713821df892
// This is for the CLI only; the library itself does not do this.
var HASH_NAME = "sha256";

// NOTE: some GPML files use lowercase "utf-8", so we need the ignore case flag to be set.
var primaryDelimiterBetweenGPMLFiles = /<\?xml\ version=["']1.0["']\ encoding=["']UTF-8["']\?>/i;
var secondaryDelimiterBetweenGPMLFiles = "</Pathway>";

var source = hl(process.stdin)
  // NOTE this splitter (the next two steps) successfully splits the stream when
  // the XML file has the XML header. We don't want to process the XML header
  // with CXML, so this splitter has the additional benefit of removing the
  // header, in addition to splitting the stream.
  .splitBy(primaryDelimiterBetweenGPMLFiles)
  .drop(1)
  // This splitter (next three steps) is needed to handle the case of a stream
  // of GPML files when one or more of those files don't have the XML header.
  // It just splits the stream at the new file boundary, without removing or
  // otherwise changing anything, because we add the separator back in.
  .splitBy(
    new RegExp(
      // TODO Do we need to replace the forward slash to make it work in the
      // RegExp? I checked the latest versions of Chrome and FF, and they don't
      // appear to need this, because they automatically do this, so both of
      // the following are equivalent:
      //secondaryDelimiterBetweenGPMLFiles.replace("/", "\\/") + "[\r\n]*$"
      secondaryDelimiterBetweenGPMLFiles + "[\r\n]*$"
    )
  )
  // NOTE: We are appending the secondary delimiter back in, because the
  // splitter removed it, but we need to include it so CXML can process
  // the GPML.
  // TODO can we use intersperse or something instead of this?
  // Problem with intersperse: we get two streams where we want just one.
  .map(x => x + secondaryDelimiterBetweenGPMLFiles)
  // NOTE: w/out this filter, the final value in the stream would always just be
  // an extraneous, isolated copy of the secondary delimiter.
  .filter(
    x =>
      !new RegExp("^[\r\n]*" + secondaryDelimiterBetweenGPMLFiles + "$").test(x)
  );

hl(source).map(gpml => hl([gpml])).each(function(gpmlStream) {
  const hash = crypto.createHash(HASH_NAME);
  hash.setEncoding("hex");

  gpmlStream.observe().each(chunk => hash.update(chunk.toString()));

  gpml2pvjson(gpmlStream, id)
    .last()
    .map(function(pvjson) {
      if (!pvjson.pathway.id) {
        pvjson.pathway.id = `hash://${HASH_NAME}/${hash.digest("hex")}`;
      }
      if (!!pathwayVersion) {
        pvjson.pathway.pathwayVersion = pathwayVersion;
      }
      return pvjson;
    })
    .errors(function(err) {
      console.error(err);
      process.exit(1);
    })
    .pipe(ndjson.serialize())
    .pipe(process.stdout);
});

// TODO does the process exit on its own?
//process.exit(0);