Updates to name assignment, commenting

src/assignNames.js

@@ -18,7 +18,6 @@ fluid.setLogging(true);
 
 const parsedArgs = minimist(process.argv.slice(2));
 
-const outputFile = parsedArgs.o || "assigned.csv";
 // fluid.module.resolvePath("%imerss-bioinfo/data/Comprehensive Lists/San_Juan_Dunwiddie_List.csv");
 // fluid.module.resolvePath("%imerss-bioinfo/data/Galiano 2022/Tracheophyta_review_summary_reviewed_2022-10-29.csv")
 // fluid.module.resolvePath("%imerss-bioinfo/data/Squamish/Tracheophyta_review_summary_2022-12-14.csv")
@@ -31,6 +30,13 @@ const reader = hortis.csvReaderWithoutMap({
     inputFile: inputFile
 });
 
+hortis.assignedFromInput = function (inputFile) {
+    const lastDotPos = inputFile.lastIndexOf(".");
+    return inputFile.substring(0, lastDotPos) + "-assigned" + inputFile.substring(lastDotPos);
+};
+
+const outputFile = parsedArgs.o || hortis.assignedFromInput(inputFile);
+
 const source = hortis.iNatTaxonSource();
 
 hortis.highRanks = ["kingdom", "phylum", "subphylum", "superclass", "class", "subclass", "superorder", "order",
@@ -63,7 +69,7 @@ hortis.queryFromGBIFRow = function (row) {
     const frontRank = rawRank.substring(rawRank.indexOf(" ") + 1);
     const rank = hortis.capitalize(frontRank);
     return {name, rank};
-}
+};
 
 hortis.obsIdFromSummaryRow2022 = function (row) {
     const obsLink = row["iNaturalist Link"];
@@ -86,8 +92,8 @@ hortis.obsIdFromSummaryRow2023 = function (row) {
 // One-off script to update taxonomies and iNaturalist ids from files produced in Andrew's 2022 run of Galiano data
 
 hortis.applyName = async function (row) {
-    //const query = hortis.queryFromSummaryRow2023(row);
-    const query = hortis.queryFromGBIFRow(row);
+    const query = hortis.queryFromSummaryRow2023(row);
+    //const query = hortis.queryFromGBIFRow(row);
     const looked = await source.get(query);
     //    console.log("Got document ", looked);
     if (looked && looked.doc) {
@@ -114,10 +120,15 @@ hortis.applyName = async function (row) {
     if (obsId) {
         row["Observation Taxon Name"] = "error";
         if (+obsId > 0) {
-            const obs = await source.get({obsId: obsId});
-            if (obs.doc.results.length === 1) {
-                const name = obs.doc.results[0].taxon.name;
-                row["Observation Taxon Name"] = name;
+            try {
+                const obs = await source.get({obsId: obsId});
+                if (obs.doc.results.length === 1) {
+                    const name = obs.doc.results[0].taxon.name;
+                    row["Observation Taxon Name"] = name;
+                }
+            }
+            catch (e) {
+                console.log(`Got error for obsId ${obsId}`, e);
             }
         }
     }
@@ -141,14 +152,14 @@ Promise.all([reader.completionPromise, source.events.onCreate]).then(async funct
         const row = reader.rows[i];
         await hortis.applyName(row);
         // TODO: produce mapping functions for these older forms of data
-/*
+        /*
         if (row.infraTaxonName) {
             await hortis.applyName(row, row.infraTaxonName);
         }
         if (!row.infraTaxonName || row["Name Status"] === "unknown") {
             await hortis.applyName(row, hortis.unscrewGBIFName(row.scientificName)); // taxonName for Dunwiddie data, scientificName for DwC
         }
-*/
+        */
         mapped.push(row);
     }
     hortis.writeCSV(fluid.module.resolvePath(outputFile), Object.keys(mapped[0]), mapped, fluid.promise());

src/materialise.js

@@ -3,23 +3,23 @@
 
 "use strict";
 
-var fluid = require("infusion");
-var glob = require("glob");
-var kettle = require("kettle");
-var fs = require("fs");
+const fluid = require("infusion");
+const glob = require("glob");
+const kettle = require("kettle");
+const fs = require("fs");
 fluid.require("%imerss-bioinfo");
 
-var baseDir = fluid.module.resolvePath("%imerss-bioinfo/data/dataPaper-I-in/Animalia/");
+const baseDir = fluid.module.resolvePath("%imerss-bioinfo/data/dataPaper-I-in/Animalia/");
 
 fs.mkdirSync(baseDir, {
     recursive: true
 });
 
 // Used during preparation of data paper to materialise an entire folder full of Google Sheets as CSV.
-// Will need to updated to use Google Sheets API directrly and not rely on a Google Drive mapping
+// Will need to updated to use Google Sheets API directly and not rely on a Google Drive mapping
 
-var dir = process.argv[2] || "e:/data/Google Drive/Galiano Data Paper 2021/Marine Life/Animalia";
-var source = kettle.dataSource.URL({
+const dir = process.argv[2] || "e:/data/Google Drive/Galiano Data Paper 2021/Marine Life/Animalia";
+const source = kettle.dataSource.URL({
     url: "https://docs.google.com/spreadsheets/d/%id/export?format=csv&id=%id&gid=0",
     port: 443,
     termMap: {
@@ -35,19 +35,19 @@ var source = kettle.dataSource.URL({
     }
 });
 
-var getSegment = function (path) {
-    var lastSlash = path.lastIndexOf("/");
-    var lastDot = path.lastIndexOf(".");
+const getSegment = function (path) {
+    const lastSlash = path.lastIndexOf("/");
+    const lastDot = path.lastIndexOf(".");
     return path.substring(lastSlash + 1, lastDot);
 };
 
 fluid.setLogging(true);
 
 glob(dir + "/*.gsheet", function (er, files) {
     console.log("Got files ", files);
-    var records = files.map(function (oneFile) {
-        var promise = kettle.JSON.readFileSync(oneFile);
-        var parsed;
+    const records = files.map(function (oneFile) {
+        const promise = kettle.JSON.readFileSync(oneFile);
+        let parsed;
         promise.then(function (result) {
             parsed = result;
         });

src/reintegrate.js

@@ -19,6 +19,9 @@ var swaps = hortis.readJSONSync("data/taxon-swaps.json5", "reading taxon swaps f
 
 // Long disused script to output a taxon-swaps JSON5 file given a CSV taxon resolution file
 
+// 24/10/23 - Interesting - forgot that swaps used to be accepted in CSV - which we think we plan to do
+// again. Did AS ever generate these?
+
 hortis.copyCSVProps = function (props, target, source) {
     props.forEach(function (prop) {
         if (fluid.isValue(source[prop]) && source[prop] !== "") {