add report option (#98)

README.md

@@ -8,11 +8,10 @@ The database uses the [Data Format for Digital Linguistics][DaFoDiL] (DaFoDiL) a
 
 <!-- TOC -->
 - [Sources](#sources)
-- [Project Requirements](#project-requirements)
 - [Process](#process)
-- [Style Guide](#style-guide)
 - [The Database](#the-database)
-- [Building the Database](#building-the-database)
+- [Building & Updating the Database](#building--updating-the-database)
+- [Steps to incrementally update the production database](#steps-to-incrementally-update-the-production-database)
 - [Tests](#tests)
 <!-- /TOC -->
 
@@ -76,6 +75,10 @@ To build and/or update the database, follow the steps below. Each of these steps
 
     Entries from individual sources are **not** imported as main entries in the ALTLab database. Instead they are stored as subentries (using the `dataSources` field). The import script merely matches entries from individual sources to a main entry, or creates a main entry if none exists. An aggregation script then does the work of combining information from each of the subentries into a main entry (see the next step).
 
+    Each import step prints a table to the console, showing how many entries from the original data source were unmatched.
+
+    When importing the Maskwacîs database, you can add an `-r` or `--report` flag to output a list of unmatched entries to a file. The flag takes the file path as its argument.
+
 6. Aggregate the data from the individual data sources: `node bin/aggregate.js <inputPath> <outputPath>` (the output path can be the same as the input path; this will overwrite the original).
 
 7. For convenience, you can perform all the above steps with a single command in the terminal: `npm run build` | `yarn build`. In order for this command to work, you will need each of the following files to be present in the `/data` directory, with these exact filenames:

bin/import-MD.js

@@ -6,6 +6,7 @@ import program  from 'commander';
 program
 .arguments(`<mdPath> <databasePath> [fstPath]`)
 .usage(`convert-md <mdPath> <databasePath> [fstPath]`)
+.option(`-r, --report <reportPath>`, `generate report of unmatched entries`)
 .action(importMD);
 
 program.parse(process.argv);

lib/import/MD.js

@@ -1,8 +1,9 @@
-import createSpinner  from 'ora';
-import DatabaseIndex  from '../utilities/DatabaseIndex.js';
-import readNDJSON     from '../utilities/readNDJSON.js';
-import { Transducer } from 'hfstol';
-import writeNDJSON    from '../utilities/writeNDJSON.js';
+import createSpinner         from 'ora';
+import { createWriteStream } from 'fs';
+import DatabaseIndex         from '../utilities/DatabaseIndex.js';
+import readNDJSON            from '../utilities/readNDJSON.js';
+import { Transducer }        from 'hfstol';
+import writeNDJSON           from '../utilities/writeNDJSON.js';
 
 function getPos(str) {
   if (!str) return ``;
@@ -32,8 +33,11 @@ function updateEntry(dbEntry, mdEntry) {
  * Imports the MD entries into the ALTLab database.
  * @param  {String} mdPath
  * @param  {String} dbPath
+ * @param  {String} [fstPath]
+ * @param  {Object} [options={}]
+ * @param  {String} [report] The path where you would like the report generated.
  */
-export default async function importMD(mdPath, dbPath, fstPath) {
+export default async function importMD(mdPath, dbPath, fstPath, { report } = {}) {
 
   const readDatabaseSpinner = createSpinner(`Reading databases.`).start();
 
@@ -141,4 +145,20 @@ export default async function importMD(mdPath, dbPath, fstPath) {
     'Entries without a match:': unmatched.length,
   });
 
+  if (report) {
+
+    const reportSpinner = createSpinner(`Generating report of unmatched entries.`).start();
+    const writeStream   = createWriteStream(report);
+
+    writeStream.write(`head\tPOS\toriginal\t\n`);
+
+    for (const { head, original, pos } of unmatched) {
+      writeStream.write(`${ head.md }\t${ pos }\t${ original }`);
+    }
+
+    writeStream.end();
+    reportSpinner.succeed();
+
+  }
+
 }