Merge 51b9e01 into fb164a4

.gitignore

@@ -65,3 +65,6 @@ lib/
 
 # Secret files
 token.txt
+
+# Useless format files
+data/*.xlsx

README.md

@@ -17,8 +17,8 @@ npm install ezs-conditor
 
 ```bash
 $ ./bin/affAlign.js < data/1000-notices-conditor-hal.json | ./bin/compareRnsr.js
-recall: 0.6487068965517241
-correct: 903
+recall: 0.7162356321839081
+correct: 997
 total: 1392
 ```
 

bin/findIdsInAddresses.js

@@ -0,0 +1,51 @@
+#!/usr/bin/env node
+/* eslint-disable no-console */
+
+const csv = require('csv-string');
+const { promises } = require('fs');
+const { difference, intersection } = require('ramda');
+const { depleteString } = require('../lib/strings');
+const { isIn } = require('../lib/rnsr');
+const RNSR = require('../data/RNSR.json');
+
+promises
+    .readFile(`${__dirname}/../data/2014_AdressesCorpusTest.tsv`, {
+        encoding: 'utf8',
+    })
+    .then((tsv) => csv.parse(tsv, '\t'))
+    .then((lines) => {
+        console.log('found\texpected\taddress');
+        const rnsrAddresses = lines.slice(1);
+        let expectedNb = 0;
+        const { foundNb, wronglyFoundNb } = rnsrAddresses.reduce(({
+            foundNb: alreadyFound,
+            wronglyFoundNb: alreadyWronglyFound,
+        }, rnsrAddress) => {
+            const depletedAddress = depleteString(rnsrAddress[1]);
+            const isInAddress = isIn(depletedAddress);
+            const rnsrIds = RNSR.structures.structure
+                .filter(isInAddress)
+                .map((s) => s.num_nat_struct);
+
+            const expectedIds = rnsrAddress[0].split(' ; ');
+            expectedNb += expectedIds.length;
+
+            const foundIds = intersection(rnsrIds, expectedIds);
+            if (foundIds.length < expectedIds.length) {
+                console.log(`${rnsrIds}\t${expectedIds}\t${depletedAddress}`);
+            }
+            const wronglyFoundIds = difference(rnsrIds, expectedIds);
+            return {
+                foundNb: alreadyFound + foundIds.length,
+                wronglyFoundNb: alreadyWronglyFound + wronglyFoundIds.length,
+            };
+        }, { foundNb: 0, wronglyFoundNb: 0 });
+        const recall = foundNb / expectedNb;
+        console.log('recall:', recall);
+        console.log('found:', foundNb);
+        console.log('total:', expectedNb);
+        const precision = foundNb / (foundNb + wronglyFoundNb);
+        console.log('precision:', precision);
+        console.log('wrongs:', wronglyFoundNb);
+        console.log('total found:', foundNb + wronglyFoundNb);
+    });