From f0fe802df9fd7e98c0333a80922f6008cab7abab Mon Sep 17 00:00:00 2001
From: eug-vs <eug-vs@keemail.me>
Date: Sat, 20 Mar 2021 17:01:41 +0300
Subject: feat: parse pdf bills into transfers

---
 src/services/uploads.service.ts | 79 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 76 insertions(+), 3 deletions(-)

(limited to 'src/services/uploads.service.ts')

diff --git a/src/services/uploads.service.ts b/src/services/uploads.service.ts
index 520079f..1d39a66 100644
--- a/src/services/uploads.service.ts
+++ b/src/services/uploads.service.ts
@@ -1,13 +1,86 @@
 import { Application } from '@feathersjs/express';
+import { HookContext } from '@feathersjs/feathers';
 import blobService from 'feathers-blob';
 import fs from 'fs-blob-store';
+import moment from 'moment';
 import Bluebird from 'bluebird';
+import { PdfReader } from 'pdfreader'
 import _ from 'lodash';
 
-const blobStorage = fs('./uploads');
-
+const dir = './documents';
+const blobStorage = fs(dir);
 const uploads = blobService({ Model: blobStorage });
 
+
+const parseTransfersBill = async (context: HookContext): Promise<HookContext> => {
+  const { id } = context.result;
+  const fileName = `${dir}/${id}`;
+
+  const reader = new PdfReader();
+  const items: any[] = [];
+
+  reader.parseFileItems(fileName, (err: Error, item: any) => {
+    if (item) items.push(item);
+    else { // Finished parsing. TODO: wrap into async
+      const hash = _.groupBy(items, 'y')
+      const rows = _.map(hash, (elements: any[]) => _.map(elements, 'text'));
+
+      const transfers = rows.reduce((acc: any, cols: string[]) => {
+        if (cols[0]?.startsWith('УНП')) {
+          acc[acc.length - 1].vatId = cols[2];
+          acc[acc.length - 1].name = cols[3];
+        } else if (cols.length === 7) {
+          const [dateString, doc, op, code, account, debet, credit] = cols;
+          try {
+            const date = moment(dateString, 'DD.MM.YYYY').toISOString();
+            acc.push({ date, doc, op, code, account, debet, credit });
+          } catch (e) {
+            console.log(`Skipping row because not a date: ${dateString}`);
+          }
+        }
+
+        return acc;
+      }, []);
+
+      return Bluebird.mapSeries(transfers, async (transfer: any) => {
+        const { date, vatId, name } = transfer;
+        const debet = parseFloat(transfer.debet.replace(/ /g, ''));
+        const credit = parseFloat(transfer.credit.replace(/ /g, ''));
+
+        const operation = debet ? 'out' : 'in';
+        const amount = debet || credit;
+
+        const contractorId = await context.app
+          .service('contractors')
+          .find({ query: { vatId } })
+          .then((results: any[]) => {
+            if (results.length) return results[0];
+            return context.app
+              .service('contractors')
+              .create({ vatId, name });
+          })
+          .then((contractor: any) => contractor._id);
+
+        return context.app.service('transfers').create({
+          date,
+          operation,
+          amount,
+          contractorId,
+        });
+      }).then(() => context.service.remove(id));
+    }
+  });
+
+  return context;
+};
+
+
 export default (app: Application): void => {
-  app.use('/uploads', uploads)
+  app.use('/uploads', uploads);
+
+  app.service('uploads').hooks({
+    after: {
+      create: parseTransfersBill,
+    },
+  });
 };
-- 
cgit v1.2.3