Add tests and file comparisons

invoice2data/input/pdftotext.py

@@ -22,8 +22,6 @@ def to_text(path):
     import shutil
     from distutils import spawn #py2 compat
 
-
-
     if spawn.find_executable("pdftotext"): #shutil.which('pdftotext'):
         out, err = subprocess.Popen(
             ["pdftotext", '-layout', '-enc', 'UTF-8', path, '-'],

invoice2data/test/common.py

@@ -0,0 +1,10 @@
+import os
+import pkg_resources
+
+def get_sample_files(extension):
+    compare_files = []
+    for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'compare')):
+        for file in files:
+            if file.endswith(extension):
+                compare_files.append(os.path.join(path, file))
+    return compare_files

invoice2data/test/compare/AmazonWebServices.json

@@ -0,0 +1,31 @@
+[
+    {
+        "amount": 4.11,
+        "amount_untaxed": 4.11,
+        "currency": "USD",
+        "date": "03/08/2014",
+        "desc": "Invoice 42183017 from Amazon Web Services",
+        "invoice_number": "42183017",
+        "issuer": "Amazon Web Services",
+        "lines": [
+            {
+                "description": "AWS Data Transfer",
+                "price_unit": "0.01"
+            },
+            {
+                "description": "Amazon Elastic Compute Cloud",
+                "price_unit": "1.87"
+            },
+            {
+                "description": "Amazon Glacier",
+                "price_unit": "2.22"
+            },
+            {
+                "description": "Amazon Simple Storage Service",
+                "price_unit": "0.01"
+            }
+        ],
+        "partner_name": "Amazon Web Services, Inc.",
+        "partner_website": "aws.amazon.com"
+    }
+]

invoice2data/test/compare/FlipkartInvoice.json

@@ -0,0 +1,11 @@
+[
+    {
+        "amount": 319.0,
+        "currency": "INR",
+        "date": "20/10/2015",
+        "desc": "Invoice #BLR_WFLD20151000982590 from Flipkart",
+        "invoice_number": "#BLR_WFLD20151000982590",
+        "issuer": "Flipkart",
+        "order_id": "OD304175096047380001"
+    }
+]

invoice2data/test/compare/QualityHosting.json

@@ -0,0 +1,56 @@
+[
+    {
+        "amount": 34.73,
+        "amount_untaxed": 34.73,
+        "currency": "EUR",
+        "date": "07/05/2014",
+        "desc": "Invoice 30064443 from QualityHosting AG",
+        "invoice_number": "30064443",
+        "issuer": "QualityHosting AG",
+        "lines": [
+            {
+                "desc": "Small Business StandardExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_strukan\n01.05.14-31.05.14",
+                "pos": "1",
+                "price": 3.89,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business QualityExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_schneider\n01.05.14-31.05.14",
+                "pos": "2",
+                "price": 5.39,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business QualityExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_minar\n01.05.14-31.05.14",
+                "pos": "3",
+                "price": 5.39,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business QualityExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_mayr\n01.05.14-31.05.14",
+                "pos": "4",
+                "price": 5.39,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business QualityExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_jenewein\n01.05.14-31.05.14",
+                "pos": "5",
+                "price": 5.39,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business QualityExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_jauernik\n01.05.14-31.05.14\nQualityHosting AG - Uferweg 40-42 - D-63571 Gelnhausen\niViveLabs Ltd.\n93B Sai Yu Chung\nYuen Long, N.T.\nHong Kong\nPos.            Menge      Beschreibung                                                            Rabatt %     VK-Preis     Zeilenbetrag\nOhne      Ohne MwSt.\nMwSt.",
+                "pos": "6",
+                "price": 5.39,
+                "qty": 1.0
+            },
+            {
+                "desc": "Small Business StandardExchange 2010\nGrundgeb\u00fchr pro Einheit\nDienst: OUDJQ_office\n01.05.14-31.05.14\n",
+                "pos": "7",
+                "price": 3.89,
+                "qty": 1.0
+            }
+        ],
+        "vat": "DE 232 446 240"
+    }
+]

invoice2data/test/compare/free_fiber.json

@@ -0,0 +1,13 @@
+[
+    {
+        "amount": 29.99,
+        "amount_untaxed": 24.99,
+        "currency": "EUR",
+        "date": "02/07/2015",
+        "date_due": "2015-07-05 00:00:00",
+        "desc": "Invoice 562044387 from Free",
+        "invoice_number": "562044387",
+        "issuer": "Free",
+        "vat": "FR60421938861"
+    }
+]

invoice2data/test/compare/oyo.json

@@ -0,0 +1,10 @@
+[
+    {
+        "amount": 1939.0,
+        "currency": "INR",
+        "date": "31/12/2017",
+        "desc": "Invoice IBZY2087 from OYO",
+        "invoice_number": "IBZY2087",
+        "issuer": "OYO"
+    }
+]

invoice2data/test/test_cli.py

@@ -1,6 +1,8 @@
 import os
 import glob
 import filecmp
+import json
+import shutil
 
 try:
     from StringIO import StringIO
@@ -12,100 +14,109 @@
 from invoice2data.main import *
 from invoice2data.extract.loader import read_templates
 
+from .common import *
+
+
 class TestCLI(unittest.TestCase):
     def setUp(self):
         self.templates = read_templates()
         self.parser = create_parser()
 
-    def _get_test_file_path(self):
-        out_files = [None]*4
-        for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'pdfs')):
-            for file in files:
-                if file == "2014-05-07 Invoice 30064443 from QualityHosting.pdf":
-                    out_files[0] = os.path.join(path, file)
-                elif file == "2014-08-03 SALES Amazon Web Services  aws.amazon.coUS.pdf":
-                    out_files[1] = os.path.join(path, file)
-                elif file == "2015-01-29 PAYPAL ENVATO MKPL EN 4029357733 AU.pdf":
-                    out_files[2] = os.path.join(path, file)
-                elif file == "2015-07-02-invoice_free_fiber.pdf":
-                    out_files[3] = os.path.join(path, file)        
-        return out_files
-
-    def _get_test_file_content():
-        pass
+    def compare_json_content(self, test_file, json_file):
+        with open(test_file) as json_test_file, open(json_file) as json_json_file:
+            jdatatest = json.load(json_test_file)
+            jdatajson = json.load(json_json_file)
+        # logger.info(jdatajson)
+        # logger.info(jdatatest)
+        if jdatajson == jdatatest:
+            logger.info("True")
+            return True
+        else:
+            logger.info("False")
+            return False
 
     def test_input(self):
-        args = self.parser.parse_args(['--input-reader', 'pdftotext'] + self._get_test_file_path())
+        args = self.parser.parse_args(['--input-reader', 'pdftotext'] + get_sample_files('.pdf'))
         main(args)
 
     def test_output_name(self):
         test_file = 'inv_test_8asd89f78a9df.csv'
-        args = self.parser.parse_args(['--output-name', test_file, '--output-format', 'csv'] + self._get_test_file_path())
+        args = self.parser.parse_args(['--output-name', test_file, '--output-format', 'csv']
+                                      + get_sample_files('.pdf'))
         main(args)
         self.assertTrue(os.path.exists(test_file))
         os.remove(test_file)
 
     def test_debug(self):
-        args = self.parser.parse_args(['--debug'] + self._get_test_file_path())
+        args = self.parser.parse_args(['--debug'] + get_sample_files('.pdf'))
         main(args)
 
     # TODO: move result comparison to own test module.
     # TODO: parse output files instaed of comparing them byte-by-byte.
 
-    def test_content_csv(self):
-        for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'compare')):
-            for file in files:
-                if file.endswith(".csv"):
-                    cmp_file = os.path.join(path, file)
-
-        test_files = 'inv_test.csv'
-        args = self.parser.parse_args(['--output-name', test_files, '--output-format', 'csv'] + self._get_test_file_path())
+    def test_content_json(self):
+        pdf_files = get_sample_files('.pdf')
+        json_files = get_sample_files('.json')
+        test_files = 'test_compare.json'
+        for pfile in pdf_files:
+            for jfile in json_files:
+                if pfile[:-4] == jfile[:-5]:
+                    args = self.parser.parse_args(
+                        ['--output-name', test_files, '--output-format', 'json', pfile])
+                    main(args)
+                    compare_verified = self.compare_json_content(test_files, jfile)
+                    print(compare_verified)
+                    if not compare_verified:
+                        self.assertTrue(False)
+                    os.remove(test_files)
+        self.assertTrue(True)
+
+    def test_copy(self):
+        # folder = pkg_resources.resource_filename(__name__, 'pdfs')
+        directory = os.path.dirname("invoice2data/test/copy_test/pdf/")
+        os.makedirs(directory)
+        args = self.parser.parse_args(['--copy', 'invoice2data/test/copy_test/pdf'] + get_sample_files('.pdf'))
         main(args)
-        # self.assertTrue(filecmp.cmp(test_files, cmp_file, shallow=False))
-        os.remove(test_files)
-
-    def test_content_xml(self):
-        for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'compare')):
+        i = 0
+        for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'copy_test/pdf')):
             for file in files:
-                if file.endswith(".xml"):
-                    cmp_file = os.path.join(path, file)
+                if file.endswith(".pdf"):
+                    i += 1
+
+        shutil.rmtree('invoice2data/test/copy_test/', ignore_errors=True)
+        self.assertEqual(i, len(get_sample_files('.json')))
+        '''
+        if i != len(self._get_test_file_json_path()):
+            print(i)
+            self.assertTrue(True)
+        else:
+            print(i)
+            self.assertTrue(False, "Number of files not equal")
+        '''
 
-        test_files = 'inv_test.xml'
-        args = self.parser.parse_args(['--output-name', test_files, '--output-format', 'xml'] + self._get_test_file_path())
-        main(args)
-        # self.assertTrue(filecmp.cmp(test_files, cmp_file, shallow=False))
-        os.remove(test_files)
+    # def test_template(self):
+    #     directory = os.path.dirname("invoice2data/test/temp_test/")
+    #     os.makedirs(directory)
+    #     args = self.parser.parse_args(['--template-folder', 'ACME-templates', self._get_test_file_path()])
+    #     main(args)
+    #     shutil.rmtree('invoice2data/test/temp_test/', ignore_errors=True)
+    #     self.assertTrue(args.template_folder)
 
-    def test_content_json(self):
+    def test_exclude_template(self):
         for path, subdirs, files in os.walk(pkg_resources.resource_filename(__name__, 'compare')):
             for file in files:
-                if file.endswith(".json"):
-                    cmp_file = os.path.join(path, file)
-
-        test_files = 'inv_test.json'
-        args = self.parser.parse_args(['--output-name', test_files, '--output-format', 'json'] + self._get_test_file_path())
+                if file.endswith("oyo.pdf"):
+                    my_file = os.path.join(path, file)
+        directory = os.path.dirname("invoice2data/test/temp_test/")
+        os.makedirs(directory)
+        shutil.copy('invoice2data/extract/templates/com/com.oyo.invoice.yml', 'invoice2data/test/temp_test/')
+        args = self.parser.parse_args(['--exclude-built-in-templates',
+                                       '--template-folder',
+                                       directory,
+                                       my_file])
         main(args)
-        # self.assertTrue(filecmp.cmp(test_files, cmp_file, shallow=False))
-        os.remove(test_files)
-
-    # def test_copy(self):
-    #     parser = create_parser()
-    #     folder = pkg_resources.resource_filename(__name__, 'pdfs')      
-    #     args = parser.parse_args(['--copy', '/invoice2data/test/', self._get_test_file_path()])
-    #     self.assertTrue(args.copy)
-
-    # def test_template(self):
-    #     parser = create_parser()
-    #     folder = pkg_resources.resource_filename(__name__, 'pdfs')      
-    #     args = parser.parse_args(['--template-folder', 'ACME-templates', self._get_test_file_path()])
-    #     self.assertTrue(args.template_folder)
+        shutil.rmtree('invoice2data/test/temp_test/')
 
-    # def test_exclude_template(self):
-    #     parser = create_parser()
-    #     folder = pkg_resources.resource_filename(__name__, 'pdfs')      
-    #     args = parser.parse_args(['--exclude-built-in-templates', '--template-folder', 'ACME-templates', self._get_test_file_path()])
-    #     self.assertTrue(args.exclude_built_in_templates)
-
 
 if __name__ == '__main__':
-    unittest.main()
+    unittest.main()