From b01cbc9aa08a51a63b1719e71335d140ce73a432 Mon Sep 17 00:00:00 2001
From: phail <phail@hacknology.de>
Date: Sat, 12 Nov 2022 15:48:30 +0100
Subject: [PATCH] add conditions to unittests

---
 .../tests/samples/sample.html.pdf.webp        | Bin 0 -> 2830 bytes
 src/paperless_mail/tests/test_parsers.py      |  88 ++++++++++++++----
 src/paperless_mail/tests/test_parsers_live.py |  65 ++++++++++---
 3 files changed, 123 insertions(+), 30 deletions(-)
 create mode 100644 src/paperless_mail/tests/samples/sample.html.pdf.webp

diff --git a/src/paperless_mail/tests/samples/sample.html.pdf.webp b/src/paperless_mail/tests/samples/sample.html.pdf.webp
new file mode 100644
index 0000000000000000000000000000000000000000..534157660b25e9bf210b2ab898aff969aa8a655b
GIT binary patch
literal 2830
zcmeH|`8yN}7sqETWxb89vDAc7#HEptbyQ=Gu{R`ZgX~+jJBr5IXfl*7jA)Q3M3H4^
zkm_QxWM4*>>?6y_Fyrmsclq)C1K#I7zkHwboacO>@8|q-Y)p-ej_?BjD8tKE*Q_o|
zGx-33jc`B#P;ndB!J(gSQfyvAg33y#n@Cp+^ZEqNM7p2qua`~g97ihbD6x%Kzt<cc
zVGoCXxbN*5TeIoSqGDUc?^-Z87)Fc8<OS|k&0)Bcu6H;;{S@=iFld)u6YIDky9swY
zHaq>L^?j`GK^)VAUB#&At?w_~i|vi!h}B_#U-DcT-xV1-+!;9h2`&bDYTtPm6JQkA
zzlsOmu3*TM1YR&Pc)htp<1`>B;YvU%h4Na19?+=}pKiasD~b8UPJ`%1ZH~gw;~F!I
zKW4lBYJW+b8^DhH?DTKfHsza>91Z?9_-jG~a<+&;TJ_5xfT)(>2x2wABAGaA84d6I
zwJLQV-GKfvGu?-9cnuvirBt*fLxxlyc-UA0U-^NmPKP18(&kNQ!vtk&dK~>`T^!P>
z0o8`QI%NV<rL-;l)2Go`ECaC)<AeY{hF_w^WNrf<YYP7d0>W`3Y9v9_)1WByqIeFk
zbXIgK`?5Yns^#whe<f@LSKcb(j796t@ywmKvd_i%16OjDFbN<&N>}yCB9yidU7q<j
zW7=8X@Y*|iABdZ=&;HH*FZ}<@|MEcq005gkD0$uDdSab-wmx`euhTUel#7SPgEO;&
z&Y;wS9qUJv^3_b^(c~0x@FbTxa^MT6WXxyOtX)r+nNGJ?;0dha034y*V$(hF;KLs4
z+Y{f2#&*&KqWsEfV{K~Y<la88*2nhMW7l9)2}8;BL+Fed=J@1c>s0mSNZpXA#`%Vt
z7E1SH@#)0LXUA7)DsRKq_U`VRyC{sQCYLo)wd>cXZO_A;>tzJDy$-8fGy2}BH=s0l
zOf(AHM^h@IUp>NP-Yibj8IdC90I&yr+fn`Dini?6LEhy2D?ZIog{*i<k^@sG=|vs!
z$}+rh?({Y2SI5a?<9HFJEvk>2NpHhjnm->_2sReY@6a*Wj|xBESfKt8e>BxpuJoDo
z$SvN0oYwf#Hwr&j8+(>{ZwnPS33_EE9`);)(rZrb0ZkI`kKJs~ArI$;-oqqid%^O%
z<P^VLThbp*d|hsE_!RmeysAQYdEkyl{d?6&H0sX)pDm}FKKonU7u8;o%YsUe{LYJN
zUcoV&Jwu1`EsSAH(^Q)Mh1a!Wry>N}3<$!ViuoW7Wv=CyE@vI)iuj$!)-+-bnku%+
zVyH<k*H60(i3-m(+b$r#*Pe56{`zu_%44XG@U-U&Q}@kAi~Tv_BLill8QBnjgu&B4
zyw?lOk<`>r>UV~u9^M-tUQbYAo5N$c0|s@CZsgETC2&>5iwU4Pt{=?xa_y!pGnbzt
zjk-TZCJ?uE&lwAwsPj8&DflUqtKGRb4PVK=9;Eg*)N)&LpXn+Usv~tfDqp^V^p18*
zidvA{Eyed`ufAv%Yv4Q-sXS>as!jeTdtD`Mk}wcqiF}uQwMp*h_T4w6sc!T-Um)n#
zhTcdFBqDR|vsCRc-4lD>hUK|<aL8Fajy>_H%6)yv8y0!ZLeHk!x)I@=`7pjX@sP_4
zFGx0k{iLhAh=k<47$1dy1>F%$#nm~DfJX{5&he?MRv9A{yjuGDE83N!yvf>nneNlw
z^3*FfBV7WN4EfpaiCi(lIC1A7b6_Zrz}SQesDK9*CA`FCS0KXTq|PUi3$N~h7zjbW
zNzqGrI7~^2M=Q~^z}>`MBNS<h&T80msyO;2;&?L#=Y4_h|Micz^2t`{qGt)*GOFJ(
zbJ~ZWV29WqNm;hy!LB`TXX+r~U6GhE!bpaG4}!DWkrh*C$4aaO?>w)UNi}~tW}Y8h
zzaGva!$w^QXSem*c?mkaF<(#qIrh8$;wBDo+X3tPqc%*UTuZ89iP>p#X5nmfNN%fw
zT5!lbUz1kK@+~X;M&4Zx!!66c`sC~8^%fUfL2e8M!cLcub9^LhJXYkKjm}2o3FoW6
zr<W-SO4Vw|-f8%9xxLWaXN@YV@bkmP_<4G-4=v>>x%y3LKH(F^3>Fv_d%txxv!qW@
zX8Z#6pQp9^Z08#3NIb-!aB;y^7qafw4d7$)t@GS&uCXnNt6a*O+bw7t`0c_KZ2=cp
zp*x+N2h1VJ{n#_&fC^ztTD*=$`ifE>$-od7FHxrD&+#4^f@xb{34TL+dceG~g<@@#
zKSOzFxpsJ1Tap|iq{6XTxMR(XNZU%zh?WGHT-d~uv`QoEFW;GdvWqfqbN9n|u&y3u
z2e^An`+-0&>G?82FoADZI7D5XPGWor@0ejoOJsl9slxhan)y8nu8H=XPuvlqA0P0M
zpP3NFu`*nn@0d?L!lIjkL7&IjO>!LDTl-}n++y9N^~Wa7xtnVYw`%7B=QVMGLHj@`
zo_me6)L+ix_)RmIPC|~TbEVc~pxF{dC&RzJ_M`ahnrw)>6)^i-HS6lyPmoZ-h2HYD
zkB%(MNq01MWU4aEBjD(UEBx+`a*&87k_4DX0{RIt`iv<%j4srAAhIz1Da#}6AmwUQ
ziNUw@#j~J~<zULXTe$>F-uGAu$e>QCA_({K9hp2jcv6vs^tI>UI;kZv#pyexAF*b-
zQ2eN56lDgd#V3^N9jFZ7+$bEhi%xcO3f2^LDnbzV#aPQ}{kiu%&U|)yAD00KyEU;b
z9EU@bdX&FJJqTRoDL38q;UKCERQ?tWsP5<KXGU4Ms3<Rm#oa#MW{V7aNXWQ_h~0XF
zb-mwE@N7Rlvqm;oG4BTW##?M*I)2y>G#uZ-$#p5Kv2<%XMKz?fp0CD^Tkxn(QCb@l
z5~_CJcJO;2Vc^3L)HKaN$n9wqrAWOayE|vnj~~I{SZ(PTF;)HczMX_jWxM!EnxMoG
zv-+fBe1S<uZFh8Ej`<EwGXy6P)S_XqxBzLxC$~YV621>8&%a@|nD6UtY>6t%f`}1X
z;>&gb?+kYt#}m7Y*GfIP6*%{gKzkhW9wSQW8n6^WsT{Qn%s1NwSK;ndfVRhhS82hD
iAA>f%ws-?V1KCMAa-tm02PkX+v<O?)zkcL@t$zSMPMm%K

literal 0
HcmV?d00001

diff --git a/src/paperless_mail/tests/test_parsers.py b/src/paperless_mail/tests/test_parsers.py
index a0c19e6f5..2bc0fbdad 100644
--- a/src/paperless_mail/tests/test_parsers.py
+++ b/src/paperless_mail/tests/test_parsers.py
@@ -268,7 +268,7 @@ class TestParser(TestCase):
     @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
     def test_tika_parse(self, m):
         html = '<html><head><meta http-equiv="content-type" content="text/html; charset=UTF-8"></head><body><p>Some Text</p></body></html>'
-        expected_text = "\n\n\n\n\n\n\n\n\nSome Text\n"
+        expected_text = "Some Text"
 
         tika_server_original = self.parser.tika_server
 
@@ -285,7 +285,7 @@ class TestParser(TestCase):
 
         # Check successful parsing
         parsed = self.parser.tika_parse(html)
-        self.assertEqual(expected_text, parsed)
+        self.assertEqual(expected_text, parsed.strip())
 
     @mock.patch("paperless_mail.parsers.MailDocumentParser.generate_pdf_from_mail")
     @mock.patch("paperless_mail.parsers.MailDocumentParser.generate_pdf_from_html")
@@ -355,19 +355,46 @@ class TestParser(TestCase):
         ) as html_expected_handle:
             self.assertHTMLEqual(html_expected_handle.read(), html_handle.read())
 
-    @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
-    def test_generate_pdf_from_mail(self, m):
+    @mock.patch("paperless_mail.parsers.requests.post")
+    @mock.patch("paperless_mail.parsers.MailDocumentParser.mail_to_html")
+    def test_generate_pdf_from_mail(
+        self,
+        mock_mail_to_html: mock.MagicMock,
+        mock_post: mock.MagicMock,
+    ):
+        mock_response = mock.MagicMock()
+        mock_response.content = b"Content"
+        mock_post.return_value = mock_response
+
+        mock_mail_to_html.return_value = "Testresponse"
+
         mail = self.parser.get_parsed(os.path.join(self.SAMPLE_FILES, "html.eml"))
 
-        pdf_path = os.path.join(self.parser.tempdir, "test_generate_pdf_from_mail.pdf")
+        retval = self.parser.generate_pdf_from_mail(mail)
+        self.assertEqual(b"Content", retval)
 
-        with open(pdf_path, "wb") as file:
-            file.write(self.parser.generate_pdf_from_mail(mail))
-            file.close()
+        mock_generate_pdf_from_mail.assert_called_once_with(
+            self.parser.get_parsed(None),
+        )
+        mock_generate_pdf_from_html.assert_called_once_with(
+            self.parser.get_parsed(None).html,
+            self.parser.get_parsed(None).attachments,
+        )
+        self.assertEqual(
+            self.parser.gotenberg_server + "/forms/pdfengines/merge",
+            mock_post.call_args.args[0],
+        )
+        self.assertEqual({}, mock_post.call_args.kwargs["headers"])
+        self.assertEqual(
+            b"Mail Return",
+            mock_post.call_args.kwargs["files"]["1_mail.pdf"][1].read(),
+        )
+        self.assertEqual(
+            b"HTML Return",
+            mock_post.call_args.kwargs["files"]["2_html.pdf"][1].read(),
+        )
 
-        extracted = extract_text(pdf_path)
-        expected = "From Name <someone@example.de>\n\n2022-10-15 09:23\n\nSubject HTML Message\n\nTo someone@example.de\n\nAttachments IntM6gnXFm00FEV5.png (6.89 KiB), 600+kbfile.txt (0.59 MiB)\n\nSome Text \n\nand an embedded image.\n\n\x0c"
-        self.assertEqual(expected, extracted)
+        mock_response.raise_for_status.assert_called_once()
 
     def test_transform_inline_html(self):
         class MailAttachmentMock:
@@ -391,13 +418,18 @@ class TestParser(TestCase):
         self.assertTrue(result[0][0] in resulting_html)
         self.assertFalse("<script" in resulting_html.lower())
 
+    @mock.patch("paperless_mail.parsers.requests.post")
     @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
-    def test_generate_pdf_from_html(self, m):
+    def test_generate_pdf_from_html(self, m, mock_post: mock.MagicMock):
         class MailAttachmentMock:
             def __init__(self, payload, content_id):
                 self.payload = payload
                 self.content_id = content_id
 
+        mock_response = mock.MagicMock()
+        mock_response.content = b"Content"
+        mock_post.return_value = mock_response
+
         result = None
 
         with open(os.path.join(self.SAMPLE_FILES, "sample.html")) as html_file:
@@ -409,12 +441,30 @@ class TestParser(TestCase):
                 ]
                 result = self.parser.generate_pdf_from_html(html, attachments)
 
-        pdf_path = os.path.join(self.parser.tempdir, "test_generate_pdf_from_html.pdf")
+        self.assertEqual(
+            self.parser.gotenberg_server + "/forms/chromium/convert/html",
+            mock_post.call_args.args[0],
+        )
+        self.assertEqual({}, mock_post.call_args.kwargs["headers"])
+        self.assertEqual(
+            {
+                "marginTop": "0.1",
+                "marginBottom": "0.1",
+                "marginLeft": "0.1",
+                "marginRight": "0.1",
+                "paperWidth": "8.27",
+                "paperHeight": "11.7",
+                "scale": "1.0",
+            },
+            mock_post.call_args.kwargs["data"],
+        )
 
-        with open(pdf_path, "wb") as file:
-            file.write(result)
-            file.close()
+        # read to assert it is a file like object.
+        mock_post.call_args.kwargs["files"]["cidpart1pNdUSz0sD3NqVtPgexamplede"][
+            1
+        ].read()
+        mock_post.call_args.kwargs["files"]["index.html"][1].read()
 
-        extracted = extract_text(pdf_path)
-        expected = "Some Text\n\n  This image should not be shown.\n\nand an embedded image.\n\nParagraph unchanged.\n\n\x0c"
-        self.assertEqual(expected, extracted)
+        mock_response.raise_for_status.assert_called_once()
+
+        self.assertEqual(b"Content", result)
diff --git a/src/paperless_mail/tests/test_parsers_live.py b/src/paperless_mail/tests/test_parsers_live.py
index 6676ebc1e..a0fa1f54d 100644
--- a/src/paperless_mail/tests/test_parsers_live.py
+++ b/src/paperless_mail/tests/test_parsers_live.py
@@ -62,7 +62,7 @@ class TestParserLive(TestCase):
     @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
     def test_tika_parse(self, m):
         html = '<html><head><meta http-equiv="content-type" content="text/html; charset=UTF-8"></head><body><p>Some Text</p></body></html>'
-        expected_text = "\n\n\n\n\n\n\n\n\nSome Text\n"
+        expected_text = "Some Text"
 
         tika_server_original = self.parser.tika_server
 
@@ -79,7 +79,7 @@ class TestParserLive(TestCase):
 
         # Check successful parsing
         parsed = self.parser.tika_parse(html)
-        self.assertEqual(expected_text, parsed)
+        self.assertEqual(expected_text, parsed.strip())
 
     @pytest.mark.skipif(
         "GOTENBERG_LIVE" not in os.environ,
@@ -124,7 +124,10 @@ class TestParserLive(TestCase):
             file.write(self.parser.generate_pdf_from_mail(mail))
             file.close()
 
-        converted = os.path.join(parser.tempdir, "test_generate_pdf_from_mail.webp")
+        converted = os.path.join(
+            self.parser.tempdir,
+            "test_generate_pdf_from_mail.webp",
+        )
         run_convert(
             density=300,
             scale="500x5000>",
@@ -149,14 +152,12 @@ class TestParserLive(TestCase):
             f"PDF looks different. Check if {converted} looks weird.",
         )
 
-    # Only run if convert is available
     @pytest.mark.skipif(
-        "PAPERLESS_TEST_SKIP_CONVERT" in os.environ,
-        reason="PAPERLESS_TEST_SKIP_CONVERT set, skipping Test",
+        "GOTENBERG_LIVE" not in os.environ,
+        reason="No gotenberg server",
     )
     @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
-    def test_generate_pdf_from_html(self, m):
-        # TODO
+    def test_generate_pdf_from_html_no_convert(self, m):
         class MailAttachmentMock:
             def __init__(self, payload, content_id):
                 self.payload = payload
@@ -179,7 +180,44 @@ class TestParserLive(TestCase):
             file.write(result)
             file.close()
 
-        converted = os.path.join(parser.tempdir, "test_generate_pdf_from_html.webp")
+        extracted = extract_text(pdf_path)
+        expected = "Some Text\n\n  This image should not be shown.\n\nand an embedded image.\n\nParagraph unchanged.\n\n\x0c"
+        self.assertEqual(expected, extracted)
+
+    @pytest.mark.skipif(
+        "GOTENBERG_LIVE" not in os.environ,
+        reason="No gotenberg server",
+    )
+    # Only run if convert is available
+    @pytest.mark.skipif(
+        "PAPERLESS_TEST_SKIP_CONVERT" in os.environ,
+        reason="PAPERLESS_TEST_SKIP_CONVERT set, skipping Test",
+    )
+    @mock.patch("documents.loggers.LoggingMixin.log")  # Disable log output
+    def test_generate_pdf_from_html(self, m):
+        class MailAttachmentMock:
+            def __init__(self, payload, content_id):
+                self.payload = payload
+                self.content_id = content_id
+
+        result = None
+
+        with open(os.path.join(self.SAMPLE_FILES, "sample.html")) as html_file:
+            with open(os.path.join(self.SAMPLE_FILES, "sample.png"), "rb") as png_file:
+                html = html_file.read()
+                png = png_file.read()
+                attachments = [
+                    MailAttachmentMock(png, "part1.pNdUSz0s.D3NqVtPg@example.de"),
+                ]
+                result = self.parser.generate_pdf_from_html(html, attachments)
+
+        pdf_path = os.path.join(self.parser.tempdir, "sample.html.pdf")
+
+        with open(pdf_path, "wb") as file:
+            file.write(result)
+            file.close()
+
+        converted = os.path.join(self.parser.tempdir, "sample.html.pdf.webp")
         run_convert(
             density=300,
             scale="500x5000>",
@@ -195,9 +233,10 @@ class TestParserLive(TestCase):
         thumb_hash = self.hashfile(converted)
 
         # The created pdf is not reproducible. But the converted image should always look the same.
-        expected_hash = (
-            "267d61f0ab8f128a037002a424b2cb4bfe18a81e17f0b70f15d241688ed47d1a"
+        expected_hash = self.hashfile(
+            os.path.join(self.SAMPLE_FILES, "sample.html.pdf.webp"),
         )
+
         self.assertEqual(
             thumb_hash,
             expected_hash,
@@ -206,6 +245,10 @@ class TestParserLive(TestCase):
         )
 
     @staticmethod
+    @pytest.mark.skipif(
+        "GOTENBERG_LIVE" not in os.environ,
+        reason="No gotenberg server",
+    )
     def test_is_online_image_still_available():
         """
         A public image is used in the html sample file. We have no control