Skip to content

Commit 7c5b6c6

Browse files
alejouribesanchezalejouribesanchez
committed
Add split and classify document support to Client
Introduces traits and methods for processing and retrieving split and classified documents via base64 and URL in the Client class. Adds corresponding tests for split and classify document functionality. Also updates the User-Agent string in request headers.
1 parent 192c917 commit 7c5b6c6

10 files changed

Lines changed: 328 additions & 9 deletions

‎src/veryfi/Client.php‎

Lines changed: 17 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -72,6 +72,12 @@
7272
use veryfi\documents\lineitems\GetLineItem;
7373
use veryfi\documents\lineitems\GetLineItems;
7474
use veryfi\documents\lineitems\UpdateLineItem;
75+
use veryfi\split\GetSplitDocument;
76+
use veryfi\split\GetSplitDocuments;
77+
use veryfi\split\ProcessSplitDocumentBase64;
78+
use veryfi\split\ProcessSplitDocumentUrl;
79+
use veryfi\classify\ProcessClassifyDocumentBase64;
80+
use veryfi\classify\ProcessClassifyDocumentUrl;
7581

7682

7783

@@ -143,14 +149,15 @@ class Client
143149
* @param string $api_version Api version to use Veryfi, currently v8
144150
* @param int $api_timeout Api timeout for call Veryfi api, by default 120
145151
*/
146-
public function __construct(string $client_id,
147-
string $client_secret,
148-
string $username,
149-
string $api_key,
150-
string $base_url = 'https://api.veryfi.com/api/',
151-
string $api_version = 'v8',
152-
int $api_timeout = 120)
153-
{
152+
public function __construct(
153+
string $client_id,
154+
string $client_secret,
155+
string $username,
156+
string $api_key,
157+
string $base_url = 'https://api.veryfi.com/api/',
158+
string $api_version = 'v8',
159+
int $api_timeout = 120
160+
) {
154161
$this->client_id = $client_id;
155162
$this->client_secret = $client_secret;
156163
$this->username = $username;
@@ -173,5 +180,7 @@ public function __construct(string $client_id,
173180
use DeleteW8BENE, GetW8BENE, GetW8BENEs, ProcessW8BENE, ProcessW8BENEBase64, ProcessW8BENEUrl;
174181
use DeleteW9, GetW9, GetW9s, ProcessW9, ProcessW9Base64, ProcessW9Url;
175182
use AddLineItem, DeleteLineItem, DeleteLineItems, GetLineItem, GetLineItems, UpdateLineItem;
183+
use GetSplitDocument, GetSplitDocuments, ProcessSplitDocumentBase64, ProcessSplitDocumentUrl;
184+
use ProcessClassifyDocumentBase64, ProcessClassifyDocumentUrl;
176185

177186
}

‎src/veryfi/classify/ProcessClassifyDocumentBase64.php‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
<?php
2+
namespace veryfi\classify;
3+
4+
trait ProcessClassifyDocumentBase64
5+
{
6+
/**
7+
* Classify a document and extract all the fields from it. https://docs.veryfi.com/api/classify/classify-a-document/
8+
*
9+
* @param string $base64_encoded_string Buffer string of a file to submit for classify extraction
10+
* @param string $file_name The file name including the extension
11+
* @param array $kwargs Additional request parameters
12+
* @return string Data extracted from the document
13+
*/
14+
public function classify_document_from_base64(string $base64_encoded_string, string $file_name, array $kwargs = []): string
15+
{
16+
$endpoint_name = '/classify/';
17+
$request_arguments = [
18+
'file_name' => $file_name,
19+
'file_data' => $base64_encoded_string,
20+
];
21+
$request_arguments = array_replace($request_arguments, $kwargs);
22+
return $this->request('POST', $endpoint_name, $request_arguments);
23+
}
24+
}

‎src/veryfi/classify/ProcessClassifyDocumentUrl.php‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
<?php
2+
namespace veryfi\classify;
3+
4+
trait ProcessClassifyDocumentUrl
5+
{
6+
/**
7+
* Classify document from url and extract all the fields from it. https://docs.veryfi.com/api/classify/classify-a-document/
8+
*
9+
* @param string|null $file_url Required if file_urls isn't specified. Publicly accessible URL to a file.
10+
* @param array|null $file_urls Required if file_url isn't specified. List of publicly accessible URLs to multiple files.
11+
* @param array $kwargs Additional request parameters
12+
* @return string Data extracted from the document.
13+
*/
14+
public function classify_document_from_url(?string $file_url = null, ?array $file_urls = null, array $kwargs = []): string
15+
{
16+
$endpoint_name = '/classify/';
17+
$request_arguments = [
18+
'file_url' => $file_url,
19+
'file_urls' => $file_urls
20+
];
21+
$request_arguments = array_replace($request_arguments, $kwargs);
22+
return $this->request('POST', $endpoint_name, $request_arguments);
23+
}
24+
}

‎src/veryfi/client/GetHeaders.php‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@ trait GetHeaders
99
private function get_headers(): array
1010
{
1111
return array(
12-
'User-Agent' => 'php veryfi-php/1.1.0',
12+
'User-Agent' => 'php veryfi-php/1.0.5',
1313
'Accept' => 'application/json',
1414
'Content-Type' => 'application/json',
1515
'Client-ID' => $this->client_id,

‎src/veryfi/split/GetSplitDocument.php‎

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,20 @@
1+
<?php
2+
namespace veryfi\split;
3+
4+
trait GetSplitDocument
5+
{
6+
/**
7+
* Veryfi's Get a Documents from PDF endpoint allows you to retrieve a collection of previously processed documents. https://docs.veryfi.com/api/receipts-invoices/get-documents-from-pdf/
8+
*
9+
* @param string $document_id ID of the document you'd like to retrieve
10+
* @param array $kwargs Additional request parameters
11+
* @return string Data extracted from the Document
12+
*/
13+
public function get_split_document(string $document_id, array $kwargs = []): string
14+
{
15+
$endpoint_name = "/documents-set/$document_id/";
16+
$request_arguments = ['id' => $document_id];
17+
$request_arguments = array_replace($request_arguments, $kwargs);
18+
return $this->request('GET', $endpoint_name, $request_arguments);
19+
}
20+
}

‎src/veryfi/split/GetSplitDocuments.php‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
<?php
2+
namespace veryfi\split;
3+
4+
trait GetSplitDocuments
5+
{
6+
/**
7+
* Veryfi's Get a Submitted PDF endpoint allows you to retrieve a collection of previously processed documents. https://docs.veryfi.com/api/receipts-invoices/get-submitted-pdf/
8+
*
9+
* @param int $page The page number. The response is capped to maximum of 50 results per page.
10+
* @param int $page_size The number of Documents per page.
11+
* @param array $kwargs Additional request parameters
12+
* @return string JSON object of previously processed documents
13+
*/
14+
public function get_split_documents(int $page = 1, int $page_size = 50, array $kwargs = []): string
15+
{
16+
$endpoint_name = '/documents-set/';
17+
$request_arguments = [
18+
'page' => $page,
19+
'page_size' => $page_size
20+
];
21+
$request_arguments = array_replace($request_arguments, $kwargs);
22+
return $this->request('GET', $endpoint_name, $request_arguments);
23+
}
24+
}

‎src/veryfi/split/ProcessSplitDocumentBase64.php‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
<?php
2+
namespace veryfi\split;
3+
4+
trait ProcessSplitDocumentBase64
5+
{
6+
/**
7+
* Split document PDF from url and extract all the fields from it. https://docs.veryfi.com/api/receipts-invoices/split-and-process-a-pdf/
8+
*
9+
* @param string $base64_encoded_string Buffer string of a file to submit for classify extraction
10+
* @param string $file_name The file name including the extension
11+
* @param array $kwargs Additional request parameters
12+
* @return string Data extracted from the document
13+
*/
14+
public function split_document_from_base64(string $base64_encoded_string, string $file_name, array $kwargs = []): string
15+
{
16+
$endpoint_name = '/documents-set/';
17+
$request_arguments = [
18+
'file_name' => $file_name,
19+
'file_data' => $base64_encoded_string,
20+
];
21+
$request_arguments = array_replace($request_arguments, $kwargs);
22+
return $this->request('POST', $endpoint_name, $request_arguments);
23+
}
24+
}

‎src/veryfi/split/ProcessSplitDocumentUrl.php‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
<?php
2+
namespace veryfi\split;
3+
4+
trait ProcessSplitDocumentUrl
5+
{
6+
/**
7+
* Split document PDF from url and extract all the fields from it. https://docs.veryfi.com/api/receipts-invoices/split-and-process-a-pdf/
8+
*
9+
* @param string|null $file_url Required if file_urls isn't specified. Publicly accessible URL to a file.
10+
* @param array|null $file_urls Required if file_url isn't specified. List of publicly accessible URLs to multiple files.
11+
* @param array $kwargs Additional request parameters
12+
* @return string Data extracted from the document.
13+
*/
14+
public function split_document_from_url(?string $file_url = null, ?array $file_urls = null, array $kwargs = []): string
15+
{
16+
$endpoint_name = '/documents-set/';
17+
$request_arguments = [
18+
'file_url' => $file_url,
19+
'file_urls' => $file_urls
20+
];
21+
$request_arguments = array_replace($request_arguments, $kwargs);
22+
return $this->request('POST', $endpoint_name, $request_arguments);
23+
}
24+
}

‎tests/ClientClassifyDocumentsTest.php‎

Lines changed: 59 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,59 @@
1+
<?php
2+
3+
use veryfi\Client;
4+
5+
require_once __DIR__ . '/ClientTestCase.php';
6+
7+
class ClientClassifyDocumentsTest extends ClientTestCase
8+
{
9+
public function test_classify_document_from_base64(): void
10+
{
11+
if ($this->mock_responses) {
12+
$veryfi_client = $this->getMockBuilder(Client::class)
13+
->onlyMethods(['exec_curl'])
14+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
15+
->getMock();
16+
17+
$file_path = __DIR__ . '/resources/processDocument.json';
18+
$file = fopen($file_path, 'r');
19+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
20+
$veryfi_client->expects($this->once())
21+
->method('exec_curl')
22+
->willReturn($file_data);
23+
24+
} else {
25+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
26+
}
27+
28+
$file_path = $this->receipt_path;
29+
$file_name = 'receipt.jpg';
30+
$base64_encoded_string = base64_encode(file_get_contents($file_path));
31+
32+
$json_response = json_decode($veryfi_client->classify_document_from_base64($base64_encoded_string, $file_name), true);
33+
$this->assertIsArray($json_response);
34+
}
35+
36+
public function test_classify_document_from_url(): void
37+
{
38+
if ($this->mock_responses) {
39+
$veryfi_client = $this->getMockBuilder(Client::class)
40+
->onlyMethods(['exec_curl'])
41+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
42+
->getMock();
43+
44+
$file_path = __DIR__ . '/resources/processDocument.json';
45+
$file = fopen($file_path, 'r');
46+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
47+
$veryfi_client->expects($this->once())
48+
->method('exec_curl')
49+
->willReturn($file_data);
50+
51+
} else {
52+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
53+
}
54+
55+
$url = 'https://raw.githubusercontent.com/veryfi/veryfi-python/master/tests/assets/receipt_public.jpg';
56+
$json_response = json_decode($veryfi_client->classify_document_from_url($url), true);
57+
$this->assertIsArray($json_response);
58+
}
59+
}

‎tests/ClientSplitDocumentsTest.php‎

Lines changed: 111 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,111 @@
1+
<?php
2+
3+
use veryfi\Client;
4+
5+
require_once __DIR__ . '/ClientTestCase.php';
6+
7+
class ClientSplitDocumentsTest extends ClientTestCase
8+
{
9+
public function test_get_split_documents(): void
10+
{
11+
if ($this->mock_responses) {
12+
$veryfi_client = $this->getMockBuilder(Client::class)
13+
->onlyMethods(['exec_curl'])
14+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
15+
->getMock();
16+
17+
$file_path = __DIR__ . '/resources/getDocuments.json';
18+
$file = fopen($file_path, 'r');
19+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
20+
$veryfi_client->expects($this->once())
21+
->method('exec_curl')
22+
->willReturn($file_data);
23+
24+
} else {
25+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
26+
}
27+
$json_response = json_decode($veryfi_client->get_split_documents(), true);
28+
$this->assertIsArray($json_response);
29+
}
30+
31+
public function test_get_split_document(): void
32+
{
33+
if ($this->mock_responses) {
34+
$veryfi_client = $this->getMockBuilder(Client::class)
35+
->onlyMethods(['exec_curl'])
36+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
37+
->getMock();
38+
39+
$file_path = __DIR__ . '/resources/getDocument.json';
40+
$file = fopen($file_path, 'r');
41+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
42+
$veryfi_client->expects($this->once())
43+
->method('exec_curl')
44+
->willReturn($file_data);
45+
$document_id = '125661908';
46+
47+
} else {
48+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
49+
$documents = json_decode($veryfi_client->get_split_documents(), true);
50+
if (isset($documents['documents'][0]['id'])) {
51+
$document_id = $documents['documents'][0]['id'];
52+
} else {
53+
$this->markTestSkipped('No documents found to test get_split_document');
54+
}
55+
}
56+
57+
$json_response = json_decode($veryfi_client->get_split_document($document_id), true);
58+
$this->assertEquals($document_id, $json_response['id']);
59+
}
60+
61+
public function test_split_document_from_base64(): void
62+
{
63+
if ($this->mock_responses) {
64+
$veryfi_client = $this->getMockBuilder(Client::class)
65+
->onlyMethods(['exec_curl'])
66+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
67+
->getMock();
68+
69+
$file_path = __DIR__ . '/resources/processDocument.json';
70+
$file = fopen($file_path, 'r');
71+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
72+
$veryfi_client->expects($this->once())
73+
->method('exec_curl')
74+
->willReturn($file_data);
75+
76+
} else {
77+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
78+
}
79+
80+
$file_path = $this->receipt_path;
81+
$file_name = 'receipt.jpg';
82+
$base64_encoded_string = base64_encode(file_get_contents($file_path));
83+
84+
$json_response = json_decode($veryfi_client->split_document_from_base64($base64_encoded_string, $file_name), true);
85+
$this->assertIsArray($json_response);
86+
}
87+
88+
public function test_split_document_from_url(): void
89+
{
90+
if ($this->mock_responses) {
91+
$veryfi_client = $this->getMockBuilder(Client::class)
92+
->onlyMethods(['exec_curl'])
93+
->setConstructorArgs([$this->client_id, $this->client_secret, $this->username, $this->api_key])
94+
->getMock();
95+
96+
$file_path = __DIR__ . '/resources/processDocument.json';
97+
$file = fopen($file_path, 'r');
98+
$file_data = mb_convert_encoding(fread($file, filesize($file_path)), 'UTF-8');
99+
$veryfi_client->expects($this->once())
100+
->method('exec_curl')
101+
->willReturn($file_data);
102+
103+
} else {
104+
$veryfi_client = new Client($this->client_id, $this->client_secret, $this->username, $this->api_key);
105+
}
106+
107+
$url = 'https://raw.githubusercontent.com/veryfi/veryfi-python/master/tests/assets/receipt_public.jpg';
108+
$json_response = json_decode($veryfi_client->split_document_from_url($url), true);
109+
$this->assertIsArray($json_response);
110+
}
111+
}

0 commit comments

Comments
 (0)