@things-factory/integration-base 8.0.0-alpha.30 → 8.0.0-alpha.34
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist-server/engine/connector/headless-connector.d.ts +23 -0
- package/dist-server/engine/connector/headless-connector.js +283 -0
- package/dist-server/engine/connector/headless-connector.js.map +1 -0
- package/dist-server/engine/connector/http-connector.js +1 -1
- package/dist-server/engine/connector/http-connector.js.map +1 -1
- package/dist-server/engine/connector/index.d.ts +1 -0
- package/dist-server/engine/connector/index.js +1 -0
- package/dist-server/engine/connector/index.js.map +1 -1
- package/dist-server/engine/index.d.ts +1 -0
- package/dist-server/engine/index.js +1 -0
- package/dist-server/engine/index.js.map +1 -1
- package/dist-server/engine/resource-pool/headless-pool.d.ts +1 -0
- package/dist-server/engine/resource-pool/headless-pool.js +121 -0
- package/dist-server/engine/resource-pool/headless-pool.js.map +1 -0
- package/dist-server/engine/resource-pool/index.d.ts +1 -0
- package/dist-server/engine/resource-pool/index.js +5 -0
- package/dist-server/engine/resource-pool/index.js.map +1 -0
- package/dist-server/engine/task/headless-post.js +19 -33
- package/dist-server/engine/task/headless-post.js.map +1 -1
- package/dist-server/engine/task/headless-scrap.js +20 -13
- package/dist-server/engine/task/headless-scrap.js.map +1 -1
- package/dist-server/tsconfig.tsbuildinfo +1 -1
- package/package.json +4 -3
- package/server/engine/connector/headless-connector.ts +341 -0
- package/server/engine/connector/http-connector.ts +1 -1
- package/server/engine/connector/index.ts +1 -0
- package/server/engine/index.ts +1 -0
- package/server/engine/resource-pool/headless-pool.ts +136 -0
- package/server/engine/resource-pool/index.ts +1 -0
- package/server/engine/task/headless-post.ts +21 -40
- package/server/engine/task/headless-scrap.ts +21 -18
- package/translations/en.json +11 -4
- package/translations/ja.json +11 -4
- package/translations/ko.json +11 -4
- package/translations/ms.json +11 -4
- package/translations/zh.json +11 -4
@@ -6,17 +6,16 @@ const url_1 = require("url");
|
|
6
6
|
const utils_1 = require("@things-factory/utils");
|
7
7
|
const task_registry_1 = require("../task-registry");
|
8
8
|
const connection_manager_1 = require("../connection-manager");
|
9
|
-
const headless_pool_for_scenario_1 = require("./utils/headless-pool-for-scenario");
|
10
9
|
async function HeadlessPost(step, { logger, data, domain }) {
|
11
|
-
|
12
|
-
|
13
|
-
|
10
|
+
const { connection: connectionName, params: stepOptions } = step;
|
11
|
+
const { headers: requestHeaders, contentType, path, accessor } = stepOptions || {};
|
12
|
+
const connection = connection_manager_1.ConnectionManager.getConnectionInstanceByName(domain, connectionName);
|
14
13
|
if (!connection) {
|
15
|
-
throw new Error(`
|
14
|
+
throw new Error(`Connection '${connectionName}' is not established.`);
|
16
15
|
}
|
17
|
-
|
18
|
-
|
19
|
-
|
16
|
+
const { endpoint, params: connectionParams, acquireSessionPage, releasePage } = connection;
|
17
|
+
const headers = Object.assign({}, requestHeaders);
|
18
|
+
let body = (0, utils_1.access)(accessor, data);
|
20
19
|
if (contentType && body) {
|
21
20
|
headers['content-type'] = contentType;
|
22
21
|
switch (contentType) {
|
@@ -31,24 +30,23 @@ async function HeadlessPost(step, { logger, data, domain }) {
|
|
31
30
|
for (const prop in body) {
|
32
31
|
searchParams.set(prop, body[prop]);
|
33
32
|
}
|
34
|
-
body = searchParams;
|
33
|
+
body = searchParams.toString();
|
35
34
|
break;
|
36
35
|
}
|
37
36
|
}
|
38
|
-
|
37
|
+
const options = {
|
39
38
|
method: 'POST',
|
40
39
|
headers,
|
41
40
|
body
|
42
41
|
};
|
43
|
-
|
42
|
+
const { rejectUnauthorized } = connectionParams;
|
44
43
|
if (!rejectUnauthorized) {
|
45
44
|
const httpsAgent = new https_1.default.Agent({
|
46
45
|
rejectUnauthorized
|
47
46
|
});
|
48
47
|
options.agent = httpsAgent;
|
49
48
|
}
|
50
|
-
const
|
51
|
-
const page = await browser.newPage();
|
49
|
+
const page = await acquireSessionPage();
|
52
50
|
try {
|
53
51
|
page.on('console', async (msg) => {
|
54
52
|
console.log(`[browser ${msg.type()}] ${msg.text()}`);
|
@@ -70,12 +68,12 @@ async function HeadlessPost(step, { logger, data, domain }) {
|
|
70
68
|
data: response
|
71
69
|
};
|
72
70
|
}
|
73
|
-
catch (
|
74
|
-
|
71
|
+
catch (error) {
|
72
|
+
logger.error('Error in HeadlessPost:', error);
|
73
|
+
throw error;
|
75
74
|
}
|
76
75
|
finally {
|
77
|
-
page
|
78
|
-
(0, headless_pool_for_scenario_1.getHeadlessPool)().release(browser);
|
76
|
+
await releasePage(page);
|
79
77
|
}
|
80
78
|
}
|
81
79
|
HeadlessPost.parameterSpec = [
|
@@ -95,22 +93,10 @@ HeadlessPost.parameterSpec = [
|
|
95
93
|
label: 'content-type',
|
96
94
|
property: {
|
97
95
|
options: [
|
98
|
-
{
|
99
|
-
|
100
|
-
|
101
|
-
}
|
102
|
-
{
|
103
|
-
display: 'application/json',
|
104
|
-
value: 'application/json'
|
105
|
-
},
|
106
|
-
{
|
107
|
-
display: 'text/plain',
|
108
|
-
value: 'text/plain'
|
109
|
-
},
|
110
|
-
{
|
111
|
-
display: 'application/x-www-form-urlencoded',
|
112
|
-
value: 'application/x-www-form-urlencoded'
|
113
|
-
}
|
96
|
+
{ display: '', value: '' },
|
97
|
+
{ display: 'application/json', value: 'application/json' },
|
98
|
+
{ display: 'text/plain', value: 'text/plain' },
|
99
|
+
{ display: 'application/x-www-form-urlencoded', value: 'application/x-www-form-urlencoded' }
|
114
100
|
]
|
115
101
|
}
|
116
102
|
},
|
@@ -1 +1 @@
|
|
1
|
-
{"version":3,"file":"headless-post.js","sourceRoot":"","sources":["../../../server/engine/task/headless-post.ts"],"names":[],"mappings":";;;AAAA,0DAAyB;AACzB,6BAAyB;AAEzB,iDAA8C;AAC9C,oDAA+C;AAC/C,8DAAyD;AAEzD,
|
1
|
+
{"version":3,"file":"headless-post.js","sourceRoot":"","sources":["../../../server/engine/task/headless-post.ts"],"names":[],"mappings":";;;AAAA,0DAAyB;AACzB,6BAAyB;AAEzB,iDAA8C;AAC9C,oDAA+C;AAC/C,8DAAyD;AAEzD,KAAK,UAAU,YAAY,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE;IACxD,MAAM,EAAE,UAAU,EAAE,cAAc,EAAE,MAAM,EAAE,WAAW,EAAE,GAAG,IAAI,CAAA;IAChE,MAAM,EAAE,OAAO,EAAE,cAAc,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,WAAW,IAAI,EAAE,CAAA;IAElF,MAAM,UAAU,GAAG,sCAAiB,CAAC,2BAA2B,CAAC,MAAM,EAAE,cAAc,CAAC,CAAA;IAExF,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,MAAM,IAAI,KAAK,CAAC,eAAe,cAAc,uBAAuB,CAAC,CAAA;IACvE,CAAC;IAED,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,WAAW,EAAE,GAAG,UAAU,CAAA;IAE1F,MAAM,OAAO,qBACR,cAAc,CAClB,CAAA;IAED,IAAI,IAAI,GAAG,IAAA,cAAM,EAAC,QAAQ,EAAE,IAAI,CAAC,CAAA;IACjC,IAAI,WAAW,IAAI,IAAI,EAAE,CAAC;QACxB,OAAO,CAAC,cAAc,CAAC,GAAG,WAAW,CAAA;QACrC,QAAQ,WAAW,EAAE,CAAC;YACpB,KAAK,YAAY;gBACf,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAA;gBAC3B,MAAK;YACP,KAAK,kBAAkB;gBACrB,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAA;gBAC3B,MAAK;YACP,KAAK,mCAAmC;gBACtC,MAAM,YAAY,GAAG,IAAI,eAAe,EAAE,CAAA;gBAC1C,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;oBACxB,YAAY,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;gBACpC,CAAC;gBACD,IAAI,GAAG,YAAY,CAAC,QAAQ,EAAE,CAAA;gBAC9B,MAAK;QACT,CAAC;IACH,CAAC;IAED,MAAM,OAAO,GAAG;QACd,MAAM,EAAE,MAAM;QACd,OAAO;QACP,IAAI;KACE,CAAA;IAER,MAAM,EAAE,kBAAkB,EAAE,GAAG,gBAAgB,CAAA;IAE/C,IAAI,CAAC,kBAAkB,EAAE,CAAC;QACxB,MAAM,UAAU,GAAG,IAAI,eAAK,CAAC,KAAK,CAAC;YACjC,kBAAkB;SACnB,CAAC,CAAA;QACF,OAAO,CAAC,KAAK,GAAG,UAAU,CAAA;IAC5B,CAAC;IAED,MAAM,IAAI,GAAG,MAAM,kBAAkB,EAAE,CAAA;IAEvC,IAAI,CAAC;QACH,IAAI,CAAC,EAAE,CAAC,SAAS,EAAE,KAAK,EAAC,GAAG,EAAC,EAAE;YAC7B,OAAO,CAAC,GAAG,CAAC,YAAY,GAAG,CAAC,IAAI,EAAE,KAAK,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA;QACtD,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,EAAE,CAAC,eAAe,EAAE,OAAO,CAAC,EAAE;YACjC,OAAO,CAAC,GAAG,CAAC,iBAAiB,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC,CAAA;QAC/C,CAAC,CAAC,CAAA;QAEF,MAAM,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,CAAC,CAAA;QAExD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,QAAQ,CAClC,KAAK,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE;YAC3B,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,SAAS,EAAE,OAAO,CAAC,CAAA;YAEhD,IAAI,QAAQ,CAAC,EAAE,IAAI,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,QAAQ,CAAC,kBAAkB,CAAC,EAAE,CAAC;gBACrF,OAAO,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAA;YAC9B,CAAC;iBAAM,CAAC;gBACN,OAAO,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAA;YAC9B,CAAC;QACH,CAAC,EACD,IAAI,SAAG,CAAC,IAAI,EAAE,QAAQ,CAAC,EACvB,OAAO,CACR,CAAA;QAED,OAAO;YACL,IAAI,EAAE,QAAQ;SACf,CAAA;IACH,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,CAAC,KAAK,CAAC,wBAAwB,EAAE,KAAK,CAAC,CAAA;QAC7C,MAAM,KAAK,CAAA;IACb,CAAC;YAAS,CAAC;QACT,MAAM,WAAW,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;AACH,CAAC;AAED,YAAY,CAAC,aAAa,GAAG;IAC3B;QACE,IAAI,EAAE,QAAQ;QACd,IAAI,EAAE,MAAM;QACZ,KAAK,EAAE,MAAM;KACd;IACD;QACE,IAAI,EAAE,cAAc;QACpB,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;KACjB;IACD;QACE,IAAI,EAAE,QAAQ;QACd,IAAI,EAAE,aAAa;QACnB,KAAK,EAAE,cAAc;QACrB,QAAQ,EAAE;YACR,OAAO,EAAE;gBACP,EAAE,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,EAAE,EAAE;gBAC1B,EAAE,OAAO,EAAE,kBAAkB,EAAE,KAAK,EAAE,kBAAkB,EAAE;gBAC1D,EAAE,OAAO,EAAE,YAAY,EAAE,KAAK,EAAE,YAAY,EAAE;gBAC9C,EAAE,OAAO,EAAE,mCAAmC,EAAE,KAAK,EAAE,mCAAmC,EAAE;aAC7F;SACF;KACF;IACD;QACE,IAAI,EAAE,qBAAqB;QAC3B,IAAI,EAAE,UAAU;QAChB,KAAK,EAAE,UAAU;KAClB;CACF,CAAA;AAED,4BAAY,CAAC,mBAAmB,CAAC,eAAe,EAAE,YAAY,CAAC,CAAA","sourcesContent":["import https from 'https'\nimport { URL } from 'url'\n\nimport { access } from '@things-factory/utils'\nimport { TaskRegistry } from '../task-registry'\nimport { ConnectionManager } from '../connection-manager'\n\nasync function HeadlessPost(step, { logger, data, domain }) {\n const { connection: connectionName, params: stepOptions } = step\n const { headers: requestHeaders, contentType, path, accessor } = stepOptions || {}\n\n const connection = ConnectionManager.getConnectionInstanceByName(domain, connectionName)\n\n if (!connection) {\n throw new Error(`Connection '${connectionName}' is not established.`)\n }\n\n const { endpoint, params: connectionParams, acquireSessionPage, releasePage } = connection\n\n const headers = {\n ...requestHeaders\n }\n\n let body = access(accessor, data)\n if (contentType && body) {\n headers['content-type'] = contentType\n switch (contentType) {\n case 'text/plain':\n body = JSON.stringify(body)\n break\n case 'application/json':\n body = JSON.stringify(body)\n break\n case 'application/x-www-form-urlencoded':\n const searchParams = new URLSearchParams()\n for (const prop in body) {\n searchParams.set(prop, body[prop])\n }\n body = searchParams.toString()\n break\n }\n }\n\n const options = {\n method: 'POST',\n headers,\n body\n } as any\n\n const { rejectUnauthorized } = connectionParams\n\n if (!rejectUnauthorized) {\n const httpsAgent = new https.Agent({\n rejectUnauthorized\n })\n options.agent = httpsAgent\n }\n\n const page = await acquireSessionPage()\n\n try {\n page.on('console', async msg => {\n console.log(`[browser ${msg.type()}] ${msg.text()}`)\n })\n\n page.on('requestfailed', request => {\n console.log('Request failed:', request.url())\n })\n\n await page.goto(endpoint, { waitUntil: 'networkidle2' })\n\n const response = await page.evaluate(\n async (urlString, options) => {\n const response = await fetch(urlString, options)\n\n if (response.ok && response.headers.get('content-type').includes('application/json')) {\n return await response.json()\n } else {\n return await response.text()\n }\n },\n new URL(path, endpoint),\n options\n )\n\n return {\n data: response\n }\n } catch (error) {\n logger.error('Error in HeadlessPost:', error)\n throw error\n } finally {\n await releasePage(page)\n }\n}\n\nHeadlessPost.parameterSpec = [\n {\n type: 'string',\n name: 'path',\n label: 'path'\n },\n {\n type: 'http-headers',\n name: 'headers',\n label: 'headers'\n },\n {\n type: 'select',\n name: 'contentType',\n label: 'content-type',\n property: {\n options: [\n { display: '', value: '' },\n { display: 'application/json', value: 'application/json' },\n { display: 'text/plain', value: 'text/plain' },\n { display: 'application/x-www-form-urlencoded', value: 'application/x-www-form-urlencoded' }\n ]\n }\n },\n {\n type: 'scenario-step-input',\n name: 'accessor',\n label: 'accessor'\n }\n]\n\nTaskRegistry.registerTaskHandler('headless-post', HeadlessPost)\n"]}
|
@@ -3,24 +3,31 @@ Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
3
|
const url_1 = require("url");
|
4
4
|
const task_registry_1 = require("../task-registry");
|
5
5
|
const connection_manager_1 = require("../connection-manager");
|
6
|
-
const headless_pool_for_scenario_1 = require("./utils/headless-pool-for-scenario");
|
7
6
|
async function HeadlessScrap(step, { logger, data, domain }) {
|
8
|
-
|
9
|
-
|
10
|
-
|
7
|
+
const { connection: connectionName, params: stepOptions } = step;
|
8
|
+
const { headers: requestHeaders, path, selectors = [] } = stepOptions || {};
|
9
|
+
const connection = connection_manager_1.ConnectionManager.getConnectionInstanceByName(domain, connectionName);
|
11
10
|
if (!connection) {
|
12
|
-
throw new Error(`
|
11
|
+
throw new Error(`Connection '${connectionName}' is not established.`);
|
13
12
|
}
|
14
|
-
|
15
|
-
|
16
|
-
const
|
17
|
-
const page = await browser.newPage();
|
13
|
+
const { endpoint, params: connectionParams, acquireSessionPage, releasePage } = connection;
|
14
|
+
const headers = Object.assign({}, requestHeaders);
|
15
|
+
const page = await acquireSessionPage();
|
18
16
|
try {
|
19
17
|
page.on('console', async (msg) => {
|
20
18
|
console.log(`[browser ${msg.type()}] ${msg.text()}`);
|
21
19
|
});
|
22
20
|
page.on('requestfailed', request => {
|
23
|
-
|
21
|
+
var _a;
|
22
|
+
console.log('Request failed:');
|
23
|
+
console.log(`- URL: ${request.url()}`);
|
24
|
+
console.log(`- Method: ${request.method()}`);
|
25
|
+
console.log(`- Failure Text: ${(_a = request.failure()) === null || _a === void 0 ? void 0 : _a.errorText}`);
|
26
|
+
console.log(`- Headers:`, request.headers());
|
27
|
+
// POST 데이터 (필요한 경우)
|
28
|
+
if (request.postData()) {
|
29
|
+
console.log(`- Post Data: ${request.postData()}`);
|
30
|
+
}
|
24
31
|
});
|
25
32
|
await page.setExtraHTTPHeaders(headers);
|
26
33
|
await page.goto(new url_1.URL(path, endpoint), { waitUntil: 'networkidle2' });
|
@@ -36,11 +43,11 @@ async function HeadlessScrap(step, { logger, data, domain }) {
|
|
36
43
|
};
|
37
44
|
}
|
38
45
|
catch (e) {
|
39
|
-
|
46
|
+
logger.error('Error in HeadlessScrap:', e);
|
47
|
+
throw e;
|
40
48
|
}
|
41
49
|
finally {
|
42
|
-
page
|
43
|
-
(0, headless_pool_for_scenario_1.getHeadlessPool)().release(browser);
|
50
|
+
await releasePage(page);
|
44
51
|
}
|
45
52
|
}
|
46
53
|
HeadlessScrap.parameterSpec = [
|
@@ -1 +1 @@
|
|
1
|
-
{"version":3,"file":"headless-scrap.js","sourceRoot":"","sources":["../../../server/engine/task/headless-scrap.ts"],"names":[],"mappings":";;AAAA,6BAAyB;AAEzB,oDAA+C;AAC/C,8DAAyD;AAEzD,
|
1
|
+
{"version":3,"file":"headless-scrap.js","sourceRoot":"","sources":["../../../server/engine/task/headless-scrap.ts"],"names":[],"mappings":";;AAAA,6BAAyB;AAEzB,oDAA+C;AAC/C,8DAAyD;AAEzD,KAAK,UAAU,aAAa,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE;IACzD,MAAM,EAAE,UAAU,EAAE,cAAc,EAAE,MAAM,EAAE,WAAW,EAAE,GAAG,IAAI,CAAA;IAChE,MAAM,EAAE,OAAO,EAAE,cAAc,EAAE,IAAI,EAAE,SAAS,GAAG,EAAE,EAAE,GAAG,WAAW,IAAI,EAAE,CAAA;IAE3E,MAAM,UAAU,GAAG,sCAAiB,CAAC,2BAA2B,CAAC,MAAM,EAAE,cAAc,CAAC,CAAA;IAExF,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,MAAM,IAAI,KAAK,CAAC,eAAe,cAAc,uBAAuB,CAAC,CAAA;IACvE,CAAC;IAED,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,WAAW,EAAE,GAAG,UAAU,CAAA;IAE1F,MAAM,OAAO,qBACR,cAAc,CAClB,CAAA;IAED,MAAM,IAAI,GAAG,MAAM,kBAAkB,EAAE,CAAA;IAEvC,IAAI,CAAC;QACH,IAAI,CAAC,EAAE,CAAC,SAAS,EAAE,KAAK,EAAC,GAAG,EAAC,EAAE;YAC7B,OAAO,CAAC,GAAG,CAAC,YAAY,GAAG,CAAC,IAAI,EAAE,KAAK,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA;QACtD,CAAC,CAAC,CAAA;QAEF,IAAI,CAAC,EAAE,CAAC,eAAe,EAAE,OAAO,CAAC,EAAE;;YACjC,OAAO,CAAC,GAAG,CAAC,iBAAiB,CAAC,CAAA;YAC9B,OAAO,CAAC,GAAG,CAAC,UAAU,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAA;YACtC,OAAO,CAAC,GAAG,CAAC,aAAa,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA;YAC5C,OAAO,CAAC,GAAG,CAAC,mBAAmB,MAAA,OAAO,CAAC,OAAO,EAAE,0CAAE,SAAS,EAAE,CAAC,CAAA;YAC9D,OAAO,CAAC,GAAG,CAAC,YAAY,EAAE,OAAO,CAAC,OAAO,EAAE,CAAC,CAAA;YAE5C,oBAAoB;YACpB,IAAI,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC;gBACvB,OAAO,CAAC,GAAG,CAAC,gBAAgB,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAA;YACnD,CAAC;QACH,CAAC,CAAC,CAAA;QAEF,MAAM,IAAI,CAAC,mBAAmB,CAAC,OAAO,CAAC,CAAA;QACvC,MAAM,IAAI,CAAC,IAAI,CAAC,IAAI,SAAG,CAAC,IAAI,EAAE,QAAQ,CAAC,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,CAAC,CAAA;QAEvE,MAAM,MAAM,GAAG,EAAE,CAAA;QAEjB,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YACjC,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,QAAQ,CAAA;YAChC,MAAM,CAAC,IAAI,CAAC,GAAG,MAAM,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,EAAE;gBACjD,OAAO,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,OAAO,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAA;YAC5D,CAAC,CAAC,CAAA;QACJ,CAAC;QAED,OAAO;YACL,IAAI,EAAE,MAAM;SACb,CAAA;IACH,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACX,MAAM,CAAC,KAAK,CAAC,yBAAyB,EAAE,CAAC,CAAC,CAAA;QAC1C,MAAM,CAAC,CAAA;IACT,CAAC;YAAS,CAAC;QACT,MAAM,WAAW,CAAC,IAAI,CAAC,CAAA;IACzB,CAAC;AACH,CAAC;AAED,aAAa,CAAC,aAAa,GAAG;IAC5B;QACE,IAAI,EAAE,QAAQ;QACd,IAAI,EAAE,MAAM;QACZ,KAAK,EAAE,MAAM;KACd;IACD;QACE,IAAI,EAAE,cAAc;QACpB,IAAI,EAAE,SAAS;QACf,KAAK,EAAE,SAAS;KACjB;IACD;QACE,IAAI,EAAE,SAAS;QACf,IAAI,EAAE,WAAW;QACjB,KAAK,EAAE,WAAW;KACnB;CACF,CAAA;AAED,4BAAY,CAAC,mBAAmB,CAAC,gBAAgB,EAAE,aAAa,CAAC,CAAA","sourcesContent":["import { URL } from 'url'\n\nimport { TaskRegistry } from '../task-registry'\nimport { ConnectionManager } from '../connection-manager'\n\nasync function HeadlessScrap(step, { logger, data, domain }) {\n const { connection: connectionName, params: stepOptions } = step\n const { headers: requestHeaders, path, selectors = [] } = stepOptions || {}\n\n const connection = ConnectionManager.getConnectionInstanceByName(domain, connectionName)\n\n if (!connection) {\n throw new Error(`Connection '${connectionName}' is not established.`)\n }\n\n const { endpoint, params: connectionParams, acquireSessionPage, releasePage } = connection\n\n const headers = {\n ...requestHeaders\n }\n\n const page = await acquireSessionPage()\n\n try {\n page.on('console', async msg => {\n console.log(`[browser ${msg.type()}] ${msg.text()}`)\n })\n\n page.on('requestfailed', request => {\n console.log('Request failed:')\n console.log(`- URL: ${request.url()}`)\n console.log(`- Method: ${request.method()}`)\n console.log(`- Failure Text: ${request.failure()?.errorText}`)\n console.log(`- Headers:`, request.headers())\n\n // POST 데이터 (필요한 경우)\n if (request.postData()) {\n console.log(`- Post Data: ${request.postData()}`)\n }\n })\n\n await page.setExtraHTTPHeaders(headers)\n await page.goto(new URL(path, endpoint), { waitUntil: 'networkidle2' })\n\n const result = {}\n\n for (const selector of selectors) {\n const { text, value } = selector\n result[text] = await page.$$eval(value, elements => {\n return elements.map(element => element.textContent.trim())\n })\n }\n\n return {\n data: result\n }\n } catch (e) {\n logger.error('Error in HeadlessScrap:', e)\n throw e\n } finally {\n await releasePage(page)\n }\n}\n\nHeadlessScrap.parameterSpec = [\n {\n type: 'string',\n name: 'path',\n label: 'path'\n },\n {\n type: 'http-headers',\n name: 'headers',\n label: 'headers'\n },\n {\n type: 'options',\n name: 'selectors',\n label: 'selectors'\n }\n]\n\nTaskRegistry.registerTaskHandler('headless-scrap', HeadlessScrap)\n"]}
|