Apify Unofficial SDK

1FROM golang:1.20.10-alpine
2
3WORKDIR /app
4
5# pre-copy/cache go.mod for pre-downloading dependencies and only redownloading them in subsequent builds if they change
6
7#COPY go.mod go.sum ./
8#RUN go mod download && go mod verify
9
10COPY . .
11
12RUN cd src \
13&& go mod init example \
14&& go mod tidy \
15&& go build -o /usr/local/bin/app .
16
17# && go mod edit -replace localhost/apify=../apify \
18
19
20CMD ["app"]

1{
2    "actorSpecification": 1,
3    "name": "my-actor-4",
4    "title": "Empty Python project",
5    "description": "Empty project in Python.",
6    "version": "0.0",
7    "buildTag": "latest",
8    "meta": {
9        "templateId": "python-empty"
10    },
11    "dockerfile": "./Dockerfile"
12}

1package apify
2import (
3	"fmt"
4	"net/http"
5)
6
7////////////////////////////////////////////////////////////
8
9type clientHttp struct {
10	token 		string
11	base_url 	string
12}
13
14func (c * clientHttp) url (path string) string {
15	return c.base_url + "/" + path
16}
17
18func (c * clientHttp) request (method, path string) string {
19	cli := & http.Client {
20		//CheckRedirect: redirectPolicyFunc,
21	}
22
23	url := c.url(path)
24	req, err := http.NewRequest(method, url, nil)
25
26	if c.token != "" {
27		req.Header.Add("Authorization", c.token)
28	}
29	
30	res, err := cli.Do(req)
31
32	fmt.Println(res, err)
33	
34	return ""
35}
36
37////////////////////////////////////////////////////////////
38
39/**
40type storeBase struct {
41}
42**/
43
44type storeHttp struct {
45	clientHttp
46}
47
48type DatasetStoreHttp struct {
49	storeHttp
50}
51
52type KeyValueStore struct {
53	storeHttp
54}
55
56func (s * KeyValueStore) GetValue (key string) string {
57	res := s.request("GET", fmt.Sprintf("records/%s", key))
58	return res
59}
60
61////////////////////////////////////////////////////////////
62type _actor struct {
63	kv_store KeyValueStore
64}
65
66func newActor() * _actor {
67	return & _actor {
68		kv_store: KeyValueStore {
69			storeHttp {
70				clientHttp {
71					token		: config.TOKEN,
72					base_url	: fmt.Sprintf("%s/v2/key-value-stores/%s", config.API_PUBLIC_BASE_URL, config.DEFAULT_KEY_VALUE_STORE_ID),
73				},
74			},
75		},
76	}
77}
78
79func (a * _actor) GetInput () string {
80	return a.GetValue(config.INPUT_KEY)
81}
82func (a * _actor) GetValue (key string) string {
83	return a.kv_store.GetValue(key)
84}
85
86////////////////////////////////////////////////////////////
87
88// private
89var config 	= newConfiguration()
90
91// public
92var Actor 	= newActor()

1package apify
2
3func Hello() string {
4	a := TOKEN
5	println( a )
6	return "Hello"
7}

1package apify
2
3import (
4	"os"
5)
6
7type Configuration struct  {
8	ACTOR_BUILD_ID string
9	ACTOR_BUILD_NUMBER string
10	ACTOR_EVENTS_WS_URL string
11	ACTOR_ID string
12	ACTOR_RUN_ID string
13	ACTOR_TASK_ID string
14	API_BASE_URL string
15	API_PUBLIC_BASE_URL string
16	CHROME_EXECUTABLE_PATH string
17	CONTAINER_PORT string
18	CONTAINER_URL string
19	DEDICATED_CPUS string
20	DEFAULT_BROWSER_PATH string
21	DEFAULT_DATASET_ID string
22	DEFAULT_KEY_VALUE_STORE_ID string
23	DEFAULT_REQUEST_QUEUE_ID string
24	DISABLE_BROWSER_SANDBOX string
25	HEADLESS string
26	INPUT_KEY string
27	INPUT_SECRETS_PRIVATE_KEY_FILE string
28	INPUT_SECRETS_PRIVATE_KEY_PASSPHRASE string
29	IS_AT_HOME string
30	MAX_USED_CPU_RATIO string
31	MEMORY_MBYTES string
32	META_ORIGIN string
33	METAMORPH_AFTER_SLEEP_MILLIS string
34	PERSIST_STATE_INTERVAL_MILLIS string
35	PERSIST_STORAGE string
36	PROXY_HOSTNAME string
37	PROXY_PASSWORD string
38	PROXY_PORT string
39	PROXY_STATUS_URL string
40	PURGE_ON_START string
41	STARTED_AT string
42	TIMEOUT_AT string
43	TOKEN string
44	USER_ID string
45	XVFB string
46	SYSTEM_INFO_INTERVAL_MILLIS string
47}
48
49func _getenv_or(key string, def string) string {
50	val := os.Getenv(key)
51	if val == "" { return def }
52	return val
53}
54func _getenv(key string) string { return _getenv_or(key, "") }
55
56func newConfiguration() * Configuration {
57	return & Configuration {
58		ACTOR_BUILD_ID 				: _getenv(BUILD_ID),
59		ACTOR_BUILD_NUMBER 			: _getenv(BUILD_NUMBER),
60		ACTOR_EVENTS_WS_URL 		: _getenv(EVENTS_WEBSOCKET_URL),
61		ACTOR_ID 					: _getenv(ID),
62		ACTOR_RUN_ID 				: _getenv(RUN_ID),
63		ACTOR_TASK_ID 				: _getenv(TASK_ID),
64		API_BASE_URL 				: _getenv_or(API_BASE_URL, "https://api.apify.com"),
65		API_PUBLIC_BASE_URL 		: _getenv_or(API_PUBLIC_BASE_URL, "https://api.apify.com"),
66		//CHROME_EXECUTABLE_PATH 		: _getenv(CHROME_EXECUTABLE_PATH),
67		//CONTAINER_PORT 				: container_port or _getenv(WEB_SERVER_PORT, 4321),
68		//CONTAINER_URL 					: container_url or _getenv(WEB_SERVER_URL, "http://localhost:4321"),
69		//DEDICATED_CPUS 				: _getenv(DEDICATED_CPUS),
70		//DEFAULT_BROWSER_PATH 			: _getenv(DEFAULT_BROWSER_PATH),
71		DEFAULT_DATASET_ID				: _getenv_or(DEFAULT_DATASET_ID, "default"),
72		DEFAULT_KEY_VALUE_STORE_ID 	: _getenv_or(DEFAULT_KEY_VALUE_STORE_ID, "default"),
73		DEFAULT_REQUEST_QUEUE_ID 		: _getenv_or(DEFAULT_REQUEST_QUEUE_ID, "default"),
74		//DISABLE_BROWSER_SANDBOX 		: _getenv(DISABLE_BROWSER_SANDBOX, False),
75		//HEADLESS 						: _getenv(HEADLESS, True),
76		INPUT_KEY 					: _getenv_or(INPUT_KEY, "INPUT"),
77		//INPUT_SECRETS_PRIVATE_KEY_FILE 		: _getenv(INPUT_SECRETS_PRIVATE_KEY_FILE),
78		//INPUT_SECRETS_PRIVATE_KEY_PASSPHRASE 	: _getenv(INPUT_SECRETS_PRIVATE_KEY_PASSPHRASE),
79		IS_AT_HOME 					: _getenv(IS_AT_HOME),
80		//MAX_USED_CPU_RATIO 			: max_used_cpu_ratio or _getenv(MAX_USED_CPU_RATIO, 0.95),
81		//MEMORY_MBYTES 					: _getenv(MEMORY_MBYTES),
82		META_ORIGIN 					: _getenv(META_ORIGIN),
83		//METAMORPH_AFTER_SLEEP_MILLIS 	: metamorph_after_sleep_millis or _getenv(METAMORPH_AFTER_SLEEP_MILLIS, 300000)  # noqa: E501
84		//PERSIST_STATE_INTERVAL_MILLIS 	: persist_state_interval_millis or _getenv(PERSIST_STATE_INTERVAL_MILLIS, 60000)  # noqa: E501
85		//PERSIST_STORAGE 				: persist_storage or _getenv(PERSIST_STORAGE, True)
86		//PROXY_HOSTNAME 				: proxy_hostname or _getenv(PROXY_HOSTNAME, "proxy.apify.com")
87		//PROXY_PASSWORD 				: proxy_password or _getenv(PROXY_PASSWORD)
88		//PROXY_PORT 					: proxy_port or _getenv(PROXY_PORT, 8000)
89		//PROXY_STATUS_URL 				: proxy_status_url or _getenv(PROXY_STATUS_URL, "http://proxy.apify.com")
90		//PURGE_ON_START 				: purge_on_start or _getenv(PURGE_ON_START, False)
91		//STARTED_AT 					: _getenv(STARTED_AT)
92		//TIMEOUT_AT 					: _getenv(TIMEOUT_AT)
93		TOKEN 							: _getenv(TOKEN),
94		USER_ID 						: _getenv(USER_ID),
95		//XVFB 							: _getenv(XVFB, False)
96		//SYSTEM_INFO_INTERVAL_MILLIS 	: system_info_interval_millis or _getenv(SYSTEM_INFO_INTERVAL_MILLIS, 60000)	
97	}
98}

1package apify
2
3// ActorJobStatus
4const (
5    // """Available statuses for actor jobs (runs or builds)."""
6
7    // Actor job initialized but not started yet
8    READY = "READY"
9    // Actor job in progress
10    RUNNING = "RUNNING"
11    // Actor job finished successfully
12    SUCCEEDED = "SUCCEEDED"
13    // Actor job or build failed
14    FAILED = "FAILED"
15    // Actor job currently timing out
16    TIMING_OUT = "TIMING-OUT"
17    // Actor job timed out
18    TIMED_OUT = "TIMED-OUT"
19    // Actor job currently being aborted by user
20    ABORTING = "ABORTING"
21    // Actor job aborted by user
22    ABORTED = "ABORTED"
23	/*
24    @property
25    def _is_terminal(self) -> bool:
26        """Whether this actor job status is terminal."""
27        return self in (ActorJobStatus.SUCCEEDED, ActorJobStatus.FAILED, ActorJobStatus.TIMED_OUT, ActorJobStatus.ABORTED)
28	*/
29)
30
31// ActorSourceType
32const (
33    // """Available source types for actors."""
34
35    // Actor source code is comprised of multiple files
36    SOURCE_FILES = "SOURCE_FILES"
37    // Actor source code is cloned from a Git repository
38    GIT_REPO = "GIT_REPO"
39    // Actor source code is downloaded using a tarball or Zip file
40    TARBALL = "TARBALL"
41    // Actor source code is taken from a GitHub Gist
42    GITHUB_GIST = "GITHUB_GIST"
43)
44
45// ActorEventTypes
46/*
47const ( 
48    // """Possible values of actor event type."""
49
50    // Info about resource usage of the actor
51    SYSTEM_INFO = "systemInfo"
52    // Sent when the actor is about to migrate
53    MIGRATING = "migrating"
54    // Sent when the actor should persist its state (every minute or when migrating)
55    PERSIST_STATE = "persistState"
56    // Sent when the actor is aborting
57    ABORTING = "aborting"
58
59)
60*/
61
62// ActorEnvVars
63const (
64    // """Possible Apify-specific environment variables prefixed with "ACTOR_"."""
65
66    // TODO: document these
67
68    BUILD_ID = "ACTOR_BUILD_ID"
69    BUILD_NUMBER = "ACTOR_BUILD_NUMBER"
70    DEFAULT_DATASET_ID = "ACTOR_DEFAULT_DATASET_ID"
71    DEFAULT_KEY_VALUE_STORE_ID = "ACTOR_DEFAULT_KEY_VALUE_STORE_ID"
72    DEFAULT_REQUEST_QUEUE_ID = "ACTOR_DEFAULT_REQUEST_QUEUE_ID"
73    EVENTS_WEBSOCKET_URL = "ACTOR_EVENTS_WEBSOCKET_URL"
74    ID = "ACTOR_ID"
75    INPUT_KEY = "ACTOR_INPUT_KEY"
76    MAX_PAID_DATASET_ITEMS = "ACTOR_MAX_PAID_DATASET_ITEMS"
77    MEMORY_MBYTES = "ACTOR_MEMORY_MBYTES"
78    RUN_ID = "ACTOR_RUN_ID"
79    STARTED_AT = "ACTOR_STARTED_AT"
80    TASK_ID = "ACTOR_TASK_ID"
81    TIMEOUT_AT = "ACTOR_TIMEOUT_AT"
82    WEB_SERVER_PORT = "ACTOR_WEB_SERVER_PORT"
83    WEB_SERVER_URL = "ACTOR_WEB_SERVER_URL"
84)
85
86// ApifyEnvVars
87const (
88    // """Possible Apify-specific environment variables prefixed with "APIFY_"."""
89
90    // TODO: document these
91
92    API_BASE_URL = "APIFY_API_BASE_URL"
93    API_PUBLIC_BASE_URL = "APIFY_API_PUBLIC_BASE_URL"
94    CHROME_EXECUTABLE_PATH = "APIFY_CHROME_EXECUTABLE_PATH"
95    DEDICATED_CPUS = "APIFY_DEDICATED_CPUS"
96    DEFAULT_BROWSER_PATH = "APIFY_DEFAULT_BROWSER_PATH"
97    DISABLE_BROWSER_SANDBOX = "APIFY_DISABLE_BROWSER_SANDBOX"
98    DISABLE_OUTDATED_WARNING = "APIFY_DISABLE_OUTDATED_WARNING"
99    FACT = "APIFY_FACT"
100    HEADLESS = "APIFY_HEADLESS"
101    INPUT_SECRETS_PRIVATE_KEY_FILE = "APIFY_INPUT_SECRETS_PRIVATE_KEY_FILE"
102    INPUT_SECRETS_PRIVATE_KEY_PASSPHRASE = "APIFY_INPUT_SECRETS_PRIVATE_KEY_PASSPHRASE"
103    IS_AT_HOME = "APIFY_IS_AT_HOME"
104    LOCAL_STORAGE_DIR = "APIFY_LOCAL_STORAGE_DIR"
105    LOG_FORMAT = "APIFY_LOG_FORMAT"
106    LOG_LEVEL = "APIFY_LOG_LEVEL"
107    MAX_USED_CPU_RATIO = "APIFY_MAX_USED_CPU_RATIO"
108    META_ORIGIN = "APIFY_META_ORIGIN"
109    METAMORPH_AFTER_SLEEP_MILLIS = "APIFY_METAMORPH_AFTER_SLEEP_MILLIS"
110    PERSIST_STATE_INTERVAL_MILLIS = "APIFY_PERSIST_STATE_INTERVAL_MILLIS"
111    PERSIST_STORAGE = "APIFY_PERSIST_STORAGE"
112    PROXY_HOSTNAME = "APIFY_PROXY_HOSTNAME"
113    PROXY_PASSWORD = "APIFY_PROXY_PASSWORD"
114    PROXY_PORT = "APIFY_PROXY_PORT"
115    PROXY_STATUS_URL = "APIFY_PROXY_STATUS_URL"
116    PURGE_ON_START = "APIFY_PURGE_ON_START"
117    SDK_LATEST_VERSION = "APIFY_SDK_LATEST_VERSION"
118    SYSTEM_INFO_INTERVAL_MILLIS = "APIFY_SYSTEM_INFO_INTERVAL_MILLIS"
119    TOKEN = "APIFY_TOKEN"
120    USER_ID = "APIFY_USER_ID"
121    WORKFLOW_KEY = "APIFY_WORKFLOW_KEY"
122    XVFB = "APIFY_XVFB"
123
124    // Replaced by ActorEnvVars, kept for backward compatibility:
125	/*
126    ACTOR_BUILD_ID = "APIFY_ACTOR_BUILD_ID"
127    ACTOR_BUILD_NUMBER = "APIFY_ACTOR_BUILD_NUMBER"
128    ACTOR_EVENTS_WS_URL = "APIFY_ACTOR_EVENTS_WS_URL"
129    ACTOR_ID = "APIFY_ACTOR_ID"
130    ACTOR_RUN_ID = "APIFY_ACTOR_RUN_ID"
131    ACTOR_TASK_ID = "APIFY_ACTOR_TASK_ID"
132    CONTAINER_PORT = "APIFY_CONTAINER_PORT"
133    CONTAINER_URL = "APIFY_CONTAINER_URL"
134    DEFAULT_DATASET_ID = "APIFY_DEFAULT_DATASET_ID"
135    DEFAULT_KEY_VALUE_STORE_ID = "APIFY_DEFAULT_KEY_VALUE_STORE_ID"
136    DEFAULT_REQUEST_QUEUE_ID = "APIFY_DEFAULT_REQUEST_QUEUE_ID"
137    INPUT_KEY = "APIFY_INPUT_KEY"
138    MEMORY_MBYTES = "APIFY_MEMORY_MBYTES"
139    STARTED_AT = "APIFY_STARTED_AT"
140    TIMEOUT_AT = "APIFY_TIMEOUT_AT"
141
142    // Deprecated, kept for backward compatibility:
143    ACT_ID = "APIFY_ACT_ID"
144    ACT_RUN_ID = "APIFY_ACT_RUN_ID"
145	*/
146)
147
148// ActorExitCodes
149const (
150    // """Usual actor exit codes."""
151
152    // The actor finished successfully
153    SUCCESS = 0
154
155    // The main function of the actor threw an Exception
156    ERROR_USER_FUNCTION_THREW = 91
157)
158
159// WebhookEventType
160const (
161    // """Events that can trigger a webhook."""
162
163    // The actor run was created
164    ACTOR_RUN_CREATED = "ACTOR.RUN.CREATED"
165    // The actor run has succeeded
166    ACTOR_RUN_SUCCEEDED = "ACTOR.RUN.SUCCEEDED"
167    // The actor run has failed
168    ACTOR_RUN_FAILED = "ACTOR.RUN.FAILED"
169    // The actor run has timed out
170    ACTOR_RUN_TIMED_OUT = "ACTOR.RUN.TIMED_OUT"
171    // The actor run was aborted
172    ACTOR_RUN_ABORTED = "ACTOR.RUN.ABORTED"
173    // The actor run was resurrected
174    ACTOR_RUN_RESURRECTED = "ACTOR.RUN.RESURRECTED"
175
176    // The actor build was created
177    ACTOR_BUILD_CREATED = "ACTOR.BUILD.CREATED"
178    // The actor build has succeeded
179    ACTOR_BUILD_SUCCEEDED = "ACTOR.BUILD.SUCCEEDED"
180    // The actor build has failed
181    ACTOR_BUILD_FAILED = "ACTOR.BUILD.FAILED"
182    // The actor build has timed out
183    ACTOR_BUILD_TIMED_OUT = "ACTOR.BUILD.TIMED_OUT"
184    // The actor build was aborted
185    ACTOR_BUILD_ABORTED = "ACTOR.BUILD.ABORTED"
186)
187
188// MetaOrigin
189const (
190    // """Possible origins for actor runs, i.e. how were the jobs started."""
191
192    // Job started from Developer console in Source section of actor
193    DEVELOPMENT = "DEVELOPMENT"
194    // Job started from other place on the website (either console or task detail page)
195    WEB = "WEB"
196    // Job started through API
197    API = "API"
198    // Job started through Scheduler
199    SCHEDULER = "SCHEDULER"
200    // Job started through test actor page
201    TEST = "TEST"
202    // Job started by the webhook
203    WEBHOOK = "WEBHOOK"
204    // Job started by another actor run
205    ACTOR = "ACTOR"
206)

1package main
2
3import (
4	//"os"
5	"example/apify"
6)
7
8func main() {
9
10	input := apify.Actor.GetInput()
11	
12	println(  input )
13}