1use crate::config::{ConnectorSpec, PipelineConfig, PipelineSpec, StateStoreSpec, TransformSpec};
10use crate::error::{CliError, CliResult};
11use serde_json::{Map, Value};
12use std::collections::{BTreeMap, HashMap};
13
14pub fn extract_scope(env: &HashMap<String, String>, prefix: &str) -> CliResult<Value> {
22 let mut object: Map<String, Value> = Map::new();
23 let mut json_fields: HashMap<String, String> = HashMap::new();
25 let mut scalar_fields: HashMap<String, String> = HashMap::new();
26
27 for (key, value) in env {
28 let Some(suffix) = key.strip_prefix(prefix) else {
29 continue;
30 };
31 if suffix.is_empty() {
32 continue;
34 }
35 let lowercase = suffix.to_ascii_lowercase();
36 if let Some(field) = lowercase.strip_suffix("_json") {
37 if let Some(scalar_var) = scalar_fields.get(field) {
38 return Err(CliError::EnvConflict {
39 field: field.to_owned(),
40 scalar_var: scalar_var.clone(),
41 json_var: key.clone(),
42 });
43 }
44 let parsed: Value =
45 serde_json::from_str(value).map_err(|e| CliError::InvalidEnvJson {
46 var: key.clone(),
47 message: e.to_string(),
48 })?;
49 object.insert(field.to_owned(), parsed);
50 json_fields.insert(field.to_owned(), key.clone());
51 } else {
52 if let Some(json_var) = json_fields.get(&lowercase) {
53 return Err(CliError::EnvConflict {
54 field: lowercase.clone(),
55 scalar_var: key.clone(),
56 json_var: json_var.clone(),
57 });
58 }
59 object.insert(lowercase.clone(), coerce_scalar(value));
60 scalar_fields.insert(lowercase, key.clone());
61 }
62 }
63 Ok(Value::Object(object))
64}
65
66fn coerce_scalar(s: &str) -> Value {
80 serde_json::from_str::<Value>(s).unwrap_or_else(|_| Value::String(s.to_owned()))
81}
82
83pub fn build_source(env: &HashMap<String, String>) -> CliResult<ConnectorSpec> {
85 let kind = env
86 .get("FAUCET_SOURCE")
87 .filter(|v| !v.is_empty())
88 .ok_or_else(|| CliError::MissingEnvSelector {
89 var: "FAUCET_SOURCE".to_owned(),
90 })?
91 .clone();
92 let prefix = format!(
93 "FAUCET_SOURCE_{}_",
94 kind.to_ascii_uppercase().replace('-', "_")
95 );
96 let config = extract_scope(env, &prefix)?;
97 Ok(ConnectorSpec {
98 kind,
99 config,
100 transforms: None,
101 inherit_transforms: true,
102 })
103}
104
105pub fn build_sink(env: &HashMap<String, String>) -> CliResult<ConnectorSpec> {
107 let kind = env
108 .get("FAUCET_SINK")
109 .filter(|v| !v.is_empty())
110 .ok_or_else(|| CliError::MissingEnvSelector {
111 var: "FAUCET_SINK".to_owned(),
112 })?
113 .clone();
114 let prefix = format!(
115 "FAUCET_SINK_{}_",
116 kind.to_ascii_uppercase().replace('-', "_")
117 );
118 let config = extract_scope(env, &prefix)?;
119 Ok(ConnectorSpec {
120 kind,
121 config,
122 transforms: None,
123 inherit_transforms: true,
124 })
125}
126
127pub fn build_state(env: &HashMap<String, String>) -> CliResult<Option<StateStoreSpec>> {
130 let Some(kind) = env.get("FAUCET_STATE").filter(|v| !v.is_empty()).cloned() else {
131 return Ok(None);
132 };
133 let prefix = format!(
134 "FAUCET_STATE_{}_",
135 kind.to_ascii_uppercase().replace('-', "_")
136 );
137 let config = extract_scope(env, &prefix)?;
138 Ok(Some(StateStoreSpec { kind, config }))
139}
140
141pub fn build_transforms(env: &HashMap<String, String>) -> CliResult<Vec<TransformSpec>> {
146 let mut kinds: BTreeMap<u32, String> = BTreeMap::new();
149 for (key, value) in env {
150 let Some(rest) = key.strip_prefix("FAUCET_TRANSFORM_") else {
151 continue;
152 };
153 if rest.is_empty() || !rest.chars().all(|c| c.is_ascii_digit()) {
154 continue;
155 }
156 let Ok(idx) = rest.parse::<u32>() else {
157 continue;
158 };
159 kinds.insert(idx, value.clone());
160 }
161 if kinds.is_empty() {
162 return Ok(Vec::new());
163 }
164 for (expected, actual) in (1u32..).zip(kinds.keys().copied()) {
166 if expected != actual {
167 return Err(CliError::TransformIndexGap { missing: expected });
168 }
169 }
170 let mut out = Vec::with_capacity(kinds.len());
172 for (idx, kind) in kinds {
173 let prefix = format!("FAUCET_TRANSFORM_{idx}_");
174 let config = extract_scope(env, &prefix)?;
175 out.push(TransformSpec { kind, config });
176 }
177 Ok(out)
178}
179
180pub fn build_named_sources(
183 env: &HashMap<String, String>,
184) -> CliResult<HashMap<String, ConnectorSpec>> {
185 build_named_catalog(env, "FAUCET_SOURCES_")
186}
187
188pub fn build_named_sinks(
190 env: &HashMap<String, String>,
191) -> CliResult<HashMap<String, ConnectorSpec>> {
192 build_named_catalog(env, "FAUCET_SINKS_")
193}
194
195fn build_named_catalog(
196 env: &HashMap<String, String>,
197 prefix: &str,
198) -> CliResult<HashMap<String, ConnectorSpec>> {
199 let mut kinds: HashMap<String, String> = HashMap::new();
202 for (key, value) in env {
203 let Some(suffix) = key.strip_prefix(prefix) else {
204 continue;
205 };
206 let Some(name_upper) = suffix.strip_suffix("_TYPE") else {
207 continue;
208 };
209 if name_upper.is_empty() {
210 continue;
211 }
212 kinds.insert(name_upper.to_ascii_lowercase(), value.clone());
213 }
214 let scope_prefixes: Vec<String> = kinds
220 .keys()
221 .map(|name| format!("{prefix}{}_", name.to_ascii_uppercase()))
222 .collect();
223
224 let mut out: HashMap<String, ConnectorSpec> = HashMap::new();
226 for (name, kind) in kinds {
227 let scope_prefix = format!("{prefix}{}_", name.to_ascii_uppercase());
228 let scoped: HashMap<String, String> = env
232 .iter()
233 .filter(|(k, _)| {
234 k.starts_with(&scope_prefix)
235 && !scope_prefixes.iter().any(|other| {
236 other.len() > scope_prefix.len() && k.starts_with(other.as_str())
237 })
238 })
239 .map(|(k, v)| (k.clone(), v.clone()))
240 .collect();
241 let mut config = extract_scope(&scoped, &scope_prefix)?;
242 if let Value::Object(m) = &mut config {
244 m.remove("type");
245 }
246 out.insert(
247 name,
248 ConnectorSpec {
249 kind,
250 config,
251 transforms: None,
252 inherit_transforms: true,
253 },
254 );
255 }
256 Ok(out)
257}
258
259pub fn build_vars(env: &HashMap<String, String>) -> Option<HashMap<String, Value>> {
262 let mut out: HashMap<String, Value> = HashMap::new();
263 for (key, value) in env {
264 let Some(name_upper) = key.strip_prefix("FAUCET_VARS_") else {
265 continue;
266 };
267 if name_upper.is_empty() {
268 continue;
269 }
270 out.insert(name_upper.to_ascii_lowercase(), coerce_scalar(value));
271 }
272 if out.is_empty() { None } else { Some(out) }
273}
274
275pub fn build_pipeline_config(env: &HashMap<String, String>) -> CliResult<PipelineConfig> {
277 let source = match env.get("FAUCET_SOURCE").filter(|v| !v.is_empty()) {
278 Some(_) => Some(build_source(env)?),
279 None => None,
280 };
281 let sink = match env.get("FAUCET_SINK").filter(|v| !v.is_empty()) {
282 Some(_) => Some(build_sink(env)?),
283 None => None,
284 };
285 let sources = build_named_sources(env)?;
286 let sinks = build_named_sinks(env)?;
287 let state = build_state(env)?;
288 let transforms = build_transforms(env)?;
289 let vars = build_vars(env);
290 let name = env.get("FAUCET_NAME").cloned().filter(|s| !s.is_empty());
291
292 if source.is_none() && sources.is_empty() {
294 return Err(CliError::MissingEnvSelector {
295 var: "FAUCET_SOURCE (or FAUCET_SOURCES_<NAME>_TYPE)".to_owned(),
296 });
297 }
298 if sink.is_none() && sinks.is_empty() {
299 return Err(CliError::MissingEnvSelector {
300 var: "FAUCET_SINK (or FAUCET_SINKS_<NAME>_TYPE)".to_owned(),
301 });
302 }
303 Ok(PipelineConfig {
304 version: 1,
305 name,
306 vars,
307 auth: None,
310 pipeline: PipelineSpec {
311 source,
312 sink,
313 sources,
314 sinks,
315 transforms,
316 state,
317 dlq: None,
318 #[cfg(feature = "quality")]
319 quality: None,
320 schema: None,
321 },
322 matrix: Vec::new(),
323 execution: None,
324 observability: None,
325 delivery: faucet_core::DeliveryMode::default(),
326 resilience: None,
327 shard: None,
328 replication: None,
329 #[cfg(feature = "schedule")]
330 schedule: None,
331 #[cfg(feature = "lineage")]
332 lineage: None,
333 })
334}
335
336pub fn from_process_env() -> CliResult<PipelineConfig> {
338 let env: HashMap<String, String> = std::env::vars().collect();
339 build_pipeline_config(&env)
340}
341
342#[cfg(test)]
343mod tests {
344 use super::*;
345 use serde_json::json;
346
347 fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
348 pairs
349 .iter()
350 .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
351 .collect()
352 }
353
354 #[test]
355 fn extract_scope_lowercases_field_names() {
356 let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
357 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
358 assert_eq!(v, json!({"base_url": "https://x.example"}));
359 }
360
361 #[test]
362 fn extract_scope_ignores_unrelated_keys() {
363 let e = env(&[
364 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
365 ("PATH", "/usr/bin"),
366 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
367 ]);
368 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
369 assert_eq!(v, json!({"base_url": "https://x.example"}));
370 }
371
372 #[test]
373 fn named_templates_do_not_leak_across_prefix_overlapping_names() {
374 let e = env(&[
377 ("FAUCET_SOURCES_USERS_TYPE", "rest"),
378 ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
379 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
380 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
381 ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
382 ]);
383 let out = build_named_sources(&e).unwrap();
384
385 let users = out.get("users").expect("users template").config.clone();
386 let users = users.as_object().unwrap();
387 assert_eq!(
388 users.get("base_url").and_then(|v| v.as_str()),
389 Some("https://u")
390 );
391 assert!(
392 !users.contains_key("api_base_url"),
393 "users must not absorb users_api's vars"
394 );
395 assert!(!users.contains_key("api_timeout"));
396
397 let api = out
398 .get("users_api")
399 .expect("users_api template")
400 .config
401 .clone();
402 let api = api.as_object().unwrap();
403 assert_eq!(
404 api.get("base_url").and_then(|v| v.as_str()),
405 Some("https://api")
406 );
407 assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
408 }
409
410 #[test]
411 fn extract_scope_coerces_numbers_and_bools() {
412 let e = env(&[
413 ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
414 ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
415 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
416 ]);
417 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
418 assert_eq!(v["timeout_secs"], json!(30));
419 assert_eq!(v["follow_redirects"], json!(true));
420 assert_eq!(v["base_url"], json!("https://x.example"));
421 }
422
423 #[test]
424 fn extract_scope_handles_json_suffix() {
425 let e = env(&[(
426 "FAUCET_SOURCE_REST_AUTH_JSON",
427 r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
428 )]);
429 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
430 assert_eq!(
431 v["auth"],
432 json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
433 );
434 }
435
436 #[test]
437 fn extract_scope_rejects_invalid_json_suffix() {
438 let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
439 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
440 match err {
441 CliError::InvalidEnvJson { var, .. } => {
442 assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
443 }
444 other => panic!("expected InvalidEnvJson, got {other:?}"),
445 }
446 }
447
448 #[test]
449 fn extract_scope_conflict_scalar_then_json() {
450 let e = env(&[
451 ("FAUCET_SOURCE_REST_AUTH", "bearer"),
452 ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
453 ]);
454 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
455 match err {
456 CliError::EnvConflict {
457 field,
458 scalar_var,
459 json_var,
460 } => {
461 assert_eq!(field, "auth");
462 assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
463 assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
464 }
465 other => panic!("expected EnvConflict, got {other:?}"),
466 }
467 }
468
469 #[test]
470 fn extract_scope_conflict_detection_is_order_independent() {
471 for _ in 0..50 {
475 let e = env(&[
476 ("FAUCET_SOURCE_REST_AUTH", "bearer"),
477 ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
478 ]);
479 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
480 match err {
481 CliError::EnvConflict {
482 field,
483 scalar_var,
484 json_var,
485 } => {
486 assert_eq!(field, "auth");
487 assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
488 assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
489 }
490 other => panic!("expected EnvConflict, got {other:?}"),
491 }
492 }
493 }
494
495 #[test]
496 fn extract_scope_skips_bare_prefix() {
497 let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
498 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
499 assert_eq!(v, json!({}));
500 }
501
502 #[test]
503 fn extract_scope_empty_when_no_matches() {
504 let e = env(&[("PATH", "/usr/bin")]);
505 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
506 assert_eq!(v, json!({}));
507 }
508
509 #[test]
510 fn build_source_reads_selector_and_scope() {
511 let e = env(&[
512 ("FAUCET_SOURCE", "rest"),
513 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
514 ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
515 ]);
516 let spec = build_source(&e).unwrap();
517 assert_eq!(spec.kind, "rest");
518 assert_eq!(spec.config["base_url"], json!("https://x.example"));
519 assert_eq!(spec.config["timeout_secs"], json!(30));
520 }
521
522 #[test]
523 fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
524 let e = env(&[
525 ("FAUCET_SOURCE", "csv"),
526 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
527 ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
528 ]);
529 let spec = build_source(&e).unwrap();
530 assert_eq!(spec.kind, "csv");
531 assert_eq!(spec.config, json!({"path": "./in.csv"}));
532 }
533
534 #[test]
535 fn build_source_errors_when_selector_missing() {
536 let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
537 let err = build_source(&e).unwrap_err();
538 match err {
539 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
540 other => panic!("expected MissingEnvSelector, got {other:?}"),
541 }
542 }
543
544 #[test]
545 fn build_source_errors_when_selector_empty() {
546 let e = env(&[("FAUCET_SOURCE", "")]);
547 let err = build_source(&e).unwrap_err();
548 match err {
549 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
550 other => panic!("expected MissingEnvSelector, got {other:?}"),
551 }
552 }
553
554 #[test]
555 fn build_sink_reads_selector_and_scope() {
556 let e = env(&[
557 ("FAUCET_SINK", "jsonl"),
558 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
559 ]);
560 let spec = build_sink(&e).unwrap();
561 assert_eq!(spec.kind, "jsonl");
562 assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
563 }
564
565 #[test]
566 fn build_sink_errors_when_selector_missing() {
567 let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
568 let err = build_sink(&e).unwrap_err();
569 match err {
570 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
571 other => panic!("expected MissingEnvSelector, got {other:?}"),
572 }
573 }
574
575 #[test]
576 fn build_sink_errors_when_selector_empty() {
577 let e = env(&[("FAUCET_SINK", "")]);
578 let err = build_sink(&e).unwrap_err();
579 match err {
580 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
581 other => panic!("expected MissingEnvSelector, got {other:?}"),
582 }
583 }
584
585 #[test]
586 fn build_state_returns_none_when_unset() {
587 let e = env(&[("FAUCET_SOURCE", "rest")]);
588 let spec = build_state(&e).unwrap();
589 assert!(spec.is_none());
590 }
591
592 #[test]
593 fn build_state_returns_none_when_empty() {
594 let e = env(&[("FAUCET_STATE", "")]);
595 let spec = build_state(&e).unwrap();
596 assert!(spec.is_none());
597 }
598
599 #[test]
600 fn build_state_reads_file_backend() {
601 let e = env(&[
602 ("FAUCET_STATE", "file"),
603 ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
604 ]);
605 let spec = build_state(&e).unwrap().unwrap();
606 assert_eq!(spec.kind, "file");
607 assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
608 }
609
610 #[test]
611 fn build_state_reads_memory_with_empty_scope() {
612 let e = env(&[("FAUCET_STATE", "memory")]);
613 let spec = build_state(&e).unwrap().unwrap();
614 assert_eq!(spec.kind, "memory");
615 assert_eq!(spec.config, json!({}));
616 }
617
618 #[test]
619 fn build_transforms_empty_when_unset() {
620 let e = env(&[("FAUCET_SOURCE", "rest")]);
621 let t = build_transforms(&e).unwrap();
622 assert!(t.is_empty());
623 }
624
625 #[test]
626 fn build_transforms_single_kind_no_config() {
627 let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
628 let t = build_transforms(&e).unwrap();
629 assert_eq!(t.len(), 1);
630 assert_eq!(t[0].kind, "snake_case");
631 assert_eq!(t[0].config, json!({}));
632 }
633
634 #[test]
635 fn build_transforms_ordered_and_with_config() {
636 let e = env(&[
637 ("FAUCET_TRANSFORM_1", "snake_case"),
638 ("FAUCET_TRANSFORM_2", "flatten"),
639 ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
640 ]);
641 let t = build_transforms(&e).unwrap();
642 assert_eq!(t.len(), 2);
643 assert_eq!(t[0].kind, "snake_case");
644 assert_eq!(t[1].kind, "flatten");
645 assert_eq!(t[1].config, json!({"separator": "__"}));
646 }
647
648 #[test]
649 fn build_transforms_handles_double_digit_indices() {
650 let e = env(&[
651 ("FAUCET_TRANSFORM_1", "snake_case"),
652 ("FAUCET_TRANSFORM_2", "flatten"),
653 ("FAUCET_TRANSFORM_3", "rename_keys"),
654 ]);
655 let t = build_transforms(&e).unwrap();
656 assert_eq!(t.len(), 3);
657 assert_eq!(t[2].kind, "rename_keys");
658 }
659
660 #[test]
661 fn build_transforms_gap_errors() {
662 let e = env(&[
663 ("FAUCET_TRANSFORM_1", "snake_case"),
664 ("FAUCET_TRANSFORM_3", "flatten"),
665 ]);
666 let err = build_transforms(&e).unwrap_err();
667 match err {
668 CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
669 other => panic!("expected TransformIndexGap, got {other:?}"),
670 }
671 }
672
673 #[test]
674 fn build_transforms_must_start_at_one() {
675 let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
676 let err = build_transforms(&e).unwrap_err();
677 match err {
678 CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
679 other => panic!("expected TransformIndexGap, got {other:?}"),
680 }
681 }
682
683 #[test]
684 fn build_transforms_ignores_field_vars_when_indexing_kinds() {
685 let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
687 let t = build_transforms(&e).unwrap();
690 assert!(t.is_empty());
691 }
692
693 #[test]
694 fn picks_up_named_source_templates() {
695 let e = env(&[
696 ("FAUCET_SOURCE", "rest"),
698 ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
699 ("FAUCET_SINK", "jsonl"),
700 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
701 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
703 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
704 ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
705 ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
706 ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
707 ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
708 ]);
709 let cfg = build_pipeline_config(&e).unwrap();
710 assert!(cfg.pipeline.source.is_some());
711 assert_eq!(cfg.pipeline.sources.len(), 2);
712 assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
713 assert_eq!(
714 cfg.pipeline.sources["users_api"].config["base_url"],
715 "https://users.example"
716 );
717 assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
718 }
719
720 #[test]
721 fn picks_up_vars_block() {
722 let e = env(&[
723 ("FAUCET_SOURCE", "rest"),
724 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
725 ("FAUCET_SINK", "jsonl"),
726 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
727 ("FAUCET_VARS_API_BASE", "https://api.example.com"),
728 ("FAUCET_VARS_REGION", "us-east-1"),
729 ]);
730 let cfg = build_pipeline_config(&e).unwrap();
731 let vars = cfg.vars.unwrap();
732 assert_eq!(vars["api_base"], "https://api.example.com");
733 assert_eq!(vars["region"], "us-east-1");
734 }
735
736 #[test]
737 fn named_source_only_no_legacy_works() {
738 let e = env(&[
742 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
743 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
744 ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
745 ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
746 ]);
747 let cfg = build_pipeline_config(&e).unwrap();
748 assert!(cfg.pipeline.source.is_none());
749 assert!(cfg.pipeline.sink.is_none());
750 assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
751 assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
752 }
753
754 #[test]
755 fn no_source_anywhere_errors() {
756 let e = env(&[
758 ("FAUCET_SINK", "jsonl"),
759 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
760 ]);
761 let err = build_pipeline_config(&e).unwrap_err();
762 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
763 }
764
765 #[test]
766 fn build_pipeline_config_minimal_csv_to_jsonl() {
767 let e = env(&[
768 ("FAUCET_SOURCE", "csv"),
769 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
770 ("FAUCET_SINK", "jsonl"),
771 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
772 ]);
773 let cfg = build_pipeline_config(&e).unwrap();
774 assert_eq!(cfg.version, 1);
775 assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
776 assert_eq!(
777 cfg.pipeline.source.as_ref().unwrap().config,
778 json!({"path": "./in.csv"})
779 );
780 assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
781 assert_eq!(
782 cfg.pipeline.sink.as_ref().unwrap().config,
783 json!({"path": "./out.jsonl"})
784 );
785 assert!(cfg.pipeline.transforms.is_empty());
786 assert!(cfg.pipeline.state.is_none());
787 assert!(cfg.name.is_none());
788 }
789
790 #[test]
791 fn build_pipeline_config_uses_faucet_name_when_set() {
792 let e = env(&[
793 ("FAUCET_NAME", "github-issues"),
794 ("FAUCET_SOURCE", "csv"),
795 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
796 ("FAUCET_SINK", "jsonl"),
797 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
798 ]);
799 let cfg = build_pipeline_config(&e).unwrap();
800 assert_eq!(cfg.name.as_deref(), Some("github-issues"));
801 }
802
803 #[test]
804 fn build_pipeline_config_treats_empty_name_as_none() {
805 let e = env(&[
806 ("FAUCET_NAME", ""),
807 ("FAUCET_SOURCE", "csv"),
808 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
809 ("FAUCET_SINK", "jsonl"),
810 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
811 ]);
812 let cfg = build_pipeline_config(&e).unwrap();
813 assert!(cfg.name.is_none());
814 }
815
816 #[test]
817 fn build_pipeline_config_with_state_and_transforms() {
818 let e = env(&[
819 ("FAUCET_SOURCE", "csv"),
820 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
821 ("FAUCET_SINK", "jsonl"),
822 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
823 ("FAUCET_STATE", "file"),
824 ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
825 ("FAUCET_TRANSFORM_1", "snake_case"),
826 ("FAUCET_TRANSFORM_2", "flatten"),
827 ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
828 ]);
829 let cfg = build_pipeline_config(&e).unwrap();
830 assert_eq!(cfg.pipeline.transforms.len(), 2);
831 assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
832 }
833
834 #[test]
835 fn build_pipeline_config_missing_source_errors() {
836 let e = env(&[
837 ("FAUCET_SINK", "jsonl"),
838 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
839 ]);
840 let err = build_pipeline_config(&e).unwrap_err();
841 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
842 }
843
844 #[test]
845 fn build_pipeline_config_missing_sink_errors() {
846 let e = env(&[
847 ("FAUCET_SOURCE", "csv"),
848 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
849 ]);
850 let err = build_pipeline_config(&e).unwrap_err();
851 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
852 }
853}