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 status: None,
103 tags: Vec::new(),
104 })
105}
106
107pub fn build_sink(env: &HashMap<String, String>) -> CliResult<ConnectorSpec> {
109 let kind = env
110 .get("FAUCET_SINK")
111 .filter(|v| !v.is_empty())
112 .ok_or_else(|| CliError::MissingEnvSelector {
113 var: "FAUCET_SINK".to_owned(),
114 })?
115 .clone();
116 let prefix = format!(
117 "FAUCET_SINK_{}_",
118 kind.to_ascii_uppercase().replace('-', "_")
119 );
120 let config = extract_scope(env, &prefix)?;
121 Ok(ConnectorSpec {
122 kind,
123 config,
124 transforms: None,
125 inherit_transforms: true,
126 status: None,
127 tags: Vec::new(),
128 })
129}
130
131pub fn build_state(env: &HashMap<String, String>) -> CliResult<Option<StateStoreSpec>> {
134 let Some(kind) = env.get("FAUCET_STATE").filter(|v| !v.is_empty()).cloned() else {
135 return Ok(None);
136 };
137 let prefix = format!(
138 "FAUCET_STATE_{}_",
139 kind.to_ascii_uppercase().replace('-', "_")
140 );
141 let config = extract_scope(env, &prefix)?;
142 Ok(Some(StateStoreSpec { kind, config }))
143}
144
145pub fn build_transforms(env: &HashMap<String, String>) -> CliResult<Vec<TransformSpec>> {
150 let mut kinds: BTreeMap<u32, String> = BTreeMap::new();
153 for (key, value) in env {
154 let Some(rest) = key.strip_prefix("FAUCET_TRANSFORM_") else {
155 continue;
156 };
157 if rest.is_empty() || !rest.chars().all(|c| c.is_ascii_digit()) {
158 continue;
159 }
160 let Ok(idx) = rest.parse::<u32>() else {
161 continue;
162 };
163 kinds.insert(idx, value.clone());
164 }
165 if kinds.is_empty() {
166 return Ok(Vec::new());
167 }
168 for (expected, actual) in (1u32..).zip(kinds.keys().copied()) {
170 if expected != actual {
171 return Err(CliError::TransformIndexGap { missing: expected });
172 }
173 }
174 let mut out = Vec::with_capacity(kinds.len());
176 for (idx, kind) in kinds {
177 let prefix = format!("FAUCET_TRANSFORM_{idx}_");
178 let config = extract_scope(env, &prefix)?;
179 out.push(TransformSpec { kind, config });
180 }
181 Ok(out)
182}
183
184pub fn build_named_sources(
187 env: &HashMap<String, String>,
188) -> CliResult<HashMap<String, ConnectorSpec>> {
189 build_named_catalog(env, "FAUCET_SOURCES_")
190}
191
192pub fn build_named_sinks(
194 env: &HashMap<String, String>,
195) -> CliResult<HashMap<String, ConnectorSpec>> {
196 build_named_catalog(env, "FAUCET_SINKS_")
197}
198
199fn build_named_catalog(
200 env: &HashMap<String, String>,
201 prefix: &str,
202) -> CliResult<HashMap<String, ConnectorSpec>> {
203 let mut kinds: HashMap<String, String> = HashMap::new();
206 for (key, value) in env {
207 let Some(suffix) = key.strip_prefix(prefix) else {
208 continue;
209 };
210 let Some(name_upper) = suffix.strip_suffix("_TYPE") else {
211 continue;
212 };
213 if name_upper.is_empty() {
214 continue;
215 }
216 kinds.insert(name_upper.to_ascii_lowercase(), value.clone());
217 }
218 let scope_prefixes: Vec<String> = kinds
224 .keys()
225 .map(|name| format!("{prefix}{}_", name.to_ascii_uppercase()))
226 .collect();
227
228 let mut out: HashMap<String, ConnectorSpec> = HashMap::new();
230 for (name, kind) in kinds {
231 let scope_prefix = format!("{prefix}{}_", name.to_ascii_uppercase());
232 let scoped: HashMap<String, String> = env
236 .iter()
237 .filter(|(k, _)| {
238 k.starts_with(&scope_prefix)
239 && !scope_prefixes.iter().any(|other| {
240 other.len() > scope_prefix.len() && k.starts_with(other.as_str())
241 })
242 })
243 .map(|(k, v)| (k.clone(), v.clone()))
244 .collect();
245 let mut config = extract_scope(&scoped, &scope_prefix)?;
246 if let Value::Object(m) = &mut config {
248 m.remove("type");
249 }
250 out.insert(
251 name,
252 ConnectorSpec {
253 kind,
254 config,
255 transforms: None,
256 inherit_transforms: true,
257 status: None,
258 tags: Vec::new(),
259 },
260 );
261 }
262 Ok(out)
263}
264
265pub fn build_vars(env: &HashMap<String, String>) -> Option<HashMap<String, Value>> {
268 let mut out: HashMap<String, Value> = HashMap::new();
269 for (key, value) in env {
270 let Some(name_upper) = key.strip_prefix("FAUCET_VARS_") else {
271 continue;
272 };
273 if name_upper.is_empty() {
274 continue;
275 }
276 out.insert(name_upper.to_ascii_lowercase(), coerce_scalar(value));
277 }
278 if out.is_empty() { None } else { Some(out) }
279}
280
281pub fn build_pipeline_config(env: &HashMap<String, String>) -> CliResult<PipelineConfig> {
283 let source = match env.get("FAUCET_SOURCE").filter(|v| !v.is_empty()) {
284 Some(_) => Some(build_source(env)?),
285 None => None,
286 };
287 let sink = match env.get("FAUCET_SINK").filter(|v| !v.is_empty()) {
288 Some(_) => Some(build_sink(env)?),
289 None => None,
290 };
291 let sources = build_named_sources(env)?;
292 let sinks = build_named_sinks(env)?;
293 let state = build_state(env)?;
294 let transforms = build_transforms(env)?;
295 let vars = build_vars(env);
296 let name = env.get("FAUCET_NAME").cloned().filter(|s| !s.is_empty());
297
298 if source.is_none() && sources.is_empty() {
300 return Err(CliError::MissingEnvSelector {
301 var: "FAUCET_SOURCE (or FAUCET_SOURCES_<NAME>_TYPE)".to_owned(),
302 });
303 }
304 if sink.is_none() && sinks.is_empty() {
305 return Err(CliError::MissingEnvSelector {
306 var: "FAUCET_SINK (or FAUCET_SINKS_<NAME>_TYPE)".to_owned(),
307 });
308 }
309 Ok(PipelineConfig {
310 version: 1,
311 name,
312 vars,
313 auth: None,
316 pipeline: PipelineSpec {
317 source,
318 sink,
319 sources,
320 sinks,
321 transforms,
322 state,
323 dlq: None,
324 #[cfg(feature = "quality")]
325 quality: None,
326 #[cfg(feature = "contract")]
327 contract: None,
328 #[cfg(feature = "masking")]
329 masking: None,
330 schema: None,
331 nodes: std::collections::HashMap::new(),
333 edges: Vec::new(),
334 },
335 matrix: Vec::new(),
336 execution: None,
337 selection: None,
338 observability: None,
339 delivery: faucet_core::DeliveryMode::default(),
340 resilience: None,
341 sla: None,
343 shard: None,
344 replication: None,
345 backfill: None,
346 #[cfg(feature = "schedule")]
347 schedule: None,
348 #[cfg(feature = "lineage")]
349 lineage: None,
350 #[cfg(feature = "catalog")]
352 catalog: None,
353 #[cfg(feature = "notify")]
354 notifications: Vec::new(),
355 })
356}
357
358pub fn from_process_env() -> CliResult<PipelineConfig> {
360 let env: HashMap<String, String> = std::env::vars().collect();
361 build_pipeline_config(&env)
362}
363
364#[cfg(test)]
365mod tests {
366 use super::*;
367 use serde_json::json;
368
369 fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
370 pairs
371 .iter()
372 .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
373 .collect()
374 }
375
376 #[test]
377 fn extract_scope_lowercases_field_names() {
378 let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
379 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
380 assert_eq!(v, json!({"base_url": "https://x.example"}));
381 }
382
383 #[test]
384 fn extract_scope_ignores_unrelated_keys() {
385 let e = env(&[
386 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
387 ("PATH", "/usr/bin"),
388 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
389 ]);
390 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
391 assert_eq!(v, json!({"base_url": "https://x.example"}));
392 }
393
394 #[test]
395 fn named_templates_do_not_leak_across_prefix_overlapping_names() {
396 let e = env(&[
399 ("FAUCET_SOURCES_USERS_TYPE", "rest"),
400 ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
401 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
402 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
403 ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
404 ]);
405 let out = build_named_sources(&e).unwrap();
406
407 let users = out.get("users").expect("users template").config.clone();
408 let users = users.as_object().unwrap();
409 assert_eq!(
410 users.get("base_url").and_then(|v| v.as_str()),
411 Some("https://u")
412 );
413 assert!(
414 !users.contains_key("api_base_url"),
415 "users must not absorb users_api's vars"
416 );
417 assert!(!users.contains_key("api_timeout"));
418
419 let api = out
420 .get("users_api")
421 .expect("users_api template")
422 .config
423 .clone();
424 let api = api.as_object().unwrap();
425 assert_eq!(
426 api.get("base_url").and_then(|v| v.as_str()),
427 Some("https://api")
428 );
429 assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
430 }
431
432 #[test]
433 fn extract_scope_coerces_numbers_and_bools() {
434 let e = env(&[
435 ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
436 ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
437 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
438 ]);
439 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
440 assert_eq!(v["timeout_secs"], json!(30));
441 assert_eq!(v["follow_redirects"], json!(true));
442 assert_eq!(v["base_url"], json!("https://x.example"));
443 }
444
445 #[test]
446 fn extract_scope_handles_json_suffix() {
447 let e = env(&[(
448 "FAUCET_SOURCE_REST_AUTH_JSON",
449 r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
450 )]);
451 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
452 assert_eq!(
453 v["auth"],
454 json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
455 );
456 }
457
458 #[test]
459 fn extract_scope_rejects_invalid_json_suffix() {
460 let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
461 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
462 match err {
463 CliError::InvalidEnvJson { var, .. } => {
464 assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
465 }
466 other => panic!("expected InvalidEnvJson, got {other:?}"),
467 }
468 }
469
470 #[test]
471 fn extract_scope_conflict_scalar_then_json() {
472 let e = env(&[
473 ("FAUCET_SOURCE_REST_AUTH", "bearer"),
474 ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
475 ]);
476 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
477 match err {
478 CliError::EnvConflict {
479 field,
480 scalar_var,
481 json_var,
482 } => {
483 assert_eq!(field, "auth");
484 assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
485 assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
486 }
487 other => panic!("expected EnvConflict, got {other:?}"),
488 }
489 }
490
491 #[test]
492 fn extract_scope_conflict_detection_is_order_independent() {
493 for _ in 0..50 {
497 let e = env(&[
498 ("FAUCET_SOURCE_REST_AUTH", "bearer"),
499 ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
500 ]);
501 let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
502 match err {
503 CliError::EnvConflict {
504 field,
505 scalar_var,
506 json_var,
507 } => {
508 assert_eq!(field, "auth");
509 assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
510 assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
511 }
512 other => panic!("expected EnvConflict, got {other:?}"),
513 }
514 }
515 }
516
517 #[test]
518 fn extract_scope_skips_bare_prefix() {
519 let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
520 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
521 assert_eq!(v, json!({}));
522 }
523
524 #[test]
525 fn extract_scope_empty_when_no_matches() {
526 let e = env(&[("PATH", "/usr/bin")]);
527 let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
528 assert_eq!(v, json!({}));
529 }
530
531 #[test]
532 fn build_source_reads_selector_and_scope() {
533 let e = env(&[
534 ("FAUCET_SOURCE", "rest"),
535 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
536 ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
537 ]);
538 let spec = build_source(&e).unwrap();
539 assert_eq!(spec.kind, "rest");
540 assert_eq!(spec.config["base_url"], json!("https://x.example"));
541 assert_eq!(spec.config["timeout_secs"], json!(30));
542 }
543
544 #[test]
545 fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
546 let e = env(&[
547 ("FAUCET_SOURCE", "csv"),
548 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
549 ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
550 ]);
551 let spec = build_source(&e).unwrap();
552 assert_eq!(spec.kind, "csv");
553 assert_eq!(spec.config, json!({"path": "./in.csv"}));
554 }
555
556 #[test]
557 fn build_source_errors_when_selector_missing() {
558 let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
559 let err = build_source(&e).unwrap_err();
560 match err {
561 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
562 other => panic!("expected MissingEnvSelector, got {other:?}"),
563 }
564 }
565
566 #[test]
567 fn build_source_errors_when_selector_empty() {
568 let e = env(&[("FAUCET_SOURCE", "")]);
569 let err = build_source(&e).unwrap_err();
570 match err {
571 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
572 other => panic!("expected MissingEnvSelector, got {other:?}"),
573 }
574 }
575
576 #[test]
577 fn build_sink_reads_selector_and_scope() {
578 let e = env(&[
579 ("FAUCET_SINK", "jsonl"),
580 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
581 ]);
582 let spec = build_sink(&e).unwrap();
583 assert_eq!(spec.kind, "jsonl");
584 assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
585 }
586
587 #[test]
588 fn build_sink_errors_when_selector_missing() {
589 let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
590 let err = build_sink(&e).unwrap_err();
591 match err {
592 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
593 other => panic!("expected MissingEnvSelector, got {other:?}"),
594 }
595 }
596
597 #[test]
598 fn build_sink_errors_when_selector_empty() {
599 let e = env(&[("FAUCET_SINK", "")]);
600 let err = build_sink(&e).unwrap_err();
601 match err {
602 CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
603 other => panic!("expected MissingEnvSelector, got {other:?}"),
604 }
605 }
606
607 #[test]
608 fn build_state_returns_none_when_unset() {
609 let e = env(&[("FAUCET_SOURCE", "rest")]);
610 let spec = build_state(&e).unwrap();
611 assert!(spec.is_none());
612 }
613
614 #[test]
615 fn build_state_returns_none_when_empty() {
616 let e = env(&[("FAUCET_STATE", "")]);
617 let spec = build_state(&e).unwrap();
618 assert!(spec.is_none());
619 }
620
621 #[test]
622 fn build_state_reads_file_backend() {
623 let e = env(&[
624 ("FAUCET_STATE", "file"),
625 ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
626 ]);
627 let spec = build_state(&e).unwrap().unwrap();
628 assert_eq!(spec.kind, "file");
629 assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
630 }
631
632 #[test]
633 fn build_state_reads_memory_with_empty_scope() {
634 let e = env(&[("FAUCET_STATE", "memory")]);
635 let spec = build_state(&e).unwrap().unwrap();
636 assert_eq!(spec.kind, "memory");
637 assert_eq!(spec.config, json!({}));
638 }
639
640 #[test]
641 fn build_transforms_empty_when_unset() {
642 let e = env(&[("FAUCET_SOURCE", "rest")]);
643 let t = build_transforms(&e).unwrap();
644 assert!(t.is_empty());
645 }
646
647 #[test]
648 fn build_transforms_single_kind_no_config() {
649 let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
650 let t = build_transforms(&e).unwrap();
651 assert_eq!(t.len(), 1);
652 assert_eq!(t[0].kind, "snake_case");
653 assert_eq!(t[0].config, json!({}));
654 }
655
656 #[test]
657 fn build_transforms_ordered_and_with_config() {
658 let e = env(&[
659 ("FAUCET_TRANSFORM_1", "snake_case"),
660 ("FAUCET_TRANSFORM_2", "flatten"),
661 ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
662 ]);
663 let t = build_transforms(&e).unwrap();
664 assert_eq!(t.len(), 2);
665 assert_eq!(t[0].kind, "snake_case");
666 assert_eq!(t[1].kind, "flatten");
667 assert_eq!(t[1].config, json!({"separator": "__"}));
668 }
669
670 #[test]
671 fn build_transforms_handles_double_digit_indices() {
672 let e = env(&[
673 ("FAUCET_TRANSFORM_1", "snake_case"),
674 ("FAUCET_TRANSFORM_2", "flatten"),
675 ("FAUCET_TRANSFORM_3", "rename_keys"),
676 ]);
677 let t = build_transforms(&e).unwrap();
678 assert_eq!(t.len(), 3);
679 assert_eq!(t[2].kind, "rename_keys");
680 }
681
682 #[test]
683 fn build_transforms_gap_errors() {
684 let e = env(&[
685 ("FAUCET_TRANSFORM_1", "snake_case"),
686 ("FAUCET_TRANSFORM_3", "flatten"),
687 ]);
688 let err = build_transforms(&e).unwrap_err();
689 match err {
690 CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
691 other => panic!("expected TransformIndexGap, got {other:?}"),
692 }
693 }
694
695 #[test]
696 fn build_transforms_must_start_at_one() {
697 let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
698 let err = build_transforms(&e).unwrap_err();
699 match err {
700 CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
701 other => panic!("expected TransformIndexGap, got {other:?}"),
702 }
703 }
704
705 #[test]
706 fn build_transforms_ignores_field_vars_when_indexing_kinds() {
707 let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
709 let t = build_transforms(&e).unwrap();
712 assert!(t.is_empty());
713 }
714
715 #[test]
716 fn picks_up_named_source_templates() {
717 let e = env(&[
718 ("FAUCET_SOURCE", "rest"),
720 ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
721 ("FAUCET_SINK", "jsonl"),
722 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
723 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
725 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
726 ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
727 ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
728 ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
729 ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
730 ]);
731 let cfg = build_pipeline_config(&e).unwrap();
732 assert!(cfg.pipeline.source.is_some());
733 assert_eq!(cfg.pipeline.sources.len(), 2);
734 assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
735 assert_eq!(
736 cfg.pipeline.sources["users_api"].config["base_url"],
737 "https://users.example"
738 );
739 assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
740 }
741
742 #[test]
743 fn picks_up_vars_block() {
744 let e = env(&[
745 ("FAUCET_SOURCE", "rest"),
746 ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
747 ("FAUCET_SINK", "jsonl"),
748 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
749 ("FAUCET_VARS_API_BASE", "https://api.example.com"),
750 ("FAUCET_VARS_REGION", "us-east-1"),
751 ]);
752 let cfg = build_pipeline_config(&e).unwrap();
753 let vars = cfg.vars.unwrap();
754 assert_eq!(vars["api_base"], "https://api.example.com");
755 assert_eq!(vars["region"], "us-east-1");
756 }
757
758 #[test]
759 fn named_source_only_no_legacy_works() {
760 let e = env(&[
764 ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
765 ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
766 ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
767 ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
768 ]);
769 let cfg = build_pipeline_config(&e).unwrap();
770 assert!(cfg.pipeline.source.is_none());
771 assert!(cfg.pipeline.sink.is_none());
772 assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
773 assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
774 }
775
776 #[test]
777 fn no_source_anywhere_errors() {
778 let e = env(&[
780 ("FAUCET_SINK", "jsonl"),
781 ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
782 ]);
783 let err = build_pipeline_config(&e).unwrap_err();
784 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
785 }
786
787 #[test]
788 fn build_pipeline_config_minimal_csv_to_jsonl() {
789 let e = env(&[
790 ("FAUCET_SOURCE", "csv"),
791 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
792 ("FAUCET_SINK", "jsonl"),
793 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
794 ]);
795 let cfg = build_pipeline_config(&e).unwrap();
796 assert_eq!(cfg.version, 1);
797 assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
798 assert_eq!(
799 cfg.pipeline.source.as_ref().unwrap().config,
800 json!({"path": "./in.csv"})
801 );
802 assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
803 assert_eq!(
804 cfg.pipeline.sink.as_ref().unwrap().config,
805 json!({"path": "./out.jsonl"})
806 );
807 assert!(cfg.pipeline.transforms.is_empty());
808 assert!(cfg.pipeline.state.is_none());
809 assert!(cfg.name.is_none());
810 }
811
812 #[test]
813 fn build_pipeline_config_uses_faucet_name_when_set() {
814 let e = env(&[
815 ("FAUCET_NAME", "github-issues"),
816 ("FAUCET_SOURCE", "csv"),
817 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
818 ("FAUCET_SINK", "jsonl"),
819 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
820 ]);
821 let cfg = build_pipeline_config(&e).unwrap();
822 assert_eq!(cfg.name.as_deref(), Some("github-issues"));
823 }
824
825 #[test]
826 fn build_pipeline_config_treats_empty_name_as_none() {
827 let e = env(&[
828 ("FAUCET_NAME", ""),
829 ("FAUCET_SOURCE", "csv"),
830 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
831 ("FAUCET_SINK", "jsonl"),
832 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
833 ]);
834 let cfg = build_pipeline_config(&e).unwrap();
835 assert!(cfg.name.is_none());
836 }
837
838 #[test]
839 fn build_pipeline_config_with_state_and_transforms() {
840 let e = env(&[
841 ("FAUCET_SOURCE", "csv"),
842 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
843 ("FAUCET_SINK", "jsonl"),
844 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
845 ("FAUCET_STATE", "file"),
846 ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
847 ("FAUCET_TRANSFORM_1", "snake_case"),
848 ("FAUCET_TRANSFORM_2", "flatten"),
849 ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
850 ]);
851 let cfg = build_pipeline_config(&e).unwrap();
852 assert_eq!(cfg.pipeline.transforms.len(), 2);
853 assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
854 }
855
856 #[test]
857 fn build_pipeline_config_missing_source_errors() {
858 let e = env(&[
859 ("FAUCET_SINK", "jsonl"),
860 ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
861 ]);
862 let err = build_pipeline_config(&e).unwrap_err();
863 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
864 }
865
866 #[test]
867 fn build_pipeline_config_missing_sink_errors() {
868 let e = env(&[
869 ("FAUCET_SOURCE", "csv"),
870 ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
871 ]);
872 let err = build_pipeline_config(&e).unwrap_err();
873 assert!(matches!(err, CliError::MissingEnvSelector { .. }));
874 }
875}