Skip to main content

faucet_cli/
env_config.rs

1//! Build a `PipelineConfig` from a snapshot of `FAUCET_*` environment variables.
2//!
3//! The public surface is intentionally split between pure functions (taking a
4//! `HashMap<String, String>` env snapshot — fully testable) and a thin shell
5//! `from_process_env()` that captures `std::env::vars()`.
6//!
7//! See `cli/README.md` and issue #42 for the user-facing variable schema.
8
9use crate::config::{ConnectorSpec, PipelineConfig, PipelineSpec, StateStoreSpec, TransformSpec};
10use crate::error::{CliError, CliResult};
11use serde_json::{Map, Value};
12use std::collections::{BTreeMap, HashMap};
13
14/// Walk every env var starting with `prefix`, strip the prefix, lowercase the
15/// remainder into a field name, apply the `_JSON` precedence rule, and assemble
16/// a `Value::Object`. Returns the object verbatim — no shape validation; the
17/// connector's own `Deserialize` impl is the gate.
18///
19/// The `_JSON` suffix is the escape hatch for nested / tagged-enum fields
20/// (auth, pagination, etc.) that don't flatten cleanly into env-var names.
21pub fn extract_scope(env: &HashMap<String, String>, prefix: &str) -> CliResult<Value> {
22    let mut object: Map<String, Value> = Map::new();
23    // Track which env var supplied each field so a conflict error can name both.
24    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            // Bare prefix (e.g. exactly "FAUCET_SOURCE_REST_") — skip, no field.
33            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
66/// Try to parse `s` as a JSON value (numbers, bools, null, strings, objects,
67/// arrays). Falls back to a plain `Value::String` if the parse fails. This
68/// matches YAML's auto-typing so `30` becomes a number and `true` becomes a
69/// bool — the connector's `Deserialize` impl gets exactly what it would have
70/// gotten via YAML.
71///
72/// **Caveat:** a value that *looks* like JSON is auto-typed, so the literal
73/// strings `"true"` / `"false"` / `"null"` / `"123"` / `"1.5"` become a bool /
74/// null / number, not a string. When a connector field must stay a string with
75/// one of those values (e.g. an API key that is all digits, or a literal
76/// `"null"` token), set it via the `*_JSON` variant with the value quoted —
77/// e.g. `FAUCET_SOURCE_REST_TOKEN_JSON='"0123"'` — which bypasses this
78/// coercion (#78 LOW).
79fn coerce_scalar(s: &str) -> Value {
80    serde_json::from_str::<Value>(s).unwrap_or_else(|_| Value::String(s.to_owned()))
81}
82
83/// Construct the source [`ConnectorSpec`] from `FAUCET_SOURCE` + `FAUCET_SOURCE_<KIND>_*`.
84pub 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
105/// Construct the sink [`ConnectorSpec`] from `FAUCET_SINK` + `FAUCET_SINK_<KIND>_*`.
106pub 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
127/// Construct an optional [`StateStoreSpec`] from `FAUCET_STATE` + `FAUCET_STATE_<KIND>_*`.
128/// Returns `Ok(None)` when `FAUCET_STATE` is unset or empty (the common case).
129pub 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
141/// Construct an ordered `Vec<TransformSpec>` from `FAUCET_TRANSFORM_<N>` selectors
142/// and `FAUCET_TRANSFORM_<N>_<FIELD>` config fields. Indices must be contiguous
143/// starting at 1; any gap is an error so a misnumbered var never silently drops
144/// a transform.
145pub fn build_transforms(env: &HashMap<String, String>) -> CliResult<Vec<TransformSpec>> {
146    // Collect the kind selectors first — keys that are exactly
147    // `FAUCET_TRANSFORM_<digits>` (no trailing field name).
148    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    // Indices must be 1, 2, 3, …
165    for (expected, actual) in (1u32..).zip(kinds.keys().copied()) {
166        if expected != actual {
167            return Err(CliError::TransformIndexGap { missing: expected });
168        }
169    }
170    // Harvest per-transform config blocks.
171    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
180/// Harvest named source templates from `FAUCET_SOURCES_<NAME>_TYPE` selectors
181/// and `FAUCET_SOURCES_<NAME>_<FIELD>` config fields. `<NAME>` is lowercased.
182pub fn build_named_sources(
183    env: &HashMap<String, String>,
184) -> CliResult<HashMap<String, ConnectorSpec>> {
185    build_named_catalog(env, "FAUCET_SOURCES_")
186}
187
188/// Same as [`build_named_sources`] but for sinks via `FAUCET_SINKS_<NAME>_*`.
189pub 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    // First sweep: find each template's `<NAME>` by spotting
200    // `<prefix><NAME>_TYPE`.
201    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    // All template scope prefixes, so each env var can be assigned to its
215    // LONGEST matching prefix. Without this, a template like `users` would
216    // absorb `users_api`'s vars, since `FAUCET_SOURCES_USERS_` is a prefix of
217    // `FAUCET_SOURCES_USERS_API_` — the shorter template silently gets the
218    // longer one's fields (#146 M17).
219    let scope_prefixes: Vec<String> = kinds
220        .keys()
221        .map(|name| format!("{prefix}{}_", name.to_ascii_uppercase()))
222        .collect();
223
224    // Second sweep: harvest each template's config block.
225    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        // A view of `env` containing only the vars whose longest matching
229        // template prefix is THIS template's — i.e. drop any var that also
230        // matches a longer sibling prefix nested under this one.
231        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        // Remove the `type` field — it's the selector, not a config field.
243        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
259/// Harvest `FAUCET_VARS_<KEY>` into the top-level vars map. Returns `None`
260/// when no `FAUCET_VARS_*` variables are set.
261pub 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
275/// Construct a complete [`PipelineConfig`] from an env snapshot.
276pub 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    // At least one source and one sink must be declared somewhere.
293    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        // Pure-env mode doesn't (yet) assemble a shared `auth:` catalog; inline
308        // auth via FAUCET_*_AUTH_JSON still works.
309        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            #[cfg(feature = "contract")]
321            contract: None,
322            #[cfg(feature = "masking")]
323            masking: None,
324            schema: None,
325        },
326        matrix: Vec::new(),
327        execution: None,
328        observability: None,
329        delivery: faucet_core::DeliveryMode::default(),
330        resilience: None,
331        // Pure-env mode doesn't (yet) assemble an `sla:` block.
332        sla: None,
333        shard: None,
334        replication: None,
335        #[cfg(feature = "schedule")]
336        schedule: None,
337        #[cfg(feature = "lineage")]
338        lineage: None,
339        // Pure-env mode doesn't (yet) assemble a `catalog:` block.
340        #[cfg(feature = "catalog")]
341        catalog: None,
342        #[cfg(feature = "notify")]
343        notifications: Vec::new(),
344    })
345}
346
347/// Snapshot `std::env::vars()` and call [`build_pipeline_config`].
348pub fn from_process_env() -> CliResult<PipelineConfig> {
349    let env: HashMap<String, String> = std::env::vars().collect();
350    build_pipeline_config(&env)
351}
352
353#[cfg(test)]
354mod tests {
355    use super::*;
356    use serde_json::json;
357
358    fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
359        pairs
360            .iter()
361            .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
362            .collect()
363    }
364
365    #[test]
366    fn extract_scope_lowercases_field_names() {
367        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
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 extract_scope_ignores_unrelated_keys() {
374        let e = env(&[
375            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
376            ("PATH", "/usr/bin"),
377            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
378        ]);
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 named_templates_do_not_leak_across_prefix_overlapping_names() {
385        // M17 (#146): template `users` must NOT absorb `users_api`'s vars just
386        // because `FAUCET_SOURCES_USERS_` is a prefix of `FAUCET_SOURCES_USERS_API_`.
387        let e = env(&[
388            ("FAUCET_SOURCES_USERS_TYPE", "rest"),
389            ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
390            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
391            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
392            ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
393        ]);
394        let out = build_named_sources(&e).unwrap();
395
396        let users = out.get("users").expect("users template").config.clone();
397        let users = users.as_object().unwrap();
398        assert_eq!(
399            users.get("base_url").and_then(|v| v.as_str()),
400            Some("https://u")
401        );
402        assert!(
403            !users.contains_key("api_base_url"),
404            "users must not absorb users_api's vars"
405        );
406        assert!(!users.contains_key("api_timeout"));
407
408        let api = out
409            .get("users_api")
410            .expect("users_api template")
411            .config
412            .clone();
413        let api = api.as_object().unwrap();
414        assert_eq!(
415            api.get("base_url").and_then(|v| v.as_str()),
416            Some("https://api")
417        );
418        assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
419    }
420
421    #[test]
422    fn extract_scope_coerces_numbers_and_bools() {
423        let e = env(&[
424            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
425            ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
426            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
427        ]);
428        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
429        assert_eq!(v["timeout_secs"], json!(30));
430        assert_eq!(v["follow_redirects"], json!(true));
431        assert_eq!(v["base_url"], json!("https://x.example"));
432    }
433
434    #[test]
435    fn extract_scope_handles_json_suffix() {
436        let e = env(&[(
437            "FAUCET_SOURCE_REST_AUTH_JSON",
438            r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
439        )]);
440        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
441        assert_eq!(
442            v["auth"],
443            json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
444        );
445    }
446
447    #[test]
448    fn extract_scope_rejects_invalid_json_suffix() {
449        let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
450        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
451        match err {
452            CliError::InvalidEnvJson { var, .. } => {
453                assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
454            }
455            other => panic!("expected InvalidEnvJson, got {other:?}"),
456        }
457    }
458
459    #[test]
460    fn extract_scope_conflict_scalar_then_json() {
461        let e = env(&[
462            ("FAUCET_SOURCE_REST_AUTH", "bearer"),
463            ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
464        ]);
465        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
466        match err {
467            CliError::EnvConflict {
468                field,
469                scalar_var,
470                json_var,
471            } => {
472                assert_eq!(field, "auth");
473                assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
474                assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
475            }
476            other => panic!("expected EnvConflict, got {other:?}"),
477        }
478    }
479
480    #[test]
481    fn extract_scope_conflict_detection_is_order_independent() {
482        // HashMap iteration order is randomized per instance via the random
483        // hasher state; loop 50 times to exercise both ordering branches
484        // (scalar-arriving-first AND json-arriving-first) statistically.
485        for _ in 0..50 {
486            let e = env(&[
487                ("FAUCET_SOURCE_REST_AUTH", "bearer"),
488                ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
489            ]);
490            let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
491            match err {
492                CliError::EnvConflict {
493                    field,
494                    scalar_var,
495                    json_var,
496                } => {
497                    assert_eq!(field, "auth");
498                    assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
499                    assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
500                }
501                other => panic!("expected EnvConflict, got {other:?}"),
502            }
503        }
504    }
505
506    #[test]
507    fn extract_scope_skips_bare_prefix() {
508        let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
509        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
510        assert_eq!(v, json!({}));
511    }
512
513    #[test]
514    fn extract_scope_empty_when_no_matches() {
515        let e = env(&[("PATH", "/usr/bin")]);
516        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
517        assert_eq!(v, json!({}));
518    }
519
520    #[test]
521    fn build_source_reads_selector_and_scope() {
522        let e = env(&[
523            ("FAUCET_SOURCE", "rest"),
524            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
525            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
526        ]);
527        let spec = build_source(&e).unwrap();
528        assert_eq!(spec.kind, "rest");
529        assert_eq!(spec.config["base_url"], json!("https://x.example"));
530        assert_eq!(spec.config["timeout_secs"], json!(30));
531    }
532
533    #[test]
534    fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
535        let e = env(&[
536            ("FAUCET_SOURCE", "csv"),
537            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
538            ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
539        ]);
540        let spec = build_source(&e).unwrap();
541        assert_eq!(spec.kind, "csv");
542        assert_eq!(spec.config, json!({"path": "./in.csv"}));
543    }
544
545    #[test]
546    fn build_source_errors_when_selector_missing() {
547        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
548        let err = build_source(&e).unwrap_err();
549        match err {
550            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
551            other => panic!("expected MissingEnvSelector, got {other:?}"),
552        }
553    }
554
555    #[test]
556    fn build_source_errors_when_selector_empty() {
557        let e = env(&[("FAUCET_SOURCE", "")]);
558        let err = build_source(&e).unwrap_err();
559        match err {
560            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
561            other => panic!("expected MissingEnvSelector, got {other:?}"),
562        }
563    }
564
565    #[test]
566    fn build_sink_reads_selector_and_scope() {
567        let e = env(&[
568            ("FAUCET_SINK", "jsonl"),
569            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
570        ]);
571        let spec = build_sink(&e).unwrap();
572        assert_eq!(spec.kind, "jsonl");
573        assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
574    }
575
576    #[test]
577    fn build_sink_errors_when_selector_missing() {
578        let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
579        let err = build_sink(&e).unwrap_err();
580        match err {
581            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
582            other => panic!("expected MissingEnvSelector, got {other:?}"),
583        }
584    }
585
586    #[test]
587    fn build_sink_errors_when_selector_empty() {
588        let e = env(&[("FAUCET_SINK", "")]);
589        let err = build_sink(&e).unwrap_err();
590        match err {
591            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
592            other => panic!("expected MissingEnvSelector, got {other:?}"),
593        }
594    }
595
596    #[test]
597    fn build_state_returns_none_when_unset() {
598        let e = env(&[("FAUCET_SOURCE", "rest")]);
599        let spec = build_state(&e).unwrap();
600        assert!(spec.is_none());
601    }
602
603    #[test]
604    fn build_state_returns_none_when_empty() {
605        let e = env(&[("FAUCET_STATE", "")]);
606        let spec = build_state(&e).unwrap();
607        assert!(spec.is_none());
608    }
609
610    #[test]
611    fn build_state_reads_file_backend() {
612        let e = env(&[
613            ("FAUCET_STATE", "file"),
614            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
615        ]);
616        let spec = build_state(&e).unwrap().unwrap();
617        assert_eq!(spec.kind, "file");
618        assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
619    }
620
621    #[test]
622    fn build_state_reads_memory_with_empty_scope() {
623        let e = env(&[("FAUCET_STATE", "memory")]);
624        let spec = build_state(&e).unwrap().unwrap();
625        assert_eq!(spec.kind, "memory");
626        assert_eq!(spec.config, json!({}));
627    }
628
629    #[test]
630    fn build_transforms_empty_when_unset() {
631        let e = env(&[("FAUCET_SOURCE", "rest")]);
632        let t = build_transforms(&e).unwrap();
633        assert!(t.is_empty());
634    }
635
636    #[test]
637    fn build_transforms_single_kind_no_config() {
638        let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
639        let t = build_transforms(&e).unwrap();
640        assert_eq!(t.len(), 1);
641        assert_eq!(t[0].kind, "snake_case");
642        assert_eq!(t[0].config, json!({}));
643    }
644
645    #[test]
646    fn build_transforms_ordered_and_with_config() {
647        let e = env(&[
648            ("FAUCET_TRANSFORM_1", "snake_case"),
649            ("FAUCET_TRANSFORM_2", "flatten"),
650            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
651        ]);
652        let t = build_transforms(&e).unwrap();
653        assert_eq!(t.len(), 2);
654        assert_eq!(t[0].kind, "snake_case");
655        assert_eq!(t[1].kind, "flatten");
656        assert_eq!(t[1].config, json!({"separator": "__"}));
657    }
658
659    #[test]
660    fn build_transforms_handles_double_digit_indices() {
661        let e = env(&[
662            ("FAUCET_TRANSFORM_1", "snake_case"),
663            ("FAUCET_TRANSFORM_2", "flatten"),
664            ("FAUCET_TRANSFORM_3", "rename_keys"),
665        ]);
666        let t = build_transforms(&e).unwrap();
667        assert_eq!(t.len(), 3);
668        assert_eq!(t[2].kind, "rename_keys");
669    }
670
671    #[test]
672    fn build_transforms_gap_errors() {
673        let e = env(&[
674            ("FAUCET_TRANSFORM_1", "snake_case"),
675            ("FAUCET_TRANSFORM_3", "flatten"),
676        ]);
677        let err = build_transforms(&e).unwrap_err();
678        match err {
679            CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
680            other => panic!("expected TransformIndexGap, got {other:?}"),
681        }
682    }
683
684    #[test]
685    fn build_transforms_must_start_at_one() {
686        let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
687        let err = build_transforms(&e).unwrap_err();
688        match err {
689            CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
690            other => panic!("expected TransformIndexGap, got {other:?}"),
691        }
692    }
693
694    #[test]
695    fn build_transforms_ignores_field_vars_when_indexing_kinds() {
696        // FAUCET_TRANSFORM_1_SEPARATOR should NOT be mistaken for a kind at index 1.
697        let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
698        // No FAUCET_TRANSFORM_1 selector means no transforms at all (not a gap error,
699        // because the indices set is empty).
700        let t = build_transforms(&e).unwrap();
701        assert!(t.is_empty());
702    }
703
704    #[test]
705    fn picks_up_named_source_templates() {
706        let e = env(&[
707            // Legacy default
708            ("FAUCET_SOURCE", "rest"),
709            ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
710            ("FAUCET_SINK", "jsonl"),
711            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
712            // Named templates
713            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
714            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
715            ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
716            ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
717            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
718            ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
719        ]);
720        let cfg = build_pipeline_config(&e).unwrap();
721        assert!(cfg.pipeline.source.is_some());
722        assert_eq!(cfg.pipeline.sources.len(), 2);
723        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
724        assert_eq!(
725            cfg.pipeline.sources["users_api"].config["base_url"],
726            "https://users.example"
727        );
728        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
729    }
730
731    #[test]
732    fn picks_up_vars_block() {
733        let e = env(&[
734            ("FAUCET_SOURCE", "rest"),
735            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
736            ("FAUCET_SINK", "jsonl"),
737            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
738            ("FAUCET_VARS_API_BASE", "https://api.example.com"),
739            ("FAUCET_VARS_REGION", "us-east-1"),
740        ]);
741        let cfg = build_pipeline_config(&e).unwrap();
742        let vars = cfg.vars.unwrap();
743        assert_eq!(vars["api_base"], "https://api.example.com");
744        assert_eq!(vars["region"], "us-east-1");
745    }
746
747    #[test]
748    fn named_source_only_no_legacy_works() {
749        // Verifies: with no FAUCET_SOURCE / FAUCET_SINK but only named
750        // templates, the singular source/sink are None and the catalogs are
751        // populated. (No MissingEnvSelector.)
752        let e = env(&[
753            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
754            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
755            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
756            ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
757        ]);
758        let cfg = build_pipeline_config(&e).unwrap();
759        assert!(cfg.pipeline.source.is_none());
760        assert!(cfg.pipeline.sink.is_none());
761        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
762        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
763    }
764
765    #[test]
766    fn no_source_anywhere_errors() {
767        // Neither legacy nor named sources — still must error.
768        let e = env(&[
769            ("FAUCET_SINK", "jsonl"),
770            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
771        ]);
772        let err = build_pipeline_config(&e).unwrap_err();
773        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
774    }
775
776    #[test]
777    fn build_pipeline_config_minimal_csv_to_jsonl() {
778        let e = env(&[
779            ("FAUCET_SOURCE", "csv"),
780            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
781            ("FAUCET_SINK", "jsonl"),
782            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
783        ]);
784        let cfg = build_pipeline_config(&e).unwrap();
785        assert_eq!(cfg.version, 1);
786        assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
787        assert_eq!(
788            cfg.pipeline.source.as_ref().unwrap().config,
789            json!({"path": "./in.csv"})
790        );
791        assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
792        assert_eq!(
793            cfg.pipeline.sink.as_ref().unwrap().config,
794            json!({"path": "./out.jsonl"})
795        );
796        assert!(cfg.pipeline.transforms.is_empty());
797        assert!(cfg.pipeline.state.is_none());
798        assert!(cfg.name.is_none());
799    }
800
801    #[test]
802    fn build_pipeline_config_uses_faucet_name_when_set() {
803        let e = env(&[
804            ("FAUCET_NAME", "github-issues"),
805            ("FAUCET_SOURCE", "csv"),
806            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
807            ("FAUCET_SINK", "jsonl"),
808            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
809        ]);
810        let cfg = build_pipeline_config(&e).unwrap();
811        assert_eq!(cfg.name.as_deref(), Some("github-issues"));
812    }
813
814    #[test]
815    fn build_pipeline_config_treats_empty_name_as_none() {
816        let e = env(&[
817            ("FAUCET_NAME", ""),
818            ("FAUCET_SOURCE", "csv"),
819            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
820            ("FAUCET_SINK", "jsonl"),
821            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
822        ]);
823        let cfg = build_pipeline_config(&e).unwrap();
824        assert!(cfg.name.is_none());
825    }
826
827    #[test]
828    fn build_pipeline_config_with_state_and_transforms() {
829        let e = env(&[
830            ("FAUCET_SOURCE", "csv"),
831            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
832            ("FAUCET_SINK", "jsonl"),
833            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
834            ("FAUCET_STATE", "file"),
835            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
836            ("FAUCET_TRANSFORM_1", "snake_case"),
837            ("FAUCET_TRANSFORM_2", "flatten"),
838            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
839        ]);
840        let cfg = build_pipeline_config(&e).unwrap();
841        assert_eq!(cfg.pipeline.transforms.len(), 2);
842        assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
843    }
844
845    #[test]
846    fn build_pipeline_config_missing_source_errors() {
847        let e = env(&[
848            ("FAUCET_SINK", "jsonl"),
849            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
850        ]);
851        let err = build_pipeline_config(&e).unwrap_err();
852        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
853    }
854
855    #[test]
856    fn build_pipeline_config_missing_sink_errors() {
857        let e = env(&[
858            ("FAUCET_SOURCE", "csv"),
859            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
860        ]);
861        let err = build_pipeline_config(&e).unwrap_err();
862        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
863    }
864}