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        },
321        matrix: Vec::new(),
322        execution: None,
323        observability: None,
324        delivery: faucet_core::DeliveryMode::default(),
325        #[cfg(feature = "schedule")]
326        schedule: None,
327        #[cfg(feature = "lineage")]
328        lineage: None,
329    })
330}
331
332/// Snapshot `std::env::vars()` and call [`build_pipeline_config`].
333pub fn from_process_env() -> CliResult<PipelineConfig> {
334    let env: HashMap<String, String> = std::env::vars().collect();
335    build_pipeline_config(&env)
336}
337
338#[cfg(test)]
339mod tests {
340    use super::*;
341    use serde_json::json;
342
343    fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
344        pairs
345            .iter()
346            .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
347            .collect()
348    }
349
350    #[test]
351    fn extract_scope_lowercases_field_names() {
352        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
353        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
354        assert_eq!(v, json!({"base_url": "https://x.example"}));
355    }
356
357    #[test]
358    fn extract_scope_ignores_unrelated_keys() {
359        let e = env(&[
360            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
361            ("PATH", "/usr/bin"),
362            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
363        ]);
364        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
365        assert_eq!(v, json!({"base_url": "https://x.example"}));
366    }
367
368    #[test]
369    fn named_templates_do_not_leak_across_prefix_overlapping_names() {
370        // M17 (#146): template `users` must NOT absorb `users_api`'s vars just
371        // because `FAUCET_SOURCES_USERS_` is a prefix of `FAUCET_SOURCES_USERS_API_`.
372        let e = env(&[
373            ("FAUCET_SOURCES_USERS_TYPE", "rest"),
374            ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
375            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
376            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
377            ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
378        ]);
379        let out = build_named_sources(&e).unwrap();
380
381        let users = out.get("users").expect("users template").config.clone();
382        let users = users.as_object().unwrap();
383        assert_eq!(
384            users.get("base_url").and_then(|v| v.as_str()),
385            Some("https://u")
386        );
387        assert!(
388            !users.contains_key("api_base_url"),
389            "users must not absorb users_api's vars"
390        );
391        assert!(!users.contains_key("api_timeout"));
392
393        let api = out
394            .get("users_api")
395            .expect("users_api template")
396            .config
397            .clone();
398        let api = api.as_object().unwrap();
399        assert_eq!(
400            api.get("base_url").and_then(|v| v.as_str()),
401            Some("https://api")
402        );
403        assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
404    }
405
406    #[test]
407    fn extract_scope_coerces_numbers_and_bools() {
408        let e = env(&[
409            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
410            ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
411            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
412        ]);
413        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
414        assert_eq!(v["timeout_secs"], json!(30));
415        assert_eq!(v["follow_redirects"], json!(true));
416        assert_eq!(v["base_url"], json!("https://x.example"));
417    }
418
419    #[test]
420    fn extract_scope_handles_json_suffix() {
421        let e = env(&[(
422            "FAUCET_SOURCE_REST_AUTH_JSON",
423            r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
424        )]);
425        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
426        assert_eq!(
427            v["auth"],
428            json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
429        );
430    }
431
432    #[test]
433    fn extract_scope_rejects_invalid_json_suffix() {
434        let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
435        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
436        match err {
437            CliError::InvalidEnvJson { var, .. } => {
438                assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
439            }
440            other => panic!("expected InvalidEnvJson, got {other:?}"),
441        }
442    }
443
444    #[test]
445    fn extract_scope_conflict_scalar_then_json() {
446        let e = env(&[
447            ("FAUCET_SOURCE_REST_AUTH", "bearer"),
448            ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
449        ]);
450        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
451        match err {
452            CliError::EnvConflict {
453                field,
454                scalar_var,
455                json_var,
456            } => {
457                assert_eq!(field, "auth");
458                assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
459                assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
460            }
461            other => panic!("expected EnvConflict, got {other:?}"),
462        }
463    }
464
465    #[test]
466    fn extract_scope_conflict_detection_is_order_independent() {
467        // HashMap iteration order is randomized per instance via the random
468        // hasher state; loop 50 times to exercise both ordering branches
469        // (scalar-arriving-first AND json-arriving-first) statistically.
470        for _ in 0..50 {
471            let e = env(&[
472                ("FAUCET_SOURCE_REST_AUTH", "bearer"),
473                ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
474            ]);
475            let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
476            match err {
477                CliError::EnvConflict {
478                    field,
479                    scalar_var,
480                    json_var,
481                } => {
482                    assert_eq!(field, "auth");
483                    assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
484                    assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
485                }
486                other => panic!("expected EnvConflict, got {other:?}"),
487            }
488        }
489    }
490
491    #[test]
492    fn extract_scope_skips_bare_prefix() {
493        let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
494        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
495        assert_eq!(v, json!({}));
496    }
497
498    #[test]
499    fn extract_scope_empty_when_no_matches() {
500        let e = env(&[("PATH", "/usr/bin")]);
501        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
502        assert_eq!(v, json!({}));
503    }
504
505    #[test]
506    fn build_source_reads_selector_and_scope() {
507        let e = env(&[
508            ("FAUCET_SOURCE", "rest"),
509            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
510            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
511        ]);
512        let spec = build_source(&e).unwrap();
513        assert_eq!(spec.kind, "rest");
514        assert_eq!(spec.config["base_url"], json!("https://x.example"));
515        assert_eq!(spec.config["timeout_secs"], json!(30));
516    }
517
518    #[test]
519    fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
520        let e = env(&[
521            ("FAUCET_SOURCE", "csv"),
522            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
523            ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
524        ]);
525        let spec = build_source(&e).unwrap();
526        assert_eq!(spec.kind, "csv");
527        assert_eq!(spec.config, json!({"path": "./in.csv"}));
528    }
529
530    #[test]
531    fn build_source_errors_when_selector_missing() {
532        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
533        let err = build_source(&e).unwrap_err();
534        match err {
535            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
536            other => panic!("expected MissingEnvSelector, got {other:?}"),
537        }
538    }
539
540    #[test]
541    fn build_source_errors_when_selector_empty() {
542        let e = env(&[("FAUCET_SOURCE", "")]);
543        let err = build_source(&e).unwrap_err();
544        match err {
545            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
546            other => panic!("expected MissingEnvSelector, got {other:?}"),
547        }
548    }
549
550    #[test]
551    fn build_sink_reads_selector_and_scope() {
552        let e = env(&[
553            ("FAUCET_SINK", "jsonl"),
554            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
555        ]);
556        let spec = build_sink(&e).unwrap();
557        assert_eq!(spec.kind, "jsonl");
558        assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
559    }
560
561    #[test]
562    fn build_sink_errors_when_selector_missing() {
563        let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
564        let err = build_sink(&e).unwrap_err();
565        match err {
566            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
567            other => panic!("expected MissingEnvSelector, got {other:?}"),
568        }
569    }
570
571    #[test]
572    fn build_sink_errors_when_selector_empty() {
573        let e = env(&[("FAUCET_SINK", "")]);
574        let err = build_sink(&e).unwrap_err();
575        match err {
576            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
577            other => panic!("expected MissingEnvSelector, got {other:?}"),
578        }
579    }
580
581    #[test]
582    fn build_state_returns_none_when_unset() {
583        let e = env(&[("FAUCET_SOURCE", "rest")]);
584        let spec = build_state(&e).unwrap();
585        assert!(spec.is_none());
586    }
587
588    #[test]
589    fn build_state_returns_none_when_empty() {
590        let e = env(&[("FAUCET_STATE", "")]);
591        let spec = build_state(&e).unwrap();
592        assert!(spec.is_none());
593    }
594
595    #[test]
596    fn build_state_reads_file_backend() {
597        let e = env(&[
598            ("FAUCET_STATE", "file"),
599            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
600        ]);
601        let spec = build_state(&e).unwrap().unwrap();
602        assert_eq!(spec.kind, "file");
603        assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
604    }
605
606    #[test]
607    fn build_state_reads_memory_with_empty_scope() {
608        let e = env(&[("FAUCET_STATE", "memory")]);
609        let spec = build_state(&e).unwrap().unwrap();
610        assert_eq!(spec.kind, "memory");
611        assert_eq!(spec.config, json!({}));
612    }
613
614    #[test]
615    fn build_transforms_empty_when_unset() {
616        let e = env(&[("FAUCET_SOURCE", "rest")]);
617        let t = build_transforms(&e).unwrap();
618        assert!(t.is_empty());
619    }
620
621    #[test]
622    fn build_transforms_single_kind_no_config() {
623        let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
624        let t = build_transforms(&e).unwrap();
625        assert_eq!(t.len(), 1);
626        assert_eq!(t[0].kind, "snake_case");
627        assert_eq!(t[0].config, json!({}));
628    }
629
630    #[test]
631    fn build_transforms_ordered_and_with_config() {
632        let e = env(&[
633            ("FAUCET_TRANSFORM_1", "snake_case"),
634            ("FAUCET_TRANSFORM_2", "flatten"),
635            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
636        ]);
637        let t = build_transforms(&e).unwrap();
638        assert_eq!(t.len(), 2);
639        assert_eq!(t[0].kind, "snake_case");
640        assert_eq!(t[1].kind, "flatten");
641        assert_eq!(t[1].config, json!({"separator": "__"}));
642    }
643
644    #[test]
645    fn build_transforms_handles_double_digit_indices() {
646        let e = env(&[
647            ("FAUCET_TRANSFORM_1", "snake_case"),
648            ("FAUCET_TRANSFORM_2", "flatten"),
649            ("FAUCET_TRANSFORM_3", "rename_keys"),
650        ]);
651        let t = build_transforms(&e).unwrap();
652        assert_eq!(t.len(), 3);
653        assert_eq!(t[2].kind, "rename_keys");
654    }
655
656    #[test]
657    fn build_transforms_gap_errors() {
658        let e = env(&[
659            ("FAUCET_TRANSFORM_1", "snake_case"),
660            ("FAUCET_TRANSFORM_3", "flatten"),
661        ]);
662        let err = build_transforms(&e).unwrap_err();
663        match err {
664            CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
665            other => panic!("expected TransformIndexGap, got {other:?}"),
666        }
667    }
668
669    #[test]
670    fn build_transforms_must_start_at_one() {
671        let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
672        let err = build_transforms(&e).unwrap_err();
673        match err {
674            CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
675            other => panic!("expected TransformIndexGap, got {other:?}"),
676        }
677    }
678
679    #[test]
680    fn build_transforms_ignores_field_vars_when_indexing_kinds() {
681        // FAUCET_TRANSFORM_1_SEPARATOR should NOT be mistaken for a kind at index 1.
682        let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
683        // No FAUCET_TRANSFORM_1 selector means no transforms at all (not a gap error,
684        // because the indices set is empty).
685        let t = build_transforms(&e).unwrap();
686        assert!(t.is_empty());
687    }
688
689    #[test]
690    fn picks_up_named_source_templates() {
691        let e = env(&[
692            // Legacy default
693            ("FAUCET_SOURCE", "rest"),
694            ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
695            ("FAUCET_SINK", "jsonl"),
696            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
697            // Named templates
698            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
699            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
700            ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
701            ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
702            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
703            ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
704        ]);
705        let cfg = build_pipeline_config(&e).unwrap();
706        assert!(cfg.pipeline.source.is_some());
707        assert_eq!(cfg.pipeline.sources.len(), 2);
708        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
709        assert_eq!(
710            cfg.pipeline.sources["users_api"].config["base_url"],
711            "https://users.example"
712        );
713        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
714    }
715
716    #[test]
717    fn picks_up_vars_block() {
718        let e = env(&[
719            ("FAUCET_SOURCE", "rest"),
720            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
721            ("FAUCET_SINK", "jsonl"),
722            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
723            ("FAUCET_VARS_API_BASE", "https://api.example.com"),
724            ("FAUCET_VARS_REGION", "us-east-1"),
725        ]);
726        let cfg = build_pipeline_config(&e).unwrap();
727        let vars = cfg.vars.unwrap();
728        assert_eq!(vars["api_base"], "https://api.example.com");
729        assert_eq!(vars["region"], "us-east-1");
730    }
731
732    #[test]
733    fn named_source_only_no_legacy_works() {
734        // Verifies: with no FAUCET_SOURCE / FAUCET_SINK but only named
735        // templates, the singular source/sink are None and the catalogs are
736        // populated. (No MissingEnvSelector.)
737        let e = env(&[
738            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
739            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
740            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
741            ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
742        ]);
743        let cfg = build_pipeline_config(&e).unwrap();
744        assert!(cfg.pipeline.source.is_none());
745        assert!(cfg.pipeline.sink.is_none());
746        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
747        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
748    }
749
750    #[test]
751    fn no_source_anywhere_errors() {
752        // Neither legacy nor named sources — still must error.
753        let e = env(&[
754            ("FAUCET_SINK", "jsonl"),
755            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
756        ]);
757        let err = build_pipeline_config(&e).unwrap_err();
758        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
759    }
760
761    #[test]
762    fn build_pipeline_config_minimal_csv_to_jsonl() {
763        let e = env(&[
764            ("FAUCET_SOURCE", "csv"),
765            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
766            ("FAUCET_SINK", "jsonl"),
767            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
768        ]);
769        let cfg = build_pipeline_config(&e).unwrap();
770        assert_eq!(cfg.version, 1);
771        assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
772        assert_eq!(
773            cfg.pipeline.source.as_ref().unwrap().config,
774            json!({"path": "./in.csv"})
775        );
776        assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
777        assert_eq!(
778            cfg.pipeline.sink.as_ref().unwrap().config,
779            json!({"path": "./out.jsonl"})
780        );
781        assert!(cfg.pipeline.transforms.is_empty());
782        assert!(cfg.pipeline.state.is_none());
783        assert!(cfg.name.is_none());
784    }
785
786    #[test]
787    fn build_pipeline_config_uses_faucet_name_when_set() {
788        let e = env(&[
789            ("FAUCET_NAME", "github-issues"),
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.name.as_deref(), Some("github-issues"));
797    }
798
799    #[test]
800    fn build_pipeline_config_treats_empty_name_as_none() {
801        let e = env(&[
802            ("FAUCET_NAME", ""),
803            ("FAUCET_SOURCE", "csv"),
804            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
805            ("FAUCET_SINK", "jsonl"),
806            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
807        ]);
808        let cfg = build_pipeline_config(&e).unwrap();
809        assert!(cfg.name.is_none());
810    }
811
812    #[test]
813    fn build_pipeline_config_with_state_and_transforms() {
814        let e = env(&[
815            ("FAUCET_SOURCE", "csv"),
816            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
817            ("FAUCET_SINK", "jsonl"),
818            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
819            ("FAUCET_STATE", "file"),
820            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
821            ("FAUCET_TRANSFORM_1", "snake_case"),
822            ("FAUCET_TRANSFORM_2", "flatten"),
823            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
824        ]);
825        let cfg = build_pipeline_config(&e).unwrap();
826        assert_eq!(cfg.pipeline.transforms.len(), 2);
827        assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
828    }
829
830    #[test]
831    fn build_pipeline_config_missing_source_errors() {
832        let e = env(&[
833            ("FAUCET_SINK", "jsonl"),
834            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
835        ]);
836        let err = build_pipeline_config(&e).unwrap_err();
837        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
838    }
839
840    #[test]
841    fn build_pipeline_config_missing_sink_errors() {
842        let e = env(&[
843            ("FAUCET_SOURCE", "csv"),
844            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
845        ]);
846        let err = build_pipeline_config(&e).unwrap_err();
847        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
848    }
849}