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        #[cfg(feature = "schedule")]
325        schedule: None,
326    })
327}
328
329/// Snapshot `std::env::vars()` and call [`build_pipeline_config`].
330pub fn from_process_env() -> CliResult<PipelineConfig> {
331    let env: HashMap<String, String> = std::env::vars().collect();
332    build_pipeline_config(&env)
333}
334
335#[cfg(test)]
336mod tests {
337    use super::*;
338    use serde_json::json;
339
340    fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
341        pairs
342            .iter()
343            .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
344            .collect()
345    }
346
347    #[test]
348    fn extract_scope_lowercases_field_names() {
349        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
350        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
351        assert_eq!(v, json!({"base_url": "https://x.example"}));
352    }
353
354    #[test]
355    fn extract_scope_ignores_unrelated_keys() {
356        let e = env(&[
357            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
358            ("PATH", "/usr/bin"),
359            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
360        ]);
361        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
362        assert_eq!(v, json!({"base_url": "https://x.example"}));
363    }
364
365    #[test]
366    fn named_templates_do_not_leak_across_prefix_overlapping_names() {
367        // M17 (#146): template `users` must NOT absorb `users_api`'s vars just
368        // because `FAUCET_SOURCES_USERS_` is a prefix of `FAUCET_SOURCES_USERS_API_`.
369        let e = env(&[
370            ("FAUCET_SOURCES_USERS_TYPE", "rest"),
371            ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
372            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
373            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
374            ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
375        ]);
376        let out = build_named_sources(&e).unwrap();
377
378        let users = out.get("users").expect("users template").config.clone();
379        let users = users.as_object().unwrap();
380        assert_eq!(
381            users.get("base_url").and_then(|v| v.as_str()),
382            Some("https://u")
383        );
384        assert!(
385            !users.contains_key("api_base_url"),
386            "users must not absorb users_api's vars"
387        );
388        assert!(!users.contains_key("api_timeout"));
389
390        let api = out
391            .get("users_api")
392            .expect("users_api template")
393            .config
394            .clone();
395        let api = api.as_object().unwrap();
396        assert_eq!(
397            api.get("base_url").and_then(|v| v.as_str()),
398            Some("https://api")
399        );
400        assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
401    }
402
403    #[test]
404    fn extract_scope_coerces_numbers_and_bools() {
405        let e = env(&[
406            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
407            ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
408            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
409        ]);
410        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
411        assert_eq!(v["timeout_secs"], json!(30));
412        assert_eq!(v["follow_redirects"], json!(true));
413        assert_eq!(v["base_url"], json!("https://x.example"));
414    }
415
416    #[test]
417    fn extract_scope_handles_json_suffix() {
418        let e = env(&[(
419            "FAUCET_SOURCE_REST_AUTH_JSON",
420            r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
421        )]);
422        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
423        assert_eq!(
424            v["auth"],
425            json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
426        );
427    }
428
429    #[test]
430    fn extract_scope_rejects_invalid_json_suffix() {
431        let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
432        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
433        match err {
434            CliError::InvalidEnvJson { var, .. } => {
435                assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
436            }
437            other => panic!("expected InvalidEnvJson, got {other:?}"),
438        }
439    }
440
441    #[test]
442    fn extract_scope_conflict_scalar_then_json() {
443        let e = env(&[
444            ("FAUCET_SOURCE_REST_AUTH", "bearer"),
445            ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
446        ]);
447        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
448        match err {
449            CliError::EnvConflict {
450                field,
451                scalar_var,
452                json_var,
453            } => {
454                assert_eq!(field, "auth");
455                assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
456                assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
457            }
458            other => panic!("expected EnvConflict, got {other:?}"),
459        }
460    }
461
462    #[test]
463    fn extract_scope_conflict_detection_is_order_independent() {
464        // HashMap iteration order is randomized per instance via the random
465        // hasher state; loop 50 times to exercise both ordering branches
466        // (scalar-arriving-first AND json-arriving-first) statistically.
467        for _ in 0..50 {
468            let e = env(&[
469                ("FAUCET_SOURCE_REST_AUTH", "bearer"),
470                ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
471            ]);
472            let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
473            match err {
474                CliError::EnvConflict {
475                    field,
476                    scalar_var,
477                    json_var,
478                } => {
479                    assert_eq!(field, "auth");
480                    assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
481                    assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
482                }
483                other => panic!("expected EnvConflict, got {other:?}"),
484            }
485        }
486    }
487
488    #[test]
489    fn extract_scope_skips_bare_prefix() {
490        let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
491        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
492        assert_eq!(v, json!({}));
493    }
494
495    #[test]
496    fn extract_scope_empty_when_no_matches() {
497        let e = env(&[("PATH", "/usr/bin")]);
498        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
499        assert_eq!(v, json!({}));
500    }
501
502    #[test]
503    fn build_source_reads_selector_and_scope() {
504        let e = env(&[
505            ("FAUCET_SOURCE", "rest"),
506            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
507            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
508        ]);
509        let spec = build_source(&e).unwrap();
510        assert_eq!(spec.kind, "rest");
511        assert_eq!(spec.config["base_url"], json!("https://x.example"));
512        assert_eq!(spec.config["timeout_secs"], json!(30));
513    }
514
515    #[test]
516    fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
517        let e = env(&[
518            ("FAUCET_SOURCE", "csv"),
519            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
520            ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
521        ]);
522        let spec = build_source(&e).unwrap();
523        assert_eq!(spec.kind, "csv");
524        assert_eq!(spec.config, json!({"path": "./in.csv"}));
525    }
526
527    #[test]
528    fn build_source_errors_when_selector_missing() {
529        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
530        let err = build_source(&e).unwrap_err();
531        match err {
532            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
533            other => panic!("expected MissingEnvSelector, got {other:?}"),
534        }
535    }
536
537    #[test]
538    fn build_source_errors_when_selector_empty() {
539        let e = env(&[("FAUCET_SOURCE", "")]);
540        let err = build_source(&e).unwrap_err();
541        match err {
542            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
543            other => panic!("expected MissingEnvSelector, got {other:?}"),
544        }
545    }
546
547    #[test]
548    fn build_sink_reads_selector_and_scope() {
549        let e = env(&[
550            ("FAUCET_SINK", "jsonl"),
551            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
552        ]);
553        let spec = build_sink(&e).unwrap();
554        assert_eq!(spec.kind, "jsonl");
555        assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
556    }
557
558    #[test]
559    fn build_sink_errors_when_selector_missing() {
560        let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
561        let err = build_sink(&e).unwrap_err();
562        match err {
563            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
564            other => panic!("expected MissingEnvSelector, got {other:?}"),
565        }
566    }
567
568    #[test]
569    fn build_sink_errors_when_selector_empty() {
570        let e = env(&[("FAUCET_SINK", "")]);
571        let err = build_sink(&e).unwrap_err();
572        match err {
573            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
574            other => panic!("expected MissingEnvSelector, got {other:?}"),
575        }
576    }
577
578    #[test]
579    fn build_state_returns_none_when_unset() {
580        let e = env(&[("FAUCET_SOURCE", "rest")]);
581        let spec = build_state(&e).unwrap();
582        assert!(spec.is_none());
583    }
584
585    #[test]
586    fn build_state_returns_none_when_empty() {
587        let e = env(&[("FAUCET_STATE", "")]);
588        let spec = build_state(&e).unwrap();
589        assert!(spec.is_none());
590    }
591
592    #[test]
593    fn build_state_reads_file_backend() {
594        let e = env(&[
595            ("FAUCET_STATE", "file"),
596            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
597        ]);
598        let spec = build_state(&e).unwrap().unwrap();
599        assert_eq!(spec.kind, "file");
600        assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
601    }
602
603    #[test]
604    fn build_state_reads_memory_with_empty_scope() {
605        let e = env(&[("FAUCET_STATE", "memory")]);
606        let spec = build_state(&e).unwrap().unwrap();
607        assert_eq!(spec.kind, "memory");
608        assert_eq!(spec.config, json!({}));
609    }
610
611    #[test]
612    fn build_transforms_empty_when_unset() {
613        let e = env(&[("FAUCET_SOURCE", "rest")]);
614        let t = build_transforms(&e).unwrap();
615        assert!(t.is_empty());
616    }
617
618    #[test]
619    fn build_transforms_single_kind_no_config() {
620        let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
621        let t = build_transforms(&e).unwrap();
622        assert_eq!(t.len(), 1);
623        assert_eq!(t[0].kind, "snake_case");
624        assert_eq!(t[0].config, json!({}));
625    }
626
627    #[test]
628    fn build_transforms_ordered_and_with_config() {
629        let e = env(&[
630            ("FAUCET_TRANSFORM_1", "snake_case"),
631            ("FAUCET_TRANSFORM_2", "flatten"),
632            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
633        ]);
634        let t = build_transforms(&e).unwrap();
635        assert_eq!(t.len(), 2);
636        assert_eq!(t[0].kind, "snake_case");
637        assert_eq!(t[1].kind, "flatten");
638        assert_eq!(t[1].config, json!({"separator": "__"}));
639    }
640
641    #[test]
642    fn build_transforms_handles_double_digit_indices() {
643        let e = env(&[
644            ("FAUCET_TRANSFORM_1", "snake_case"),
645            ("FAUCET_TRANSFORM_2", "flatten"),
646            ("FAUCET_TRANSFORM_3", "rename_keys"),
647        ]);
648        let t = build_transforms(&e).unwrap();
649        assert_eq!(t.len(), 3);
650        assert_eq!(t[2].kind, "rename_keys");
651    }
652
653    #[test]
654    fn build_transforms_gap_errors() {
655        let e = env(&[
656            ("FAUCET_TRANSFORM_1", "snake_case"),
657            ("FAUCET_TRANSFORM_3", "flatten"),
658        ]);
659        let err = build_transforms(&e).unwrap_err();
660        match err {
661            CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
662            other => panic!("expected TransformIndexGap, got {other:?}"),
663        }
664    }
665
666    #[test]
667    fn build_transforms_must_start_at_one() {
668        let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
669        let err = build_transforms(&e).unwrap_err();
670        match err {
671            CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
672            other => panic!("expected TransformIndexGap, got {other:?}"),
673        }
674    }
675
676    #[test]
677    fn build_transforms_ignores_field_vars_when_indexing_kinds() {
678        // FAUCET_TRANSFORM_1_SEPARATOR should NOT be mistaken for a kind at index 1.
679        let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
680        // No FAUCET_TRANSFORM_1 selector means no transforms at all (not a gap error,
681        // because the indices set is empty).
682        let t = build_transforms(&e).unwrap();
683        assert!(t.is_empty());
684    }
685
686    #[test]
687    fn picks_up_named_source_templates() {
688        let e = env(&[
689            // Legacy default
690            ("FAUCET_SOURCE", "rest"),
691            ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
692            ("FAUCET_SINK", "jsonl"),
693            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
694            // Named templates
695            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
696            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
697            ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
698            ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
699            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
700            ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
701        ]);
702        let cfg = build_pipeline_config(&e).unwrap();
703        assert!(cfg.pipeline.source.is_some());
704        assert_eq!(cfg.pipeline.sources.len(), 2);
705        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
706        assert_eq!(
707            cfg.pipeline.sources["users_api"].config["base_url"],
708            "https://users.example"
709        );
710        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
711    }
712
713    #[test]
714    fn picks_up_vars_block() {
715        let e = env(&[
716            ("FAUCET_SOURCE", "rest"),
717            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
718            ("FAUCET_SINK", "jsonl"),
719            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
720            ("FAUCET_VARS_API_BASE", "https://api.example.com"),
721            ("FAUCET_VARS_REGION", "us-east-1"),
722        ]);
723        let cfg = build_pipeline_config(&e).unwrap();
724        let vars = cfg.vars.unwrap();
725        assert_eq!(vars["api_base"], "https://api.example.com");
726        assert_eq!(vars["region"], "us-east-1");
727    }
728
729    #[test]
730    fn named_source_only_no_legacy_works() {
731        // Verifies: with no FAUCET_SOURCE / FAUCET_SINK but only named
732        // templates, the singular source/sink are None and the catalogs are
733        // populated. (No MissingEnvSelector.)
734        let e = env(&[
735            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
736            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
737            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
738            ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
739        ]);
740        let cfg = build_pipeline_config(&e).unwrap();
741        assert!(cfg.pipeline.source.is_none());
742        assert!(cfg.pipeline.sink.is_none());
743        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
744        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
745    }
746
747    #[test]
748    fn no_source_anywhere_errors() {
749        // Neither legacy nor named sources — still must error.
750        let e = env(&[
751            ("FAUCET_SINK", "jsonl"),
752            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
753        ]);
754        let err = build_pipeline_config(&e).unwrap_err();
755        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
756    }
757
758    #[test]
759    fn build_pipeline_config_minimal_csv_to_jsonl() {
760        let e = env(&[
761            ("FAUCET_SOURCE", "csv"),
762            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
763            ("FAUCET_SINK", "jsonl"),
764            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
765        ]);
766        let cfg = build_pipeline_config(&e).unwrap();
767        assert_eq!(cfg.version, 1);
768        assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
769        assert_eq!(
770            cfg.pipeline.source.as_ref().unwrap().config,
771            json!({"path": "./in.csv"})
772        );
773        assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
774        assert_eq!(
775            cfg.pipeline.sink.as_ref().unwrap().config,
776            json!({"path": "./out.jsonl"})
777        );
778        assert!(cfg.pipeline.transforms.is_empty());
779        assert!(cfg.pipeline.state.is_none());
780        assert!(cfg.name.is_none());
781    }
782
783    #[test]
784    fn build_pipeline_config_uses_faucet_name_when_set() {
785        let e = env(&[
786            ("FAUCET_NAME", "github-issues"),
787            ("FAUCET_SOURCE", "csv"),
788            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
789            ("FAUCET_SINK", "jsonl"),
790            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
791        ]);
792        let cfg = build_pipeline_config(&e).unwrap();
793        assert_eq!(cfg.name.as_deref(), Some("github-issues"));
794    }
795
796    #[test]
797    fn build_pipeline_config_treats_empty_name_as_none() {
798        let e = env(&[
799            ("FAUCET_NAME", ""),
800            ("FAUCET_SOURCE", "csv"),
801            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
802            ("FAUCET_SINK", "jsonl"),
803            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
804        ]);
805        let cfg = build_pipeline_config(&e).unwrap();
806        assert!(cfg.name.is_none());
807    }
808
809    #[test]
810    fn build_pipeline_config_with_state_and_transforms() {
811        let e = env(&[
812            ("FAUCET_SOURCE", "csv"),
813            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
814            ("FAUCET_SINK", "jsonl"),
815            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
816            ("FAUCET_STATE", "file"),
817            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
818            ("FAUCET_TRANSFORM_1", "snake_case"),
819            ("FAUCET_TRANSFORM_2", "flatten"),
820            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
821        ]);
822        let cfg = build_pipeline_config(&e).unwrap();
823        assert_eq!(cfg.pipeline.transforms.len(), 2);
824        assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
825    }
826
827    #[test]
828    fn build_pipeline_config_missing_source_errors() {
829        let e = env(&[
830            ("FAUCET_SINK", "jsonl"),
831            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
832        ]);
833        let err = build_pipeline_config(&e).unwrap_err();
834        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
835    }
836
837    #[test]
838    fn build_pipeline_config_missing_sink_errors() {
839        let e = env(&[
840            ("FAUCET_SOURCE", "csv"),
841            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
842        ]);
843        let err = build_pipeline_config(&e).unwrap_err();
844        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
845    }
846}