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        status: None,
103        tags: Vec::new(),
104        complete_for: None,
105    })
106}
107
108/// Construct the sink [`ConnectorSpec`] from `FAUCET_SINK` + `FAUCET_SINK_<KIND>_*`.
109pub fn build_sink(env: &HashMap<String, String>) -> CliResult<ConnectorSpec> {
110    let kind = env
111        .get("FAUCET_SINK")
112        .filter(|v| !v.is_empty())
113        .ok_or_else(|| CliError::MissingEnvSelector {
114            var: "FAUCET_SINK".to_owned(),
115        })?
116        .clone();
117    let prefix = format!(
118        "FAUCET_SINK_{}_",
119        kind.to_ascii_uppercase().replace('-', "_")
120    );
121    let config = extract_scope(env, &prefix)?;
122    Ok(ConnectorSpec {
123        kind,
124        config,
125        transforms: None,
126        inherit_transforms: true,
127        status: None,
128        tags: Vec::new(),
129        complete_for: None,
130    })
131}
132
133/// Construct an optional [`StateStoreSpec`] from `FAUCET_STATE` + `FAUCET_STATE_<KIND>_*`.
134/// Returns `Ok(None)` when `FAUCET_STATE` is unset or empty (the common case).
135pub fn build_state(env: &HashMap<String, String>) -> CliResult<Option<StateStoreSpec>> {
136    let Some(kind) = env.get("FAUCET_STATE").filter(|v| !v.is_empty()).cloned() else {
137        return Ok(None);
138    };
139    let prefix = format!(
140        "FAUCET_STATE_{}_",
141        kind.to_ascii_uppercase().replace('-', "_")
142    );
143    let config = extract_scope(env, &prefix)?;
144    Ok(Some(StateStoreSpec { kind, config }))
145}
146
147/// Construct an ordered `Vec<TransformSpec>` from `FAUCET_TRANSFORM_<N>` selectors
148/// and `FAUCET_TRANSFORM_<N>_<FIELD>` config fields. Indices must be contiguous
149/// starting at 1; any gap is an error so a misnumbered var never silently drops
150/// a transform.
151pub fn build_transforms(env: &HashMap<String, String>) -> CliResult<Vec<TransformSpec>> {
152    // Collect the kind selectors first — keys that are exactly
153    // `FAUCET_TRANSFORM_<digits>` (no trailing field name).
154    let mut kinds: BTreeMap<u32, String> = BTreeMap::new();
155    for (key, value) in env {
156        let Some(rest) = key.strip_prefix("FAUCET_TRANSFORM_") else {
157            continue;
158        };
159        if rest.is_empty() || !rest.chars().all(|c| c.is_ascii_digit()) {
160            continue;
161        }
162        let Ok(idx) = rest.parse::<u32>() else {
163            continue;
164        };
165        kinds.insert(idx, value.clone());
166    }
167    if kinds.is_empty() {
168        return Ok(Vec::new());
169    }
170    // Indices must be 1, 2, 3, …
171    for (expected, actual) in (1u32..).zip(kinds.keys().copied()) {
172        if expected != actual {
173            return Err(CliError::TransformIndexGap { missing: expected });
174        }
175    }
176    // Harvest per-transform config blocks.
177    let mut out = Vec::with_capacity(kinds.len());
178    for (idx, kind) in kinds {
179        let prefix = format!("FAUCET_TRANSFORM_{idx}_");
180        let config = extract_scope(env, &prefix)?;
181        out.push(TransformSpec { kind, config });
182    }
183    Ok(out)
184}
185
186/// Harvest named source templates from `FAUCET_SOURCES_<NAME>_TYPE` selectors
187/// and `FAUCET_SOURCES_<NAME>_<FIELD>` config fields. `<NAME>` is lowercased.
188pub fn build_named_sources(
189    env: &HashMap<String, String>,
190) -> CliResult<HashMap<String, ConnectorSpec>> {
191    build_named_catalog(env, "FAUCET_SOURCES_")
192}
193
194/// Same as [`build_named_sources`] but for sinks via `FAUCET_SINKS_<NAME>_*`.
195pub fn build_named_sinks(
196    env: &HashMap<String, String>,
197) -> CliResult<HashMap<String, ConnectorSpec>> {
198    build_named_catalog(env, "FAUCET_SINKS_")
199}
200
201fn build_named_catalog(
202    env: &HashMap<String, String>,
203    prefix: &str,
204) -> CliResult<HashMap<String, ConnectorSpec>> {
205    // First sweep: find each template's `<NAME>` by spotting
206    // `<prefix><NAME>_TYPE`.
207    let mut kinds: HashMap<String, String> = HashMap::new();
208    for (key, value) in env {
209        let Some(suffix) = key.strip_prefix(prefix) else {
210            continue;
211        };
212        let Some(name_upper) = suffix.strip_suffix("_TYPE") else {
213            continue;
214        };
215        if name_upper.is_empty() {
216            continue;
217        }
218        kinds.insert(name_upper.to_ascii_lowercase(), value.clone());
219    }
220    // All template scope prefixes, so each env var can be assigned to its
221    // LONGEST matching prefix. Without this, a template like `users` would
222    // absorb `users_api`'s vars, since `FAUCET_SOURCES_USERS_` is a prefix of
223    // `FAUCET_SOURCES_USERS_API_` — the shorter template silently gets the
224    // longer one's fields (#146 M17).
225    let scope_prefixes: Vec<String> = kinds
226        .keys()
227        .map(|name| format!("{prefix}{}_", name.to_ascii_uppercase()))
228        .collect();
229
230    // Second sweep: harvest each template's config block.
231    let mut out: HashMap<String, ConnectorSpec> = HashMap::new();
232    for (name, kind) in kinds {
233        let scope_prefix = format!("{prefix}{}_", name.to_ascii_uppercase());
234        // A view of `env` containing only the vars whose longest matching
235        // template prefix is THIS template's — i.e. drop any var that also
236        // matches a longer sibling prefix nested under this one.
237        let scoped: HashMap<String, String> = env
238            .iter()
239            .filter(|(k, _)| {
240                k.starts_with(&scope_prefix)
241                    && !scope_prefixes.iter().any(|other| {
242                        other.len() > scope_prefix.len() && k.starts_with(other.as_str())
243                    })
244            })
245            .map(|(k, v)| (k.clone(), v.clone()))
246            .collect();
247        let mut config = extract_scope(&scoped, &scope_prefix)?;
248        // Remove the `type` field — it's the selector, not a config field.
249        if let Value::Object(m) = &mut config {
250            m.remove("type");
251        }
252        out.insert(
253            name,
254            ConnectorSpec {
255                kind,
256                config,
257                transforms: None,
258                inherit_transforms: true,
259                status: None,
260                tags: Vec::new(),
261                complete_for: None,
262            },
263        );
264    }
265    Ok(out)
266}
267
268/// Harvest `FAUCET_VARS_<KEY>` into the top-level vars map. Returns `None`
269/// when no `FAUCET_VARS_*` variables are set.
270pub fn build_vars(env: &HashMap<String, String>) -> Option<HashMap<String, Value>> {
271    let mut out: HashMap<String, Value> = HashMap::new();
272    for (key, value) in env {
273        let Some(name_upper) = key.strip_prefix("FAUCET_VARS_") else {
274            continue;
275        };
276        if name_upper.is_empty() {
277            continue;
278        }
279        out.insert(name_upper.to_ascii_lowercase(), coerce_scalar(value));
280    }
281    if out.is_empty() { None } else { Some(out) }
282}
283
284/// Construct a complete [`PipelineConfig`] from an env snapshot.
285pub fn build_pipeline_config(env: &HashMap<String, String>) -> CliResult<PipelineConfig> {
286    let source = match env.get("FAUCET_SOURCE").filter(|v| !v.is_empty()) {
287        Some(_) => Some(build_source(env)?),
288        None => None,
289    };
290    let sink = match env.get("FAUCET_SINK").filter(|v| !v.is_empty()) {
291        Some(_) => Some(build_sink(env)?),
292        None => None,
293    };
294    let sources = build_named_sources(env)?;
295    let sinks = build_named_sinks(env)?;
296    let state = build_state(env)?;
297    let transforms = build_transforms(env)?;
298    let vars = build_vars(env);
299    let name = env.get("FAUCET_NAME").cloned().filter(|s| !s.is_empty());
300
301    // At least one source and one sink must be declared somewhere.
302    if source.is_none() && sources.is_empty() {
303        return Err(CliError::MissingEnvSelector {
304            var: "FAUCET_SOURCE (or FAUCET_SOURCES_<NAME>_TYPE)".to_owned(),
305        });
306    }
307    if sink.is_none() && sinks.is_empty() {
308        return Err(CliError::MissingEnvSelector {
309            var: "FAUCET_SINK (or FAUCET_SINKS_<NAME>_TYPE)".to_owned(),
310        });
311    }
312    Ok(PipelineConfig {
313        version: 1,
314        name,
315        vars,
316        // Pure-env mode has no `params:` surface: every value already comes from
317        // the environment, which is what params would be overriding.
318        params: Default::default(),
319        // Pure-env mode doesn't (yet) assemble a shared `auth:` catalog; inline
320        // auth via FAUCET_*_AUTH_JSON still works.
321        auth: None,
322        pipeline: PipelineSpec {
323            source,
324            sink,
325            sources,
326            sinks,
327            transforms,
328            state,
329            dlq: None,
330            #[cfg(feature = "quality")]
331            quality: None,
332            #[cfg(feature = "contract")]
333            contract: None,
334            #[cfg(feature = "masking")]
335            masking: None,
336            schema: None,
337            // Pure-env mode does not assemble a topology graph.
338            nodes: std::collections::HashMap::new(),
339            edges: Vec::new(),
340        },
341        matrix: Vec::new(),
342        execution: None,
343        selection: None,
344        observability: None,
345        delivery: faucet_core::DeliveryMode::default(),
346        resilience: None,
347        // Pure-env mode doesn't (yet) assemble an `sla:` block.
348        sla: None,
349        shard: None,
350        replication: None,
351        backfill: None,
352        partition: None,
353        #[cfg(feature = "schedule")]
354        schedule: None,
355        #[cfg(feature = "lineage")]
356        lineage: None,
357        // Pure-env mode doesn't (yet) assemble a `catalog:` block.
358        #[cfg(feature = "catalog")]
359        catalog: None,
360        #[cfg(feature = "notify")]
361        notifications: Vec::new(),
362    })
363}
364
365/// Snapshot `std::env::vars()` and call [`build_pipeline_config`].
366pub fn from_process_env() -> CliResult<PipelineConfig> {
367    let env: HashMap<String, String> = std::env::vars().collect();
368    build_pipeline_config(&env)
369}
370
371#[cfg(test)]
372mod tests {
373    use super::*;
374    use serde_json::json;
375
376    fn env(pairs: &[(&str, &str)]) -> HashMap<String, String> {
377        pairs
378            .iter()
379            .map(|(k, v)| ((*k).to_owned(), (*v).to_owned()))
380            .collect()
381    }
382
383    #[test]
384    fn extract_scope_lowercases_field_names() {
385        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
386        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
387        assert_eq!(v, json!({"base_url": "https://x.example"}));
388    }
389
390    #[test]
391    fn extract_scope_ignores_unrelated_keys() {
392        let e = env(&[
393            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
394            ("PATH", "/usr/bin"),
395            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
396        ]);
397        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
398        assert_eq!(v, json!({"base_url": "https://x.example"}));
399    }
400
401    #[test]
402    fn named_templates_do_not_leak_across_prefix_overlapping_names() {
403        // M17 (#146): template `users` must NOT absorb `users_api`'s vars just
404        // because `FAUCET_SOURCES_USERS_` is a prefix of `FAUCET_SOURCES_USERS_API_`.
405        let e = env(&[
406            ("FAUCET_SOURCES_USERS_TYPE", "rest"),
407            ("FAUCET_SOURCES_USERS_BASE_URL", "https://u"),
408            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
409            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://api"),
410            ("FAUCET_SOURCES_USERS_API_TIMEOUT", "30"),
411        ]);
412        let out = build_named_sources(&e).unwrap();
413
414        let users = out.get("users").expect("users template").config.clone();
415        let users = users.as_object().unwrap();
416        assert_eq!(
417            users.get("base_url").and_then(|v| v.as_str()),
418            Some("https://u")
419        );
420        assert!(
421            !users.contains_key("api_base_url"),
422            "users must not absorb users_api's vars"
423        );
424        assert!(!users.contains_key("api_timeout"));
425
426        let api = out
427            .get("users_api")
428            .expect("users_api template")
429            .config
430            .clone();
431        let api = api.as_object().unwrap();
432        assert_eq!(
433            api.get("base_url").and_then(|v| v.as_str()),
434            Some("https://api")
435        );
436        assert_eq!(api.get("timeout").and_then(|v| v.as_i64()), Some(30));
437    }
438
439    #[test]
440    fn extract_scope_coerces_numbers_and_bools() {
441        let e = env(&[
442            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
443            ("FAUCET_SOURCE_REST_FOLLOW_REDIRECTS", "true"),
444            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
445        ]);
446        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
447        assert_eq!(v["timeout_secs"], json!(30));
448        assert_eq!(v["follow_redirects"], json!(true));
449        assert_eq!(v["base_url"], json!("https://x.example"));
450    }
451
452    #[test]
453    fn extract_scope_handles_json_suffix() {
454        let e = env(&[(
455            "FAUCET_SOURCE_REST_AUTH_JSON",
456            r#"{"type":"ApiKey","header":"Authorization","value":"Bearer x"}"#,
457        )]);
458        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
459        assert_eq!(
460            v["auth"],
461            json!({"type": "ApiKey", "header": "Authorization", "value": "Bearer x"})
462        );
463    }
464
465    #[test]
466    fn extract_scope_rejects_invalid_json_suffix() {
467        let e = env(&[("FAUCET_SOURCE_REST_AUTH_JSON", "not-json")]);
468        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
469        match err {
470            CliError::InvalidEnvJson { var, .. } => {
471                assert_eq!(var, "FAUCET_SOURCE_REST_AUTH_JSON")
472            }
473            other => panic!("expected InvalidEnvJson, got {other:?}"),
474        }
475    }
476
477    #[test]
478    fn extract_scope_conflict_scalar_then_json() {
479        let e = env(&[
480            ("FAUCET_SOURCE_REST_AUTH", "bearer"),
481            ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
482        ]);
483        let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
484        match err {
485            CliError::EnvConflict {
486                field,
487                scalar_var,
488                json_var,
489            } => {
490                assert_eq!(field, "auth");
491                assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
492                assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
493            }
494            other => panic!("expected EnvConflict, got {other:?}"),
495        }
496    }
497
498    #[test]
499    fn extract_scope_conflict_detection_is_order_independent() {
500        // HashMap iteration order is randomized per instance via the random
501        // hasher state; loop 50 times to exercise both ordering branches
502        // (scalar-arriving-first AND json-arriving-first) statistically.
503        for _ in 0..50 {
504            let e = env(&[
505                ("FAUCET_SOURCE_REST_AUTH", "bearer"),
506                ("FAUCET_SOURCE_REST_AUTH_JSON", r#"{"type":"ApiKey"}"#),
507            ]);
508            let err = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap_err();
509            match err {
510                CliError::EnvConflict {
511                    field,
512                    scalar_var,
513                    json_var,
514                } => {
515                    assert_eq!(field, "auth");
516                    assert_eq!(scalar_var, "FAUCET_SOURCE_REST_AUTH");
517                    assert_eq!(json_var, "FAUCET_SOURCE_REST_AUTH_JSON");
518                }
519                other => panic!("expected EnvConflict, got {other:?}"),
520            }
521        }
522    }
523
524    #[test]
525    fn extract_scope_skips_bare_prefix() {
526        let e = env(&[("FAUCET_SOURCE_REST_", "ignored")]);
527        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
528        assert_eq!(v, json!({}));
529    }
530
531    #[test]
532    fn extract_scope_empty_when_no_matches() {
533        let e = env(&[("PATH", "/usr/bin")]);
534        let v = extract_scope(&e, "FAUCET_SOURCE_REST_").unwrap();
535        assert_eq!(v, json!({}));
536    }
537
538    #[test]
539    fn build_source_reads_selector_and_scope() {
540        let e = env(&[
541            ("FAUCET_SOURCE", "rest"),
542            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
543            ("FAUCET_SOURCE_REST_TIMEOUT_SECS", "30"),
544        ]);
545        let spec = build_source(&e).unwrap();
546        assert_eq!(spec.kind, "rest");
547        assert_eq!(spec.config["base_url"], json!("https://x.example"));
548        assert_eq!(spec.config["timeout_secs"], json!(30));
549    }
550
551    #[test]
552    fn build_source_uses_kind_scope_so_other_kinds_dont_leak() {
553        let e = env(&[
554            ("FAUCET_SOURCE", "csv"),
555            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
556            ("FAUCET_SOURCE_REST_BASE_URL", "https://other.example"),
557        ]);
558        let spec = build_source(&e).unwrap();
559        assert_eq!(spec.kind, "csv");
560        assert_eq!(spec.config, json!({"path": "./in.csv"}));
561    }
562
563    #[test]
564    fn build_source_errors_when_selector_missing() {
565        let e = env(&[("FAUCET_SOURCE_REST_BASE_URL", "https://x.example")]);
566        let err = build_source(&e).unwrap_err();
567        match err {
568            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
569            other => panic!("expected MissingEnvSelector, got {other:?}"),
570        }
571    }
572
573    #[test]
574    fn build_source_errors_when_selector_empty() {
575        let e = env(&[("FAUCET_SOURCE", "")]);
576        let err = build_source(&e).unwrap_err();
577        match err {
578            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SOURCE"),
579            other => panic!("expected MissingEnvSelector, got {other:?}"),
580        }
581    }
582
583    #[test]
584    fn build_sink_reads_selector_and_scope() {
585        let e = env(&[
586            ("FAUCET_SINK", "jsonl"),
587            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
588        ]);
589        let spec = build_sink(&e).unwrap();
590        assert_eq!(spec.kind, "jsonl");
591        assert_eq!(spec.config, json!({"path": "./out.jsonl"}));
592    }
593
594    #[test]
595    fn build_sink_errors_when_selector_missing() {
596        let e = env(&[("FAUCET_SINK_JSONL_PATH", "./out.jsonl")]);
597        let err = build_sink(&e).unwrap_err();
598        match err {
599            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
600            other => panic!("expected MissingEnvSelector, got {other:?}"),
601        }
602    }
603
604    #[test]
605    fn build_sink_errors_when_selector_empty() {
606        let e = env(&[("FAUCET_SINK", "")]);
607        let err = build_sink(&e).unwrap_err();
608        match err {
609            CliError::MissingEnvSelector { var } => assert_eq!(var, "FAUCET_SINK"),
610            other => panic!("expected MissingEnvSelector, got {other:?}"),
611        }
612    }
613
614    #[test]
615    fn build_state_returns_none_when_unset() {
616        let e = env(&[("FAUCET_SOURCE", "rest")]);
617        let spec = build_state(&e).unwrap();
618        assert!(spec.is_none());
619    }
620
621    #[test]
622    fn build_state_returns_none_when_empty() {
623        let e = env(&[("FAUCET_STATE", "")]);
624        let spec = build_state(&e).unwrap();
625        assert!(spec.is_none());
626    }
627
628    #[test]
629    fn build_state_reads_file_backend() {
630        let e = env(&[
631            ("FAUCET_STATE", "file"),
632            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
633        ]);
634        let spec = build_state(&e).unwrap().unwrap();
635        assert_eq!(spec.kind, "file");
636        assert_eq!(spec.config, json!({"path": "./.faucet-state"}));
637    }
638
639    #[test]
640    fn build_state_reads_memory_with_empty_scope() {
641        let e = env(&[("FAUCET_STATE", "memory")]);
642        let spec = build_state(&e).unwrap().unwrap();
643        assert_eq!(spec.kind, "memory");
644        assert_eq!(spec.config, json!({}));
645    }
646
647    #[test]
648    fn build_transforms_empty_when_unset() {
649        let e = env(&[("FAUCET_SOURCE", "rest")]);
650        let t = build_transforms(&e).unwrap();
651        assert!(t.is_empty());
652    }
653
654    #[test]
655    fn build_transforms_single_kind_no_config() {
656        let e = env(&[("FAUCET_TRANSFORM_1", "snake_case")]);
657        let t = build_transforms(&e).unwrap();
658        assert_eq!(t.len(), 1);
659        assert_eq!(t[0].kind, "snake_case");
660        assert_eq!(t[0].config, json!({}));
661    }
662
663    #[test]
664    fn build_transforms_ordered_and_with_config() {
665        let e = env(&[
666            ("FAUCET_TRANSFORM_1", "snake_case"),
667            ("FAUCET_TRANSFORM_2", "flatten"),
668            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
669        ]);
670        let t = build_transforms(&e).unwrap();
671        assert_eq!(t.len(), 2);
672        assert_eq!(t[0].kind, "snake_case");
673        assert_eq!(t[1].kind, "flatten");
674        assert_eq!(t[1].config, json!({"separator": "__"}));
675    }
676
677    #[test]
678    fn build_transforms_handles_double_digit_indices() {
679        let e = env(&[
680            ("FAUCET_TRANSFORM_1", "snake_case"),
681            ("FAUCET_TRANSFORM_2", "flatten"),
682            ("FAUCET_TRANSFORM_3", "rename_keys"),
683        ]);
684        let t = build_transforms(&e).unwrap();
685        assert_eq!(t.len(), 3);
686        assert_eq!(t[2].kind, "rename_keys");
687    }
688
689    #[test]
690    fn build_transforms_gap_errors() {
691        let e = env(&[
692            ("FAUCET_TRANSFORM_1", "snake_case"),
693            ("FAUCET_TRANSFORM_3", "flatten"),
694        ]);
695        let err = build_transforms(&e).unwrap_err();
696        match err {
697            CliError::TransformIndexGap { missing } => assert_eq!(missing, 2),
698            other => panic!("expected TransformIndexGap, got {other:?}"),
699        }
700    }
701
702    #[test]
703    fn build_transforms_must_start_at_one() {
704        let e = env(&[("FAUCET_TRANSFORM_2", "snake_case")]);
705        let err = build_transforms(&e).unwrap_err();
706        match err {
707            CliError::TransformIndexGap { missing } => assert_eq!(missing, 1),
708            other => panic!("expected TransformIndexGap, got {other:?}"),
709        }
710    }
711
712    #[test]
713    fn build_transforms_ignores_field_vars_when_indexing_kinds() {
714        // FAUCET_TRANSFORM_1_SEPARATOR should NOT be mistaken for a kind at index 1.
715        let e = env(&[("FAUCET_TRANSFORM_1_SEPARATOR", "__")]);
716        // No FAUCET_TRANSFORM_1 selector means no transforms at all (not a gap error,
717        // because the indices set is empty).
718        let t = build_transforms(&e).unwrap();
719        assert!(t.is_empty());
720    }
721
722    #[test]
723    fn picks_up_named_source_templates() {
724        let e = env(&[
725            // Legacy default
726            ("FAUCET_SOURCE", "rest"),
727            ("FAUCET_SOURCE_REST_BASE_URL", "https://default.example"),
728            ("FAUCET_SINK", "jsonl"),
729            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
730            // Named templates
731            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
732            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://users.example"),
733            ("FAUCET_SOURCES_POSTS_API_TYPE", "rest"),
734            ("FAUCET_SOURCES_POSTS_API_BASE_URL", "https://posts.example"),
735            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
736            ("FAUCET_SINKS_ARCHIVE_PATH", "./archive.jsonl"),
737        ]);
738        let cfg = build_pipeline_config(&e).unwrap();
739        assert!(cfg.pipeline.source.is_some());
740        assert_eq!(cfg.pipeline.sources.len(), 2);
741        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
742        assert_eq!(
743            cfg.pipeline.sources["users_api"].config["base_url"],
744            "https://users.example"
745        );
746        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
747    }
748
749    #[test]
750    fn picks_up_vars_block() {
751        let e = env(&[
752            ("FAUCET_SOURCE", "rest"),
753            ("FAUCET_SOURCE_REST_BASE_URL", "https://x.example"),
754            ("FAUCET_SINK", "jsonl"),
755            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
756            ("FAUCET_VARS_API_BASE", "https://api.example.com"),
757            ("FAUCET_VARS_REGION", "us-east-1"),
758        ]);
759        let cfg = build_pipeline_config(&e).unwrap();
760        let vars = cfg.vars.unwrap();
761        assert_eq!(vars["api_base"], "https://api.example.com");
762        assert_eq!(vars["region"], "us-east-1");
763    }
764
765    #[test]
766    fn named_source_only_no_legacy_works() {
767        // Verifies: with no FAUCET_SOURCE / FAUCET_SINK but only named
768        // templates, the singular source/sink are None and the catalogs are
769        // populated. (No MissingEnvSelector.)
770        let e = env(&[
771            ("FAUCET_SOURCES_USERS_API_TYPE", "rest"),
772            ("FAUCET_SOURCES_USERS_API_BASE_URL", "https://x.example"),
773            ("FAUCET_SINKS_ARCHIVE_TYPE", "jsonl"),
774            ("FAUCET_SINKS_ARCHIVE_PATH", "./o.jsonl"),
775        ]);
776        let cfg = build_pipeline_config(&e).unwrap();
777        assert!(cfg.pipeline.source.is_none());
778        assert!(cfg.pipeline.sink.is_none());
779        assert_eq!(cfg.pipeline.sources["users_api"].kind, "rest");
780        assert_eq!(cfg.pipeline.sinks["archive"].kind, "jsonl");
781    }
782
783    #[test]
784    fn no_source_anywhere_errors() {
785        // Neither legacy nor named sources — still must error.
786        let e = env(&[
787            ("FAUCET_SINK", "jsonl"),
788            ("FAUCET_SINK_JSONL_PATH", "./o.jsonl"),
789        ]);
790        let err = build_pipeline_config(&e).unwrap_err();
791        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
792    }
793
794    #[test]
795    fn build_pipeline_config_minimal_csv_to_jsonl() {
796        let e = env(&[
797            ("FAUCET_SOURCE", "csv"),
798            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
799            ("FAUCET_SINK", "jsonl"),
800            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
801        ]);
802        let cfg = build_pipeline_config(&e).unwrap();
803        assert_eq!(cfg.version, 1);
804        assert_eq!(cfg.pipeline.source.as_ref().unwrap().kind, "csv");
805        assert_eq!(
806            cfg.pipeline.source.as_ref().unwrap().config,
807            json!({"path": "./in.csv"})
808        );
809        assert_eq!(cfg.pipeline.sink.as_ref().unwrap().kind, "jsonl");
810        assert_eq!(
811            cfg.pipeline.sink.as_ref().unwrap().config,
812            json!({"path": "./out.jsonl"})
813        );
814        assert!(cfg.pipeline.transforms.is_empty());
815        assert!(cfg.pipeline.state.is_none());
816        assert!(cfg.name.is_none());
817    }
818
819    #[test]
820    fn build_pipeline_config_uses_faucet_name_when_set() {
821        let e = env(&[
822            ("FAUCET_NAME", "github-issues"),
823            ("FAUCET_SOURCE", "csv"),
824            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
825            ("FAUCET_SINK", "jsonl"),
826            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
827        ]);
828        let cfg = build_pipeline_config(&e).unwrap();
829        assert_eq!(cfg.name.as_deref(), Some("github-issues"));
830    }
831
832    #[test]
833    fn build_pipeline_config_treats_empty_name_as_none() {
834        let e = env(&[
835            ("FAUCET_NAME", ""),
836            ("FAUCET_SOURCE", "csv"),
837            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
838            ("FAUCET_SINK", "jsonl"),
839            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
840        ]);
841        let cfg = build_pipeline_config(&e).unwrap();
842        assert!(cfg.name.is_none());
843    }
844
845    #[test]
846    fn build_pipeline_config_with_state_and_transforms() {
847        let e = env(&[
848            ("FAUCET_SOURCE", "csv"),
849            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
850            ("FAUCET_SINK", "jsonl"),
851            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
852            ("FAUCET_STATE", "file"),
853            ("FAUCET_STATE_FILE_PATH", "./.faucet-state"),
854            ("FAUCET_TRANSFORM_1", "snake_case"),
855            ("FAUCET_TRANSFORM_2", "flatten"),
856            ("FAUCET_TRANSFORM_2_SEPARATOR", "__"),
857        ]);
858        let cfg = build_pipeline_config(&e).unwrap();
859        assert_eq!(cfg.pipeline.transforms.len(), 2);
860        assert_eq!(cfg.pipeline.state.as_ref().unwrap().kind, "file");
861    }
862
863    #[test]
864    fn build_pipeline_config_missing_source_errors() {
865        let e = env(&[
866            ("FAUCET_SINK", "jsonl"),
867            ("FAUCET_SINK_JSONL_PATH", "./out.jsonl"),
868        ]);
869        let err = build_pipeline_config(&e).unwrap_err();
870        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
871    }
872
873    #[test]
874    fn build_pipeline_config_missing_sink_errors() {
875        let e = env(&[
876            ("FAUCET_SOURCE", "csv"),
877            ("FAUCET_SOURCE_CSV_PATH", "./in.csv"),
878        ]);
879        let err = build_pipeline_config(&e).unwrap_err();
880        assert!(matches!(err, CliError::MissingEnvSelector { .. }));
881    }
882}