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