Skip to main content

oxdock_process/
expand.rs

1use std::collections::HashMap;
2
3use anyhow::{Result, anyhow, bail};
4
5use crate::contract::CommandContext;
6
7/// Maximum bytes to buffer while scanning for closing delimiter.
8/// If exceeded without finding closing delimiter, buffered bytes are flushed as literals.
9const MAX_PLACEHOLDER_SCAN: usize = 1024;
10
11/// Configurable delimiter syntax for template expansion.
12pub struct TemplateDelimiters {
13    pub open: &'static [u8],
14    pub close: &'static [u8],
15}
16
17impl Default for TemplateDelimiters {
18    fn default() -> Self {
19        Self {
20            open: b"{{",
21            close: b"}}",
22        }
23    }
24}
25
26/// Streaming template expansion state machine.
27///
28/// Processes input bytes incrementally, expanding `{{ env:KEY }}` placeholders.
29/// At most `MAX_PLACEHOLDER_SCAN` bytes are held in buffer. Plain text streams
30/// flush immediately with zero buffering.
31pub struct StreamingExpand {
32    /// Bytes accumulated as key payload inside `{{ ... }}` (no open delimiter prefix).
33    buffer: Vec<u8>,
34    /// Explicit key=value overrides (take precedence over env).
35    overrides: HashMap<String, String>,
36    /// Environment variable lookup.
37    env: HashMap<String, String>,
38    /// Structured variable lookup (for key-path evaluation).
39    vars: HashMap<String, oxdock_parser::Value>,
40    /// State: are we currently inside a placeholder?
41    in_placeholder: bool,
42    /// Trailing opening byte from previous chunk — deferred across chunks.
43    pending_brace: bool,
44    /// Trailing closing byte from previous chunk — deferred across chunks.
45    pending_close_brace: bool,
46    /// Trailing backslash from previous chunk — deferred across chunks so
47    /// `\{{` split across a boundary still emits a literal opener.
48    pending_escape: bool,
49    /// Trailing `\` + `{` from previous chunk — deferred across chunks so
50    /// a `\{` split across a boundary still resolves as an escape pair.
51    pending_escape_brace: bool,
52    /// Configurable delimiter syntax.
53    delimiters: TemplateDelimiters,
54}
55
56impl StreamingExpand {
57    /// Create with env vars and optional explicit overrides.
58    /// Overrides take precedence over env vars.
59    pub fn new(overrides: &[(String, String)], env: &HashMap<String, String>) -> Self {
60        Self {
61            buffer: Vec::with_capacity(256),
62            overrides: overrides.iter().cloned().collect(),
63            env: env.clone(),
64            vars: HashMap::new(),
65            in_placeholder: false,
66            pending_brace: false,
67            pending_close_brace: false,
68            pending_escape: false,
69            pending_escape_brace: false,
70            delimiters: TemplateDelimiters::default(),
71        }
72    }
73
74    /// Create with env vars, structured variables, and optional explicit overrides.
75    /// Enables key-path evaluation in template tags (e.g., `{{ pkg.package.name }}`).
76    pub fn with_vars(mut self, vars: &HashMap<String, oxdock_parser::Value>) -> Self {
77        self.vars = vars.clone();
78        self
79    }
80
81    /// Process a chunk of input bytes, writing expanded output to `out`.
82    /// Returns early on empty input to preserve pending boundary state.
83    pub fn process_bytes(&mut self, input: &[u8], out: &mut Vec<u8>) -> Result<usize> {
84        if input.is_empty() {
85            return Ok(0);
86        }
87
88        let start_len = out.len();
89        let mut i = 0;
90
91        // Handle pending close brace from previous chunk
92        if self.pending_close_brace {
93            self.pending_close_brace = false;
94            if input[0] == self.delimiters.close[1] {
95                // Confirmed close delimiter across boundary — extract key, lookup, emit
96                let key = extract_key(&self.buffer);
97                let value = lookup(&key, &self.overrides, &self.env, &self.vars)?;
98                out.extend_from_slice(value.as_bytes());
99                self.buffer.clear();
100                self.in_placeholder = false;
101                i = self.delimiters.close.len() - 1; // Skip input[0] (the second byte)
102            } else {
103                // Lone closing byte — treat as literal part of key
104                // Push it to buffer, then let scan_placeholder process input[0]
105                self.buffer.push(self.delimiters.close[0]);
106                i = 0; // Do NOT skip input[0] — let scan_placeholder handle it
107            }
108        }
109
110        // Handle pending open brace from previous chunk
111        if self.pending_brace {
112            self.pending_brace = false;
113            if input[0] == self.delimiters.open[1] {
114                // Confirmed `{{` across boundary — enter PlaceholderScan
115                self.in_placeholder = true;
116                self.buffer.clear();
117                i = 1; // Skip the second open byte
118            } else {
119                // Single open byte was just a literal — flush it
120                out.push(self.delimiters.open[0]);
121            }
122        }
123
124        // Handle pending escape pair from previous chunk: a trailing `\{`
125        // combines with a leading `{` into a literal opener. Anything
126        // else means both bytes were literal; re-emit them, then let the
127        // scan below handle this chunk from the start.
128        if self.pending_escape_brace {
129            self.pending_escape_brace = false;
130            if !input.is_empty() && input[0] == self.delimiters.open[1] {
131                out.extend_from_slice(b"{{");
132                i = 1;
133            } else {
134                out.push(b'\\');
135                out.push(self.delimiters.open[0]);
136            }
137        }
138
139        // Handle pending escape from previous chunk: a lone trailing `\`
140        // combines with a leading `{{` into a literal opener, or with a
141        // leading `\` into a literal backslash (the pair is complete, so
142        // whatever follows processes normally). A lone trailing `{`
143        // cannot decide yet, so it joins the deferred pair above.
144        if self.pending_escape {
145            self.pending_escape = false;
146            if input.len() >= 2 && input[0] == b'{' && input[1] == b'{' {
147                out.extend_from_slice(b"{{");
148                i = 2;
149            } else if input.len() == 1 && input[0] == self.delimiters.open[0] {
150                self.pending_escape_brace = true;
151                i = 1;
152            } else if !input.is_empty() && input[0] == b'\\' {
153                out.push(b'\\');
154                i = 1;
155            } else {
156                // Not an escape — the backslash was literal; reprocess
157                // this chunk from the start (a lone `{` still defers via
158                // the pending_brace path below).
159                out.push(b'\\');
160            }
161        }
162
163        if self.in_placeholder {
164            // We're inside a placeholder — scan for closing delimiter
165            i = self.scan_placeholder(input, i, out)?;
166        }
167
168        // Normal state — scan for opening byte or flush literals
169        while i < input.len() {
170            if input[i] == b'\\' {
171                if i + 2 < input.len() && input[i + 1] == b'{' && input[i + 2] == b'{' {
172                    // Escaped opener — emit a literal `{{`, consume all three
173                    out.extend_from_slice(b"{{");
174                    i += 3;
175                } else if i + 2 == input.len() && input[i + 1] == self.delimiters.open[0] {
176                    // `\{` ends the chunk — defer the pair; the next
177                    // chunk decides literal `{{` vs literal `\` + `{`.
178                    self.pending_escape_brace = true;
179                    i += 2;
180                } else if i + 1 < input.len() && input[i + 1] == b'\\' {
181                    // Escaped backslash — emit one `\`, consume both (this
182                    // keeps `\\{{ ... }}` expanding, matching the lenient
183                    // interpolator used for command arguments)
184                    out.push(b'\\');
185                    i += 2;
186                } else if i + 1 == input.len() {
187                    // Trailing backslash — defer across the chunk boundary
188                    self.pending_escape = true;
189                    i += 1;
190                } else {
191                    // Ordinary backslash — literal, reprocess what follows
192                    out.push(b'\\');
193                    i += 1;
194                }
195            } else if input[i] == self.delimiters.open[0] {
196                if i + 1 < input.len() && input[i + 1] == self.delimiters.open[1] {
197                    // Found open delimiter — enter PlaceholderScan
198                    self.in_placeholder = true;
199                    self.buffer.clear();
200                    i += 2;
201                    i = self.scan_placeholder(input, i, out)?;
202                } else if i + 1 == input.len() {
203                    // Opening byte is the LAST byte of chunk — defer
204                    self.pending_brace = true;
205                    i += 1;
206                } else {
207                    // Single opening byte in the middle — flush as literal
208                    out.push(self.delimiters.open[0]);
209                    i += 1;
210                }
211            } else {
212                // Flush literal bytes until we find an opening byte, a
213                // backslash (potential `\{{` escape), or end of chunk
214                let start = i;
215                while i < input.len() && input[i] != self.delimiters.open[0] && input[i] != b'\\' {
216                    i += 1;
217                }
218                out.extend_from_slice(&input[start..i]);
219            }
220        }
221
222        Ok(out.len() - start_len)
223    }
224
225    /// Flush remaining buffer. Incomplete placeholders are treated as literals.
226    pub fn flush(mut self, out: &mut Vec<u8>) -> Result<()> {
227        // Emit deferred closing byte if present
228        if self.pending_close_brace {
229            self.buffer.push(self.delimiters.close[0]);
230            self.pending_close_brace = false;
231        }
232        // Emit deferred opening byte if present
233        if self.pending_brace {
234            out.push(self.delimiters.open[0]);
235            self.pending_brace = false;
236        }
237        // Emit deferred backslash if present
238        if self.pending_escape {
239            out.push(b'\\');
240            self.pending_escape = false;
241        }
242        // Emit deferred escape pair if present
243        if self.pending_escape_brace {
244            out.push(b'\\');
245            out.push(self.delimiters.open[0]);
246            self.pending_escape_brace = false;
247        }
248        // If inside placeholder, emit open delimiter prefix ONCE before buffer
249        if self.in_placeholder {
250            out.extend_from_slice(self.delimiters.open);
251            self.in_placeholder = false;
252        }
253        // Flush remaining buffer as literal text
254        out.extend_from_slice(&self.buffer);
255        self.buffer.clear();
256        Ok(())
257    }
258
259    /// Process a complete string (convenience for short command arguments).
260    pub fn expand_string(self, input: &str) -> Result<String> {
261        let mut out = Vec::with_capacity(input.len());
262        let mut expander = self;
263        expander.process_bytes(input.as_bytes(), &mut out)?;
264        expander.flush(&mut out)?;
265        Ok(String::from_utf8(out).unwrap_or_default())
266    }
267
268    /// Scan for closing delimiter starting at position `i`.
269    /// Returns the next position to process after the placeholder.
270    fn scan_placeholder(&mut self, input: &[u8], mut i: usize, out: &mut Vec<u8>) -> Result<usize> {
271        while i < input.len() {
272            if input[i] == self.delimiters.close[0] {
273                if i + 1 < input.len() && input[i + 1] == self.delimiters.close[1] {
274                    // Found closing delimiter — extract key, lookup, emit expansion
275                    let key = extract_key(&self.buffer);
276                    let value = lookup(&key, &self.overrides, &self.env, &self.vars)?;
277                    out.extend_from_slice(value.as_bytes());
278                    self.buffer.clear();
279                    self.in_placeholder = false;
280                    return Ok(i + 2);
281                }
282                if i + 1 == input.len() {
283                    // Closing byte is the LAST byte — defer to next chunk
284                    self.pending_close_brace = true;
285                    return Ok(i + 1);
286                }
287            }
288            self.buffer.push(input[i]);
289            i += 1;
290
291            // Buffer limit exceeded — flush as literal
292            if self.buffer.len() > MAX_PLACEHOLDER_SCAN {
293                out.extend_from_slice(self.delimiters.open);
294                out.extend_from_slice(&self.buffer);
295                self.buffer.clear();
296                self.in_placeholder = false;
297                return Ok(i);
298            }
299        }
300        Ok(i)
301    }
302}
303
304/// Extract and trim key from buffer content (the bytes between delimiters).
305fn extract_key(buffer: &[u8]) -> String {
306    String::from_utf8_lossy(buffer).trim().to_string()
307}
308
309/// Lookup a key in overrides and env, stripping namespace prefixes.
310///
311/// Strict resolution contract:
312/// - `{{ KEY }}` (bare) → overrides only
313/// - `{{ env:KEY }}` → overrides then env
314/// - `{{ $var }}` → script vars
315/// - `{{ $var.field }}` → script var key-path
316///
317/// All missing or invalid references return an error.
318fn lookup(
319    raw_key: &str,
320    overrides: &HashMap<String, String>,
321    env: &HashMap<String, String>,
322    vars: &HashMap<String, oxdock_parser::Value>,
323) -> Result<String> {
324    let key = raw_key.trim();
325
326    // 1. Explicit overrides (command-level CLI flags: KEY=val)
327    if let Some(val) = overrides.get(key) {
328        return Ok(val.clone());
329    }
330
331    // 2. Environment variables: must be prefixed with "env:"
332    if let Some(env_key) = key.strip_prefix("env:") {
333        if let Some(val) = overrides.get(env_key).or_else(|| env.get(env_key)) {
334            return Ok(val.clone());
335        }
336        let hint = if vars.contains_key(env_key) {
337            format!("; did you mean '${env_key}' (script variable)?")
338        } else {
339            String::new()
340        };
341        bail!("undefined environment variable: '{env_key}'{hint}");
342    }
343
344    // 3. Script variables: must be prefixed with "$"
345    if let Some(var_key) = key.strip_prefix('$') {
346        if var_key.contains('.') {
347            let parts: Vec<&str> = var_key.split('.').collect();
348            return resolve_key_path_strict(&parts, vars);
349        }
350        if let Some(val) = vars.get(var_key) {
351            return Ok(format_value_for_string(val));
352        }
353        let hint = if env.contains_key(var_key) {
354            format!("; did you mean 'env:{var_key}' (environment variable)?")
355        } else if overrides.contains_key(var_key) {
356            format!("; did you mean '{var_key}' (step override)?")
357        } else {
358            String::new()
359        };
360        bail!("undefined script variable: '${var_key}'{hint}");
361    }
362
363    // 4. Unprefixed key: could be a missing step override or invalid syntax
364    if !key.is_empty()
365        && key
366            .chars()
367            .all(|c| c.is_alphanumeric() || c == '_' || c == '-')
368    {
369        let hint = if vars.contains_key(key) {
370            format!("; did you mean '${key}' (script variable)?")
371        } else if env.contains_key(key) {
372            format!("; did you mean 'env:{key}' (environment variable)?")
373        } else {
374            String::new()
375        };
376        bail!("missing required step override argument: '{key}'{hint}");
377    }
378    bail!(
379        "invalid placeholder format '{key}': script variables must start with '$' and environment variables with 'env:'"
380    );
381}
382
383/// Resolve nested key-paths against the vars map.
384///
385/// Fails explicitly on missing object keys, out-of-bounds array indices,
386/// or type mismatches (e.g. trying to access a property on a primitive).
387fn resolve_key_path_strict(
388    parts: &[&str],
389    vars: &HashMap<String, oxdock_parser::Value>,
390) -> Result<String> {
391    let root_key = parts[0];
392    let mut current = vars
393        .get(root_key)
394        .ok_or_else(|| anyhow!("undefined script variable: '${root_key}'"))?;
395
396    for &segment in &parts[1..] {
397        match current {
398            oxdock_parser::Value::Map(map) => {
399                current = map.get(segment).ok_or_else(|| {
400                    anyhow!("property '{segment}' not found on object '${root_key}'")
401                })?;
402            }
403            oxdock_parser::Value::List(list) => {
404                let idx: usize = segment.parse().map_err(|_| {
405                    anyhow!("invalid array index '{segment}' on list '${root_key}'")
406                })?;
407                current = list.get(idx).ok_or_else(|| {
408                    anyhow!(
409                        "index {idx} out of bounds for list '${root_key}' (len: {})",
410                        list.len()
411                    )
412                })?;
413            }
414            _ => bail!("cannot access property '{segment}' on primitive value of '${root_key}'"),
415        }
416    }
417
418    Ok(format_value_for_string(current))
419}
420
421/// Format a Value as a string for inline interpolation.
422fn format_value_for_string(val: &oxdock_parser::Value) -> String {
423    match val {
424        oxdock_parser::Value::String(s) => s.clone(),
425        oxdock_parser::Value::Int(i) => i.to_string(),
426        oxdock_parser::Value::Bool(b) => b.to_string(),
427        oxdock_parser::Value::List(items) => items
428            .iter()
429            .map(format_value_for_string)
430            .collect::<Vec<_>>()
431            .join(" "),
432        oxdock_parser::Value::Map(map) => map
433            .iter()
434            .map(|(k, v)| format!("\"{}\": {}", k, format_value_for_string(v)))
435            .collect::<Vec<_>>()
436            .join(", "),
437        oxdock_parser::Value::TaskHandle(id) => format!("task#{}", id),
438    }
439}
440
441// Legacy functions for backward compatibility
442
443pub(crate) fn expand_with_lookup<F>(input: &str, mut lookup_fn: F) -> String
444where
445    F: FnMut(&str) -> Option<String>,
446{
447    let mut out = String::with_capacity(input.len());
448    let mut chars = input.chars().peekable();
449    while let Some(c) = chars.next() {
450        if c == '{' {
451            if let Some(&'{') = chars.peek() {
452                chars.next(); // consume second '{'
453                let mut content = String::new();
454                let mut closed = false;
455                // Look ahead for closing }}
456                let mut inner_chars = chars.clone();
457                while let Some(ch) = inner_chars.next() {
458                    if ch == '}'
459                        && let Some(&'}') = inner_chars.peek()
460                    {
461                        closed = true;
462                        break;
463                    }
464                    content.push(ch);
465                }
466
467                if closed {
468                    // Advance main iterator past content and closing braces.
469                    // Count chars, not bytes: content may contain multi-byte
470                    // UTF-8 (e.g. non-ASCII placeholder names).
471                    for _ in 0..content.chars().count() {
472                        chars.next();
473                    }
474                    chars.next(); // first }
475                    chars.next(); // second }
476
477                    let key = content.trim();
478                    if !key.is_empty() {
479                        out.push_str(&lookup_fn(key).unwrap_or_default());
480                    }
481                } else {
482                    out.push('{');
483                    out.push('{');
484                }
485            } else {
486                out.push('{');
487            }
488        } else {
489            out.push(c);
490        }
491    }
492    out
493}
494
495pub fn expand_script_env(input: &str, script_envs: &HashMap<String, String>) -> String {
496    expand_with_lookup(input, |name| {
497        if let Some(key) = name.strip_prefix("env:") {
498            script_envs
499                .get(key)
500                .cloned()
501                .or_else(|| std::env::var(key).ok())
502        } else {
503            None
504        }
505    })
506}
507
508pub fn expand_command_env(input: &str, ctx: &CommandContext) -> String {
509    expand_with_lookup(input, |name| {
510        if let Some(key) = name.strip_prefix("env:") {
511            ctx.envs().get(key).cloned()
512        } else {
513            None
514        }
515    })
516}
517
518#[cfg(test)]
519mod tests {
520    use super::*;
521    use std::collections::HashMap;
522
523    #[test]
524    fn basic_expansion() {
525        let mut env = HashMap::new();
526        env.insert("NAME".into(), "World".into());
527        let expander = StreamingExpand::new(&[], &env);
528        let result = expander.expand_string("Hello {{ env:NAME }}").unwrap();
529        assert_eq!(result, "Hello World");
530    }
531
532    #[test]
533    fn multiple_vars() {
534        let mut env = HashMap::new();
535        env.insert("A".into(), "X".into());
536        env.insert("B".into(), "Y".into());
537        let expander = StreamingExpand::new(&[], &env);
538        let result = expander
539            .expand_string("{{ env:A }} and {{ env:B }}")
540            .unwrap();
541        assert_eq!(result, "X and Y");
542    }
543
544    #[test]
545    fn missing_var() {
546        let env = HashMap::new();
547        let expander = StreamingExpand::new(&[], &env);
548        let result = expander.expand_string("{{ env:MISSING }}");
549        assert!(result.is_err());
550        assert!(
551            result
552                .unwrap_err()
553                .to_string()
554                .contains("undefined environment variable"),
555            "error should mention undefined environment variable"
556        );
557    }
558
559    #[test]
560    fn no_placeholders() {
561        let env = HashMap::new();
562        let expander = StreamingExpand::new(&[], &env);
563        let result = expander.expand_string("plain text").unwrap();
564        assert_eq!(result, "plain text");
565    }
566
567    #[test]
568    fn empty_input() {
569        let env = HashMap::new();
570        let expander = StreamingExpand::new(&[], &env);
571        let result = expander.expand_string("").unwrap();
572        assert_eq!(result, "");
573    }
574
575    #[test]
576    fn override_precedence() {
577        let mut env = HashMap::new();
578        env.insert("KEY".into(), "envval".into());
579        let overrides = vec![("KEY".into(), "override".into())];
580        let expander = StreamingExpand::new(&overrides, &env);
581        let result = expander.expand_string("{{ env:KEY }}").unwrap();
582        assert_eq!(result, "override");
583    }
584
585    #[test]
586    fn override_with_namespace() {
587        let mut env = HashMap::new();
588        env.insert("CRATE".into(), "envval".into());
589        let overrides = vec![("CRATE".into(), "override".into())];
590        let expander = StreamingExpand::new(&overrides, &env);
591        let result = expander.expand_string("{{ env:CRATE }}").unwrap();
592        assert_eq!(result, "override");
593    }
594
595    #[test]
596    fn override_raw_key_match() {
597        let mut env = HashMap::new();
598        env.insert("CRATE".into(), "envval".into());
599        let overrides = vec![("env:CRATE".into(), "override".into())];
600        let expander = StreamingExpand::new(&overrides, &env);
601        let result = expander.expand_string("{{ env:CRATE }}").unwrap();
602        assert_eq!(result, "override");
603    }
604
605    #[test]
606    fn unclosed_placeholder() {
607        let env = HashMap::new();
608        let expander = StreamingExpand::new(&[], &env);
609        let result = expander.expand_string("{{ env:KEY").unwrap();
610        assert_eq!(result, "{{ env:KEY");
611    }
612
613    #[test]
614    fn unclosed_with_prefix() {
615        let env = HashMap::new();
616        let expander = StreamingExpand::new(&[], &env);
617        let result = expander.expand_string("Hello {{ env:KEY").unwrap();
618        assert_eq!(result, "Hello {{ env:KEY");
619    }
620
621    #[test]
622    fn buffer_limit_exceeded() {
623        let env = HashMap::new();
624        let expander = StreamingExpand::new(&[], &env);
625        // Create input with `{{` followed by >1024 bytes without `}}`
626        let mut input = b"{{ ".to_vec();
627        input.extend(std::iter::repeat_n(b'x', 2000));
628        let result = expander
629            .expand_string(&String::from_utf8_lossy(&input))
630            .unwrap();
631        // Should flush as literal with `{{` prefix
632        assert!(result.starts_with("{{ "));
633        assert!(result.len() > 1024);
634    }
635
636    #[test]
637    fn escaped_opener_emits_literal() {
638        let mut env = HashMap::new();
639        env.insert("PROJECT".into(), "OxDock".into());
640        let expander = StreamingExpand::new(&[], &env);
641        let result = expander
642            .expand_string("Built with \\{{ env:PROJECT }}")
643            .unwrap();
644        assert_eq!(result, "Built with {{ env:PROJECT }}");
645    }
646
647    #[test]
648    fn escaped_opener_across_chunks() {
649        let mut env = HashMap::new();
650        env.insert("PROJECT".into(), "OxDock".into());
651        let mut expander = StreamingExpand::new(&[], &env);
652        let mut out = Vec::new();
653
654        // Split `\` / `{{ env:PROJECT }}` across chunks
655        expander.process_bytes(b"Built with \\", &mut out).unwrap();
656        expander
657            .process_bytes(b"{{ env:PROJECT }}", &mut out)
658            .unwrap();
659        expander.flush(&mut out).unwrap();
660
661        assert_eq!(
662            String::from_utf8_lossy(&out),
663            "Built with {{ env:PROJECT }}"
664        );
665    }
666
667    #[test]
668    fn double_backslash_then_placeholder_expands() {
669        let mut env = HashMap::new();
670        env.insert("PROJECT".into(), "OxDock".into());
671        let expander = StreamingExpand::new(&[], &env);
672        let result = expander.expand_string("\\\\{{ env:PROJECT }}").unwrap();
673        assert_eq!(result, "\\OxDock");
674    }
675
676    #[test]
677    fn trailing_backslash_flushes_literal() {
678        let env = HashMap::new();
679        let expander = StreamingExpand::new(&[], &env);
680        let result = expander.expand_string("end\\").unwrap();
681        assert_eq!(result, "end\\");
682    }
683
684    #[test]
685    fn backslash_before_other_text_is_literal() {
686        let env = HashMap::new();
687        let expander = StreamingExpand::new(&[], &env);
688        let result = expander.expand_string("a\\b \\{ once }").unwrap();
689        assert_eq!(result, "a\\b \\{ once }");
690    }
691
692    #[test]
693    fn escaped_pair_split_across_chunks() {
694        let mut env = HashMap::new();
695        env.insert("PROJECT".into(), "OxDock".into());
696        let mut expander = StreamingExpand::new(&[], &env);
697        let mut out = Vec::new();
698
699        // Split `\\` / `{{ env:PROJECT }}` across chunks: the pair
700        // completes to one backslash, then the placeholder expands
701        expander.process_bytes(b"\\", &mut out).unwrap();
702        expander
703            .process_bytes(b"\\{{ env:PROJECT }}", &mut out)
704            .unwrap();
705        expander.flush(&mut out).unwrap();
706
707        assert_eq!(String::from_utf8_lossy(&out), "\\OxDock");
708    }
709
710    #[test]
711    fn partial_across_chunks() {
712        let mut env = HashMap::new();
713        env.insert("NAME".into(), "World".into());
714        let mut expander = StreamingExpand::new(&[], &env);
715        let mut out = Vec::new();
716
717        // Split `{{ env:NA` / `ME }}` across chunks
718        expander.process_bytes(b"{{ env:NA", &mut out).unwrap();
719        expander.process_bytes(b"ME }}", &mut out).unwrap();
720        expander.flush(&mut out).unwrap();
721
722        assert_eq!(String::from_utf8_lossy(&out), "World");
723    }
724
725    #[test]
726    fn trailing_brace_across_chunks() {
727        let mut env = HashMap::new();
728        env.insert("NAME".into(), "World".into());
729        let mut expander = StreamingExpand::new(&[], &env);
730        let mut out = Vec::new();
731
732        // Split `...{` / `{env:NAME}}` across chunks
733        expander.process_bytes(b"...", &mut out).unwrap();
734        expander.process_bytes(b"{", &mut out).unwrap();
735        expander.process_bytes(b"{env:NAME}}", &mut out).unwrap();
736        expander.flush(&mut out).unwrap();
737
738        assert_eq!(String::from_utf8_lossy(&out), "...World");
739    }
740
741    #[test]
742    fn trailing_brace_at_eof() {
743        let env = HashMap::new();
744        let mut expander = StreamingExpand::new(&[], &env);
745        let mut out = Vec::new();
746
747        expander.process_bytes(b"hello{", &mut out).unwrap();
748        expander.flush(&mut out).unwrap();
749
750        assert_eq!(String::from_utf8_lossy(&out), "hello{");
751    }
752
753    #[test]
754    fn immediate_flush_guarantee() {
755        let env = HashMap::new();
756        let mut expander = StreamingExpand::new(&[], &env);
757        let mut out = Vec::new();
758
759        // 1MB of plain text with no placeholders
760        let input = std::iter::repeat_n(b'x', 1024 * 1024).collect::<Vec<_>>();
761        expander.process_bytes(&input, &mut out).unwrap();
762        expander.flush(&mut out).unwrap();
763
764        assert_eq!(out.len(), 1024 * 1024);
765    }
766
767    #[test]
768    fn nested_braces() {
769        let mut env = HashMap::new();
770        env.insert("KEY{1}".into(), "val".into());
771        let expander = StreamingExpand::new(&[], &env);
772        let result = expander.expand_string("{{ env:KEY{1} }}").unwrap();
773        assert_eq!(result, "val");
774    }
775
776    #[test]
777    fn split_close_delimiter_across_chunks() {
778        let mut env = HashMap::new();
779        env.insert("NAME".into(), "World".into());
780        let mut expander = StreamingExpand::new(&[], &env);
781        let mut out = Vec::new();
782
783        // Split `}}` across chunks: `{{ env:NAME` / `}}`
784        expander.process_bytes(b"{{ env:NAME", &mut out).unwrap();
785        expander.process_bytes(b"}}", &mut out).unwrap();
786        expander.flush(&mut out).unwrap();
787
788        assert_eq!(String::from_utf8_lossy(&out), "World");
789    }
790
791    #[test]
792    fn split_close_delimiter_with_trailing_content() {
793        let mut env = HashMap::new();
794        env.insert("NAME".into(), "World".into());
795        let mut expander = StreamingExpand::new(&[], &env);
796        let mut out = Vec::new();
797
798        // Split `}}` across chunks with content after
799        expander.process_bytes(b"{{ env:NAME", &mut out).unwrap();
800        expander.process_bytes(b"}} rest", &mut out).unwrap();
801        expander.flush(&mut out).unwrap();
802
803        assert_eq!(String::from_utf8_lossy(&out), "World rest");
804    }
805
806    #[test]
807    fn empty_input_preserves_pending_state() {
808        let mut env = HashMap::new();
809        env.insert("NAME".into(), "World".into());
810        let mut expander = StreamingExpand::new(&[], &env);
811        let mut out = Vec::new();
812
813        // End chunk with closing byte, then empty input, then confirm
814        expander.process_bytes(b"{{ env:NAME", &mut out).unwrap();
815        expander.process_bytes(b"", &mut out).unwrap(); // empty — should preserve state
816        expander.process_bytes(b"}}", &mut out).unwrap();
817        expander.flush(&mut out).unwrap();
818
819        assert_eq!(String::from_utf8_lossy(&out), "World");
820    }
821
822    #[test]
823    fn missing_env_var_errors() {
824        let env = HashMap::new();
825        let expander = StreamingExpand::new(&[], &env);
826        let result = expander.expand_string("{{ env:UNDEFINED_VAR }}");
827        assert!(result.is_err());
828        let msg = result.unwrap_err().to_string();
829        assert!(msg.contains("undefined environment variable"), "got: {msg}");
830        assert!(msg.contains("UNDEFINED_VAR"), "got: {msg}");
831    }
832
833    #[test]
834    fn missing_script_var_errors() {
835        let env = HashMap::new();
836        let expander = StreamingExpand::new(&[], &env);
837        let result = expander.expand_string("{{ $undefined_var }}");
838        assert!(result.is_err());
839        let msg = result.unwrap_err().to_string();
840        assert!(msg.contains("undefined script variable"), "got: {msg}");
841        assert!(msg.contains("$undefined_var"), "got: {msg}");
842    }
843
844    #[test]
845    fn missing_key_in_map_errors() {
846        let mut vars = HashMap::new();
847        vars.insert(
848            "cfg".into(),
849            oxdock_parser::Value::Map(std::collections::BTreeMap::from([(
850                "server".into(),
851                oxdock_parser::Value::Map(std::collections::BTreeMap::from([(
852                    "port".into(),
853                    oxdock_parser::Value::Int(8080),
854                )])),
855            )])),
856        );
857        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
858        let result = expander.expand_string("{{ $cfg.missing_key }}");
859        assert!(result.is_err());
860        let msg = result.unwrap_err().to_string();
861        assert!(
862            msg.contains("property 'missing_key' not found"),
863            "got: {msg}"
864        );
865    }
866
867    #[test]
868    fn out_of_bounds_array_index_errors() {
869        let mut vars = HashMap::new();
870        vars.insert(
871            "arr".into(),
872            oxdock_parser::Value::List(vec![oxdock_parser::Value::String("a".into())]),
873        );
874        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
875        let result = expander.expand_string("{{ $arr.5 }}");
876        assert!(result.is_err());
877        let msg = result.unwrap_err().to_string();
878        assert!(msg.contains("index 5 out of bounds"), "got: {msg}");
879    }
880
881    #[test]
882    fn type_mismatch_navigation_errors() {
883        let mut vars = HashMap::new();
884        vars.insert("name".into(), oxdock_parser::Value::String("alice".into()));
885        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
886        let result = expander.expand_string("{{ $name.sub_field }}");
887        assert!(result.is_err());
888        let msg = result.unwrap_err().to_string();
889        assert!(
890            msg.contains("cannot access property 'sub_field' on primitive"),
891            "got: {msg}"
892        );
893    }
894
895    #[test]
896    fn unprefixed_identifier_errors() {
897        let env = HashMap::new();
898        let expander = StreamingExpand::new(&[], &env);
899        let result = expander.expand_string("{{ bare_word }}");
900        assert!(result.is_err());
901        let msg = result.unwrap_err().to_string();
902        assert!(msg.contains("missing required step override"), "got: {msg}");
903    }
904
905    // ── Strict namespace isolation tests ─────────────────────────────────────
906
907    #[test]
908    fn script_var_does_not_fall_back_to_env() {
909        let mut env = HashMap::new();
910        env.insert("WHO".into(), "from-env".into());
911        let expander = StreamingExpand::new(&[], &env);
912        // $WHO queries vars, NOT env — should error even though env has WHO
913        let result = expander.expand_string("{{ $WHO }}");
914        assert!(result.is_err());
915        let msg = result.unwrap_err().to_string();
916        assert!(msg.contains("undefined script variable"), "got: {msg}");
917        assert!(
918            msg.contains("did you mean 'env:WHO'"),
919            "hint should suggest env: prefix, got: {msg}"
920        );
921    }
922
923    #[test]
924    fn env_var_does_not_fall_back_to_vars() {
925        let mut vars = HashMap::new();
926        vars.insert(
927            "HOST".into(),
928            oxdock_parser::Value::String("from-var".into()),
929        );
930        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
931        // env:HOST queries env, NOT vars — should error even though vars has HOST
932        let result = expander.expand_string("{{ env:HOST }}");
933        assert!(result.is_err());
934        let msg = result.unwrap_err().to_string();
935        assert!(msg.contains("undefined environment variable"), "got: {msg}");
936        assert!(
937            msg.contains("did you mean '$HOST'"),
938            "hint should suggest $ prefix, got: {msg}"
939        );
940    }
941
942    #[test]
943    fn step_override_does_not_fall_back_to_vars() {
944        let mut vars = HashMap::new();
945        vars.insert("PORT".into(), oxdock_parser::Value::Int(8080));
946        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
947        // PORT (bare) queries overrides, NOT vars — should error
948        let result = expander.expand_string("{{ PORT }}");
949        assert!(result.is_err());
950        let msg = result.unwrap_err().to_string();
951        assert!(msg.contains("missing required step override"), "got: {msg}");
952        assert!(
953            msg.contains("did you mean '$PORT'"),
954            "hint should suggest $ prefix, got: {msg}"
955        );
956    }
957
958    #[test]
959    fn env_prefix_isolated_from_script_vars() {
960        let mut vars = HashMap::new();
961        vars.insert("MODE".into(), oxdock_parser::Value::String("dev".into()));
962        let expander = StreamingExpand::new(&[], &HashMap::new()).with_vars(&vars);
963        // env:MODE looks in env, not vars — should error
964        let result = expander.expand_string("{{ env:MODE }}");
965        assert!(result.is_err());
966        let msg = result.unwrap_err().to_string();
967        assert!(msg.contains("undefined environment variable"), "got: {msg}");
968        assert!(
969            msg.contains("did you mean '$MODE'"),
970            "hint should suggest $ prefix, got: {msg}"
971        );
972    }
973
974    #[test]
975    fn dollar_prefix_isolated_from_env() {
976        let mut env = HashMap::new();
977        env.insert("PORT".into(), "3000".into());
978        let expander = StreamingExpand::new(&[], &env);
979        // $PORT looks in vars, not env — should error
980        let result = expander.expand_string("{{ $PORT }}");
981        assert!(result.is_err());
982        let msg = result.unwrap_err().to_string();
983        assert!(msg.contains("undefined script variable"), "got: {msg}");
984        assert!(
985            msg.contains("did you mean 'env:PORT'"),
986            "hint should suggest env: prefix, got: {msg}"
987        );
988    }
989
990    #[test]
991    fn empty_placeholder_errors() {
992        let expander = StreamingExpand::new(&[], &HashMap::new());
993        let result = expander.expand_string("{{ }}");
994        assert!(result.is_err());
995        let msg = result.unwrap_err().to_string();
996        assert!(msg.contains("invalid placeholder format"), "got: {msg}");
997    }
998
999    #[test]
1000    fn malformed_symbol_placeholder_errors() {
1001        let expander = StreamingExpand::new(&[], &HashMap::new());
1002        let result = expander.expand_string("{{ @invalid! }}");
1003        assert!(result.is_err());
1004        let msg = result.unwrap_err().to_string();
1005        assert!(msg.contains("invalid placeholder format"), "got: {msg}");
1006    }
1007
1008    /// Feed input in fixed-size chunks, then flush: exercises every
1009    /// chunk-boundary alignment for the given split size.
1010    fn render_split(
1011        input: &[u8],
1012        size: usize,
1013        env: &HashMap<String, String>,
1014    ) -> Result<String, anyhow::Error> {
1015        let mut expander = StreamingExpand::new(&[], env);
1016        let mut out = Vec::new();
1017        for chunk in input.chunks(size) {
1018            expander.process_bytes(chunk, &mut out)?;
1019        }
1020        expander.flush(&mut out)?;
1021        Ok(String::from_utf8_lossy(&out).into_owned())
1022    }
1023
1024    #[test]
1025    fn escaped_opener_one_byte_chunks_stays_literal() {
1026        let mut env = HashMap::new();
1027        env.insert("NAME".into(), "World".into());
1028        // Every byte arrives alone: `\` | `{` | `{` | ... must still
1029        // resolve to a literal opener, never an expansion.
1030        let out = render_split(b"\\{{ env:NAME }}", 1, &env).unwrap();
1031        assert_eq!(out, "{{ env:NAME }}");
1032    }
1033
1034    #[test]
1035    fn escaped_opener_two_byte_chunks_stays_literal() {
1036        let mut env = HashMap::new();
1037        env.insert("NAME".into(), "World".into());
1038        // Two-byte splits land `\{` and lone `{` at chunk ends.
1039        let out = render_split(b"a\\{{ env:NAME }}b", 2, &env).unwrap();
1040        assert_eq!(out, "a{{ env:NAME }}b");
1041    }
1042
1043    #[test]
1044    fn escaped_opener_split_variants_stay_literal() {
1045        let mut env = HashMap::new();
1046        env.insert("NAME".into(), "World".into());
1047        for chunks in [
1048            vec![b"\\".as_slice(), b"{", b"{ env:NAME }}"],
1049            vec![b"\\{".as_slice(), b"{ env:NAME }}"],
1050            vec![b"\\{{ env:NAME ".as_slice(), b"}}"],
1051        ] {
1052            let mut expander = StreamingExpand::new(&[], &env);
1053            let mut out = Vec::new();
1054            for chunk in chunks {
1055                expander.process_bytes(chunk, &mut out).unwrap();
1056            }
1057            expander.flush(&mut out).unwrap();
1058            assert_eq!(
1059                String::from_utf8_lossy(&out),
1060                "{{ env:NAME }}",
1061                "chunks must not corrupt the escape"
1062            );
1063        }
1064    }
1065
1066    #[test]
1067    fn plain_opener_split_chunks_still_expands() {
1068        let mut env = HashMap::new();
1069        env.insert("NAME".into(), "World".into());
1070        // The fix must not swallow genuine openers fragmented the
1071        // same way: `{` | `{` opens, then the key resolves.
1072        let out = render_split(b"{{ env:NAME }}", 1, &env).unwrap();
1073        assert_eq!(out, "World");
1074    }
1075
1076    #[test]
1077    fn double_backslash_split_chunks_still_expands() {
1078        let mut env = HashMap::new();
1079        env.insert("NAME".into(), "World".into());
1080        // `\\` completes to one backslash; the opener after it expands.
1081        let out = render_split(b"\\\\{{ env:NAME }}", 1, &env).unwrap();
1082        assert_eq!(out, "\\World");
1083    }
1084
1085    #[test]
1086    fn trailing_escape_brace_flushes_literal() {
1087        let env = HashMap::new();
1088        let expander = StreamingExpand::new(&[], &env);
1089        let result = expander.expand_string("end\\{").unwrap();
1090        assert_eq!(result, "end\\{");
1091    }
1092}