visi_core/core/vba/mod.rs
1// The syntax layer. These are `#[doc(hidden)] pub` for the same reason
2// `ovba` and `vba_xlsx` are: `visi-core/fuzz`'s `vba_parse` target needs to
3// reach `parse_module` from outside the crate. The supported surface is
4// [`check_syntax`] and [`ModuleSyntax`] below, which is what `core`'s
5// `pub use` list carries -- the AST is an implementation detail until the
6// interpreter phases need it, and pinning its shape now would be a semver
7// commitment made a phase too early.
8#[doc(hidden)]
9pub mod ast;
10pub(crate) mod builtin_names;
11#[doc(hidden)]
12pub mod builtins;
13pub(crate) mod color;
14#[doc(hidden)]
15pub mod host;
16#[doc(hidden)]
17pub mod interp;
18#[doc(hidden)]
19pub mod lexer;
20#[doc(hidden)]
21pub mod parser;
22pub(crate) mod resolve;
23#[doc(hidden)]
24pub mod value;
25
26use crate::{Error, ObjectKind};
27use serde::{Deserialize, Serialize};
28
29/// What [`check_syntax`] found in a module that parsed.
30#[derive(Debug, Clone, PartialEq, Eq, Default)]
31#[non_exhaustive]
32pub struct ModuleSyntax {
33 /// The names of every `Sub`, `Function` and `Property` declared, in source
34 /// order. Procedures inside a `#If` branch are all included: which branch
35 /// is live depends on `#Const` values, which parsing alone cannot decide.
36 pub procedures: Vec<String>,
37}
38
39/// Checks a VBA module's source for syntax errors.
40///
41/// Phase 0 of the plan in `docs/vba-macro-support.md`, plus the narrow
42/// name-resolution pass in [`resolve`]: it answers
43/// whether the source *compiles*, as far as parsing and resolving the names
44/// it can see will show. It does not check types or evaluate anything, so it
45/// will still accept a module that fails at run time -- and, being an
46/// independent implementation, may differ from Excel's compiler at the edges.
47///
48/// **`source` is treated as a self-contained project.** A name used with
49/// call syntax that resolves nowhere -- not in this module, not a VBA or
50/// Excel built-in -- is reported, which is right for a standalone `.bas` and
51/// for the single generated module the differential harness compiles, but
52/// would be wrong for one module of a larger project, where the name may
53/// live in a sibling. Use [`VbaProject::check_modules`] for that case -- it
54/// supplies each module the others' names -- or [`check_syntax_partial`]
55/// when the siblings are not available at all.
56///
57/// ```
58/// use visi_core::core::check_syntax;
59/// assert!(check_syntax("Sub Hello()\n MsgBox \"hi\"\nEnd Sub\n").is_ok());
60/// assert!(check_syntax("Sub Hello()\n").is_err());
61/// ```
62pub fn check_syntax(source: &str) -> Result<ModuleSyntax, Error> {
63 let empty = std::collections::HashSet::new();
64 check_source(source, None, &resolve::Scope::self_contained(&empty))
65}
66
67/// [`check_syntax`] for source that is **one module of a larger project**
68/// whose other modules are not available.
69///
70/// Same parse and the same rules, with one exception: a name that resolves
71/// nowhere is accepted rather than reported, since a sibling module this
72/// call cannot see may well declare it. Everything the module's own text
73/// disproves -- a syntax error, a duplicate declaration, a plain local used
74/// as a call target -- is still reported.
75///
76/// This is strictly the weaker check, and is the scope
77/// [`VbaModule::check_syntax`] already uses. Prefer
78/// [`VbaProject::check_modules`] wherever the whole project is in hand;
79/// reach for this only when it genuinely is not, as for a `.bas` file cut
80/// out of a project that lives elsewhere.
81///
82/// ```
83/// use visi_core::core::{check_syntax, check_syntax_partial};
84/// // `DoWork` is declared by some other module of the project.
85/// let src = "Sub Caller()\n DoWork 1\nEnd Sub\n";
86/// assert!(check_syntax(src).is_err());
87/// assert!(check_syntax_partial(src).is_ok());
88/// // A fragment is still held to what its own text shows.
89/// assert!(check_syntax_partial("Sub Caller()\n").is_err());
90/// ```
91pub fn check_syntax_partial(source: &str) -> Result<ModuleSyntax, Error> {
92 let empty = std::collections::HashSet::new();
93 check_source(source, None, &resolve::Scope::partial(&empty))
94}
95
96/// [`check_syntax`]'s body, with the resolution scope chosen by the caller.
97fn check_source(
98 source: &str,
99 module_name: Option<&str>,
100 scope: &resolve::Scope<'_>,
101) -> Result<ModuleSyntax, Error> {
102 let to_err = |e: parser::ParseError| Error::VbaSyntax {
103 message: e.message,
104 module: module_name.map(str::to_string),
105 line: e.pos.line,
106 column: e.pos.col,
107 };
108 let module = parser::parse_module(source).map_err(to_err)?;
109 resolve::check_module(&module, scope).map_err(to_err)?;
110 Ok(ModuleSyntax {
111 procedures: module.procedures().iter().map(|p| p.name.clone()).collect(),
112 })
113}
114
115/// The outcome of running a VBA procedure: its return value, rendered the way
116/// VBA would render it, plus the subtype name `TypeName()` reports.
117///
118/// Both halves matter. An interpreter that computes the right number with the
119/// wrong subtype has a real bug -- `1 + 1` is an `Integer` and `1 / 1` is a
120/// `Double` -- so the differential fuzzer compares the type as well as the
121/// value.
122#[derive(Debug, Clone, PartialEq, Eq)]
123#[non_exhaustive]
124pub struct RunOutcome {
125 /// `TypeName()` of the returned value.
126 pub type_name: String,
127 /// `CStr()` of the returned value, or `None` where VBA itself cannot
128 /// stringify it (`Null`).
129 pub value: Option<String>,
130 /// Whether the run changed the workbook.
131 ///
132 /// Always `false` from [`run_macro`], which has no workbook to change.
133 /// From [`crate::core::WorkbookManager::run_macro`] this is what tells a caller
134 /// whether it has something worth saving -- and, for the `visi` CLI,
135 /// whether discarding the result silently would be a data loss rather
136 /// than a no-op.
137 pub mutated: bool,
138}
139
140/// Turns command-line argument text into the `Variant`s a procedure receives.
141///
142/// Arguments arrive as text -- they come from a CLI or a fuzz harness -- and
143/// are given the type VBA would give the same literal, so `-a 1` is an
144/// `Integer` and `-a 1.5` a `Double`.
145fn parse_args(args: &[&str]) -> Vec<value::Variant> {
146 args.iter()
147 .map(|a| match value::parse_vba_number(a) {
148 Ok(n) if !a.trim().is_empty() => {
149 value::Variant::from_literal(n, a.contains('.') || a.contains(['e', 'E']))
150 }
151 _ => value::Variant::Str((*a).to_string()),
152 })
153 .collect()
154}
155
156fn to_outcome(result: value::Variant, mutated: bool, interp: &interp::Interpreter) -> RunOutcome {
157 RunOutcome {
158 type_name: interp.type_name_of(&result),
159 value: result.to_vba_string().ok(),
160 mutated,
161 }
162}
163
164fn parse_or_error(source: &str, module: Option<&str>) -> Result<ast::Module, Error> {
165 parser::parse_module(source).map_err(|e| Error::VbaSyntax {
166 message: e.message,
167 module: module.map(str::to_string),
168 line: e.pos.line,
169 column: e.pos.col,
170 })
171}
172
173fn to_runtime_error(e: value::VbaError) -> Error {
174 Error::VbaRuntime {
175 message: e.description,
176 number: e.number,
177 }
178}
179
180impl crate::core::WorkbookManager {
181 /// Runs one of this workbook's own VBA procedures **against** this
182 /// workbook.
183 ///
184 /// Phase 2 of `docs/vba-macro-support.md`, and the entry point that
185 /// separates it from Phase 1: the interpreter borrows the workbook for
186 /// the duration, so a macro can read and write cells, walk the sheets,
187 /// and call worksheet functions. [`run_macro`] stays as the text-only
188 /// form -- it is what `visi_core.run_macro` and `fuzz/fuzz_vba.py` drive,
189 /// and a macro that touches no workbook has no reason to need one.
190 ///
191 /// `module` picks which module to take the procedure from; `None`
192 /// searches every module for one that declares it, which is the common
193 /// single-module case. Resolving it here rather than in each caller is
194 /// Runs a VBA procedure in the workbook's project.
195 ///
196 /// The workbook is left recalculated, so a caller that saves afterwards
197 /// writes the values the macro itself would have read.
198 pub fn run_macro(
199 &mut self,
200 module: Option<&str>,
201 procedure: &str,
202 args: &[&str],
203 ) -> Result<RunOutcome, Error> {
204 let args = parse_args(args);
205
206 let interp = if let Some(project) = &self.vba_project {
207 if let Some(name) = module
208 && project.find_module(name).is_none()
209 {
210 let available = project.modules.iter().map(|m| m.name.clone()).collect();
211 return Err(Error::not_found_among(
212 ObjectKind::VbaModule,
213 name,
214 available,
215 ));
216 }
217 interp::Interpreter::from_project(project, module).map_err(to_runtime_error)?
218 } else {
219 let source = self.macro_source_for(module, procedure)?;
220 let parsed = parse_or_error(&source, module)?;
221 interp::Interpreter::new(parsed)
222 };
223
224 let host = host::Host::new(self).map_err(to_runtime_error)?;
225 let mut interp = interp.with_host(host);
226
227 let result = interp.run(procedure, args);
228 // The recalculation runs whether or not the procedure succeeded: a
229 // macro that wrote three cells and then raised has still written
230 // them, and leaving the workbook holding stale computed values would
231 // make the failure look like corruption.
232 interp.finish();
233 let mutated = interp.mutated();
234 let result = result.map_err(to_runtime_error)?;
235 Ok(to_outcome(result, mutated, &interp))
236 }
237
238 /// Runs startup macro events (`Workbook_Open` in `ThisWorkbook` then `Auto_Open` in standard modules).
239 pub fn run_open_events(&mut self) -> Result<RunOutcome, Error> {
240 let interp = if let Some(project) = &self.vba_project {
241 interp::Interpreter::from_project(project, None).map_err(to_runtime_error)?
242 } else {
243 return Err(Error::not_found(
244 ObjectKind::VbaModule,
245 "Workbook_Open or Auto_Open",
246 ));
247 };
248
249 let host = host::Host::new(self).map_err(to_runtime_error)?;
250 let mut interp = interp.with_host(host);
251
252 interp.run_open_events().map_err(to_runtime_error)?;
253 interp.finish();
254 let mutated = interp.mutated();
255 Ok(RunOutcome {
256 type_name: "Empty".to_string(),
257 value: Some(String::new()),
258 mutated,
259 })
260 }
261
262 /// The source text to run, resolving `module` the way
263 /// [`WorkbookManager::run_macro`] documents.
264 fn macro_source_for(&self, module: Option<&str>, procedure: &str) -> Result<String, Error> {
265 let project = self
266 .vba_project
267 .as_ref()
268 .ok_or_else(|| Error::not_found(ObjectKind::VbaModule, module.unwrap_or(procedure)))?;
269 let available = || project.modules.iter().map(|m| m.name.clone()).collect();
270 if let Some(name) = module {
271 return project
272 .find_module(name)
273 .map(|m| m.source.clone())
274 .ok_or_else(|| Error::not_found_among(ObjectKind::VbaModule, name, available()));
275 }
276 project
277 .modules
278 .iter()
279 // A module that does not parse is skipped rather than fatal: it
280 // cannot be the one declaring the procedure, and reporting its
281 // syntax error here would blame the wrong module entirely.
282 //
283 // Deliberately `parse_module` rather than `check_syntax`: the
284 // only question is which module *declares* this procedure, which
285 // is answered by parsing alone. Going through the name-resolution
286 // pass as well would let an unrelated unresolved name elsewhere
287 // in the module hide a procedure that is really there.
288 .find(|m| {
289 parser::parse_module(&m.source).is_ok_and(|module| {
290 module
291 .procedures()
292 .iter()
293 .any(|p| p.name.eq_ignore_ascii_case(procedure))
294 })
295 })
296 .map(|m| m.source.clone())
297 .ok_or_else(|| {
298 Error::not_found_among(
299 ObjectKind::VbaModule,
300 format!("a module declaring '{procedure}'"),
301 available(),
302 )
303 })
304 }
305}
306
307/// Parses `source` and runs one of its procedures.
308///
309/// Phase 1 of `docs/vba-macro-support.md`: expressions, control flow,
310/// `Sub`/`Function` calls and `On Error`. There is **no host object model**,
311/// so anything touching a workbook raises a run-time error naming what it
312/// was rather than silently doing nothing.
313///
314/// Execution is bounded -- a statement budget stops a runaway loop and a
315/// depth limit stops unbounded recursion -- because this runs source the
316/// caller did not necessarily write.
317///
318/// ```
319/// use visi_core::core::run_macro;
320/// let src = "Function Add2(a, b)\n Add2 = a + b\nEnd Function\n";
321/// let out = run_macro(src, "Add2", &["1", "2"]).unwrap();
322/// assert_eq!(out.type_name, "Integer");
323/// assert_eq!(out.value.as_deref(), Some("3"));
324/// ```
325pub fn run_macro(source: &str, procedure: &str, args: &[&str]) -> Result<RunOutcome, Error> {
326 let module = parser::parse_module(source).map_err(|e| Error::VbaSyntax {
327 message: e.message,
328 module: None,
329 line: e.pos.line,
330 column: e.pos.col,
331 })?;
332 let mut interp = interp::Interpreter::new(module);
333 let result = interp
334 .run(procedure, parse_args(args))
335 .map_err(to_runtime_error)?;
336
337 Ok(to_outcome(result, false, &interp))
338}
339
340impl VbaModule {
341 /// Checks this module's source, naming it in any error.
342 ///
343 /// The name matters more than it looks: a workbook can hold many modules
344 /// and `visi macro check` reports on all of them, so an error that does
345 /// not say which one it came from is close to useless.
346 ///
347 /// A `VbaModule` does not know its project, so unlike the free
348 /// [`check_syntax`] this **cannot** conclude anything from a name it
349 /// fails to resolve -- a sibling module may well declare it. Reach for
350 /// [`VbaProject::check_modules`] when the project is available; it is
351 /// strictly the better check.
352 pub fn check_syntax(&self) -> Result<ModuleSyntax, Error> {
353 let empty = std::collections::HashSet::new();
354 check_source(
355 &self.source,
356 Some(&self.name),
357 &resolve::Scope::partial(&empty),
358 )
359 }
360}
361
362/// What kind of VBA module a [`VbaModule`] is, which decides how it binds to
363/// the workbook.
364#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq)]
365pub enum VbaModuleKind {
366 /// A `.bas`-equivalent module with no host object binding.
367 Standard,
368 /// A `.cls`-equivalent module (not validated end-to-end against real
369 /// Excel yet -- see the feature plan's open-risk notes).
370 Class,
371 /// `ThisWorkbook` or a worksheet's code-behind module. Must correspond
372 /// 1:1 with an existing sheet (or the workbook itself) via
373 /// `bound_sheet_id`, mirroring Excel's own codeName wiring.
374 Document,
375}
376
377/// A single VBA module's editable content plus the opaque bytes needed to
378/// keep Excel happy on export.
379#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
380pub struct VbaModule {
381 /// VB_Name -- must satisfy `validate_vba_module_name`.
382 pub name: String,
383 /// What kind of module this is, and so how it binds to the workbook.
384 pub kind: VbaModuleKind,
385 /// Plain VBA source text (no compression, no Attribute-line management
386 /// beyond what the caller writes -- callers are expected to include the
387 /// `Attribute VB_Name = "..."` line themselves, matching how real
388 /// Excel-authored module streams are shaped).
389 pub source: String,
390 /// Required iff `kind == Document`: the sheet this module's code
391 /// belongs to (or `None`/ignored for `ThisWorkbook`, which isn't tied to
392 /// a specific sheet). Kept as a stable id (not a name) so sheet renames
393 /// don't silently orphan the binding -- deliberately NOT cascaded the
394 /// other direction (renaming this module does not rename the sheet, and
395 /// vice versa; Excel allows the two names to diverge).
396 pub bound_sheet_id: Option<u64>,
397 /// Opaque bytes forming the pre-TextOffset "p-code prefix" of this
398 /// module's stream. Never reparsed or validated by this codebase --
399 /// proven (via the POC) that its *content* doesn't need to correspond
400 /// to this module's actual source, only its presence matters, as long
401 /// as it's shaped the way real Excel's module loader expects (a
402 /// naively zero-filled placeholder of the same length is NOT enough).
403 /// For an imported module these are the real bytes read back from the
404 /// original file; for a module created in this codebase they're
405 /// `vba_synth::synthetic_module_prefix()`'s from-scratch, self-consistent
406 /// zero-procedure cache -- see that module's doc comment.
407 #[serde(default)]
408 pub prefix_bytes: Vec<u8>,
409 /// The module stream's MODULECOOKIE record (`0x002C`) value. MS-OVBA
410 /// documents this as implementation-specific and ignorable on read.
411 /// Preserved here so an imported module's original value survives re-export.
412 #[serde(default = "default_module_cookie")]
413 pub module_cookie: u16,
414 /// This module stream's already-compressed source, as read back
415 /// verbatim from an imported file -- `None` for a module created fresh
416 /// in this session (nothing to cache yet). `set_vba_module_source`
417 /// clears this whenever `source` is replaced. Export reuses the cached
418 /// bytes instead of recompressing `source` from scratch for every
419 /// module untouched by the CRUD operation that triggered the save.
420 #[serde(default)]
421 pub cached_compressed_source: Option<Vec<u8>>,
422}
423
424fn default_module_cookie() -> u16 {
425 0xFFFF
426}
427
428impl VbaModule {
429 /// Whether this is a document module -- `ThisWorkbook` or a worksheet's
430 /// code-behind -- as opposed to a standard or class module.
431 pub fn is_document(&self) -> bool {
432 self.kind == VbaModuleKind::Document
433 }
434}
435
436/// A workbook's VBA project: its modules plus the raw material needed to
437/// patch (not rebuild from scratch) a `vbaProject.bin` on export.
438#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
439pub struct VbaProject {
440 /// Project ID GUID, e.g. `"{7B4E3A2C-1F5D-4A6B-9C8E-2D3F4A5B6C7D}"`.
441 /// Must stay internally consistent with `protection_lines` -- never
442 /// mutated after import/creation, so it always is. If `CMG`/`DPB`/`GC`
443 /// protection-state lines are ever made independently settable, they
444 /// must correspond to this exact ID or Excel reports the whole project
445 /// "unviewable" (a real finding from the POC, not a hypothetical).
446 pub project_id: String,
447 /// The project's modules, in no particular order. Names are unique
448 /// case-insensitively.
449 pub modules: Vec<VbaModule>,
450 /// The full original `vbaProject.bin` bytes this project was imported
451 /// from, or (for a project created fresh in this session)
452 /// `vba_synth::synthetic_raw_donor()`'s from-scratch bytes -- export's
453 /// patch base. See `vba_xlsx.rs`.
454 #[serde(default)]
455 pub raw_donor: Vec<u8>,
456 /// P-code prefix bytes to donate to the first module ever added to a
457 /// project that started with none -- kept separate from `modules`
458 /// rather than as a phantom placeholder module, so it never shows up in
459 /// `list_vba_modules`/export. Once a project has at least one real
460 /// module, new modules instead borrow prefix bytes from an existing
461 /// one, and this field goes unused.
462 #[serde(default)]
463 pub seed_prefix_bytes: Vec<u8>,
464 /// `VbaModule::module_cookie` to donate to the first module ever added
465 /// to a project that started with none -- same donation scheme as
466 /// `seed_prefix_bytes`, see there for why.
467 #[serde(default = "default_module_cookie")]
468 pub seed_module_cookie: u16,
469 /// The donor's original `PROJECT` stream `CMG=`/`DPB=`/`GC=` lines
470 /// (joined with `\r\n`), reproduced verbatim on export -- `None` for a
471 /// project created fresh in this session, which never had any. See
472 /// `vba_xlsx::build_project_stream` for why these must be preserved
473 /// rather than dropped.
474 #[serde(default)]
475 pub protection_lines: Option<String>,
476}
477
478impl VbaProject {
479 /// A brand-new, empty VBA project with no real Excel-authored file
480 /// behind it anywhere -- `raw_donor` and `seed_prefix_bytes` are built
481 /// by `vba_synth` entirely from scratch. See `vba_synth`'s doc comment
482 /// for why that's now possible.
483 pub fn new_empty() -> Self {
484 VbaProject {
485 project_id: new_project_guid(),
486 modules: Vec::new(),
487 raw_donor: crate::core::vba_synth::synthetic_raw_donor(),
488 seed_prefix_bytes: crate::core::vba_synth::synthetic_module_prefix(),
489 seed_module_cookie: default_module_cookie(),
490 protection_lines: None,
491 }
492 }
493
494 /// Finds a module by name, matched case-insensitively as VBA does.
495 pub fn find_module(&self, name: &str) -> Option<&VbaModule> {
496 self.modules
497 .iter()
498 .find(|m| m.name.eq_ignore_ascii_case(name))
499 }
500
501 /// [`VbaProject::find_module`], mutably.
502 pub fn find_module_mut(&mut self, name: &str) -> Option<&mut VbaModule> {
503 self.modules
504 .iter_mut()
505 .find(|m| m.name.eq_ignore_ascii_case(name))
506 }
507
508 /// Whether a module of this name already exists, matched
509 /// case-insensitively.
510 pub fn module_name_taken(&self, name: &str) -> bool {
511 self.find_module(name).is_some()
512 }
513
514 /// Checks every module, resolving names against the **whole project**.
515 ///
516 /// This is the check to prefer wherever the project is in hand.
517 /// [`VbaModule::check_syntax`] sees one module and so has to accept any
518 /// name it cannot resolve, since a sibling may declare it; here the
519 /// siblings are known, so `x = arr(1)` with no `arr` anywhere is
520 /// reported the way Excel reports it -- Excel compiles a project, not a
521 /// file.
522 ///
523 /// Returns one entry per module, in `modules` order, pairing the
524 /// module's name with its result. A module whose *source* does not parse
525 /// still contributes whatever names it declares to the others, since a
526 /// parse failure in one module is not evidence about another.
527 pub fn check_modules(&self) -> Vec<(String, Result<ModuleSyntax, Error>)> {
528 self.check_modules_scoped(true)
529 }
530
531 /// [`check_modules`](Self::check_modules) for a project that is **not**
532 /// the whole story -- one whose procedures may live in a referenced
533 /// project this `VbaProject` does not model.
534 ///
535 /// Modules still resolve against each other; the only thing that
536 /// changes is that a name resolving nowhere is accepted rather than
537 /// reported, as in [`check_syntax_partial`]. Nothing in a workbook
538 /// records whether such a reference exists, so this is a caller's
539 /// assertion, not something to infer.
540 pub fn check_modules_partial(&self) -> Vec<(String, Result<ModuleSyntax, Error>)> {
541 self.check_modules_scoped(false)
542 }
543
544 /// The body both of the above share, `complete` being
545 /// [`resolve::Scope::complete_project`].
546 fn check_modules_scoped(&self, complete: bool) -> Vec<(String, Result<ModuleSyntax, Error>)> {
547 let mut declared: std::collections::HashSet<String> = std::collections::HashSet::new();
548 let parsed: Vec<_> = self
549 .modules
550 .iter()
551 .map(|m| (m, parser::parse_module(&m.source).ok()))
552 .collect();
553 for (_, module) in &parsed {
554 if let Some(module) = module {
555 declared.extend(resolve::declared_names(module));
556 }
557 }
558
559 parsed
560 .iter()
561 .map(|(m, _)| {
562 let scope = resolve::Scope {
563 external: &declared,
564 complete_project: complete,
565 };
566 (
567 m.name.clone(),
568 check_source(&m.source, Some(&m.name), &scope),
569 )
570 })
571 .collect()
572 }
573}
574
575/// A GUID-shaped project id (`{XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX}`) for
576/// a brand-new project, built from two `generate_unique_id()` draws rather
577/// than duplicating its getrandom/fallback logic.
578fn new_project_guid() -> String {
579 let hi = crate::core::engine::generate_unique_id();
580 let lo = crate::core::engine::generate_unique_id();
581 format!(
582 "{{{:08X}-{:04X}-{:04X}-{:04X}-{:012X}}}",
583 (hi >> 32) as u32,
584 (hi >> 16) as u16,
585 hi as u16,
586 (lo >> 48) as u16,
587 lo & 0xFFFF_FFFF_FFFF,
588 )
589}
590
591/// VBA identifiers: must start with a letter, contain only letters/digits/
592/// underscore, and be at most 31 characters (the real VBE module-name
593/// limit).
594pub fn validate_vba_module_name(name: &str) -> Result<(), String> {
595 let trimmed = name.trim();
596 if trimmed.is_empty() {
597 return Err("Module name cannot be empty".to_string());
598 }
599 if trimmed.chars().count() > 31 {
600 return Err(format!(
601 "Module name '{}' exceeds VBA's 31-character limit",
602 name
603 ));
604 }
605 let first = trimmed.chars().next().unwrap();
606 if !first.is_alphabetic() {
607 return Err(format!("Module name '{}' must start with a letter", name));
608 }
609 if !trimmed.chars().all(|c| c.is_alphanumeric() || c == '_') {
610 return Err(format!(
611 "Module name '{}' may only contain letters, digits, and underscores",
612 name
613 ));
614 }
615 Ok(())
616}
617
618#[cfg(test)]
619mod tests {
620 use super::*;
621
622 fn sample_project() -> VbaProject {
623 VbaProject {
624 project_id: "{00000000-0000-0000-0000-000000000000}".to_string(),
625 modules: vec![
626 VbaModule {
627 name: "ThisWorkbook".to_string(),
628 kind: VbaModuleKind::Document,
629 source: "Attribute VB_Name = \"ThisWorkbook\"\r\n".to_string(),
630 bound_sheet_id: None,
631 prefix_bytes: vec![0xAA; 16],
632 module_cookie: 0xFFFF,
633 cached_compressed_source: None,
634 },
635 VbaModule {
636 name: "Module1".to_string(),
637 kind: VbaModuleKind::Standard,
638 source: "Attribute VB_Name = \"Module1\"\r\nSub Foo()\r\nEnd Sub\r\n"
639 .to_string(),
640 bound_sheet_id: None,
641 prefix_bytes: vec![0xBB; 16],
642 module_cookie: 0xFFFF,
643 cached_compressed_source: None,
644 },
645 ],
646 raw_donor: Vec::new(),
647 seed_prefix_bytes: Vec::new(),
648 seed_module_cookie: 0xFFFF,
649 protection_lines: None,
650 }
651 }
652
653 #[test]
654 fn validate_name_rules() {
655 assert!(validate_vba_module_name("Module1").is_ok());
656 assert!(validate_vba_module_name("_Bad").is_err());
657 assert!(validate_vba_module_name("1Bad").is_err());
658 assert!(validate_vba_module_name("").is_err());
659 assert!(validate_vba_module_name("Has Space").is_err());
660 assert!(validate_vba_module_name("Has-Dash").is_err());
661 assert!(validate_vba_module_name(&"A".repeat(32)).is_err());
662 assert!(validate_vba_module_name(&"A".repeat(31)).is_ok());
663 }
664
665 #[test]
666 fn find_module_case_insensitive() {
667 let project = sample_project();
668 assert!(project.find_module("module1").is_some());
669 assert!(project.find_module("MODULE1").is_some());
670 assert!(project.find_module("Module2").is_none());
671 }
672
673 #[test]
674 fn module_name_taken_case_insensitive() {
675 let project = sample_project();
676 assert!(project.module_name_taken("module1"));
677 assert!(!project.module_name_taken("Module2"));
678 }
679
680 /// `sample_project()`'s shape with the sources the caller cares about,
681 /// one standard module per `(name, source)` pair.
682 fn project_of(sources: &[(&str, &str)]) -> VbaProject {
683 let mut project = sample_project();
684 project.modules = sources
685 .iter()
686 .map(|(name, source)| VbaModule {
687 name: (*name).to_string(),
688 kind: VbaModuleKind::Standard,
689 source: (*source).to_string(),
690 bound_sheet_id: None,
691 prefix_bytes: vec![0xBB; 16],
692 module_cookie: 0xFFFF,
693 cached_compressed_source: None,
694 })
695 .collect();
696 project
697 }
698
699 const CALLER: &str = "Public Sub Caller()\n DoWork 1\nEnd Sub\n";
700 const CALLEE: &str = "Public Sub DoWork(n As Long)\nEnd Sub\n";
701
702 /// The two scopes differ on exactly one thing, and only on it: a name
703 /// no supplied module declares.
704 #[test]
705 fn partial_scope_accepts_a_call_into_source_not_supplied() {
706 // A fragment on its own: reported by default, accepted as partial.
707 assert!(check_syntax(CALLER).is_err());
708 assert!(check_syntax_partial(CALLER).is_ok());
709
710 // Nothing else moves. A duplicate declaration is disproved by the
711 // module's own text, so the partial scope still reports it.
712 let dup = "Sub Test()\n Dim x As Long\n Dim x As Long\nEnd Sub\n";
713 assert!(check_syntax(dup).is_err());
714 assert!(check_syntax_partial(dup).is_err());
715 }
716
717 #[test]
718 fn check_modules_resolves_across_siblings() {
719 let project = project_of(&[("Module1", CALLER), ("Module2", CALLEE)]);
720 for (name, result) in project.check_modules() {
721 assert!(result.is_ok(), "{name} should be clean: {result:?}");
722 }
723
724 // Drop the sibling and the same call is a whole-project error.
725 let alone = project_of(&[("Module1", CALLER)]);
726 let results = alone.check_modules();
727 assert_eq!(results.len(), 1);
728 match &results[0].1 {
729 Err(Error::VbaSyntax {
730 message, module, ..
731 }) => {
732 assert!(message.contains("DoWork"), "{message}");
733 assert_eq!(module.as_deref(), Some("Module1"));
734 }
735 other => panic!("expected a syntax error, got {other:?}"),
736 }
737
738 // ...and clean again under `--partial`, where the missing declaration
739 // may be in a project this one merely references.
740 assert!(alone.check_modules_partial()[0].1.is_ok());
741 }
742
743 #[test]
744 fn set_source_leaves_prefix_bytes_untouched() {
745 let mut project = sample_project();
746 let original_prefix = project.find_module("Module1").unwrap().prefix_bytes.clone();
747 project.find_module_mut("Module1").unwrap().source =
748 "Attribute VB_Name = \"Module1\"\r\nSub Bar()\r\nEnd Sub\r\n".to_string();
749 assert_eq!(
750 project.find_module("Module1").unwrap().prefix_bytes,
751 original_prefix
752 );
753 }
754}