polyxml_cli/
config.rs

1use std::collections::{BTreeMap, HashMap, HashSet};
2use std::fs;
3use std::path::{Path, PathBuf};
4
5use glob::glob;
6use serde::{Deserialize, Serialize};
7use thiserror::Error;
8
9#[derive(Debug, Error)]
10pub enum ConfigError {
11    #[error("I/O error reading configuration: {0}")]
12    Io(#[from] std::io::Error),
13
14    #[error("TOML syntax error: {0}")]
15    Toml(#[from] toml::de::Error),
16
17    #[error("Invalid glob pattern '{pattern}': {error}")]
18    GlobPattern {
19        pattern: String,
20        error: glob::PatternError,
21    },
22
23    #[error("Failed to read glob path: {0}")]
24    Glob(#[from] glob::GlobError),
25
26    #[error("Module '{module}' depends on unknown module '{dependency}'")]
27    UnknownModuleDependency { module: String, dependency: String },
28
29    #[error("Cycle in workspace module dependencies involving '{0}'")]
30    ModuleCycle(String),
31
32    #[error("Invalid module name '{0}': use ASCII letters, digits, and underscores, starting with a letter or underscore")]
33    InvalidModuleName(String),
34}
35
36/// The top-level `polyxml.toml` workspace manifest.
37#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
38pub struct WorkspaceManifest {
39    pub workspace: Option<WorkspaceSection>,
40    #[serde(default)]
41    pub generate: Vec<TargetConfig>,
42    pub codegen: Option<HashMap<String, CodegenTargetConfig>>,
43    #[serde(default)]
44    pub modules: BTreeMap<String, ModuleConfig>,
45}
46
47/// A schema package compiled once and imported by dependent packages.
48#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
49#[serde(deny_unknown_fields)]
50pub struct ModuleConfig {
51    #[serde(default)]
52    pub schemas: Vec<String>,
53    #[serde(default)]
54    pub root_elements: Vec<String>,
55    #[serde(default)]
56    pub depends_on: Vec<String>,
57}
58
59#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
60pub struct WorkspaceSection {
61    pub name: Option<String>,
62    #[serde(default)]
63    pub schemas: Vec<String>,
64    #[serde(default)]
65    pub root_elements: Vec<String>,
66    pub include_dirs: Option<Vec<String>>,
67    pub output_base_dir: Option<String>,
68    pub custom_header: Option<String>,
69    pub go_module: Option<String>,
70}
71
72/// Target configuration from either `[[generate]]` or `[codegen.<target>]`.
73/// Unknown keys are rejected so misspelled or unsupported options fail
74/// instead of being silently ignored.
75#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
76#[serde(deny_unknown_fields)]
77pub struct TargetConfig {
78    pub target: String,
79    pub output: String,
80    pub enabled: Option<bool>,
81    pub backend: Option<String>,
82    #[serde(default)]
83    pub features: Vec<String>,
84    pub package: Option<String>,
85    pub namespace: Option<String>,
86    pub strict_facets: Option<bool>,
87    pub slots: Option<bool>,
88    pub kw_only: Option<bool>,
89    pub zero_copy: Option<bool>,
90    pub codecs: Option<bool>,
91    pub standard: Option<String>,
92    pub derive_traits: Option<Vec<String>>,
93    pub box_cycles: Option<bool>,
94    pub modules: Option<bool>,
95    pub mode: Option<String>,
96    pub serializer: Option<String>,
97    pub style: Option<String>,
98    pub custom_header: Option<String>,
99}
100
101#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
102#[serde(deny_unknown_fields)]
103pub struct CodegenTargetConfig {
104    pub enabled: Option<bool>,
105    pub output: Option<String>,
106    pub backend: Option<String>,
107    #[serde(default)]
108    pub features: Vec<String>,
109    pub package: Option<String>,
110    pub namespace: Option<String>,
111    pub strict_facets: Option<bool>,
112    pub slots: Option<bool>,
113    pub kw_only: Option<bool>,
114    pub zero_copy: Option<bool>,
115    pub codecs: Option<bool>,
116    pub standard: Option<String>,
117    pub derive_traits: Option<Vec<String>>,
118    pub box_cycles: Option<bool>,
119    pub modules: Option<bool>,
120    pub mode: Option<String>,
121    pub serializer: Option<String>,
122    pub style: Option<String>,
123    pub custom_header: Option<String>,
124}
125
126impl std::str::FromStr for WorkspaceManifest {
127    type Err = ConfigError;
128
129    fn from_str(toml_str: &str) -> Result<Self, Self::Err> {
130        let manifest: WorkspaceManifest = toml::from_str(toml_str)?;
131        Ok(manifest)
132    }
133}
134
135impl WorkspaceManifest {
136    pub fn from_file(path: impl AsRef<Path>) -> Result<Self, ConfigError> {
137        let content = fs::read_to_string(path)?;
138        content.parse()
139    }
140
141    /// Retrieve all configured target configurations, combining `[[generate]]`
142    /// and `[codegen.<target>]` definitions.
143    pub fn resolved_targets(&self) -> Vec<TargetConfig> {
144        let mut targets = Vec::new();
145
146        let ws_header = self
147            .workspace
148            .as_ref()
149            .and_then(|w| w.custom_header.clone());
150
151        // 1. Array of tables [[generate]]
152        for gen in &self.generate {
153            if gen.enabled.unwrap_or(true) {
154                let mut target = gen.clone();
155                if target.custom_header.is_none() {
156                    target.custom_header = ws_header.clone();
157                }
158                targets.push(target);
159            }
160        }
161
162        // 2. Table-based [codegen.<lang>]
163        if let Some(ref codegen_map) = self.codegen {
164            for (lang, cfg) in codegen_map {
165                if cfg.enabled.unwrap_or(true) {
166                    let output = cfg
167                        .output
168                        .clone()
169                        .unwrap_or_else(|| format!("generated/{}", lang));
170
171                    targets.push(TargetConfig {
172                        target: lang.clone(),
173                        output,
174                        enabled: cfg.enabled,
175                        backend: cfg.backend.clone(),
176                        features: cfg.features.clone(),
177                        package: cfg.package.clone(),
178                        namespace: cfg.namespace.clone(),
179                        strict_facets: cfg.strict_facets,
180                        slots: cfg.slots,
181                        kw_only: cfg.kw_only,
182                        zero_copy: cfg.zero_copy,
183                        codecs: cfg.codecs,
184                        standard: cfg.standard.clone(),
185                        derive_traits: cfg.derive_traits.clone(),
186                        box_cycles: cfg.box_cycles,
187                        modules: cfg.modules,
188                        mode: cfg.mode.clone(),
189                        serializer: cfg.serializer.clone(),
190                        style: cfg.style.clone(),
191                        custom_header: cfg.custom_header.clone().or_else(|| ws_header.clone()),
192                    });
193                }
194            }
195        }
196
197        targets
198    }
199
200    /// Expand all schema glob patterns in `workspace.schemas` relative to base directory.
201    pub fn expand_schemas(&self, base_dir: &Path) -> Result<Vec<PathBuf>, ConfigError> {
202        let Some(ref ws) = self.workspace else {
203            return Ok(Vec::new());
204        };
205
206        expand_patterns(&ws.schemas, base_dir)
207    }
208
209    pub fn expand_module_schemas(
210        &self,
211        name: &str,
212        base_dir: &Path,
213    ) -> Result<Vec<PathBuf>, ConfigError> {
214        let Some(module) = self.modules.get(name) else {
215            return Ok(Vec::new());
216        };
217        expand_patterns(&module.schemas, base_dir)
218    }
219
220    pub fn module_order(&self) -> Result<Vec<String>, ConfigError> {
221        for name in self.modules.keys() {
222            let mut chars = name.chars();
223            if !chars
224                .next()
225                .is_some_and(|ch| ch.is_ascii_alphabetic() || ch == '_')
226                || !chars.all(|ch| ch.is_ascii_alphanumeric() || ch == '_')
227            {
228                return Err(ConfigError::InvalidModuleName(name.clone()));
229            }
230        }
231        let mut order = Vec::new();
232        let mut visiting = HashSet::new();
233        let mut visited = HashSet::new();
234        fn visit(
235            name: &str,
236            manifest: &WorkspaceManifest,
237            visiting: &mut HashSet<String>,
238            visited: &mut HashSet<String>,
239            order: &mut Vec<String>,
240        ) -> Result<(), ConfigError> {
241            if visited.contains(name) {
242                return Ok(());
243            }
244            if !visiting.insert(name.to_string()) {
245                return Err(ConfigError::ModuleCycle(name.to_string()));
246            }
247            let module = &manifest.modules[name];
248            for dependency in &module.depends_on {
249                if !manifest.modules.contains_key(dependency) {
250                    return Err(ConfigError::UnknownModuleDependency {
251                        module: name.to_string(),
252                        dependency: dependency.clone(),
253                    });
254                }
255                visit(dependency, manifest, visiting, visited, order)?;
256            }
257            visiting.remove(name);
258            visited.insert(name.to_string());
259            order.push(name.to_string());
260            Ok(())
261        }
262        for name in self.modules.keys() {
263            visit(name, self, &mut visiting, &mut visited, &mut order)?;
264        }
265        Ok(order)
266    }
267}
268
269fn expand_patterns(patterns: &[String], base_dir: &Path) -> Result<Vec<PathBuf>, ConfigError> {
270    let mut paths = Vec::new();
271    for pattern in patterns {
272        let full_pattern = if Path::new(pattern).is_absolute() {
273            pattern.clone()
274        } else {
275            base_dir.join(pattern).to_string_lossy().to_string()
276        };
277
278        let entries = glob(&full_pattern).map_err(|e| ConfigError::GlobPattern {
279            pattern: full_pattern.clone(),
280            error: e,
281        })?;
282
283        for entry in entries {
284            let path = entry?;
285            if path.is_file() {
286                paths.push(path);
287            }
288        }
289    }
290
291    paths.sort();
292    paths.dedup();
293    Ok(paths)
294}
polyxml_cli/config.rs

polyxml_cli/
config.rs
