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

polyxml_cli/
config.rs
