1use super::convert;
4use super::key;
5use super::map::ImportMap;
6use super::owner;
7use super::sources::{self, Unit};
8use crate::authoring::Plan;
9use crate::corpus::Corpus;
10use crate::files::hash;
11use crate::markdown::{body, frontmatter};
12use crate::model::doctype::{Form, KeyKind};
13use crate::model::{DocId, Tier, stamp};
14use crate::placement::paths;
15use anyhow::{Result, anyhow};
16use regex::Regex;
17use serde::Serialize;
18use serde_json::{Map, Value, json};
19use std::collections::{BTreeMap, BTreeSet};
20
21#[derive(Debug, Clone, Default, Serialize)]
22pub struct Row {
23 pub source: String,
24 pub kind: String,
26 pub id: String,
27 pub path: String,
28 pub owner_reason: String,
29 pub status_from: String,
30 pub status_to: String,
31 pub aliases: Vec<String>,
32 pub unresolved: Vec<String>,
33}
34
35#[derive(Debug, Default, Serialize)]
36pub struct ImportPlan {
37 pub rows: Vec<Row>,
38 #[serde(skip)]
39 pub plan: Plan,
40 pub alias_rows: Vec<(String, String)>,
41 pub uploads: Vec<(String, String)>,
43}
44
45fn name_of(p: &str) -> &str {
46 p.rsplit('/').next().unwrap_or(p)
47}
48
49fn stem_of(p: &str) -> &str {
51 name_of(p).trim_end_matches(".md")
52}
53
54pub fn build(c: &Corpus, map: &ImportMap) -> Result<ImportPlan> {
55 let units = sources::expand(&c.cfg.root, map);
56 let mut out = ImportPlan::default();
57 let mut taken: BTreeSet<String> = c.docs.iter().map(|d| d.id.to_string()).collect();
58 let mut by_source: BTreeMap<String, String> = BTreeMap::new();
59 let mut redirects = vec![];
60 for u in &units {
61 let rule = &map.sources[u.rule];
62 if let Some(re) = &rule.redirect {
63 redirects.push((u.clone(), Regex::new(re).map_err(|e| anyhow!("bad redirect regex: {e}"))?));
64 continue;
65 }
66 if rule.drop {
67 out.plan.remove(u.source.clone());
68 out.rows.push(Row { source: u.source.clone(), kind: "drop".into(), ..Row::default() });
69 continue;
70 }
71 let row = doc_row(c, map, u, &mut taken, &mut out)?;
72 for p in [&u.main, &u.source].into_iter().chain(&u.chapters).chain(&u.attachments) {
73 by_source.insert(p.clone(), row.id.clone());
74 }
75 if map.stem_aliases {
76 out.alias_rows.push((stem_of(&u.source).to_string(), row.id.clone()));
77 }
78 out.rows.push(row);
79 }
80 for (u, re) in redirects {
81 let text = std::fs::read_to_string(c.cfg.root.join(&u.main)).unwrap_or_default();
82 let target = re
83 .captures(&text)
84 .and_then(|cap| cap.name("target"))
85 .map(|m| m.as_str().trim().trim_start_matches("./").to_string());
86 let resolved = target.as_ref().and_then(|t| {
87 by_source.get(t).cloned().or_else(|| c.docs.iter().find(|d| &d.path == t).map(|d| d.id.to_string()))
88 });
89 let mut unresolved = vec![];
90 match &resolved {
91 Some(id) => {
92 out.alias_rows.push((u.source.clone(), id.clone()));
93 if map.stem_aliases {
94 out.alias_rows.push((stem_of(&u.source).to_string(), id.clone()));
95 }
96 }
97 None => unresolved.push(format!("stub target {} is not imported", target.unwrap_or_else(|| "?".into()))),
98 }
99 out.plan.remove(u.source.clone());
100 out.rows.push(Row {
101 source: u.source.clone(),
102 kind: "redirect".into(),
103 id: resolved.unwrap_or_default(),
104 path: String::new(),
105 owner_reason: String::new(),
106 status_from: String::new(),
107 status_to: String::new(),
108 aliases: vec![],
109 unresolved,
110 });
111 }
112 Ok(out)
113}
114
115fn doc_row(c: &Corpus, map: &ImportMap, u: &Unit, taken: &mut BTreeSet<String>, out: &mut ImportPlan) -> Result<Row> {
116 let rule = &map.sources[u.rule];
117 let ty_name = rule.ty.clone().ok_or_else(|| anyhow!("source {} needs a type", rule.glob))?;
118 let ty = c.cfg.registry.get(&ty_name).ok_or_else(|| anyhow!("unknown type {ty_name}"))?;
119 let generated = u.main.is_empty();
121 let text = if generated { String::new() } else { std::fs::read_to_string(c.cfg.root.join(&u.main))? };
122 let split = frontmatter::split(&text);
123 let mut unresolved = vec![];
124 let old = match frontmatter::parse(split.raw) {
125 frontmatter::Parsed::Map(m) => m,
126 frontmatter::Parsed::None => Map::new(),
127 frontmatter::Parsed::Invalid(e) => {
128 unresolved.push(format!("old frontmatter does not parse: {e}"));
129 Map::new()
130 }
131 };
132 let scanned = body::scan(split.body, split.body_line, &body::path_cite_re(c.cfg.spec_dir()));
133 let name = name_of(&u.source).trim_end_matches(".md");
134 let name_re = rule.name.as_deref().map(Regex::new).transpose().map_err(|e| anyhow!("bad name regex: {e}"))?;
135 let caps = name_re.as_ref().and_then(|r| r.captures(name_of(&u.source)));
136 let named = caps.as_ref().and_then(|c| c.name("slug")).map(|m| m.as_str()).unwrap_or(name);
137 let fallback_title = scanned.first_heading.clone().unwrap_or_else(|| named.replace(['-', '_'], " "));
138 let mut conv = convert::frontmatter(map, rule, &old, &ty.statuses, &fallback_title);
139 unresolved.append(&mut conv.unresolved);
140 let title = conv.fm.get("title").and_then(Value::as_str).unwrap_or(&fallback_title).to_string();
141 let created = conv.fm.get("created").and_then(Value::as_str).map(str::to_string);
142 let src =
143 key::Source { path: &u.main, caps: caps.as_ref(), created: created.as_deref(), title: &title, name_stem: name };
144 let (mut k, slug) = key::choose(&c.cfg.root, rule, ty, &src);
145 let governs: Vec<String> = conv
146 .fm
147 .get("rel")
148 .and_then(|r| r.get("governs"))
149 .map(|v| crate::model::doc::list_of(Some(v)))
150 .unwrap_or_default();
151 let (owner, owner_reason) = match rule.owner.as_str() {
152 "auto" => {
153 let home = c.owners.storage_owner(c.owners.containing(&u.source));
154 owner::choose(&c.owners, &governs, &text, home)
155 }
156 fixed => {
157 let o = c.owners.by_id(fixed).ok_or_else(|| anyhow!("source {}: no owner '{fixed}'", rule.glob))?;
158 (o, format!("fixed to {fixed} by the import map"))
159 }
160 };
161 loop {
162 let id = DocId::new(&ty.name, &owner.id, &k).to_string();
163 if !taken.contains(&id) {
164 break;
165 }
166 k = match ty.key {
167 KeyKind::Stamp => stamp::next_stamp(&k).unwrap_or(k),
168 KeyKind::Slug => format!("{k}-2"),
169 };
170 }
171 let slug = if ty.key == KeyKind::Slug { k.clone() } else { slug };
172 let id = DocId::new(&ty.name, &owner.id, &k);
173 taken.insert(id.to_string());
174 let bundle = !u.chapters.is_empty() || !u.attachments.is_empty() || ty.form == Form::Dir;
175 let dir = paths::type_dir(&c.cfg, &c.owners, owner, Tier::Committed, ty);
176 let main = paths::main_file(&dir, ty, &paths::stem(ty, &k, &slug), bundle);
177 let bundle_dir = main.rsplit_once('/').map(|x| x.0.to_string()).unwrap_or_default();
178 let base = if u.source == u.main { u.main.trim_end_matches(".md").to_string() } else { u.source.clone() };
179 let mut files = vec![];
180 for ch in &u.chapters {
183 let rel = ch.strip_prefix(&format!("{base}/")).unwrap_or(name_of(ch));
184 out.plan.copy(ch.clone(), format!("{bundle_dir}/{}", rel.replace('/', "-")));
185 out.plan.remove(ch.clone());
186 }
187 for a in &u.attachments {
188 let rel = a.strip_prefix(&format!("{base}/")).unwrap_or(name_of(a)).to_string();
189 let (sha, bytes) = hash::file(&c.cfg.root.join(a))?;
190 let store = if bytes > c.cfg.raw.files.inline_max_bytes { "store" } else { "repo" };
191 if store == "store" {
192 out.uploads.push((format!("{bundle_dir}/{rel}"), sha.clone()));
193 }
194 out.plan.copy(a.clone(), format!("{bundle_dir}/{rel}"));
195 out.plan.remove(a.clone());
196 files.push(json!({"path": rel, "sha256": sha, "bytes": bytes, "store": store}));
197 }
198 let mut fm = conv.fm;
199 fm.insert("id".into(), Value::String(id.to_string()));
200 fm.entry("created").or_insert_with(|| Value::String(stamp::normalize_date(&k).unwrap_or_else(stamp::today)));
201 let created_v = fm["created"].clone();
202 let updated = fm.entry("updated").or_insert(created_v.clone());
203 if updated.as_str() < created_v.as_str() {
205 *updated = created_v;
206 }
207 let mut aliases = conv.aliases;
208 if let Some(a) = caps.as_ref().and_then(|c| c.name("alias")) {
209 aliases.push(a.as_str().to_string());
210 }
211 if !aliases.is_empty() {
212 fm.insert("aliases".into(), json!(aliases));
213 }
214 if !files.is_empty() {
215 fm.insert("files".into(), Value::Array(files));
216 }
217 let body = if generated { format!("\n# {title}\n") } else { split.body.to_string() };
218 let content = format!("{}{}", frontmatter::render(&crate::markdown::frontmatter::edit::ordered(fm)), body);
219 out.plan.write(main.clone(), content, true);
220 out.plan.remove(u.source.clone());
221 for old in std::iter::once(u.source.clone())
223 .chain((u.main != u.source).then(|| u.main.clone()))
224 .chain(u.chapters.iter().cloned())
225 .chain(u.attachments.iter().cloned())
226 .chain(aliases.iter().cloned())
227 {
228 out.alias_rows.push((old, id.to_string()));
229 }
230 Ok(Row {
231 source: u.source.clone(),
232 kind: "doc".into(),
233 id: id.to_string(),
234 path: main,
235 owner_reason,
236 status_from: conv.status_from,
237 status_to: fm_status(&out.plan, &id.to_string()),
238 aliases,
239 unresolved,
240 })
241}
242
243fn fm_status(plan: &Plan, _id: &str) -> String {
244 plan.changes
245 .iter()
246 .rev()
247 .find_map(|c| match c {
248 crate::authoring::Change::Write { content, .. } => {
249 content.lines().find_map(|l| l.strip_prefix("status: ").map(str::to_string))
250 }
251 _ => None,
252 })
253 .unwrap_or_default()
254}
255
256pub fn alias_table(existing: &str, rows: &[(String, String)]) -> String {
258 let mut seen: BTreeSet<String> = existing.lines().map(str::to_string).collect();
259 let mut out = if existing.is_empty() {
260 "# old\tnew-id (written by spec import; frozen afterwards)\n".to_string()
261 } else {
262 existing.to_string()
263 };
264 for (old, new) in rows {
265 let line = format!("{old}\t{new}");
266 if seen.insert(line.clone()) {
267 out.push_str(&line);
268 out.push('\n');
269 }
270 }
271 out
272}