use super::convert;
use super::key;
use super::map::ImportMap;
use super::owner;
use super::sources::{self, Unit};
use crate::authoring::Plan;
use crate::corpus::Corpus;
use crate::files::hash;
use crate::markdown::{body, frontmatter};
use crate::model::doctype::{Form, KeyKind};
use crate::model::{DocId, Tier, stamp};
use crate::placement::paths;
use anyhow::{Result, anyhow};
use regex::Regex;
use serde::Serialize;
use serde_json::{Map, Value, json};
use std::collections::{BTreeMap, BTreeSet};
#[derive(Debug, Clone, Serialize)]
pub struct Row {
pub source: String,
pub kind: String,
pub id: String,
pub path: String,
pub owner_reason: String,
pub status_from: String,
pub status_to: String,
pub aliases: Vec<String>,
pub unresolved: Vec<String>,
}
#[derive(Debug, Default, Serialize)]
pub struct ImportPlan {
pub rows: Vec<Row>,
#[serde(skip)]
pub plan: Plan,
pub alias_rows: Vec<(String, String)>,
pub uploads: Vec<(String, String)>,
}
fn name_of(p: &str) -> &str {
p.rsplit('/').next().unwrap_or(p)
}
pub fn build(c: &Corpus, map: &ImportMap) -> Result<ImportPlan> {
let units = sources::expand(&c.cfg.root, map);
let mut out = ImportPlan::default();
let mut taken: BTreeSet<String> = c.docs.iter().map(|d| d.id.to_string()).collect();
let mut by_source: BTreeMap<String, String> = BTreeMap::new();
let mut redirects = vec![];
for u in &units {
let rule = &map.sources[u.rule];
if let Some(re) = &rule.redirect {
redirects.push((u.clone(), Regex::new(re).map_err(|e| anyhow!("bad redirect regex: {e}"))?));
continue;
}
let row = doc_row(c, map, u, &mut taken, &mut out)?;
by_source.insert(u.main.clone(), row.id.clone());
by_source.insert(u.source.clone(), row.id.clone());
out.rows.push(row);
}
for (u, re) in redirects {
let text = std::fs::read_to_string(c.cfg.root.join(&u.main)).unwrap_or_default();
let target = re
.captures(&text)
.and_then(|cap| cap.name("target"))
.map(|m| m.as_str().trim().trim_start_matches("./").to_string());
let resolved = target.as_ref().and_then(|t| {
by_source.get(t).cloned().or_else(|| c.docs.iter().find(|d| &d.path == t).map(|d| d.id.to_string()))
});
let mut unresolved = vec![];
match &resolved {
Some(id) => out.alias_rows.push((u.source.clone(), id.clone())),
None => unresolved.push(format!("stub target {} is not imported", target.unwrap_or_else(|| "?".into()))),
}
out.plan.remove(u.source.clone());
out.rows.push(Row {
source: u.source.clone(),
kind: "redirect".into(),
id: resolved.unwrap_or_default(),
path: String::new(),
owner_reason: String::new(),
status_from: String::new(),
status_to: String::new(),
aliases: vec![],
unresolved,
});
}
Ok(out)
}
fn doc_row(c: &Corpus, map: &ImportMap, u: &Unit, taken: &mut BTreeSet<String>, out: &mut ImportPlan) -> Result<Row> {
let rule = &map.sources[u.rule];
let ty_name = rule.ty.clone().ok_or_else(|| anyhow!("source {} needs a type", rule.glob))?;
let ty = c.cfg.registry.get(&ty_name).ok_or_else(|| anyhow!("unknown type {ty_name}"))?;
let text = std::fs::read_to_string(c.cfg.root.join(&u.main))?;
let split = frontmatter::split(&text);
let mut unresolved = vec![];
let old = match frontmatter::parse(split.raw) {
frontmatter::Parsed::Map(m) => m,
frontmatter::Parsed::None => Map::new(),
frontmatter::Parsed::Invalid(e) => {
unresolved.push(format!("old frontmatter does not parse: {e}"));
Map::new()
}
};
let scanned = body::scan(split.body, split.body_line, &body::path_cite_re(c.cfg.spec_dir()));
let name = name_of(&u.source).trim_end_matches(".md");
let name_re = rule.name.as_deref().map(Regex::new).transpose().map_err(|e| anyhow!("bad name regex: {e}"))?;
let caps = name_re.as_ref().and_then(|r| r.captures(name_of(&u.source)));
let fallback_title = scanned.first_heading.clone().unwrap_or_else(|| name.replace(['-', '_'], " "));
let mut conv = convert::frontmatter(map, rule, &old, &ty.statuses, &fallback_title);
unresolved.append(&mut conv.unresolved);
let title = conv.fm.get("title").and_then(Value::as_str).unwrap_or(&fallback_title).to_string();
let created = conv.fm.get("created").and_then(Value::as_str).map(str::to_string);
let src =
key::Source { path: &u.main, caps: caps.as_ref(), created: created.as_deref(), title: &title, name_stem: name };
let (mut k, slug) = key::choose(&c.cfg.root, rule, ty, &src);
let governs: Vec<String> = conv
.fm
.get("rel")
.and_then(|r| r.get("governs"))
.map(|v| crate::model::doc::list_of(Some(v)))
.unwrap_or_default();
let (owner, owner_reason) = match rule.owner.as_str() {
"auto" => owner::choose(&c.owners, &governs, &text),
fixed => {
let o = c.owners.by_id(fixed).ok_or_else(|| anyhow!("source {}: no owner '{fixed}'", rule.glob))?;
(o, format!("fixed to {fixed} by the import map"))
}
};
loop {
let id = DocId::new(&ty.name, &owner.id, &k).to_string();
if !taken.contains(&id) {
break;
}
k = match ty.key {
KeyKind::Stamp => stamp::next_stamp(&k).unwrap_or(k),
KeyKind::Slug => format!("{k}-2"),
};
}
let slug = if ty.key == KeyKind::Slug { k.clone() } else { slug };
let id = DocId::new(&ty.name, &owner.id, &k);
taken.insert(id.to_string());
let bundle = !u.chapters.is_empty() || !u.attachments.is_empty() || ty.form == Form::Dir;
let dir = paths::type_dir(&c.cfg, &c.owners, owner, Tier::Committed, ty);
let main = paths::main_file(&dir, ty, &paths::stem(ty, &k, &slug), bundle);
let bundle_dir = main.rsplit_once('/').map(|x| x.0.to_string()).unwrap_or_default();
let base = if u.source == u.main { u.main.trim_end_matches(".md").to_string() } else { u.source.clone() };
let mut files = vec![];
for ch in &u.chapters {
let rel = ch.strip_prefix(&format!("{base}/")).unwrap_or(name_of(ch));
out.plan.copy(ch.clone(), format!("{bundle_dir}/{}", rel.replace('/', "-")));
}
for a in &u.attachments {
let rel = a.strip_prefix(&format!("{base}/")).unwrap_or(name_of(a)).to_string();
let (sha, bytes) = hash::file(&c.cfg.root.join(a))?;
let store = if bytes > c.cfg.raw.files.inline_max_bytes { "store" } else { "repo" };
if store == "store" {
out.uploads.push((format!("{bundle_dir}/{rel}"), sha.clone()));
}
out.plan.copy(a.clone(), format!("{bundle_dir}/{rel}"));
files.push(json!({"path": rel, "sha256": sha, "bytes": bytes, "store": store}));
}
let mut fm = conv.fm;
fm.insert("id".into(), Value::String(id.to_string()));
fm.entry("created").or_insert_with(|| Value::String(stamp::normalize_date(&k).unwrap_or_else(stamp::today)));
let created_v = fm["created"].clone();
fm.entry("updated").or_insert(created_v);
let mut aliases = conv.aliases;
if let Some(a) = caps.as_ref().and_then(|c| c.name("alias")) {
aliases.push(a.as_str().to_string());
}
if !aliases.is_empty() {
fm.insert("aliases".into(), json!(aliases));
}
if !files.is_empty() {
fm.insert("files".into(), Value::Array(files));
}
let content = format!("{}{}", frontmatter::render(&crate::markdown::frontmatter::edit::ordered(fm)), split.body);
out.plan.write(main.clone(), content, true);
out.plan.remove(u.source.clone());
for old in std::iter::once(u.source.clone())
.chain((u.main != u.source).then(|| u.main.clone()))
.chain(aliases.iter().cloned())
{
out.alias_rows.push((old, id.to_string()));
}
Ok(Row {
source: u.source.clone(),
kind: "doc".into(),
id: id.to_string(),
path: main,
owner_reason,
status_from: conv.status_from,
status_to: fm_status(&out.plan, &id.to_string()),
aliases,
unresolved,
})
}
fn fm_status(plan: &Plan, _id: &str) -> String {
plan.changes
.iter()
.rev()
.find_map(|c| match c {
crate::authoring::Change::Write { content, .. } => {
content.lines().find_map(|l| l.strip_prefix("status: ").map(str::to_string))
}
_ => None,
})
.unwrap_or_default()
}
pub fn alias_table(existing: &str, rows: &[(String, String)]) -> String {
let mut seen: BTreeSet<String> = existing.lines().map(str::to_string).collect();
let mut out = if existing.is_empty() {
"# old\tnew-id (written by spec import; frozen afterwards)\n".to_string()
} else {
existing.to_string()
};
for (old, new) in rows {
let line = format!("{old}\t{new}");
if seen.insert(line.clone()) {
out.push_str(&line);
out.push('\n');
}
}
out
}