mod.rs

crates/cortex-core/src/export/mod.rs

1. //! A conversation as a Markdown note with YAML frontmatter, for an Obsidian
2. //! vault: title, dates, models, tokens and cost up top, then the exchange.
4. use chrono::TimeZone;
6. use crate::cost::to_usd;
7. use crate::storage::{Conversation, Message};
8. use crate::types::UsageSource;
10. #[derive(Debug, Clone, Copy, PartialEq, Eq)]
11. pub enum ImageLinks {
12. /// `![[file.png]]`, Obsidian's embed.
13. Wiki,
14. /// `![file.png](attachments/file.png)`, plain Markdown.
15. Markdown,
16. }
18. pub struct ExportFile {
19. /// Where to copy it, relative to the note's folder.
20. pub relative_path: String,
21. pub source_path: String,
22. }
24. pub struct Export {
25. pub filename: String,
26. pub markdown: String,
27. pub files: Vec<ExportFile>,
28. }
30. /// A YAML double-quoted scalar, escaped.
31. pub fn yaml_str(s: &str) -> String {
32. let mut out = String::with_capacity(s.len() + 2);
33. out.push('"');
34. for c in s.chars() {
35. match c {
36. '"' => out.push_str("\\\""),
37. '\\' => out.push_str("\\\\"),
38. '\n' => out.push_str("\\n"),
39. '\r' => out.push_str("\\r"),
40. '\t' => out.push_str("\\t"),
41. c if (c as u32) < 0x20 => out.push_str(&format!("\\x{:02x}", c as u32)),
42. c => out.push(c),
43. }
44. }
45. out.push('"');
46. out
47. }
49. /// A filename Obsidian and every filesystem accept.
50. pub fn safe_filename(title: &str) -> String {
51. let cleaned: String =
52. title.chars().map(|c| if "[]#^|\\/:*?\"<>".contains(c) || c.is_control() { ' ' } else { c }).collect();
53. let squashed = cleaned.split_whitespace().collect::<Vec<_>>().join(" ");
54. let trimmed = squashed.trim_matches(|c: char| c == '.' || c == ' ');
55. let mut name: String = trimmed.chars().take(80).collect();
56. if name.is_empty() {
57. name = "Cortex chat".into();
58. }
59. name
60. }
62. fn iso<Tz: TimeZone>(ms: i64, tz: &Tz) -> String
63. where
64. Tz::Offset: std::fmt::Display,
65. {
66. tz.timestamp_millis_opt(ms).single().map(|d| d.format("%Y-%m-%dT%H:%M:%S%:z").to_string()).unwrap_or_default()
67. }
69. pub fn to_markdown<Tz: TimeZone>(
70. conv: &Conversation,
71. messages: &[Message],
72. attachment_paths: &dyn Fn(&str) -> Option<String>,
73. links: ImageLinks,
74. tz: &Tz,
75. ) -> Export
76. where
77. Tz::Offset: std::fmt::Display,
78. {
79. let mut unique: Vec<&str> = Vec::new();
80. for m in messages.iter().filter(|m| m.role == "assistant").filter_map(|m| m.model_id.as_deref()) {
81. if !unique.contains(&m) {
82. unique.push(m);
83. }
84. }
85. let assistant = || messages.iter().filter(|m| m.role == "assistant");
86. let tokens_in: i64 = assistant().map(|m| m.usage.prompt_tokens).sum();
87. let tokens_out: i64 = assistant().map(|m| m.usage.completion_tokens).sum();
88. let cost: i64 = assistant().map(|m| m.cost_nanousd).sum();
89. let estimated = assistant().any(|m| m.usage_source == UsageSource::Estimated);
91. let mut fm = vec!["---".to_string(), format!("title: {}", yaml_str(&conv.title))];
92. fm.push(format!("created: {}", iso(conv.created_at, tz)));
93. fm.push(format!("updated: {}", iso(conv.updated_at, tz)));
94. fm.push("models:".into());
95. for m in &unique {
96. fm.push(format!(" - {}", yaml_str(m)));
97. }
98. if unique.is_empty() {
99. fm.pop();
100. fm.push("models: []".into());
101. }
102. fm.push(format!("tokens_in: {tokens_in}"));
103. fm.push(format!("tokens_out: {tokens_out}"));
104. fm.push(format!("cost_usd: {:.6}", to_usd(cost)));
105. fm.push(format!("cost_estimated: {estimated}"));
106. fm.push(format!("cortex_id: {}", yaml_str(&conv.id)));
107. fm.push("tags:".into());
108. fm.push(" - cortex".into());
109. fm.push("---".into());
111. let mut body = vec![String::new(), format!("# {}", conv.title), String::new()];
112. if !conv.system_prompt.trim().is_empty() {
113. body.push("> [!note] System prompt".into());
114. for line in conv.system_prompt.trim().lines() {
115. body.push(format!("> {line}"));
116. }
117. body.push(String::new());
118. }
119. let mut files = Vec::new();
120. let mut used_names: Vec<String> = Vec::new();
121. for m in messages.iter().filter(|m| m.role != "system") {
122. let who =
123. if m.role == "user" { "You".to_string() } else { m.model_id.clone().unwrap_or_else(|| "Assistant".into()) };
124. body.push(format!("## {who}"));
125. body.push(String::new());
126. if !m.content.trim().is_empty() {
127. body.push(m.content.trim_end().to_string());
128. body.push(String::new());
129. }
130. for a in &m.attachments {
131. match a.kind.as_str() {
132. "image" => {
133. let Some(src) = attachment_paths(&a.id) else { continue };
134. let mut name = safe_filename(&a.filename);
135. let mut n = 2;
136. while used_names.contains(&name) {
137. name = format!("{n}-{}", safe_filename(&a.filename));
138. n += 1;
139. }
140. used_names.push(name.clone());
141. body.push(match links {
142. ImageLinks::Wiki => format!("![[{name}]]"),
143. ImageLinks::Markdown => format!("![{name}](attachments/{})", name.replace(' ', "%20")),
144. });
145. body.push(String::new());
146. files.push(ExportFile { relative_path: format!("attachments/{name}"), source_path: src });
147. }
148. _ => {
149. body.push(format!("*Attached: {}*", a.filename));
150. body.push(String::new());
151. }
152. }
153. }
154. match m.status.as_str() {
155. "cancelled" => body.push("*Stopped before it finished.*\n".into()),
156. "error" => body.push(format!("*Error: {}*\n", m.error_message.clone().unwrap_or_default())),
157. _ => {}
158. }
159. }
160. Export {
161. filename: format!("{}.md", safe_filename(&conv.title)),
162. markdown: format!("{}\n{}\n", fm.join("\n"), body.join("\n").trim_end()),
163. files,
164. }
165. }
167. #[cfg(test)]
168. mod tests {
169. use super::*;
170. use crate::storage::{AttachmentMeta, Conversation, Message};
171. use crate::types::{Usage, UsageSource};
173. fn msg(role: &str, content: &str, model: Option<&str>, cost: i64, src: UsageSource) -> Message {
174. Message {
175. id: format!("m-{role}-{cost}"),
176. conversation_id: "c1".into(),
177. role: role.into(),
178. content: content.into(),
179. reasoning: None,
180. status: "complete".into(),
181. error_kind: None,
182. error_message: None,
183. provider_id: None,
184. model_id: model.map(str::to_string),
185. usage: if role == "assistant" {
186. Usage { prompt_tokens: 120, completion_tokens: 80, cached_tokens: 0 }
187. } else {
188. Usage::default()
189. },
190. usage_source: src,
191. prices: None,
192. cost_nanousd: cost,
193. finish_reason: None,
194. ttft_ms: None,
195. duration_ms: None,
196. created_at: 0,
197. attachments: vec![],
198. }
199. }
201. #[test]
202. fn a_whole_note() {
203. let conv = Conversation {
204. id: "0199-abc".into(),
205. title: "Compare two \"pricing\" tiers: a/b".into(),
206. provider_id: "deepinfra".into(),
207. model_id: "m".into(),
208. preset_id: None,
209. system_prompt: "Be concise.\nUse tables.".into(),
210. origin: "main".into(),
211. pinned: false,
212. archived: false,
213. created_at: 1_790_000_000_000,
214. updated_at: 1_790_000_600_000,
215. };
216. let mut user = msg("user", "Which tier is cheaper per seat?", None, 0, UsageSource::None);
217. user.attachments.push(AttachmentMeta {
218. id: "a1".into(),
219. kind: "image".into(),
220. filename: "chart: q3.png".into(),
221. mime: "image/png".into(),
222. byte_size: 10,
223. width: Some(800),
224. height: Some(600),
225. });
226. user.attachments.push(AttachmentMeta {
227. id: "a2".into(),
228. kind: "text".into(),
229. filename: "notes.txt".into(),
230. mime: "text/plain".into(),
231. byte_size: 10,
232. width: None,
233. height: None,
234. });
235. let mut cut =
236. msg("assistant", "Tier B, by about", Some("google/gemma-3-27b-it"), 1_000, UsageSource::Estimated);
237. cut.status = "cancelled".into();
238. let msgs = vec![
239. user,
240. msg(
241. "assistant",
242. "| Tier | Per seat |\n|---|---|\n| A | $12 |\n| B | $9 |",
243. Some("Qwen/Qwen3-VL-30B-A3B-Instruct"),
244. 180_000,
245. UsageSource::Reported,
246. ),
247. msg("user", "And yearly?", None, 0, UsageSource::None),
248. cut,
249. ];
250. let paths = |id: &str| (id == "a1").then(|| "/data/attachments/abc.png".to_string());
251. let tz = chrono::FixedOffset::west_opt(5 * 3600).unwrap();
252. let wiki = to_markdown(&conv, &msgs, &paths, ImageLinks::Wiki, &tz);
253. assert_eq!(wiki.filename, "Compare two pricing tiers a b.md");
254. assert_eq!(wiki.files.len(), 1);
255. assert_eq!(wiki.files[0].relative_path, "attachments/chart q3.png");
256. insta::assert_snapshot!("export_wiki", wiki.markdown);
257. let md = to_markdown(&conv, &msgs, &paths, ImageLinks::Markdown, &tz);
258. assert!(md.markdown.contains("![chart q3.png](attachments/chart%20q3.png)"));
259. }
261. #[test]
262. fn yaml_escaping() {
263. assert_eq!(yaml_str("a \"b\" \\ c\nd"), r#""a \"b\" \\ c\nd""#);
264. assert_eq!(yaml_str("tab\there"), r#""tab\there""#);
265. assert_eq!(yaml_str("\u{7}"), r#""\x07""#);
266. }
268. #[test]
269. fn filenames() {
270. assert_eq!(safe_filename(" ..weird [[name]] #tag ^x | y "), "weird name tag x y");
271. assert_eq!(safe_filename("///"), "Cortex chat");
272. assert_eq!(safe_filename(&"x".repeat(200)).len(), 80);
273. }
274. }