From dff2e45c51404713460080e7d42b9d7c7869f91c Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 28 Jan 2019 20:39:34 -0800 Subject: WIP on a hack-y changelog export --- rust/src/bin/fatcat-export.rs | 95 ++++++++++++++++++++++++++++++++++++++----- 1 file changed, 84 insertions(+), 11 deletions(-) diff --git a/rust/src/bin/fatcat-export.rs b/rust/src/bin/fatcat-export.rs index 6d2d396c..182c8228 100644 --- a/rust/src/bin/fatcat-export.rs +++ b/rust/src/bin/fatcat-export.rs @@ -36,7 +36,8 @@ arg_enum! { Fileset, Webcapture, Release, - Work + Work, + Changelog, } } @@ -44,6 +45,7 @@ struct IdentRow { ident_id: FatcatId, rev_id: Option, redirect_id: Option, + index: Option, } macro_rules! generic_loop_work { @@ -87,6 +89,39 @@ generic_loop_work!(loop_work_webcapture, WebcaptureEntity); generic_loop_work!(loop_work_release, ReleaseEntity); generic_loop_work!(loop_work_work, WorkEntity); +fn look_work_changelog(row_receiver: channel::Receiver, output_sender: channel::Sender, db_conn: &DbConn, expand: Option) { + let result: Result<()> = (|| { + for row in row_receiver { + + let cl_row: ChangelogRow = changelog::table.find(row.index.unwrap()).first(conn)?; + let editgroup = + self.get_editgroup_handler(conn, FatcatId::from_uuid(&cl_row.editgroup_id))?; + + let mut entry = cl_row.into_model(); + entry.editgroup = Some(editgroup); + Ok(entry) + + let mut entity = ChangelogEntry::db_get_rev( + db_conn, + row.rev_id.expect("valid, non-deleted row"), + HideFlags::none(), + )?; // .chain_err(|| "reading entity from database")?; + entity.state = Some("active".to_string()); // only active lines were passed + entity.ident = Some(row.ident_id.to_string()); + if let Some(expand) = expand { + entity.db_expand(db_conn, expand)? + // chain_err(|| "expanding sub-entities from database")?; + } + output_sender.send(serde_json::to_string(&entity)?); + } + Ok(()) + })(); + if let Err(ref e) = result { + error!("{}", e); // e.display_chain()) + } + result.unwrap() +} + fn loop_printer( output_receiver: channel::Receiver, done_sender: channel::Sender<()>, @@ -119,6 +154,20 @@ fn parse_line(s: &str) -> Result { "" => None, val => Some(FatcatId::from_uuid(&Uuid::from_str(&val)?)), }, + index: None, + }) +} + +fn parse_line_changelog(s: &str) -> Result { + let fields: Vec = s.split('\t').map(|v| v.to_string()).collect(); + if fields.len() != 3 { + bail!("Invalid input line"); + } + Ok(IdentRow { + ident_id: FatcatId::from_uuid(&Uuid::from_str(&fields[1])?), + rev_id: None, + redirect_id: None, + index: fields[0].parse::()?, }) } @@ -139,6 +188,22 @@ fn test_parse_line() { assert!(parse_line("00000000-0000-0000-3333-000000000002\t00000000-0000-0000-3333-fff000000001\t00000000-0000-0000-3333-000000000001").is_ok()); } +#[test] +fn test_parse_line_changelog() { + assert!(parse_line( + "8\t94f11d0d-32c0-49fb-a71e-f3835af56582\t2019-01-23 20:51:50.321105-08" + ) + .is_ok()); + assert!(parse_line( + "\t94f11d0d-32c0-49fb-a71e-f3835af56582\t2019-01-23 20:51:50.321105-08" + ) + .is_err()); + assert!(parse_line( + "8\t94f11d0d-32c0-49fb-a71e-f3835af56582\t" + ) + .is_err()); +} + // Use num_cpus/2, or CLI arg for worker count // // 1. open buffered reader, buffered writer, and database pool. create channels @@ -177,8 +242,8 @@ pub fn do_export( ExportEntityType::Creator => thread::spawn(move || { loop_work_creator(row_receiver, output_sender, &db_conn, expand) }), - ExportEntityType::File => { - thread::spawn(move || loop_work_file(row_receiver, output_sender, &db_conn, expand)) + ExportEntityType::File => { thread::spawn(move || + loop_work_file(row_receiver, output_sender, &db_conn, expand)) } ExportEntityType::Fileset => thread::spawn(move || { loop_work_fileset(row_receiver, output_sender, &db_conn, expand) @@ -189,9 +254,12 @@ pub fn do_export( ExportEntityType::Release => thread::spawn(move || { loop_work_release(row_receiver, output_sender, &db_conn, expand) }), - ExportEntityType::Work => { - thread::spawn(move || loop_work_work(row_receiver, output_sender, &db_conn, expand)) - } + ExportEntityType::Work => { thread::spawn(move || + loop_work_work(row_receiver, output_sender, &db_conn, expand)) + }, + ExportEntityType::Release => thread::spawn(move || { + loop_work_changelog(row_receiver, output_sender, &db_conn, expand) + }), }; } drop(output_sender); @@ -201,11 +269,16 @@ pub fn do_export( let mut count = 0; for line in buf_input.lines() { let line = line?; - let row = parse_line(&line)?; - match (row.rev_id, row.redirect_id, redirects) { - (None, _, _) => (), - (Some(_), Some(_), false) => (), - _ => row_sender.send(row), + if ExportEntityType == ExportEntityType::Changelog { + let change_row = parse_line_changelog(&line)?; + row_sender.send(change_row); + } else { + let row = parse_line(&line)?; + match (row.rev_id, row.redirect_id, redirects) { + (None, _, _, _) => (), + (Some(_), Some(_), false, None) => (), + _ => row_sender.send(row), + } } count += 1; if count % 1000 == 0 { -- cgit v1.2.3