From cb580f83eb5db5b26b045921f5d58485aa57710f Mon Sep 17 00:00:00 2001
From: Stuart Axelbrooke <stuart@axelbrooke.com>
Date: Thu, 16 Oct 2025 19:27:25 -0700
Subject: [PATCH] lay groundwork for log line parsing

---
 databuild/data_deps.rs        | 21 +++++++++++++++++++++
 databuild/databuild.proto     |  3 +++
 databuild/event_transforms.rs |  2 +-
 databuild/job_run.rs          | 12 ++++++------
 databuild/lib.rs              |  1 +
 5 files changed, 32 insertions(+), 7 deletions(-)
 create mode 100644 databuild/data_deps.rs

diff --git a/databuild/data_deps.rs b/databuild/data_deps.rs
new file mode 100644
index 0000000..4bc9b6e
--- /dev/null
+++ b/databuild/data_deps.rs
@@ -0,0 +1,21 @@
+use crate::JobRunMissingDeps;
+
+pub struct LogLine(String);
+
+const DATABUILD_JSON: &str = "DATABUILD_JSON:";
+const DATABUILD_JSON_LEN: usize = DATABUILD_JSON.len();
+
+impl From<LogLine> for Option<JobRunMissingDeps> {
+    fn from(value: LogLine) -> Self {
+        line_matches(&value.0).and_then(json_to_missing_deps)
+    }
+}
+
+fn line_matches(line: &str) -> Option<&str> {
+    line.trim().strip_prefix(DATABUILD_JSON)
+}
+
+fn json_to_missing_deps(line: &str) -> Option<JobRunMissingDeps> {
+    todo!()
+}
+
diff --git a/databuild/databuild.proto b/databuild/databuild.proto
index 2dc8353..26b9d36 100644
--- a/databuild/databuild.proto
+++ b/databuild/databuild.proto
@@ -86,6 +86,9 @@ message JobRunMissingDepsEventV1 {
   string job_run_id = 1;
   repeated MissingDeps missing_deps = 2;
 }
+message JobRunMissingDeps {
+  repeated MissingDeps missing_deps = 2;
+}
 message MissingDeps {
   // The list of partition refs that are prevented from building by these missing deps (can be just 1)
   repeated PartitionRef impacted = 1;
diff --git a/databuild/event_transforms.rs b/databuild/event_transforms.rs
index 59d34dc..d3bfd1b 100644
--- a/databuild/event_transforms.rs
+++ b/databuild/event_transforms.rs
@@ -1,6 +1,6 @@
+use crate::data_build_event::Event;
 use crate::util::current_timestamp;
 use crate::{event_source, EventSource, JobRunStatus, JobRunStatusCode, ManuallyTriggeredEvent, PartitionRef, PartitionStatus, PartitionStatusCode, WantCancelEventV1, WantCreateEventV1, WantDetail, WantStatus, WantStatusCode};
-use crate::data_build_event::Event;
 
 impl From<&WantCreateEventV1> for WantDetail {
     fn from(e: &WantCreateEventV1) -> Self {
diff --git a/databuild/job_run.rs b/databuild/job_run.rs
index c975286..3dd4f14 100644
--- a/databuild/job_run.rs
+++ b/databuild/job_run.rs
@@ -1,9 +1,9 @@
-use std::collections::HashMap;
-use std::marker::PhantomData;
 use crate::data_build_event::Event;
-use crate::{JobRunHeartbeatEventV1, JobRunStatus, JobRunStatusCode, JobRunSuccessEventV1, JobRunFailureEventV1, JobRunCancelEventV1, EventSource};
+use crate::{EventSource, JobRunCancelEventV1, JobRunFailureEventV1, JobRunStatus, JobRunSuccessEventV1};
+use std::collections::HashMap;
 use std::error::Error;
 use std::io::{BufRead, BufReader};
+use std::marker::PhantomData;
 use std::process::{Child, Command, Stdio};
 use uuid::Uuid;
 
@@ -314,10 +314,10 @@ pub struct JobRunPollResult {
 }
 
 mod tests {
-    use std::collections::HashMap;
     use crate::data_build_event::Event;
-    use crate::job_run::{JobRunBackend, JobRunVisitResult, NotStartedJobRun, SubProcessBackend};
-    use crate::{ManuallyTriggeredEvent};
+    use crate::job_run::{JobRunBackend, JobRunVisitResult, SubProcessBackend};
+    use crate::ManuallyTriggeredEvent;
+    use std::collections::HashMap;
 
     fn test_helper_path() -> String {
         std::env::var("TEST_SRCDIR")
diff --git a/databuild/lib.rs b/databuild/lib.rs
index 259f1e3..6639e42 100644
--- a/databuild/lib.rs
+++ b/databuild/lib.rs
@@ -6,6 +6,7 @@ mod util;
 mod build_state;
 mod event_transforms;
 mod event_defaults;
+mod data_deps;
 
 // Include generated protobuf code
 include!("databuild.rs");