feat(git): extract issue references from commit messages

Extract URLs, GitHub-style (#123), and Jira-style (PROJ-456) issue
references from commit messages and store them as :commit/issue-refs
(cardinality-many). Extracted deterministically during import via
regex. Existing commits are backfilled during sync/update.

Adds two named queries: commits-by-issue and issue-refs.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: leifericf

resources/queries/commits-by-issue.edn

@@ -0,0 +1,12 @@
+{:name "commits-by-issue"
+ :description "Commits that reference a given issue (by key or URL substring)"
+ :query [:find ?sha ?msg ?name ?date
+         :in $ ?ref
+         :where
+         [?commit :commit/issue-refs ?ref]
+         [?commit :git/sha ?sha]
+         [?commit :commit/message ?msg]
+         [?commit :commit/author ?person]
+         [?person :person/name ?name]
+         [?commit :commit/authored-at ?date]]
+ :params {"ref" "Issue reference to search for, e.g. #123 or PROJ-456"}}

resources/queries/index.edn

@@ -13,6 +13,7 @@
  "co-changed-files"
  "commit-kinds"
  "commit-spread"
+ "commits-by-issue"
  "complex-hotspots"
  "cross-dir-imports"
  "dependency-drift"
@@ -30,6 +31,7 @@
  "fix-authors"
  "hotspots"
  "import-hotspots"
+ "issue-refs"
  "llm-cost-by-file"
  "llm-cost-by-model"
  "llm-cost-total"

resources/queries/issue-refs.edn

@@ -0,0 +1,5 @@
+{:name "issue-refs"
+ :description "All unique issue references with commit count"
+ :query [:find ?ref (count ?commit)
+         :where
+         [?commit :commit/issue-refs ?ref]]}

resources/schema/core.edn

@@ -83,6 +83,11 @@
   :db/cardinality :db.cardinality/one
   :db/doc         "Total lines deleted across all files in this commit"}
 
+ {:db/ident       :commit/issue-refs
+  :db/valueType   :db.type/string
+  :db/cardinality :db.cardinality/many
+  :db/doc         "Issue references extracted from commit message — keys (#123, PROJ-456) and URLs"}
+
  ;; --- Persons ---
  {:db/ident       :person/email
   :db/valueType   :db.type/string

src/noumenon/git.clj

@@ -165,6 +165,26 @@
                 :message         (str/trim body)}
                numstat)))))
 
+;; --- Issue reference extraction ---
+
+(def ^:private issue-ref-patterns
+  "Patterns for extracting issue references from commit messages."
+  [#"https?://[^\s\)\]>\"']+"          ;; URLs
+   #"(?<![A-Za-z0-9/])[A-Z][A-Z0-9]+-\d+" ;; Jira-style: PROJ-123
+   #"(?<=\s|^|\()#\d+"                 ;; GitHub-style: #123
+   #"(?<=\s|^|\()GH-\d+"])             ;; GitHub alt: GH-42
+
+(defn extract-issue-refs
+  "Extract unique issue references (URLs, keys like #123, PROJ-456) from text.
+   Returns a sorted set of strings, or nil if none found."
+  [text]
+  (when-not (str/blank? text)
+    (let [refs (->> issue-ref-patterns
+                    (mapcat #(re-seq % text))
+                    (map str/trim)
+                    (into (sorted-set)))]
+      (when (seq refs) refs))))
+
 ;; --- Commit classification ---
 
 (def ^:private conventional-prefix-re
@@ -241,7 +261,9 @@
                         (pos? (or deletions 0)) (assoc :commit/deletions deletions)
                         (seq parent-shas)        (assoc :commit/parents
                                                         (mapv #(vector :git/sha %) parent-shas))
-                        (seq file-tids)          (assoc :commit/changed-files file-tids))]
+                        (seq file-tids)          (assoc :commit/changed-files file-tids)
+                        (extract-issue-refs message) (assoc :commit/issue-refs
+                                                            (extract-issue-refs message)))]
     (-> (person-tx-data author-tid author-email author-name
                         committer-tid committer-email committer-name)
         (into [commit

src/noumenon/sync.clj

@@ -195,6 +195,27 @@
                                               :tx/source :deterministic})}))
     (count stale)))
 
+(defn backfill-issue-refs!
+  "Extract issue references from all commit messages that don't have any yet.
+   Returns count of commits updated."
+  [conn]
+  (let [db       (d/db conn)
+        commits  (d/q '[:find ?e ?msg
+                        :where
+                        [?e :commit/message ?msg]
+                        (not [?e :commit/issue-refs])]
+                      db)
+        updates  (->> commits
+                      (keep (fn [[eid msg]]
+                              (when-let [refs (git/extract-issue-refs msg)]
+                                {:db/id eid :commit/issue-refs refs})))
+                      vec)]
+    (when (seq updates)
+      (d/transact conn {:tx-data (conj updates {:db/id "datomic.tx"
+                                                :tx/op :import
+                                                :tx/source :deterministic})}))
+    (count updates)))
+
 ;; --- Sync orchestration ---
 
 (defn update-repo!
@@ -229,6 +250,9 @@
               reclass-n (reclassify-commits! conn)
               _         (when (pos? reclass-n)
                           (log! (str "Reclassified " reclass-n " commit kinds")))
+              issues-n  (backfill-issue-refs! conn)
+              _         (when (pos? issues-n)
+                          (log! (str "Extracted issue refs for " issues-n " commits")))
               files-r   (files/import-files! conn repo-path repo-uri)
               post-r    (when (or fresh?
                                   (seq (:added changes))

test/noumenon/git_test.clj

@@ -186,6 +186,49 @@
     (is (nil? (:commit/parents entity)))
     (is (nil? (:commit/changed-files entity)))))
 
+;; --- Issue reference extraction ---
+
+(deftest extract-issue-refs-github-style
+  (is (= #{"#123" "#456"}
+         (git/extract-issue-refs "fix(auth): resolve login bug (#123) and (#456)"))))
+
+(deftest extract-issue-refs-jira-style
+  (is (= #{"PROJ-42" "PROJ-999"}
+         (git/extract-issue-refs "PROJ-42 implement feature, see also PROJ-999"))))
+
+(deftest extract-issue-refs-urls
+  (is (= #{"https://github.com/foo/bar/issues/123"}
+         (git/extract-issue-refs "Closes https://github.com/foo/bar/issues/123"))))
+
+(deftest extract-issue-refs-mixed
+  (let [refs (git/extract-issue-refs "Fix #55 (JIRA-100) see https://jira.example.com/browse/JIRA-100")]
+    (is (contains? refs "#55"))
+    (is (contains? refs "JIRA-100"))
+    (is (some #(str/starts-with? % "https://") refs))))
+
+(deftest extract-issue-refs-none
+  (is (nil? (git/extract-issue-refs "chore: bump deps")))
+  (is (nil? (git/extract-issue-refs nil)))
+  (is (nil? (git/extract-issue-refs ""))))
+
+(deftest tx-data-includes-issue-refs
+  (let [commit  {:sha "abc" :parent-shas [] :message "fix(api): handle timeout (#42, PROJ-7)"
+                 :author-name "A" :author-email "a@x.com" :authored-at #inst "2024-01-01"
+                 :committer-name "A" :committer-email "a@x.com" :committed-at #inst "2024-01-01"
+                 :changed-files []}
+        tx-data (git/commit->tx-data "test://repo" commit)
+        entity  (first (filter :git/sha tx-data))]
+    (is (= #{"#42" "PROJ-7"} (:commit/issue-refs entity)))))
+
+(deftest tx-data-omits-issue-refs-when-none
+  (let [commit  {:sha "abc" :parent-shas [] :message "chore: update deps"
+                 :author-name "A" :author-email "a@x.com" :authored-at #inst "2024-01-01"
+                 :committer-name "A" :committer-email "a@x.com" :committed-at #inst "2024-01-01"
+                 :changed-files []}
+        tx-data (git/commit->tx-data "test://repo" commit)
+        entity  (first (filter :git/sha tx-data))]
+    (is (nil? (:commit/issue-refs entity)))))
+
 ;; --- Rename path resolution (regression: Flask import crash) ---
 
 (deftest resolve-rename-path-directory