hinge 0.1.0__py3-none-any.whl

hinge/dbt/models/intermediate/int_user_artifact_incidence.sql

@@ -0,0 +1,21 @@
+{{ config(materialized='table') }}
+
+{{ build_incidence(
+    left_type='user',
+    right_type='artifact',
+    edge_types=[
+        'authored',
+        'committed',
+        'coauthored',
+        'opened',
+        'reviewed',
+        'merged',
+        'closed',
+        'reopened',
+        'commented_on',
+        'review_commented_on',
+        'pushed',
+        'touched'
+    ],
+    weight_mode='event_count'
+) }}

hinge/dbt/models/networks/artifact_reference.sql

@@ -0,0 +1,27 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('artifact_reference') }}
+
+WITH reference_edges AS (
+    {{ slice_edges(
+        edge_types=['references', 'closes', 'fixes', 'duplicates', 'relates_to'],
+        source_types=['artifact'],
+        target_types=['artifact']
+    ) }}
+)
+
+{{ network_edges(
+    relation='reference_edges',
+    recipe_name='artifact_reference',
+    source_node_id='source_node_id',
+    source_node_type="'artifact'",
+    target_node_id='target_node_id',
+    target_node_type="'artifact'",
+    edge_type='edge_type',
+    directed='true',
+    weight='coalesce(weight, 1.0)',
+    weight_kind="'binary'",
+    first_seen_at='occurred_at',
+    last_seen_at='occurred_at',
+    properties="to_json({'source_record_id': source_record_id, 'reference_type': edge_type})"
+) }}

hinge/dbt/models/networks/co_commit_user_user.sql

@@ -0,0 +1,48 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('co_commit_user_user') }}
+
+WITH commit_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'commit'
+      AND role IN ('authored', 'committed', 'coauthored')
+),
+
+user_commit AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS commit_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight
+    FROM commit_events
+    GROUP BY left_node_id, right_node_id
+),
+
+co_commits AS (
+    {{ project_bipartite(
+        incidence_relation='user_commit',
+        left_col='user_node_id',
+        right_col='commit_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='co_commits',
+    recipe_name='co_commit_user_user',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'co_committed'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_commits': n_contexts, 'commits': context_node_ids})"
+) }}

hinge/dbt/models/networks/co_edit_file_user_user.sql

@@ -0,0 +1,48 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('co_edit_file_user_user') }}
+
+WITH file_touch_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'file'
+      AND role = 'touched'
+),
+
+user_file AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS file_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight
+    FROM file_touch_events
+    GROUP BY left_node_id, right_node_id
+),
+
+co_edits AS (
+    {{ project_bipartite(
+        incidence_relation='user_file',
+        left_col='user_node_id',
+        right_col='file_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='co_edits',
+    recipe_name='co_edit_file_user_user',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'co_edited_file'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_files': n_contexts, 'files': context_node_ids})"
+) }}

hinge/dbt/models/networks/co_edit_line_user_user.sql

@@ -0,0 +1,48 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('co_edit_line_user_user') }}
+
+WITH line_touch_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'line_span'
+      AND role = 'touched'
+),
+
+user_line AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS line_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight
+    FROM line_touch_events
+    GROUP BY left_node_id, right_node_id
+),
+
+co_edits AS (
+    {{ project_bipartite(
+        incidence_relation='user_line',
+        left_col='user_node_id',
+        right_col='line_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='co_edits',
+    recipe_name='co_edit_line_user_user',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'co_edited_line'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_lines': n_contexts, 'line_spans': context_node_ids})"
+) }}

hinge/dbt/models/networks/dev_interaction.sql

@@ -0,0 +1,98 @@
+-- ============================================================================
+-- dev_interaction.sql  —  developer collaboration network
+--
+-- READ THIS BEFORE WRITING A NEW PROJECTION. Every .sql file in this folder
+-- must satisfy the contract documented in the header below.
+-- ============================================================================
+--
+-- WHAT THIS PROJECTION DOES
+--   Builds an undirected user-to-user "collaborator" graph. Two developers
+--   are collaborators if they both made code contributions to the same
+--   repository. A contribution is any of: opening a PR or issue, reviewing,
+--   merging, commenting on a PR or issue, or pushing commits.
+--
+--   "Shared a repo" is derived from the HIN structure using two hops:
+--     user → <contribution edge> → artifact ← contains ← repo
+--   The self-join on repo then finds all pairs of contributors per repo.
+--
+--   Starring, forking, and wiki edits are intentionally excluded — they
+--   do not constitute active development collaboration.
+--
+-- ─── INPUT CONTRACT ─────────────────────────────────────────────────────────
+--   Every projection reads from exactly two dbt sources:
+--
+--     {{ ref('hin_nodes') }}  canonical HIN node model
+--     {{ ref('hin_edges') }}  canonical HIN edge model
+--
+--   The store (DuckDBStore) rewrites these views before each projection run
+--   to point at the requested dataset_id. Do NOT touch any other table.
+--   If you need helper logic, put it in a CTE — do not create intermediate
+--   dbt models unless the projection genuinely needs to be split.
+--
+-- ─── OUTPUT CONTRACT ────────────────────────────────────────────────────────
+--   Every projection model should use the network_edges macro. It emits the
+--   standard cookbook shape:
+--
+--     recipe_name, recipe_version,
+--     source_node_id, source_node_type, target_node_id, target_node_type,
+--     directed, edge_type,
+--     weight, weight_kind, n_contexts, n_events,
+--     first_seen_at, last_seen_at, time_bin, bot_policy, properties
+--
+--   DbtProjection converts this richer table into TypedEdge objects for the
+--   current exporter API, merging standard fields and properties into attrs.
+--
+--   The DbtProjection stage discovers the result table by its dbt model
+--   name (this file's stem, `dev_interaction`). dbt's default
+--   materialisation is `table`, set globally in dbt_project.yml.
+--
+-- ─── HOW TO REGISTER THIS FILE ──────────────────────────────────────────────
+--   1. Save the .sql file in hinge/dbt/models/networks/<name>.sql.
+--   2. Add a sibling Python module under
+--      hinge/stages/projection/specs/<name>.py exposing a `SPEC` constant
+--      (see specs/dev_interaction.py for the template).
+--   3. Register the entry-point in pyproject.toml under
+--      [project.entry-points."hinge.projection_specs"].
+--   4. Write a test in tests/stages/projection/test_<name>.py against a
+--      seeded :memory: DuckDB fixture.
+--
+-- ─── PARAMETERS ─────────────────────────────────────────────────────────────
+--   Projections can accept run-time parameters via dbt vars (passed by
+--   DbtProjection from the ProjectionParams object). Read them with
+--   {{ var('name', default) }}.
+--   This template ignores all vars — keep it that way until a real use case
+--   appears.
+-- ============================================================================
+
+{{ assert_recipe_supported('dev_interaction') }}
+
+WITH
+
+-- Collaborator pairs: users who share at least one active development repo.
+-- This is the generic bipartite projection pattern: user -> repo <- user.
+collaborators AS (
+    {{ project_bipartite(
+        incidence_relation=ref('int_developer_repo_affiliation'),
+        left_col='user_node_id',
+        right_col='repo_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='collaborators',
+    recipe_name='dev_interaction',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'collaborates_with'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_repos': n_contexts, 'repos': context_node_ids})"
+) }}

hinge/dbt/models/networks/developer_repo_affiliation.sql

@@ -0,0 +1,20 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('developer_repo_affiliation') }}
+
+{{ network_edges(
+    relation=ref('int_developer_repo_affiliation'),
+    recipe_name='developer_repo_affiliation',
+    source_node_id='user_node_id',
+    source_node_type="'user'",
+    target_node_id='repo_node_id',
+    target_node_type="'repo'",
+    edge_type="'affiliated_with'",
+    directed='true',
+    weight='weight',
+    weight_kind="'event_count'",
+    n_events='n_events',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'roles': roles})"
+) }}

hinge/dbt/models/networks/follow_user_user.sql

@@ -0,0 +1,27 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('follow_user_user') }}
+
+WITH follow_edges AS (
+    {{ slice_edges(
+        edge_types=['follows'],
+        source_types=['user'],
+        target_types=['user']
+    ) }}
+)
+
+{{ network_edges(
+    relation='follow_edges',
+    recipe_name='follow_user_user',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'follows'",
+    directed='true',
+    weight='coalesce(weight, 1.0)',
+    weight_kind="'binary'",
+    first_seen_at='occurred_at',
+    last_seen_at='occurred_at',
+    properties="to_json({'source_record_id': source_record_id})"
+) }}

hinge/dbt/models/networks/fork_repo_repo.sql

@@ -0,0 +1,27 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('fork_repo_repo') }}
+
+WITH fork_edges AS (
+    {{ slice_edges(
+        edge_types=['fork_of'],
+        source_types=['repo'],
+        target_types=['repo']
+    ) }}
+)
+
+{{ network_edges(
+    relation='fork_edges',
+    recipe_name='fork_repo_repo',
+    source_node_id='source_node_id',
+    source_node_type="'repo'",
+    target_node_id='target_node_id',
+    target_node_type="'repo'",
+    edge_type="'fork_of'",
+    directed='true',
+    weight='coalesce(weight, 1.0)',
+    weight_kind="'binary'",
+    first_seen_at='occurred_at',
+    last_seen_at='occurred_at',
+    properties="to_json({'source_record_id': source_record_id})"
+) }}

hinge/dbt/models/networks/issue_co_participation.sql

@@ -0,0 +1,50 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('issue_co_participation') }}
+
+WITH issue_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'issue'
+      AND role IN ('opened', 'commented_on', 'closed')
+),
+
+issue_participation AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS issue_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight,
+        count(DISTINCT evidence_edge_id) AS n_events,
+        list_distinct(list(role)) AS roles
+    FROM issue_events
+    GROUP BY left_node_id, right_node_id
+),
+
+co_participation AS (
+    {{ project_bipartite(
+        incidence_relation='issue_participation',
+        left_col='user_node_id',
+        right_col='issue_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='co_participation',
+    recipe_name='issue_co_participation',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'co_participates_issue'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_issues': n_contexts, 'issues': context_node_ids})"
+) }}

hinge/dbt/models/networks/issue_participation.sql

@@ -0,0 +1,40 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('issue_participation') }}
+
+WITH issue_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'issue'
+      AND role IN ('opened', 'commented_on', 'closed')
+),
+
+participation AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS issue_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight,
+        count(DISTINCT evidence_edge_id) AS n_events,
+        list_distinct(list(role)) AS roles
+    FROM issue_events
+    GROUP BY left_node_id, right_node_id
+)
+
+{{ network_edges(
+    relation='participation',
+    recipe_name='issue_participation',
+    source_node_id='user_node_id',
+    source_node_type="'user'",
+    target_node_id='issue_node_id',
+    target_node_type="'artifact'",
+    edge_type="'participated_in_issue'",
+    directed='true',
+    weight='weight',
+    weight_kind="'event_count'",
+    n_events='n_events',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'roles': roles})"
+) }}

hinge/dbt/models/networks/pr_author_reviewer.sql

@@ -0,0 +1,66 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('pr_author_reviewer') }}
+
+WITH pr_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'pull_request'
+      AND role IN ('opened', 'reviewed')
+),
+
+openers AS (
+    SELECT
+        left_node_id AS author_node_id,
+        right_node_id AS pr_node_id,
+        min(occurred_at) AS first_seen_at
+    FROM pr_events
+    WHERE role = 'opened'
+    GROUP BY left_node_id, right_node_id
+),
+
+reviewers AS (
+    SELECT
+        left_node_id AS reviewer_node_id,
+        right_node_id AS pr_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        count(DISTINCT evidence_edge_id) AS n_reviews
+    FROM pr_events
+    WHERE role = 'reviewed'
+    GROUP BY left_node_id, right_node_id
+),
+
+author_reviewer AS (
+    SELECT
+        openers.author_node_id,
+        reviewers.reviewer_node_id,
+        count(DISTINCT openers.pr_node_id) AS n_contexts,
+        sum(reviewers.n_reviews) AS n_events,
+        list_distinct(list(openers.pr_node_id)) AS pull_requests,
+        min(least(openers.first_seen_at, reviewers.first_seen_at)) AS first_seen_at,
+        max(reviewers.last_seen_at) AS last_seen_at
+    FROM openers
+    JOIN reviewers
+        ON reviewers.pr_node_id = openers.pr_node_id
+       AND reviewers.reviewer_node_id != openers.author_node_id
+    GROUP BY openers.author_node_id, reviewers.reviewer_node_id
+)
+
+{{ network_edges(
+    relation='author_reviewer',
+    recipe_name='pr_author_reviewer',
+    source_node_id='author_node_id',
+    source_node_type="'user'",
+    target_node_id='reviewer_node_id',
+    target_node_type="'user'",
+    edge_type="'reviewed_pr_from'",
+    directed='true',
+    weight='n_contexts',
+    weight_kind="'shared_count'",
+    n_contexts='n_contexts',
+    n_events='n_events',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'pull_requests': pull_requests})"
+) }}

hinge/dbt/models/networks/pr_participation.sql

@@ -0,0 +1,40 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('pr_participation') }}
+
+WITH pr_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'pull_request'
+      AND role IN ('opened', 'reviewed', 'merged', 'commented_on', 'review_commented_on')
+),
+
+participation AS (
+    SELECT
+        left_node_id AS user_node_id,
+        right_node_id AS pr_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight,
+        count(DISTINCT evidence_edge_id) AS n_events,
+        list_distinct(list(role)) AS roles
+    FROM pr_events
+    GROUP BY left_node_id, right_node_id
+)
+
+{{ network_edges(
+    relation='participation',
+    recipe_name='pr_participation',
+    source_node_id='user_node_id',
+    source_node_type="'user'",
+    target_node_id='pr_node_id',
+    target_node_type="'artifact'",
+    edge_type="'participated_in_pr'",
+    directed='true',
+    weight='weight',
+    weight_kind="'event_count'",
+    n_events='n_events',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'roles': roles})"
+) }}

hinge/dbt/models/networks/pr_reviewer_coreview.sql

@@ -0,0 +1,48 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('pr_reviewer_coreview') }}
+
+WITH review_events AS (
+    SELECT *
+    FROM {{ ref('int_user_artifact_incidence') }}
+    WHERE right_node_subtype = 'pull_request'
+      AND role = 'reviewed'
+),
+
+reviewer_pr AS (
+    SELECT
+        left_node_id AS reviewer_node_id,
+        right_node_id AS pr_node_id,
+        min(occurred_at) AS first_seen_at,
+        max(occurred_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight
+    FROM review_events
+    GROUP BY left_node_id, right_node_id
+),
+
+co_reviews AS (
+    {{ project_bipartite(
+        incidence_relation='reviewer_pr',
+        left_col='reviewer_node_id',
+        right_col='pr_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='co_reviews',
+    recipe_name='pr_reviewer_coreview',
+    source_node_id='source_node_id',
+    source_node_type="'user'",
+    target_node_id='target_node_id',
+    target_node_type="'user'",
+    edge_type="'co_reviewed_pr'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_pull_requests': n_contexts, 'pull_requests': context_node_ids})"
+) }}

hinge/dbt/models/networks/repo_shared_contributors.sql

@@ -0,0 +1,41 @@
+{{ config(materialized='table') }}
+
+{{ assert_recipe_supported('repo_shared_contributors') }}
+
+WITH contributor_repo AS (
+    SELECT
+        repo_node_id,
+        user_node_id,
+        min(first_seen_at) AS first_seen_at,
+        max(last_seen_at) AS last_seen_at,
+        sum(weight)::DOUBLE AS weight
+    FROM {{ ref('int_developer_repo_affiliation') }}
+    GROUP BY repo_node_id, user_node_id
+),
+
+shared AS (
+    {{ project_bipartite(
+        incidence_relation='contributor_repo',
+        left_col='repo_node_id',
+        right_col='user_node_id',
+        directed=false,
+        weight_mode='shared_count'
+    ) }}
+)
+
+{{ network_edges(
+    relation='shared',
+    recipe_name='repo_shared_contributors',
+    source_node_id='source_node_id',
+    source_node_type="'repo'",
+    target_node_id='target_node_id',
+    target_node_type="'repo'",
+    edge_type="'shared_contributors'",
+    directed='false',
+    weight='weight',
+    weight_kind='weight_kind',
+    n_contexts='n_contexts',
+    first_seen_at='first_seen_at',
+    last_seen_at='last_seen_at',
+    properties="to_json({'shared_contributors': n_contexts, 'contributors': context_node_ids})"
+) }}