wip: a few dot examples

1 files changed, 73 insertions, 0 deletions

diff --git a/extra/dot/levels_inbound.py b/extra/dot/levels_inbound.py
new file mode 100644
index 0000000..290ab6f
--- /dev/null
+++ b/extra/dot/levels_inbound.py
@@ -0,0 +1,73 @@
+#!/usr/bin/env python
+
+"""
+Take a document fatcat id and emit the citation structure up to X levels as graphviz dot file.
+"""
+
+from elasticsearch import Elasticsearch
+from elasticsearch_dsl import Search
+import sys
+import requests
+
+seed = sys.argv[1] if len(sys.argv) > 1 else "rgzyfbx7t5d53m5amvbllewzqi"
+max_nodes = 50
+
+client = Elasticsearch("http://localhost:9200")
+S = Search(using=client, index="fatcat_ref_v01")
+
+def pretty_label(s, k=10):
+    parts = s.split()
+    cur, result =[], []
+    for p in parts:
+        if len(" ".join(cur)) < k:
+            cur.append(p)
+        else:
+            result.append(" ".join(cur))
+            cur = []
+    result.append(" ".join(cur))
+    return "\\n".join(result)
+
+def get_dot_label(ident):
+    title = requests.get("https://api.fatcat.wiki/v0/release/{}".format(ident)).json().get("title")
+    return pretty_label(title, k=10)
+
+
+def unique_sources(ident):
+    """
+    Unique inbound references.
+    """
+    s = S.query("match", target_release_ident=ident)
+    unique_source_idents = set()
+    for hit in s.scan():
+        unique_source_idents.add(hit["source_release_ident"])
+    return list(unique_source_idents)
+
+queue = set([seed])
+i = 0
+edges = set()
+while queue:
+    node = queue.pop()
+    i += 1
+    if i == max_nodes:
+        break
+    for n in unique_sources(node)[:5]:
+        if n == node:
+            print("skipping self ref: {}".format(n), file=sys.stderr)
+            continue
+        edges.add((node, n))
+        queue.add(n)
+
+nodes = set()
+for a, b in edges:
+    nodes.add(a)
+    nodes.add(b)
+
+print("digraph G {")
+# add sensible labels
+for n in nodes:
+    print(""" "{}"  [label="{}"]; """.format(n, get_dot_label(n)))
+
+for a, b in edges:
+    print(""" "{}" -> "{}"; """.format(a, b))
+print("}")
+