update notes on dot

3 files changed, 80 insertions, 61 deletions

diff --git a/extra/dot/README.md b/extra/dot/README.md
index 6914dc1..bad41b8 100644
--- a/extra/dot/README.md
+++ b/extra/dot/README.md
@@ -2,3 +2,8 @@
 
 Examples using ref index plus fatcat API to generate citation graph snippets
 via graphviz.
+
+![](example.png)
+
+![](example_inbound.png)
+
diff --git a/extra/dot/levels.py b/extra/dot/levels.py
index 96af9c4..ab4bf85 100644
--- a/extra/dot/levels.py
+++ b/extra/dot/levels.py
@@ -1,7 +1,10 @@
 #!/usr/bin/env python
-
 """
 Take a document fatcat id and emit the citation structure up to X levels as graphviz dot file.
+
+    $ python levels.py [RELEASE IDENT]
+
+Throwaway script, expected to be run on the same machine as the ref index currently.
 """
 
 from elasticsearch import Elasticsearch
@@ -10,14 +13,16 @@ import sys
 import requests
 
 seed = sys.argv[1] if len(sys.argv) > 1 else "rgzyfbx7t5d53m5amvbllewzqi"
-max_nodes = 30
+max_nodes_total = 30
+max_nodes = 5
 
 client = Elasticsearch("http://localhost:9200")
 S = Search(using=client, index="fatcat_ref_v01")
 
+
 def pretty_label(s, k=10):
     parts = s.split()
-    cur, result =[], []
+    cur, result = [], []
     for p in parts:
         if len(" ".join(cur)) < k:
             cur.append(p)
@@ -27,8 +32,10 @@ def pretty_label(s, k=10):
     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")
+    title = (requests.get("https://api.fatcat.wiki/v0/release/{}".format(
+        ident)).json().get("title"))
     return pretty_label(title, k=10)
 
 
@@ -39,32 +46,33 @@ def unique_targets(ident):
         unique_target_idents.add(hit["target_release_ident"])
     return list(unique_target_idents)
 
-queue = set([seed])
-i = 0
-edges = set()
-while queue:
-    node = queue.pop()
-    i += 1
-    if i == max_nodes:
-        break
-    for n in unique_targets(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("}")
 
+if __name__ == "__main__":
+    queue = set([seed])
+    edges = set()
+    i = 0
+    while queue:
+        node = queue.pop()
+        i += 1
+        if i == max_nodes_total:
+            break
+        for n in unique_targets(node)[:max_nodes]:
+            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("}")
diff --git a/extra/dot/levels_inbound.py b/extra/dot/levels_inbound.py
index 290ab6f..9a5d398 100644
--- a/extra/dot/levels_inbound.py
+++ b/extra/dot/levels_inbound.py
@@ -1,7 +1,8 @@
 #!/usr/bin/env python
-
 """
 Take a document fatcat id and emit the citation structure up to X levels as graphviz dot file.
+
+Copy-pasted from levels.py - throwaway code.
 """
 
 from elasticsearch import Elasticsearch
@@ -10,14 +11,16 @@ import sys
 import requests
 
 seed = sys.argv[1] if len(sys.argv) > 1 else "rgzyfbx7t5d53m5amvbllewzqi"
-max_nodes = 50
+max_nodes_total = 50
+max_nodes = 5
 
 client = Elasticsearch("http://localhost:9200")
 S = Search(using=client, index="fatcat_ref_v01")
 
+
 def pretty_label(s, k=10):
     parts = s.split()
-    cur, result =[], []
+    cur, result = [], []
     for p in parts:
         if len(" ".join(cur)) < k:
             cur.append(p)
@@ -27,8 +30,10 @@ def pretty_label(s, k=10):
     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")
+    title = requests.get("https://api.fatcat.wiki/v0/release/{}".format(
+        ident)).json().get("title")
     return pretty_label(title, k=10)
 
 
@@ -42,32 +47,33 @@ def unique_sources(ident):
         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)
+if __name__ == '__main__':
+    queue = set([seed])
+    edges = set()
+    i = 0
+    while queue:
+        node = queue.pop()
+        i += 1
+        if i == max_nodes_total:
+            break
+        for n in unique_sources(node)[:max_nodes]:
+            if n == node:
+                print("skipping self ref: {}".format(n), file=sys.stderr)
+                continue
+            edges.add((node, n))
+            queue.add(n)
 
-print("digraph G {")
-# add sensible labels
-for n in nodes:
-    print(""" "{}"  [label="{}"]; """.format(n, get_dot_label(n)))
+    nodes = set()
+    for a, b in edges:
+        nodes.add(a)
+        nodes.add(b)
 
-for a, b in edges:
-    print(""" "{}" -> "{}"; """.format(a, b))
-print("}")
+    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("}")