gtfs_df 0.5.0 → 0.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 140458a6ce1013bef475e0a6cdcab6364cff04b8a18eedb5e5d0244e3bccf38a
-  data.tar.gz: c420a34f7004eca9267f32f53038632f822224924eac5b77aa98957bb3149e20
+  metadata.gz: c7c1d87e57bbb44ceb4ce8112da7172c5a8c76f8e88d3b1e8fbb610aad850cf3
+  data.tar.gz: 6a1b68dfc723d3c70b779687a70100b735c9a57e91603b302685318660473c66
 SHA512:
-  metadata.gz: 032d24ed1df3ed43e5e6953abebbeda70ba5450cab731e931e8548bd058e37d9472edd688a78b0c941e8ff995111b9f616c8aae23e7daba9f3a610813aade528
-  data.tar.gz: b808c05aeedea83faf728feded28a38a78ac7a1d2ff139ea36cb0f474f310228792f926a97160df637f9d85dca83272921ed9dff72a632c4aab575163643610d
+  metadata.gz: 85684d79eac1479bac56995cd00a4d22106e5247979e2b54b07dee0d2b1948c4e86bf005be674b57656c0a9c14acf731a11d17163edde72d0302561a2d489159
+  data.tar.gz: 7033b279730614870ff8710e51b29ff20f2b399aff9cd1b13a8a2a3202940a88221d9f0edc5dc82a5c3593c9f2314fdc6082fd8c89673f8ec956ba9eada11e2a

data/CHANGELOG.md

@@ -1,3 +1,7 @@
+## [0.6.0] - 2025-12-09
+### Fixed
+
+- visit nodes multiple times
 ## [0.5.0] - 2025-12-08
 
 ### Added

data/lib/gtfs_df/feed.rb

@@ -157,59 +157,76 @@ module GtfsDf
     # Traverses the grah to prune unreferenced entities from child dataframes
     # based on parent relationships. See GtfsDf::Graph::STOP_NODES
     def prune!(root, filtered, filter_only_children: false)
+      seen_edges = Set.new
       maybe_digraph = filter_only_children ? graph : graph.to_undirected
-      maybe_digraph.each_bfs_edge(root) do |parent_node_id, child_node_id|
-        parent_node = Graph::NODES[parent_node_id]
-        child_node = Graph::NODES[child_node_id]
-        parent_df = filtered[parent_node.fetch(:file)]
-        next unless parent_df
-
-        child_df = filtered[child_node.fetch(:file)]
-        # Certain nodes are pre-filtered because they reference only
-        # a piece of the dataframe
-        filter_attrs = child_node[:filter_attrs]
-        if filter_attrs && child_df.columns.include?(filter_attrs.fetch(:filter_col))
-          filter = filter_attrs.fetch(:filter)
-          # Temporarily remove rows that do not match node filter criteria to process them
-          # separately (e.g., when filtering stops, parent stations that should be preserved
-          # regardless of direct references)
-          saved_vals = child_df.filter(filter.is_not)
-          child_df = child_df.filter(filter)
-        end
-        next unless child_df && child_df.height > 0
-
-        attrs = maybe_digraph.get_edge_data(parent_node_id, child_node_id)
-
-        attrs[:dependencies].each do |dep|
-          parent_col = dep[parent_node_id]
-          child_col = dep[child_node_id]
-          allow_null = !!dep[:allow_null]
-
-          next unless parent_col && child_col &&
-            parent_df.columns.include?(parent_col) && child_df.columns.include?(child_col)
-
-          # Get valid values from parent
-          valid_values = parent_df[parent_col].to_a.uniq.compact
 
-          # Filter child to only include rows that reference valid parent values
-          before = child_df.height
-          filter = Polars.col(child_col).is_in(valid_values)
-          if allow_null
-            filter = (filter | Polars.col(child_col).is_null)
+      queue = [root]
+
+      while queue.length > 0
+        parent_node_id = queue.shift
+        maybe_digraph.adj[parent_node_id].each do |child_node_id, attrs|
+          edge = edge_id(parent_node_id, child_node_id)
+
+          next if seen_edges.include?(edge)
+          seen_edges.add(edge)
+
+          parent_node = Graph::NODES[parent_node_id]
+          child_node = Graph::NODES[child_node_id]
+          parent_df = filtered[parent_node.fetch(:file)]
+          next unless parent_df
+
+          child_df = filtered[child_node.fetch(:file)]
+          # Certain nodes are pre-filtered because they reference only
+          # a piece of the dataframe
+          filter_attrs = child_node[:filter_attrs]
+          if filter_attrs && child_df.columns.include?(filter_attrs.fetch(:filter_col))
+            filter = filter_attrs.fetch(:filter)
+            # Temporarily remove rows that do not match node filter criteria to process them
+            # separately (e.g., when filtering stops, parent stations that should be preserved
+            # regardless of direct references)
+            saved_vals = child_df.filter(filter.is_not)
+            child_df = child_df.filter(filter)
           end
-          child_df = child_df.filter(filter)
-          changed = child_df.height < before
-
-          # If we removed a part of the child_df earlier, concat it back on
-          if saved_vals
-            child_df = Polars.concat([child_df, saved_vals], how: "vertical")
-          end
-
-          if changed
-            filtered[child_node.fetch(:file)] = child_df
+          next unless child_df && child_df.height > 0
+
+          queue << child_node_id
+
+          attrs[:dependencies].each do |dep|
+            parent_col = dep[parent_node_id]
+            child_col = dep[child_node_id]
+            allow_null = !!dep[:allow_null]
+
+            next unless parent_col && child_col &&
+              parent_df.columns.include?(parent_col) && child_df.columns.include?(child_col)
+
+            # Get valid values from parent
+            valid_values = parent_df[parent_col].to_a.uniq.compact
+
+            # Filter child to only include rows that reference valid parent values
+            before = child_df.height
+            filter = Polars.col(child_col).is_in(valid_values)
+            if allow_null
+              filter = (filter | Polars.col(child_col).is_null)
+            end
+            child_df = child_df.filter(filter)
+            changed = child_df.height < before
+
+            # If we removed a part of the child_df earlier, concat it back on
+            if saved_vals
+              child_df = Polars.concat([child_df, saved_vals], how: "vertical")
+            end
+
+            if changed
+              filtered[child_node.fetch(:file)] = child_df
+            end
           end
         end
       end
     end
+
+    def edge_id(parent, child)
+      # Alphabetize to make sure this works with undirected graph
+      [parent, child].sort.join("-")
+    end
   end
 end

data/lib/gtfs_df/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module GtfsDf
-  VERSION = "0.5.0"
+  VERSION = "0.6.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: gtfs_df
 version: !ruby/object:Gem::Version
-  version: 0.5.0
+  version: 0.6.0
 platform: ruby
 authors:
 - David Mejorado