From 2c38e7c9611db018a0e7fd61be33d1d027e83545 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 13:37:35 +0200
Subject: [PATCH 01/16] Add number of root nodes to graph storage statistics

Default to "1" for already imported corpora.
---
 CHANGELOG.md                             |  1 +
 core/src/graph/storage/adjacencylist.rs  |  2 ++
 core/src/graph/storage/disk_adjacency.rs |  2 ++
 core/src/graph/storage/mod.rs            | 12 ++++++++++--
 4 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index bbdc41385..7497f2ac4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -12,6 +12,7 @@ configuration. Can be used to additionally output all log messages to the given
 file.
 - `Graph:ensure_loaded_parallel` returns the actually loaded components that did
 exist.
+- Add number of root nodes to graph storage statistics.
 
 ### Fixed
 
diff --git a/core/src/graph/storage/adjacencylist.rs b/core/src/graph/storage/adjacencylist.rs
index 9a105acc9..6972c33a2 100644
--- a/core/src/graph/storage/adjacencylist.rs
+++ b/core/src/graph/storage/adjacencylist.rs
@@ -334,6 +334,7 @@ impl WriteableGraphStorage for AdjacencyListStorage {
             cyclic: false,
             rooted_tree: true,
             nodes: 0,
+            root_nodes: 0,
             dfs_visit_ratio: 0.0,
         };
 
@@ -370,6 +371,7 @@ impl WriteableGraphStorage for AdjacencyListStorage {
                 }
             }
         }
+        stats.root_nodes = roots.len();
 
         let fan_outs = get_fan_outs(&self.edges);
         let sum_fan_out: usize = fan_outs.iter().sum();
diff --git a/core/src/graph/storage/disk_adjacency.rs b/core/src/graph/storage/disk_adjacency.rs
index d86fcee22..b61a22a26 100644
--- a/core/src/graph/storage/disk_adjacency.rs
+++ b/core/src/graph/storage/disk_adjacency.rs
@@ -368,6 +368,7 @@ impl WriteableGraphStorage for DiskAdjacencyListStorage {
             cyclic: false,
             rooted_tree: true,
             nodes: 0,
+            root_nodes: 0,
             dfs_visit_ratio: 0.0,
         };
 
@@ -404,6 +405,7 @@ impl WriteableGraphStorage for DiskAdjacencyListStorage {
                 roots.remove(&e.target);
             }
         }
+        stats.root_nodes = roots.len();
 
         let fan_outs = get_fan_outs(&self.edges)?;
         let sum_fan_out: usize = fan_outs.iter().sum();
diff --git a/core/src/graph/storage/mod.rs b/core/src/graph/storage/mod.rs
index f9a426e33..f2c885398 100644
--- a/core/src/graph/storage/mod.rs
+++ b/core/src/graph/storage/mod.rs
@@ -29,6 +29,10 @@ pub struct GraphStatistic {
     /// Number of nodes in this graph storage (both source and target nodes).
     pub nodes: usize,
 
+    /// Number of root nodes in this graph storage.
+    #[serde(default = "default_number_root_nodes")]
+    pub root_nodes: usize,
+
     /// Average fan out.
     pub avg_fan_out: f64,
     /// Max fan-out of 99% of the data.
@@ -46,12 +50,16 @@ pub struct GraphStatistic {
     pub dfs_visit_ratio: f64,
 }
 
+fn default_number_root_nodes() -> usize {
+    1
+}
+
 impl std::fmt::Display for GraphStatistic {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(
             f,
-            "nodes={}, avg_fan_out={:.2}, max_fan_out={}, max_depth={}",
-            self.nodes, self.avg_fan_out, self.max_fan_out, self.max_depth
+            "nodes={}, avg_fan_out={:.2}, max_fan_out={}, fan_out_99%={}, inv_fan_out_99%={}, max_depth={}",
+            self.nodes, self.avg_fan_out, self.max_fan_out, self.fan_out_99_percentile, self.inverse_fan_out_99_percentile, self.max_depth
         )?;
         if self.cyclic {
             write!(f, ", cyclic")?;

From bfb66e14eb537ce4892d21de167b13bfc7432e4e Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 14:50:01 +0200
Subject: [PATCH 02/16] Adjust disk based graph storage to serialize their
 graph statistics as TOML file

---
 core/src/graph/storage/disk_adjacency.rs | 30 ++++++++-----
 core/src/graph/storage/disk_path.rs      | 30 ++++++++-----
 core/src/graph/storage/legacy.rs         | 54 ++++++++++++++++++++++++
 core/src/graph/storage/mod.rs            |  2 +
 4 files changed, 96 insertions(+), 20 deletions(-)
 create mode 100644 core/src/graph/storage/legacy.rs

diff --git a/core/src/graph/storage/disk_adjacency.rs b/core/src/graph/storage/disk_adjacency.rs
index b61a22a26..65846aab9 100644
--- a/core/src/graph/storage/disk_adjacency.rs
+++ b/core/src/graph/storage/disk_adjacency.rs
@@ -13,6 +13,7 @@ use std::ops::Bound;
 use transient_btree_index::BtreeConfig;
 
 pub const SERIALIZATION_ID: &str = "DiskAdjacencyListV1";
+const STATISTICS_FILE_NAME: &str = "edge_stats.toml";
 
 pub struct DiskAdjacencyListStorage {
     edges: DiskMap<Edge, bool>,
@@ -143,11 +144,22 @@ impl GraphStorage for DiskAdjacencyListStorage {
     where
         Self: std::marker::Sized,
     {
-        // Read stats
-        let stats_path = location.join("edge_stats.bin");
-        let f_stats = std::fs::File::open(stats_path)?;
-        let input = std::io::BufReader::new(f_stats);
-        let stats = bincode::deserialize_from(input)?;
+        // Read stats from file.
+        let stats_path_toml = location.join(STATISTICS_FILE_NAME);
+        let legacy_stats_path_bin = location.join("edge_stats.bin");
+
+        let stats = if stats_path_toml.is_file() {
+            let file_content = std::fs::read_to_string(stats_path_toml)?;
+            toml::from_str(&file_content)?
+        } else if legacy_stats_path_bin.is_file() {
+            let f_stats = std::fs::File::open(legacy_stats_path_bin)?;
+            let input = std::io::BufReader::new(f_stats);
+            // This is a legacy file which needs an older version of the struct
+            let legacy_stats: Option<legacy::GraphStatisticV1> = bincode::deserialize_from(input)?;
+            legacy_stats.map(|s| s.into())
+        } else {
+            None
+        };
 
         let result = DiskAdjacencyListStorage {
             edges: DiskMap::new(
@@ -179,11 +191,9 @@ impl GraphStorage for DiskAdjacencyListStorage {
         self.inverse_edges
             .write_to(&location.join("inverse_edges.bin"))?;
         self.annos.save_annotations_to(location)?;
-        // Write stats with bincode
-        let stats_path = location.join("edge_stats.bin");
-        let f_stats = std::fs::File::create(stats_path)?;
-        let mut writer = std::io::BufWriter::new(f_stats);
-        bincode::serialize_into(&mut writer, &self.stats)?;
+        // Write stats as TOML file
+        let file_content = toml::to_string(&self.stats)?;
+        std::fs::write(location.join(STATISTICS_FILE_NAME), file_content)?;
 
         Ok(())
     }
diff --git a/core/src/graph/storage/disk_path.rs b/core/src/graph/storage/disk_path.rs
index 04292c522..11457974c 100644
--- a/core/src/graph/storage/disk_path.rs
+++ b/core/src/graph/storage/disk_path.rs
@@ -16,12 +16,13 @@ use crate::{
     util::disk_collections::{DiskMap, EvictionStrategy, DEFAULT_BLOCK_CACHE_CAPACITY},
 };
 
-use super::{EdgeContainer, GraphStatistic, GraphStorage};
+use super::{legacy, EdgeContainer, GraphStatistic, GraphStorage};
 use binary_layout::prelude::*;
 
 pub(crate) const MAX_DEPTH: usize = 15;
 pub(crate) const SERIALIZATION_ID: &str = "DiskPathV1_D15";
 const ENTRY_SIZE: usize = (MAX_DEPTH * 8) + 1;
+const STATISTICS_FILE_NAME: &str = "edge_stats.toml";
 
 binary_layout!(node_path, LittleEndian, {
     length: u8,
@@ -372,10 +373,21 @@ impl GraphStorage for DiskPathStorage {
         ))?;
 
         // Read stats
-        let stats_path = location.join("edge_stats.bin");
-        let f_stats = std::fs::File::open(stats_path)?;
-        let input = std::io::BufReader::new(f_stats);
-        let stats = bincode::deserialize_from(input)?;
+        let stats_path_toml = location.join(STATISTICS_FILE_NAME);
+        let legacy_stats_path_bin = location.join("edge_stats.bin");
+
+        let stats = if stats_path_toml.is_file() {
+            let file_content = std::fs::read_to_string(stats_path_toml)?;
+            toml::from_str(&file_content)?
+        } else if legacy_stats_path_bin.is_file() {
+            let f_stats = std::fs::File::open(legacy_stats_path_bin)?;
+            let input = std::io::BufReader::new(f_stats);
+            // This is a legacy file which needs an older version of the struct
+            let legacy_stats: Option<legacy::GraphStatisticV1> = bincode::deserialize_from(input)?;
+            legacy_stats.map(|s| s.into())
+        } else {
+            None
+        };
 
         Ok(Self {
             paths,
@@ -413,11 +425,9 @@ impl GraphStorage for DiskPathStorage {
         // Save edge annotations
         self.annos.save_annotations_to(location)?;
 
-        // Write stats with bincode
-        let stats_path = location.join("edge_stats.bin");
-        let f_stats = std::fs::File::create(stats_path)?;
-        let mut writer = std::io::BufWriter::new(f_stats);
-        bincode::serialize_into(&mut writer, &self.stats)?;
+        // Write stats as TOML file
+        let file_content = toml::to_string(&self.stats)?;
+        std::fs::write(location.join(STATISTICS_FILE_NAME), file_content)?;
 
         Ok(())
     }
diff --git a/core/src/graph/storage/legacy.rs b/core/src/graph/storage/legacy.rs
new file mode 100644
index 000000000..3340b8a6d
--- /dev/null
+++ b/core/src/graph/storage/legacy.rs
@@ -0,0 +1,54 @@
+//! Legacy structures of graph storages. Old versions of graph storages need to
+//! be kept for compatibility reasons, but are not further developed. If
+//! possible, only the legacy data structure is kept, the graph storage is
+//! converted into a newer version and there is no specific implementation for
+//! the old data structure.
+
+use super::GraphStatistic;
+
+/// Some general statistical numbers specific to a graph component
+#[derive(Serialize, Deserialize, Clone)]
+pub(crate) struct GraphStatisticV1 {
+    /// True if the component contains any cycle.
+    pub cyclic: bool,
+
+    /// True if the component consists of [rooted trees](https://en.wikipedia.org/wiki/Tree_(graph_theory)).
+    pub rooted_tree: bool,
+
+    /// Number of nodes in this graph storage (both source and target nodes).
+    pub nodes: usize,
+
+    /// Average fan out.
+    pub avg_fan_out: f64,
+    /// Max fan-out of 99% of the data.
+    pub fan_out_99_percentile: usize,
+
+    /// Max inverse fan-out of 99% of the data.
+    pub inverse_fan_out_99_percentile: usize,
+
+    /// Maximal number of children of a node.
+    pub max_fan_out: usize,
+    /// Maximum length from a root node to a terminal node.
+    pub max_depth: usize,
+
+    /// Only valid for acyclic graphs: the average number of times a DFS will visit each node.
+    pub dfs_visit_ratio: f64,
+}
+
+impl From<GraphStatisticV1> for GraphStatistic {
+    fn from(value: GraphStatisticV1) -> Self {
+        let root_nodes = if value.nodes > 0 { 1 } else { 0 };
+        Self {
+            cyclic: value.cyclic,
+            rooted_tree: value.rooted_tree,
+            nodes: value.nodes,
+            root_nodes,
+            avg_fan_out: value.avg_fan_out,
+            fan_out_99_percentile: value.fan_out_99_percentile,
+            inverse_fan_out_99_percentile: value.inverse_fan_out_99_percentile,
+            max_fan_out: value.max_fan_out,
+            max_depth: value.max_depth,
+            dfs_visit_ratio: value.dfs_visit_ratio,
+        }
+    }
+}
diff --git a/core/src/graph/storage/mod.rs b/core/src/graph/storage/mod.rs
index f2c885398..0e2de8b48 100644
--- a/core/src/graph/storage/mod.rs
+++ b/core/src/graph/storage/mod.rs
@@ -7,6 +7,8 @@ pub mod prepost;
 pub mod registry;
 pub mod union;
 
+pub(crate) mod legacy;
+
 use crate::annostorage::{EdgeAnnotationStorage, NodeAnnotationStorage};
 use crate::{
     annostorage::AnnotationStorage,

From 686e4e29c5f7cd0f96915f07ae6967cd3cd668d0 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 16:05:52 +0200
Subject: [PATCH 03/16] Serialize all fields of the graph storages separate

Also add a fallback to deserialize the old component.bin with the legacy graph statistic field
---
 .../snapshots/cli__show_corpus_info.snap      | 20 +++----
 core/src/graph/storage/adjacencylist.rs       | 43 +++++++++++++--
 core/src/graph/storage/dense_adjacency.rs     | 31 +++++++++--
 core/src/graph/storage/disk_adjacency.rs      | 24 +--------
 core/src/graph/storage/disk_path.rs           | 27 +++-------
 core/src/graph/storage/legacy.rs              | 52 ++++++++++++++++++-
 core/src/graph/storage/linear.rs              | 33 ++++++++++--
 core/src/graph/storage/mod.rs                 | 50 +++++++++++++++---
 core/src/graph/storage/prepost.rs             | 34 ++++++++++--
 9 files changed, 238 insertions(+), 76 deletions(-)

diff --git a/cli/tests/snapshots/cli__show_corpus_info.snap b/cli/tests/snapshots/cli__show_corpus_info.snap
index e99870165..c663216a9 100644
--- a/cli/tests/snapshots/cli__show_corpus_info.snap
+++ b/cli/tests/snapshots/cli__show_corpus_info.snap
@@ -21,52 +21,52 @@ Status: "fully loaded"
 Token search shortcut possible: true
 ------------
 Component Coverage//: 0 annnotations
-Stats: nodes=92, avg_fan_out=2.17, max_fan_out=11, max_depth=1
+Stats: nodes=92, avg_fan_out=2.17, max_fan_out=11, fan_out_99%=11, inv_fan_out_99%=9, max_depth=1
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/annis/: 0 annnotations
-Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, max_depth=1, tree
+Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/default_ns/: 0 annnotations
-Stats: nodes=56, avg_fan_out=0.93, max_fan_out=10, max_depth=1
+Stats: nodes=56, avg_fan_out=0.93, max_fan_out=10, fan_out_99%=10, inv_fan_out_99%=2, max_depth=1
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/annis/inherited-coverage: 0 annnotations
-Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, max_depth=1, tree
+Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Dominance/syntax/: 0 annnotations
-Stats: nodes=92, avg_fan_out=0.96, max_fan_out=3, max_depth=9, tree
+Stats: nodes=92, avg_fan_out=0.96, max_fan_out=3, fan_out_99%=3, inv_fan_out_99%=1, max_depth=9, tree
 Implementation: PrePostOrderO16L8V1
 Status: "fully loaded"
 ------------
 Component Pointing/default_ns/anaphoric: 0 annnotations
-Stats: nodes=8, avg_fan_out=0.50, max_fan_out=1, max_depth=1, tree
+Stats: nodes=8, avg_fan_out=0.50, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=1, tree
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component Ordering/annis/: 0 annnotations
-Stats: nodes=44, avg_fan_out=0.91, max_fan_out=1, max_depth=10, tree
+Stats: nodes=44, avg_fan_out=0.91, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=10, tree
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component LeftToken/annis/: 0 annnotations
-Stats: nodes=92, avg_fan_out=0.65, max_fan_out=1, max_depth=1
+Stats: nodes=92, avg_fan_out=0.65, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=3, max_depth=1
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component RightToken/annis/: 0 annnotations
-Stats: nodes=84, avg_fan_out=0.71, max_fan_out=1, max_depth=1
+Stats: nodes=84, avg_fan_out=0.71, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=8, max_depth=1
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component PartOf/annis/: 0 annnotations
-Stats: nodes=115, avg_fan_out=0.99, max_fan_out=1, max_depth=4
+Stats: nodes=115, avg_fan_out=0.99, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=26, max_depth=4
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
diff --git a/core/src/graph/storage/adjacencylist.rs b/core/src/graph/storage/adjacencylist.rs
index 6972c33a2..49c8a798d 100644
--- a/core/src/graph/storage/adjacencylist.rs
+++ b/core/src/graph/storage/adjacencylist.rs
@@ -7,7 +7,12 @@ use crate::{
     types::{AnnoKey, Annotation, Edge, NodeID},
 };
 
-use super::{EdgeContainer, GraphStatistic, GraphStorage, WriteableGraphStorage};
+use super::{
+    deserialize_gs_field,
+    legacy::{self, AdjacencyListStorageV1},
+    load_statistics_from_location, save_statistics_to_toml, serialize_gs_field, EdgeContainer,
+    GraphStatistic, GraphStorage, WriteableGraphStorage,
+};
 use itertools::Itertools;
 use rustc_hash::FxHashSet;
 use serde::Deserialize;
@@ -123,13 +128,34 @@ impl GraphStorage for AdjacencyListStorage {
     where
         for<'de> Self: std::marker::Sized + Deserialize<'de>,
     {
-        let mut result: Self = super::default_deserialize_gs(location)?;
+        let legacy_path = location.join("component.bin");
+        let mut result: Self = if legacy_path.is_file() {
+            let component: AdjacencyListStorageV1 = deserialize_gs_field(location, "component")?;
+            Self {
+                stats: component.stats.map(|s| GraphStatistic::from(s)),
+                edges: component.edges,
+                inverse_edges: component.inverse_edges,
+                annos: component.annos,
+            }
+        } else {
+            let stats = load_statistics_from_location(location)?;
+            Self {
+                edges: deserialize_gs_field(location, "edges")?,
+                inverse_edges: deserialize_gs_field(location, "inverse_edges")?,
+                annos: deserialize_gs_field(location, "annos")?,
+                stats: stats,
+            }
+        };
+
         result.annos.after_deserialization();
         Ok(result)
     }
 
     fn save_to(&self, location: &Path) -> Result<()> {
-        super::default_serialize_gs(self, location)?;
+        serialize_gs_field(&self.edges, "edges", location)?;
+        serialize_gs_field(&self.inverse_edges, "inverse_edges", location)?;
+        serialize_gs_field(&self.annos, "annos", location)?;
+        save_statistics_to_toml(location, self.stats.as_ref())?;
         Ok(())
     }
 
@@ -448,5 +474,16 @@ impl WriteableGraphStorage for AdjacencyListStorage {
     }
 }
 
+impl From<legacy::AdjacencyListStorageV1> for AdjacencyListStorage {
+    fn from(value: legacy::AdjacencyListStorageV1) -> Self {
+        Self {
+            edges: value.edges,
+            inverse_edges: value.inverse_edges,
+            annos: value.annos,
+            stats: value.stats.map(|s| GraphStatistic::from(s)),
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests;
diff --git a/core/src/graph/storage/dense_adjacency.rs b/core/src/graph/storage/dense_adjacency.rs
index 1f5d6355e..c48b55026 100644
--- a/core/src/graph/storage/dense_adjacency.rs
+++ b/core/src/graph/storage/dense_adjacency.rs
@@ -1,4 +1,7 @@
-use super::{EdgeContainer, GraphStatistic, GraphStorage};
+use super::{
+    deserialize_gs_field, legacy::DenseAdjacencyListStorageV1, load_statistics_from_location,
+    save_statistics_to_toml, serialize_gs_field, EdgeContainer, GraphStatistic, GraphStorage,
+};
 use crate::{
     annostorage::{
         inmemory::AnnoStorageImpl, AnnotationStorage, EdgeAnnotationStorage, NodeAnnotationStorage,
@@ -219,13 +222,35 @@ impl GraphStorage for DenseAdjacencyListStorage {
     where
         for<'de> Self: std::marker::Sized + Deserialize<'de>,
     {
-        let mut result: Self = super::default_deserialize_gs(location)?;
+        let legacy_path = location.join("component.bin");
+        let mut result: Self = if legacy_path.is_file() {
+            let component: DenseAdjacencyListStorageV1 =
+                deserialize_gs_field(location, "component")?;
+            Self {
+                edges: component.edges,
+                inverse_edges: component.inverse_edges,
+                annos: component.annos,
+                stats: component.stats.map(|s| GraphStatistic::from(s)),
+            }
+        } else {
+            let stats = load_statistics_from_location(location)?;
+            Self {
+                edges: deserialize_gs_field(location, "edges")?,
+                inverse_edges: deserialize_gs_field(location, "inverse_edges")?,
+                annos: deserialize_gs_field(location, "annos")?,
+                stats: stats,
+            }
+        };
+
         result.annos.after_deserialization();
         Ok(result)
     }
 
     fn save_to(&self, location: &Path) -> Result<()> {
-        super::default_serialize_gs(self, location)?;
+        serialize_gs_field(&self.edges, "edges", location)?;
+        serialize_gs_field(&self.inverse_edges, "inverse_edges", location)?;
+        serialize_gs_field(&self.annos, "annos", location)?;
+        save_statistics_to_toml(location, self.stats.as_ref())?;
         Ok(())
     }
 }
diff --git a/core/src/graph/storage/disk_adjacency.rs b/core/src/graph/storage/disk_adjacency.rs
index 65846aab9..f38b9bdfb 100644
--- a/core/src/graph/storage/disk_adjacency.rs
+++ b/core/src/graph/storage/disk_adjacency.rs
@@ -13,7 +13,6 @@ use std::ops::Bound;
 use transient_btree_index::BtreeConfig;
 
 pub const SERIALIZATION_ID: &str = "DiskAdjacencyListV1";
-const STATISTICS_FILE_NAME: &str = "edge_stats.toml";
 
 pub struct DiskAdjacencyListStorage {
     edges: DiskMap<Edge, bool>,
@@ -144,23 +143,7 @@ impl GraphStorage for DiskAdjacencyListStorage {
     where
         Self: std::marker::Sized,
     {
-        // Read stats from file.
-        let stats_path_toml = location.join(STATISTICS_FILE_NAME);
-        let legacy_stats_path_bin = location.join("edge_stats.bin");
-
-        let stats = if stats_path_toml.is_file() {
-            let file_content = std::fs::read_to_string(stats_path_toml)?;
-            toml::from_str(&file_content)?
-        } else if legacy_stats_path_bin.is_file() {
-            let f_stats = std::fs::File::open(legacy_stats_path_bin)?;
-            let input = std::io::BufReader::new(f_stats);
-            // This is a legacy file which needs an older version of the struct
-            let legacy_stats: Option<legacy::GraphStatisticV1> = bincode::deserialize_from(input)?;
-            legacy_stats.map(|s| s.into())
-        } else {
-            None
-        };
-
+        let stats = load_statistics_from_location(location)?;
         let result = DiskAdjacencyListStorage {
             edges: DiskMap::new(
                 Some(&location.join("edges.bin")),
@@ -191,10 +174,7 @@ impl GraphStorage for DiskAdjacencyListStorage {
         self.inverse_edges
             .write_to(&location.join("inverse_edges.bin"))?;
         self.annos.save_annotations_to(location)?;
-        // Write stats as TOML file
-        let file_content = toml::to_string(&self.stats)?;
-        std::fs::write(location.join(STATISTICS_FILE_NAME), file_content)?;
-
+        save_statistics_to_toml(location, self.stats.as_ref())?;
         Ok(())
     }
 
diff --git a/core/src/graph/storage/disk_path.rs b/core/src/graph/storage/disk_path.rs
index 11457974c..9264bd616 100644
--- a/core/src/graph/storage/disk_path.rs
+++ b/core/src/graph/storage/disk_path.rs
@@ -16,13 +16,15 @@ use crate::{
     util::disk_collections::{DiskMap, EvictionStrategy, DEFAULT_BLOCK_CACHE_CAPACITY},
 };
 
-use super::{legacy, EdgeContainer, GraphStatistic, GraphStorage};
+use super::{
+    load_statistics_from_location, save_statistics_to_toml, EdgeContainer, GraphStatistic,
+    GraphStorage,
+};
 use binary_layout::prelude::*;
 
 pub(crate) const MAX_DEPTH: usize = 15;
 pub(crate) const SERIALIZATION_ID: &str = "DiskPathV1_D15";
 const ENTRY_SIZE: usize = (MAX_DEPTH * 8) + 1;
-const STATISTICS_FILE_NAME: &str = "edge_stats.toml";
 
 binary_layout!(node_path, LittleEndian, {
     length: u8,
@@ -372,22 +374,7 @@ impl GraphStorage for DiskPathStorage {
             location.join(crate::annostorage::ondisk::SUBFOLDER_NAME),
         ))?;
 
-        // Read stats
-        let stats_path_toml = location.join(STATISTICS_FILE_NAME);
-        let legacy_stats_path_bin = location.join("edge_stats.bin");
-
-        let stats = if stats_path_toml.is_file() {
-            let file_content = std::fs::read_to_string(stats_path_toml)?;
-            toml::from_str(&file_content)?
-        } else if legacy_stats_path_bin.is_file() {
-            let f_stats = std::fs::File::open(legacy_stats_path_bin)?;
-            let input = std::io::BufReader::new(f_stats);
-            // This is a legacy file which needs an older version of the struct
-            let legacy_stats: Option<legacy::GraphStatisticV1> = bincode::deserialize_from(input)?;
-            legacy_stats.map(|s| s.into())
-        } else {
-            None
-        };
+        let stats = load_statistics_from_location(location)?;
 
         Ok(Self {
             paths,
@@ -425,9 +412,7 @@ impl GraphStorage for DiskPathStorage {
         // Save edge annotations
         self.annos.save_annotations_to(location)?;
 
-        // Write stats as TOML file
-        let file_content = toml::to_string(&self.stats)?;
-        std::fs::write(location.join(STATISTICS_FILE_NAME), file_content)?;
+        save_statistics_to_toml(location, self.stats.as_ref())?;
 
         Ok(())
     }
diff --git a/core/src/graph/storage/legacy.rs b/core/src/graph/storage/legacy.rs
index 3340b8a6d..ad011bd48 100644
--- a/core/src/graph/storage/legacy.rs
+++ b/core/src/graph/storage/legacy.rs
@@ -4,7 +4,20 @@
 //! converted into a newer version and there is no specific implementation for
 //! the old data structure.
 
-use super::GraphStatistic;
+use std::collections::HashMap;
+
+use rustc_hash::FxHashMap;
+
+use crate::{
+    annostorage::inmemory::AnnoStorageImpl,
+    types::{Edge, NodeID, NumValue},
+};
+
+use super::{
+    linear::RelativePosition,
+    prepost::{OrderVecEntry, PrePost},
+    GraphStatistic,
+};
 
 /// Some general statistical numbers specific to a graph component
 #[derive(Serialize, Deserialize, Clone)]
@@ -52,3 +65,40 @@ impl From<GraphStatisticV1> for GraphStatistic {
         }
     }
 }
+
+/// An adjacency list based storage that uses the [`GraphStatisticV1`]
+#[derive(Serialize, Deserialize, Clone)]
+pub(crate) struct AdjacencyListStorageV1 {
+    pub(crate) edges: HashMap<NodeID, Vec<NodeID>>,
+    pub(crate) inverse_edges: HashMap<NodeID, Vec<NodeID>>,
+    pub(crate) annos: AnnoStorageImpl<Edge>,
+    pub(crate) stats: Option<GraphStatisticV1>,
+}
+
+/// An adjacency list based storage that uses the [`GraphStatisticV1`] and is
+/// optimized for graphs where almost all nodes have an outgoing edge.
+#[derive(Serialize, Deserialize, Clone)]
+pub(crate) struct DenseAdjacencyListStorageV1 {
+    pub(crate) edges: Vec<Option<NodeID>>,
+    pub(crate) inverse_edges: HashMap<NodeID, Vec<NodeID>>,
+    pub(crate) annos: AnnoStorageImpl<Edge>,
+    pub(crate) stats: Option<GraphStatisticV1>,
+}
+
+/// A graph storage for linar graphs that uses the [`GraphStatisticV1`].
+#[derive(Serialize, Deserialize, Clone)]
+pub(crate) struct LinearGraphStorageV1<PosT: NumValue> {
+    pub(crate) node_to_pos: HashMap<NodeID, RelativePosition<PosT>>,
+    pub(crate) node_chains: HashMap<NodeID, Vec<NodeID>>,
+    pub(crate) annos: AnnoStorageImpl<Edge>,
+    pub(crate) stats: Option<GraphStatisticV1>,
+}
+
+/// A graph storage for trees that uses the [`GraphStatisticV1`] and indexes graphs using the pre/post order.
+#[derive(Serialize, Deserialize, Clone)]
+pub(crate) struct PrePostOrderStorageV1<OrderT: NumValue, LevelT: NumValue> {
+    pub(crate) node_to_order: FxHashMap<NodeID, Vec<PrePost<OrderT, LevelT>>>,
+    pub(crate) order_to_node: Vec<OrderVecEntry<OrderT, LevelT>>,
+    pub(crate) annos: AnnoStorageImpl<Edge>,
+    pub(crate) stats: Option<GraphStatisticV1>,
+}
diff --git a/core/src/graph/storage/linear.rs b/core/src/graph/storage/linear.rs
index f3a9b155d..4b6776c1b 100644
--- a/core/src/graph/storage/linear.rs
+++ b/core/src/graph/storage/linear.rs
@@ -1,4 +1,7 @@
-use super::{EdgeContainer, GraphStatistic, GraphStorage};
+use super::{
+    deserialize_gs_field, legacy::LinearGraphStorageV1, load_statistics_from_location,
+    save_statistics_to_toml, serialize_gs_field, EdgeContainer, GraphStatistic, GraphStorage,
+};
 use crate::{
     annostorage::{
         inmemory::AnnoStorageImpl, AnnotationStorage, EdgeAnnotationStorage, NodeAnnotationStorage,
@@ -12,7 +15,7 @@ use serde::{Deserialize, Serialize};
 use std::{clone::Clone, collections::HashMap, path::Path};
 
 #[derive(Serialize, Deserialize, Clone)]
-struct RelativePosition<PosT> {
+pub(crate) struct RelativePosition<PosT> {
     pub root: NodeID,
     pub pos: PosT,
 }
@@ -165,13 +168,35 @@ where
     where
         for<'de> Self: std::marker::Sized + Deserialize<'de>,
     {
-        let mut result: Self = super::default_deserialize_gs(location)?;
+        let legacy_path = location.join("component.bin");
+        let mut result: Self = if legacy_path.is_file() {
+            let component: LinearGraphStorageV1<PosT> =
+                deserialize_gs_field(location, "component")?;
+            Self {
+                node_to_pos: component.node_to_pos,
+                node_chains: component.node_chains,
+                annos: component.annos,
+                stats: component.stats.map(|s| GraphStatistic::from(s)),
+            }
+        } else {
+            let stats = load_statistics_from_location(location)?;
+            Self {
+                node_to_pos: deserialize_gs_field(location, "node_to_pos")?,
+                node_chains: deserialize_gs_field(location, "node_chains")?,
+                annos: deserialize_gs_field(location, "annos")?,
+                stats: stats,
+            }
+        };
+
         result.annos.after_deserialization();
         Ok(result)
     }
 
     fn save_to(&self, location: &Path) -> Result<()> {
-        super::default_serialize_gs(self, location)?;
+        serialize_gs_field(&self.node_to_pos, "node_to_pos", location)?;
+        serialize_gs_field(&self.node_chains, "node_chains", location)?;
+        serialize_gs_field(&self.annos, "annos", location)?;
+        save_statistics_to_toml(location, self.stats.as_ref())?;
         Ok(())
     }
 
diff --git a/core/src/graph/storage/mod.rs b/core/src/graph/storage/mod.rs
index 0e2de8b48..04aefdd27 100644
--- a/core/src/graph/storage/mod.rs
+++ b/core/src/graph/storage/mod.rs
@@ -200,30 +200,64 @@ pub trait GraphStorage: EdgeContainer {
     fn save_to(&self, location: &Path) -> Result<()>;
 }
 
-pub fn default_serialize_gs<GS>(gs: &GS, location: &Path) -> Result<()>
+pub fn serialize_gs_field<T>(field: &T, field_name: &str, location: &Path) -> Result<()>
 where
-    GS: Serialize,
+    T: Serialize,
 {
-    let data_path = location.join("component.bin");
+    let data_path = location.join(format!("{field_name}.bin"));
     let f_data = std::fs::File::create(data_path)?;
     let mut writer = std::io::BufWriter::new(f_data);
-    bincode::serialize_into(&mut writer, gs)?;
+    bincode::serialize_into(&mut writer, field)?;
     Ok(())
 }
 
-pub fn default_deserialize_gs<GS>(location: &Path) -> Result<GS>
+pub fn deserialize_gs_field<T>(location: &Path, field_name: &str) -> Result<T>
 where
-    for<'de> GS: std::marker::Sized + Deserialize<'de>,
+    for<'de> T: std::marker::Sized + Deserialize<'de>,
 {
-    let data_path = location.join("component.bin");
+    let data_path = location.join(format!("{field_name}.bin"));
     let f_data = std::fs::File::open(data_path)?;
     let input = std::io::BufReader::new(f_data);
 
     let result = bincode::deserialize_from(input)?;
-
     Ok(result)
 }
 
+const STATISTICS_FILE_NAME: &str = "stats.toml";
+
+pub fn load_statistics_from_location(location: &Path) -> Result<Option<GraphStatistic>> {
+    let stats_path_toml = location.join(STATISTICS_FILE_NAME);
+    let legacy_stats_path_bin = location.join("edge_stats.bin");
+
+    let stats = if stats_path_toml.is_file() {
+        let file_content = std::fs::read_to_string(stats_path_toml)?;
+        let stats: GraphStatistic = toml::from_str(&file_content)?;
+        Some(stats)
+    } else if legacy_stats_path_bin.is_file() {
+        let f_stats = std::fs::File::open(legacy_stats_path_bin)?;
+        let input = std::io::BufReader::new(f_stats);
+        // This is a legacy file which needs an older version of the struct
+        let legacy_stats: Option<legacy::GraphStatisticV1> = bincode::deserialize_from(input)?;
+        legacy_stats.map(|s| s.into())
+    } else {
+        None
+    };
+    Ok(stats)
+}
+
+pub fn save_statistics_to_toml(location: &Path, stats: Option<&GraphStatistic>) -> Result<()> {
+    let file_path = location.join(STATISTICS_FILE_NAME);
+    if file_path.is_file() {
+        std::fs::remove_file(&file_path)?;
+    }
+
+    if let Some(stats) = stats {
+        let file_content = toml::to_string(stats)?;
+        std::fs::write(file_path, file_content)?;
+    }
+    Ok(())
+}
+
 /// Trait for accessing graph storages which can be written to.
 pub trait WriteableGraphStorage: GraphStorage {
     /// Add an edge to this graph storage.
diff --git a/core/src/graph/storage/prepost.rs b/core/src/graph/storage/prepost.rs
index 3dd22cb81..0550dac34 100644
--- a/core/src/graph/storage/prepost.rs
+++ b/core/src/graph/storage/prepost.rs
@@ -1,4 +1,7 @@
-use super::{EdgeContainer, GraphStatistic, GraphStorage};
+use super::{
+    deserialize_gs_field, legacy::PrePostOrderStorageV1, load_statistics_from_location,
+    save_statistics_to_toml, serialize_gs_field, EdgeContainer, GraphStatistic, GraphStorage,
+};
 use crate::{
     annostorage::{
         inmemory::AnnoStorageImpl, AnnotationStorage, EdgeAnnotationStorage, NodeAnnotationStorage,
@@ -20,7 +23,7 @@ pub struct PrePost<OrderT, LevelT> {
 }
 
 #[derive(Serialize, Deserialize, Clone, Debug)]
-enum OrderVecEntry<OrderT, LevelT> {
+pub(crate) enum OrderVecEntry<OrderT, LevelT> {
     None,
     Pre {
         post: OrderT,
@@ -197,13 +200,36 @@ where
     where
         for<'de> Self: std::marker::Sized + Deserialize<'de>,
     {
-        let mut result: Self = super::default_deserialize_gs(location)?;
+        let legacy_path = location.join("component.bin");
+        let mut result: Self = if legacy_path.is_file() {
+            let component: PrePostOrderStorageV1<OrderT, LevelT> =
+                deserialize_gs_field(location, "component")?;
+            Self {
+                node_to_order: component.node_to_order,
+                order_to_node: component.order_to_node,
+                annos: component.annos,
+                stats: component.stats.map(|s| GraphStatistic::from(s)),
+            }
+        } else {
+            let stats = load_statistics_from_location(location)?;
+            Self {
+                node_to_order: deserialize_gs_field(location, "node_to_order")?,
+                order_to_node: deserialize_gs_field(location, "order_to_node")?,
+                annos: deserialize_gs_field(location, "annos")?,
+                stats: stats,
+            }
+        };
+
         result.annos.after_deserialization();
+
         Ok(result)
     }
 
     fn save_to(&self, location: &Path) -> Result<()> {
-        super::default_serialize_gs(self, location)?;
+        serialize_gs_field(&self.node_to_order, "node_to_order", location)?;
+        serialize_gs_field(&self.order_to_node, "order_to_node", location)?;
+        serialize_gs_field(&self.annos, "annos", location)?;
+        save_statistics_to_toml(location, self.stats.as_ref())?;
         Ok(())
     }
 

From 0b6d18098e90b6e5c83f38a7f34cf71d2f6f734e Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 16:57:41 +0200
Subject: [PATCH 04/16] Use "corpus" as node estimation for PartOf component

---
 core/src/graph/storage/mod.rs                    |  4 ++--
 graphannis/src/annis/db/aql/operators/edge_op.rs | 13 +++++++++++--
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/core/src/graph/storage/mod.rs b/core/src/graph/storage/mod.rs
index 04aefdd27..3bb627af8 100644
--- a/core/src/graph/storage/mod.rs
+++ b/core/src/graph/storage/mod.rs
@@ -60,8 +60,8 @@ impl std::fmt::Display for GraphStatistic {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(
             f,
-            "nodes={}, avg_fan_out={:.2}, max_fan_out={}, fan_out_99%={}, inv_fan_out_99%={}, max_depth={}",
-            self.nodes, self.avg_fan_out, self.max_fan_out, self.fan_out_99_percentile, self.inverse_fan_out_99_percentile, self.max_depth
+            "nodes={}, root nodes={}, avg_fan_out={:.2}, max_fan_out={}, fan_out_99%={}, inv_fan_out_99%={}, max_depth={}",
+            self.nodes, self.root_nodes, self.avg_fan_out, self.max_fan_out, self.fan_out_99_percentile, self.inverse_fan_out_99_percentile, self.max_depth
         )?;
         if self.cyclic {
             write!(f, ", cyclic")?;
diff --git a/graphannis/src/annis/db/aql/operators/edge_op.rs b/graphannis/src/annis/db/aql/operators/edge_op.rs
index 3c56d0a60..dd9473cbb 100644
--- a/graphannis/src/annis/db/aql/operators/edge_op.rs
+++ b/graphannis/src/annis/db/aql/operators/edge_op.rs
@@ -43,14 +43,23 @@ impl BaseEdgeOp {
             })?;
             gs.push(gs_for_component);
         }
+        let all_part_of_components = spec
+            .components
+            .iter()
+            .all(|c| c.get_type() == AnnotationComponentType::PartOf);
+        let node_type = if all_part_of_components {
+            "corpus"
+        } else {
+            "node"
+        };
         Ok(BaseEdgeOp {
             gs,
             spec,
             max_nodes_estimate: db.get_node_annos().guess_max_count(
                 Some(&NODE_TYPE_KEY.ns),
                 &NODE_TYPE_KEY.name,
-                "node",
-                "node",
+                node_type,
+                node_type,
             )?,
             inverse: false,
         })

From bfc07784b8f53f155ba120c77cf5360e5eddd3d3 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:03:09 +0200
Subject: [PATCH 05/16] Fix clippy issues

---
 core/src/graph/storage/adjacencylist.rs   | 6 +++---
 core/src/graph/storage/dense_adjacency.rs | 4 ++--
 core/src/graph/storage/linear.rs          | 4 ++--
 core/src/graph/storage/prepost.rs         | 4 ++--
 4 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/core/src/graph/storage/adjacencylist.rs b/core/src/graph/storage/adjacencylist.rs
index 49c8a798d..c70f19f2f 100644
--- a/core/src/graph/storage/adjacencylist.rs
+++ b/core/src/graph/storage/adjacencylist.rs
@@ -132,7 +132,7 @@ impl GraphStorage for AdjacencyListStorage {
         let mut result: Self = if legacy_path.is_file() {
             let component: AdjacencyListStorageV1 = deserialize_gs_field(location, "component")?;
             Self {
-                stats: component.stats.map(|s| GraphStatistic::from(s)),
+                stats: component.stats.map(GraphStatistic::from),
                 edges: component.edges,
                 inverse_edges: component.inverse_edges,
                 annos: component.annos,
@@ -143,7 +143,7 @@ impl GraphStorage for AdjacencyListStorage {
                 edges: deserialize_gs_field(location, "edges")?,
                 inverse_edges: deserialize_gs_field(location, "inverse_edges")?,
                 annos: deserialize_gs_field(location, "annos")?,
-                stats: stats,
+                stats,
             }
         };
 
@@ -480,7 +480,7 @@ impl From<legacy::AdjacencyListStorageV1> for AdjacencyListStorage {
             edges: value.edges,
             inverse_edges: value.inverse_edges,
             annos: value.annos,
-            stats: value.stats.map(|s| GraphStatistic::from(s)),
+            stats: value.stats.map(GraphStatistic::from),
         }
     }
 }
diff --git a/core/src/graph/storage/dense_adjacency.rs b/core/src/graph/storage/dense_adjacency.rs
index c48b55026..3d86425a6 100644
--- a/core/src/graph/storage/dense_adjacency.rs
+++ b/core/src/graph/storage/dense_adjacency.rs
@@ -230,7 +230,7 @@ impl GraphStorage for DenseAdjacencyListStorage {
                 edges: component.edges,
                 inverse_edges: component.inverse_edges,
                 annos: component.annos,
-                stats: component.stats.map(|s| GraphStatistic::from(s)),
+                stats: component.stats.map(GraphStatistic::from),
             }
         } else {
             let stats = load_statistics_from_location(location)?;
@@ -238,7 +238,7 @@ impl GraphStorage for DenseAdjacencyListStorage {
                 edges: deserialize_gs_field(location, "edges")?,
                 inverse_edges: deserialize_gs_field(location, "inverse_edges")?,
                 annos: deserialize_gs_field(location, "annos")?,
-                stats: stats,
+                stats,
             }
         };
 
diff --git a/core/src/graph/storage/linear.rs b/core/src/graph/storage/linear.rs
index 4b6776c1b..d3f037002 100644
--- a/core/src/graph/storage/linear.rs
+++ b/core/src/graph/storage/linear.rs
@@ -176,7 +176,7 @@ where
                 node_to_pos: component.node_to_pos,
                 node_chains: component.node_chains,
                 annos: component.annos,
-                stats: component.stats.map(|s| GraphStatistic::from(s)),
+                stats: component.stats.map(GraphStatistic::from),
             }
         } else {
             let stats = load_statistics_from_location(location)?;
@@ -184,7 +184,7 @@ where
                 node_to_pos: deserialize_gs_field(location, "node_to_pos")?,
                 node_chains: deserialize_gs_field(location, "node_chains")?,
                 annos: deserialize_gs_field(location, "annos")?,
-                stats: stats,
+                stats,
             }
         };
 
diff --git a/core/src/graph/storage/prepost.rs b/core/src/graph/storage/prepost.rs
index 0550dac34..6f3ee559a 100644
--- a/core/src/graph/storage/prepost.rs
+++ b/core/src/graph/storage/prepost.rs
@@ -208,7 +208,7 @@ where
                 node_to_order: component.node_to_order,
                 order_to_node: component.order_to_node,
                 annos: component.annos,
-                stats: component.stats.map(|s| GraphStatistic::from(s)),
+                stats: component.stats.map(GraphStatistic::from),
             }
         } else {
             let stats = load_statistics_from_location(location)?;
@@ -216,7 +216,7 @@ where
                 node_to_order: deserialize_gs_field(location, "node_to_order")?,
                 order_to_node: deserialize_gs_field(location, "order_to_node")?,
                 annos: deserialize_gs_field(location, "annos")?,
-                stats: stats,
+                stats,
             }
         };
 

From 73d81996d7bdcce13e7c631795475cbe7877f48f Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:10:33 +0200
Subject: [PATCH 06/16] Add test corpus that uses the new statistic

---
 cli/tests/cli.rs                              |   2 +-
 .../cli__list_corpora_fully_loaded.snap       |   1 +
 .../cli__list_corpora_not_loaded.snap         |   2 +-
 .../cli__list_corpora_partially_loaded.snap   |   1 +
 .../snapshots/cli__show_corpus_info.snap      |  26 +++---
 .../sample-disk-based-3.8/corpus-config.toml  |  78 ++++++++++++++++++
 .../current/global_statistics.toml            |   6 ++
 .../current/gs/Coverage/annis/edges.bin       | Bin 0 -> 123 bytes
 .../current/gs/Coverage/annis/impl.cfg        |   1 +
 .../annis/inherited-coverage/edges.bin        | Bin 0 -> 123 bytes
 .../annis/inherited-coverage/impl.cfg         |   1 +
 .../inherited-coverage/inverse_edges.bin      | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/by_anno_qname.bin        | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/by_container.bin         | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/custom.bin               | Bin 0 -> 33 bytes
 .../annis/inherited-coverage/stats.toml       |  10 +++
 .../gs/Coverage/annis/inverse_edges.bin       | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_anno_qname.bin  | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_container.bin   | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/custom.bin         | Bin 0 -> 33 bytes
 .../current/gs/Coverage/annis/stats.toml      |  10 +++
 .../gs/Coverage/default_layer/edges.bin       | Bin 0 -> 2211 bytes
 .../gs/Coverage/default_layer/impl.cfg        |   1 +
 .../Coverage/default_layer/inverse_edges.bin  | Bin 0 -> 2163 bytes
 .../nodes_diskmap_v1/by_anno_qname.bin        | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/by_container.bin         | Bin 0 -> 123 bytes
 .../default_layer/nodes_diskmap_v1/custom.bin | Bin 0 -> 33 bytes
 .../gs/Coverage/default_layer/stats.toml      |  10 +++
 .../current/gs/Coverage/default_ns/edges.bin  | Bin 0 -> 694 bytes
 .../current/gs/Coverage/default_ns/impl.cfg   |   1 +
 .../gs/Coverage/default_ns/inverse_edges.bin  | Bin 0 -> 950 bytes
 .../nodes_diskmap_v1/by_anno_qname.bin        | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/by_container.bin         | Bin 0 -> 123 bytes
 .../default_ns/nodes_diskmap_v1/custom.bin    | Bin 0 -> 33 bytes
 .../current/gs/Coverage/default_ns/stats.toml |  10 +++
 .../current/gs/Dominance/syntax/annos.bin     | Bin 0 -> 73 bytes
 .../current/gs/Dominance/syntax/impl.cfg      |   1 +
 .../gs/Dominance/syntax/node_to_order.bin     | Bin 0 -> 1940 bytes
 .../gs/Dominance/syntax/order_to_node.bin     | Bin 0 -> 2768 bytes
 .../current/gs/Dominance/syntax/stats.toml    |  10 +++
 .../current/gs/LeftToken/annis/impl.cfg       |   1 +
 .../gs/LeftToken/annis/inverse_edges.bin      | Bin 0 -> 912 bytes
 .../annis/nodes_diskmap_v1/by_anno_qname.bin  | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_container.bin   | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/custom.bin         | Bin 0 -> 33 bytes
 .../current/gs/LeftToken/annis/paths.bin      | Bin 0 -> 13068 bytes
 .../current/gs/LeftToken/annis/stats.toml     |  10 +++
 .../current/gs/Ordering/annis/impl.cfg        |   1 +
 .../gs/Ordering/annis/inverse_edges.bin       | Bin 0 -> 804 bytes
 .../annis/nodes_diskmap_v1/by_anno_qname.bin  | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_container.bin   | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/custom.bin         | Bin 0 -> 33 bytes
 .../current/gs/Ordering/annis/paths.bin       | Bin 0 -> 10527 bytes
 .../current/gs/Ordering/annis/stats.toml      |  10 +++
 .../current/gs/PartOf/annis/impl.cfg          |   1 +
 .../current/gs/PartOf/annis/inverse_edges.bin | Bin 0 -> 1213 bytes
 .../annis/nodes_diskmap_v1/by_anno_qname.bin  | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_container.bin   | Bin 0 -> 123 bytes
 .../PartOf/annis/nodes_diskmap_v1/custom.bin  | Bin 0 -> 33 bytes
 .../current/gs/PartOf/annis/paths.bin         | Bin 0 -> 13915 bytes
 .../current/gs/PartOf/annis/stats.toml        |  10 +++
 .../gs/Pointing/default_ns/anaphoric/impl.cfg |   1 +
 .../default_ns/anaphoric/inverse_edges.bin    | Bin 0 -> 234 bytes
 .../nodes_diskmap_v1/by_anno_qname.bin        | Bin 0 -> 123 bytes
 .../nodes_diskmap_v1/by_container.bin         | Bin 0 -> 123 bytes
 .../anaphoric/nodes_diskmap_v1/custom.bin     | Bin 0 -> 33 bytes
 .../Pointing/default_ns/anaphoric/paths.bin   | Bin 0 -> 5687 bytes
 .../Pointing/default_ns/anaphoric/stats.toml  |  10 +++
 .../current/gs/RightToken/annis/impl.cfg      |   1 +
 .../gs/RightToken/annis/inverse_edges.bin     | Bin 0 -> 832 bytes
 .../annis/nodes_diskmap_v1/by_anno_qname.bin  | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/by_container.bin   | Bin 0 -> 123 bytes
 .../annis/nodes_diskmap_v1/custom.bin         | Bin 0 -> 33 bytes
 .../current/gs/RightToken/annis/paths.bin     | Bin 0 -> 13068 bytes
 .../current/gs/RightToken/annis/stats.toml    |  10 +++
 .../nodes_diskmap_v1/by_anno_qname.bin        | Bin 0 -> 7167 bytes
 .../current/nodes_diskmap_v1/by_container.bin | Bin 0 -> 11833 bytes
 .../current/nodes_diskmap_v1/custom.bin       | Bin 0 -> 11732 bytes
 78 files changed, 211 insertions(+), 15 deletions(-)
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/corpus-config.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/global_statistics.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/annos.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/node_to_order.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/order_to_node.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/paths.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/paths.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/paths.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/paths.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/impl.cfg
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/inverse_edges.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/custom.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/paths.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/stats.toml
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_anno_qname.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_container.bin
 create mode 100644 graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/custom.bin

diff --git a/cli/tests/cli.rs b/cli/tests/cli.rs
index 4830293c2..bf94d1680 100644
--- a/cli/tests/cli.rs
+++ b/cli/tests/cli.rs
@@ -26,7 +26,7 @@ fn show_corpus_info() -> Result<(), Box<dyn std::error::Error>> {
 
     cmd.arg("../graphannis/tests/data/")
         .arg("-c")
-        .arg("corpus sample-disk-based-3.3")
+        .arg("corpus sample-disk-based-3.8")
         .arg("-c")
         .arg("preload")
         .arg("-c")
diff --git a/cli/tests/snapshots/cli__list_corpora_fully_loaded.snap b/cli/tests/snapshots/cli__list_corpora_fully_loaded.snap
index fac1ea16c..b7a158610 100644
--- a/cli/tests/snapshots/cli__list_corpora_fully_loaded.snap
+++ b/cli/tests/snapshots/cli__list_corpora_fully_loaded.snap
@@ -20,6 +20,7 @@ exit_code: 0
 sample-disk-based-1.5 (not loaded)
 sample-disk-based-3.2 (not loaded)
 sample-disk-based-3.3 (fully loaded)
+sample-disk-based-3.8 (not loaded)
 sample-memory-based-1.5 (not loaded)
 sample-memory-based-3.2 (not loaded)
 sample-memory-based-3.3 (not loaded)
diff --git a/cli/tests/snapshots/cli__list_corpora_not_loaded.snap b/cli/tests/snapshots/cli__list_corpora_not_loaded.snap
index 091f4f73c..7fd24b45b 100644
--- a/cli/tests/snapshots/cli__list_corpora_not_loaded.snap
+++ b/cli/tests/snapshots/cli__list_corpora_not_loaded.snap
@@ -13,10 +13,10 @@ exit_code: 0
 sample-disk-based-1.5 (not loaded)
 sample-disk-based-3.2 (not loaded)
 sample-disk-based-3.3 (not loaded)
+sample-disk-based-3.8 (not loaded)
 sample-memory-based-1.5 (not loaded)
 sample-memory-based-3.2 (not loaded)
 sample-memory-based-3.3 (not loaded)
 graphANNIS says good-bye!
 
 ----- stderr -----
-
diff --git a/cli/tests/snapshots/cli__list_corpora_partially_loaded.snap b/cli/tests/snapshots/cli__list_corpora_partially_loaded.snap
index 58256a6c3..f8b2994b4 100644
--- a/cli/tests/snapshots/cli__list_corpora_partially_loaded.snap
+++ b/cli/tests/snapshots/cli__list_corpora_partially_loaded.snap
@@ -21,6 +21,7 @@ result: 44 matches in 4 documents
 sample-disk-based-1.5 (not loaded)
 sample-disk-based-3.2 (not loaded)
 sample-disk-based-3.3 (partially loaded)
+sample-disk-based-3.8 (not loaded)
 sample-memory-based-1.5 (not loaded)
 sample-memory-based-3.2 (not loaded)
 sample-memory-based-3.3 (not loaded)
diff --git a/cli/tests/snapshots/cli__show_corpus_info.snap b/cli/tests/snapshots/cli__show_corpus_info.snap
index c663216a9..876b412df 100644
--- a/cli/tests/snapshots/cli__show_corpus_info.snap
+++ b/cli/tests/snapshots/cli__show_corpus_info.snap
@@ -5,7 +5,7 @@ info:
   args:
     - "../graphannis/tests/data/"
     - "-c"
-    - corpus sample-disk-based-3.3
+    - corpus sample-disk-based-3.8
     - "-c"
     - preload
     - "-c"
@@ -14,59 +14,59 @@ info:
 success: true
 exit_code: 0
 ----- stdout -----
-12:00:00[INFO] Loaded corpus sample-disk-based-3.3
-12:00:00[INFO] Corpus cache after preloading sample-disk-based-3.3: 100MB / 300MB - loaded corpora [sample-disk-based-3.3]
+12:00:00[INFO] Loaded corpus sample-disk-based-3.8
+12:00:00[INFO] Corpus cache after preloading sample-disk-based-3.8: 100MB / 300MB - loaded corpora [sample-disk-based-3.8]
 12:00:00[INFO] Preloaded corpus in 10 ms
 Status: "fully loaded"
 Token search shortcut possible: true
 ------------
 Component Coverage//: 0 annnotations
-Stats: nodes=92, avg_fan_out=2.17, max_fan_out=11, fan_out_99%=11, inv_fan_out_99%=9, max_depth=1
+Stats: nodes=92, root nodes=48, avg_fan_out=2.17, max_fan_out=11, fan_out_99%=11, inv_fan_out_99%=9, max_depth=1
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/annis/: 0 annnotations
-Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
+Stats: nodes=0, root nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/default_ns/: 0 annnotations
-Stats: nodes=56, avg_fan_out=0.93, max_fan_out=10, fan_out_99%=10, inv_fan_out_99%=2, max_depth=1
+Stats: nodes=56, root nodes=12, avg_fan_out=0.93, max_fan_out=10, fan_out_99%=10, inv_fan_out_99%=2, max_depth=1
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Coverage/annis/inherited-coverage: 0 annnotations
-Stats: nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
+Stats: nodes=0, root nodes=0, avg_fan_out=0.00, max_fan_out=0, fan_out_99%=0, inv_fan_out_99%=0, max_depth=1, tree
 Implementation: DiskAdjacencyListV1
 Status: "fully loaded"
 ------------
 Component Dominance/syntax/: 0 annnotations
-Stats: nodes=92, avg_fan_out=0.96, max_fan_out=3, fan_out_99%=3, inv_fan_out_99%=1, max_depth=9, tree
+Stats: nodes=92, root nodes=4, avg_fan_out=0.96, max_fan_out=3, fan_out_99%=3, inv_fan_out_99%=1, max_depth=9, tree
 Implementation: PrePostOrderO16L8V1
 Status: "fully loaded"
 ------------
 Component Pointing/default_ns/anaphoric: 0 annnotations
-Stats: nodes=8, avg_fan_out=0.50, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=1, tree
+Stats: nodes=8, root nodes=4, avg_fan_out=0.50, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=1, tree
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component Ordering/annis/: 0 annnotations
-Stats: nodes=44, avg_fan_out=0.91, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=10, tree
+Stats: nodes=44, root nodes=4, avg_fan_out=0.91, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=1, max_depth=10, tree
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component LeftToken/annis/: 0 annnotations
-Stats: nodes=92, avg_fan_out=0.65, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=3, max_depth=1
+Stats: nodes=92, root nodes=60, avg_fan_out=0.65, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=3, max_depth=1
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component RightToken/annis/: 0 annnotations
-Stats: nodes=84, avg_fan_out=0.71, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=8, max_depth=1
+Stats: nodes=84, root nodes=60, avg_fan_out=0.71, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=8, max_depth=1
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
 Component PartOf/annis/: 0 annnotations
-Stats: nodes=115, avg_fan_out=0.99, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=26, max_depth=4
+Stats: nodes=115, root nodes=104, avg_fan_out=0.99, max_fan_out=1, fan_out_99%=1, inv_fan_out_99%=26, max_depth=4
 Implementation: DiskPathV1_D15
 Status: "fully loaded"
 ------------
diff --git a/graphannis/tests/data/sample-disk-based-3.8/corpus-config.toml b/graphannis/tests/data/sample-disk-based-3.8/corpus-config.toml
new file mode 100644
index 000000000..97cc8d4f2
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/corpus-config.toml
@@ -0,0 +1,78 @@
+[context]
+default = 5
+sizes = [0, 1, 2, 5, 10, 20, 25, 50]
+
+[view]
+page_size = 10
+
+[[visualizers]]
+vis_type = "kwic"
+display_name = "kwic"
+visibility = "permanent"
+
+[[visualizers]]
+element = "node"
+layer = "default_ns"
+vis_type = "grid"
+display_name = "grid (default_ns)"
+visibility = "hidden"
+
+[[visualizers]]
+element = "node"
+layer = "syntax"
+vis_type = "tree"
+display_name = "tree (syntax)"
+visibility = "hidden"
+
+[visualizers.mappings]
+edge_type = "null"
+node_anno_ns = "default_ns"
+node_key = "const"
+
+[[visualizers]]
+element = "edge"
+layer = "default_ns"
+vis_type = "arch_dependency"
+display_name = "anaphoric (default_ns)"
+visibility = "hidden"
+
+[[visualizers]]
+element = "node"
+layer = "tiger"
+vis_type = "tree"
+display_name = "tree"
+visibility = "hidden"
+
+[[visualizers]]
+element = "node"
+layer = "exmaralda"
+vis_type = "grid"
+display_name = "exmaralda"
+visibility = "hidden"
+
+[[visualizers]]
+element = "node"
+layer = "mmax"
+vis_type = "grid"
+display_name = "mmax"
+visibility = "hidden"
+
+[[visualizers]]
+element = "edge"
+layer = "mmax"
+vis_type = "discourse"
+display_name = "coref"
+visibility = "hidden"
+
+[[visualizers]]
+element = "node"
+layer = "urml"
+vis_type = "grid"
+display_name = "urml"
+visibility = "hidden"
+
+[corpus_size]
+quantity = 44
+
+[corpus_size.unit]
+name = "tokens"
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/global_statistics.toml b/graphannis/tests/data/sample-disk-based-3.8/current/global_statistics.toml
new file mode 100644
index 000000000..9054d2c28
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/global_statistics.toml
@@ -0,0 +1,6 @@
+all_token_in_order_component = true
+
+[corpus_size.Token]
+base_token_count = 44
+
+[corpus_size.Token.segmentation_count]
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/impl.cfg
new file mode 100644
index 000000000..449496de3
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/impl.cfg
@@ -0,0 +1 @@
+DiskAdjacencyListV1
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/impl.cfg
new file mode 100644
index 000000000..449496de3
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/impl.cfg
@@ -0,0 +1 @@
+DiskAdjacencyListV1
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/stats.toml
new file mode 100644
index 000000000..5dc1a4696
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inherited-coverage/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = true
+nodes = 0
+root_nodes = 0
+avg_fan_out = 0.0
+fan_out_99_percentile = 0
+inverse_fan_out_99_percentile = 0
+max_fan_out = 0
+max_depth = 1
+dfs_visit_ratio = 0.0
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/stats.toml
new file mode 100644
index 000000000..5dc1a4696
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/annis/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = true
+nodes = 0
+root_nodes = 0
+avg_fan_out = 0.0
+fan_out_99_percentile = 0
+inverse_fan_out_99_percentile = 0
+max_fan_out = 0
+max_depth = 1
+dfs_visit_ratio = 0.0
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cd98d32e7aea5775d4686a46279409bce48665ce
GIT binary patch
literal 2211
zcmbtVYe<t(7(SbKopa8lGRtYx)O6^|%goZeYwEmYYH7LYZZ<DTnv{k=49t;8AQ=6q
zpy)^Ws{|nkqmYQG=#O-R6v+}Ii7tduy}sZ1zT@UnqYutA-sgSJbDr~l8zV9?;hr%1
zDHn<SMB+fi^@tk~HzEchZbA%3+>E#dF+?Qt@Dgub-H8at^5PYY!BR;rg+RcC+a0rN
zJlGLIcL-V*W1F3s)BAxzq@Bv){?LzNeW;>owOqHIc7)l1I)-s97Ei)>K@v+PTop$v
zHyphOq9>vkqBo)sq6E<wQHtn?C_~?P=T0mJFIb~7L{(EuAr^2uD7QuzcKlAqjbN5Q
zxfSjQszj=CdVlCAu|8DEv~rjebTqUBvtzM%5=IG<6efY+56A7I99NaoN`a%#N8E>4
zh**SJjHpE{K`ccqLp*?I?PpFb4p088F=tspExAA(SILfnD>)6xj_1WfW|&D8H4wG}
zS8ZqN^tPZnXs2?R2J{cHK2$Zd5_BvU--026u$Bs!hCJ+qJ3={bR~@ZX_zHF-rXi*y
z?nTT%%tXvW%tp*v`kmpeaxsqXe~sZvdg>tr0<PZ9SBS@^!V8?H1%n1^AZ!KBU}yTb
zoklyA!yLfH#QH3bnUc%p7TOU;2^Nh`xTBO?95y^BVEuwM1XVEn!(D`>b|K^*taMjG
zCSe)f2>A%B%bgJ4#ZAA@6@(-xT`v154bypv;m5KRGahrr(TdR08|EiUtxgt`Nh0fS
z$DS{5{1NXlXt`xor$$E)h>0<vBr{2TMwKO<pE0<FS>K)tY|fFJUk@}*oV|W2v`VZS
zPJW&(^G}uZ+FBK#Y_%`r<{}DAol&Pn<7HJ!?TfsgD@tjn+<ht_`%AZO`rUo~e1r7+
zu%})=?_SszX0$x))BH3Klxo-Y`A>8wde=|rN8hLGzxIbu_1*n8m@kbhwpwRR`9T%4
zrZ;K%pC#TmQ>2ClO{OMba&+Xe#J#TCFgK=n=65pKb<gb8&^l4swOn;ip?Rb+W~?>$
zQBk{12U8$q0hSNB^I^P;gt;0m*5igYmDSK;ur}yb8SPM8nlr2|EloMtQV&XdVX%bY
s%pVL5MiGA)Fh0TyA;czKWKh#zW9hgci+?H$;Z!SO<wB25*`E9A55il^+yDRo

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/impl.cfg
new file mode 100644
index 000000000..449496de3
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/impl.cfg
@@ -0,0 +1 @@
+DiskAdjacencyListV1
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..24c24b63faaf672a04c3b1787ed4346a9f838710
GIT binary patch
literal 2163
zcmZP&U}As(Rw!-2$jHygWXQ<K&dJ1%E(Vk_0?TmVl;Ol8V+=Oa1gw(_pF9KHIovpv
znSzbs!6y%NrWx2ubFf-oqSRV2G9s+!BUTI0Doe0cR$#06$<c?f7N*@AY?uJTFkoZ|
zL3v7G>y*KyDwtFUlNw-B6RcAdT`f?*IJy|n=_+8=64+%B&XUBU0AjTo*Z?U4^3piv
zfldcHO$MJl!Wl4S5T9y+waVht3iKgdEs(7Z4j~<|J~?9bA*_dKfg1xfQWtEPJi;)b
zZxo=sV6b%|U@{y`MuN#GFc}TjsRvb?02WIGlgVH*6-=gq$#mqTX^XA}Xr(>67|>}+
zU<({@$~a<?fmoda*6D;#9uX?eIF$i?0CbuQK6#)sAwJ9it92z-Ey8-3mP~L!Wr2-x
zBi0z8^>BSab~aeIJ3=?m`JPZ7JTie|-soZo=fI>w!AAL@D+F2sNeK~Pd0ztZemLcU
zDGBHhf9&!I=fD&oC%hQ2)&T5Ufeu7d3lxb3hix2KUm#(92<u^rfmR|L17yd84GTgT
z2J}rBf(KLq$v41o1c~+oF&~iC3dAzN8f!5SD={-LtOQ~U76yjxK<oyr!L9;v2pa<f
zxWu39C6siL;n3=;=Q8#t%_&_g4@R+OitRbOs<6|Mi#0IN$@$iKu>}FFmMXPxYTH^A
z#nmPFy82f6HJ!2$)o*LyT4*?@mP`2uleeS#Z@Hfmdk*VxD6nuboUn*kFw^X*x4wjv
z&`RUum20f^xjegDT})MbwYK)8tq>P%T$Se|e*Nj8uB8*FzcH;UNYZXwJV&OEN&J^^
zXQ1xlE$wj=rT%dxOHNq9%DPymsQR5f<Ma2%mKv$O0*lzboY|p$bXiQx$J0?0Em%DE
z`OSIYr{c=6y`kB8;Yab5RKYcqRXV&2RQsKM9xV9OeO2PduJvaaJ%p2%D5{h_mMHVz
zv6??k-Qo!_6c`wO12H!P^G<6X1|^oX%$$<cBE6i{vecZEBt556AU`wDDJMTa*9|OX
z^nkYo+#CRh`7O>vH4K8l00skC<b*OX+y;j(Sk24{h5ugi8_pMim4Hyy*r@Q|4c#iG
G?zaJ(Sldki

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/stats.toml
new file mode 100644
index 000000000..59af7ef1f
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_layer/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = false
+nodes = 92
+root_nodes = 48
+avg_fan_out = 2.1739130434782608
+fan_out_99_percentile = 11
+inverse_fan_out_99_percentile = 9
+max_fan_out = 11
+max_depth = 1
+dfs_visit_ratio = 2.6956521739130435
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bd7597dcef50c4b8e785770a8d4745bb61cf105
GIT binary patch
literal 694
zcmZP&U}As(7AVci$jHvg#0KTDGcxisGI4-OPB6&@Cb_{R518ZylYC&3A502>%@9O4
z18A-=R7@PKP6Dh>1S%%V$jAV9ju=!>2&@pU9OxKHu!t0xlm?SBU{V%L%7ICFu)&H@
zGrYlKVPG+RsF*!i%mFNB0~NCc^X#BJJ!I#=1mW6%4siq<;RGg~!K4eAbOn=cVA36I
zqzBYUPcY95%2NQVfH=+v%=QJ7eqhoc*?O3k0I*abm<$4wKwp83>jz>vAZsoVgJV*@
z{ziDn1tAlb8JB*D9pFACAr+7yG@<dsXM;IMenl22O!b(2Z-T%C$>7L?A+s(;T$>OO
za6`5=_)eW`Lqx^tpea9+M1>}SQ-KrEXl@4m*TEJHO3Z1QIVGt@dO4|OsW~Z0dQPQ4
zerBFiPJVu_8(7NdF0(htL10%M`f<#kL68~52LhNU!WbBCfW^RS%<OKlu4OTJ#3Bz-
Pg^vvX-O#O4>V6vl)qzax

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/impl.cfg
new file mode 100644
index 000000000..449496de3
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/impl.cfg
@@ -0,0 +1 @@
+DiskAdjacencyListV1
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..986cbd5ef6d8b2c819db546c539676f746edd918
GIT binary patch
literal 950
zcmZP&U}As(Rw&KF$jHvg#17@LF*5QqG6{mkIIxIuBE*2YxKMc9C_Ek%9xn=y4~55%
z!V^H@2|;;cU>Au(c_Lt*IFu*M$jAWq7EBPLK?0!!=q-qtB#MzzC_HHto(u|47KJB=
z!jnhgDL{E%U}x$<d3Io)EtF>i=GjAe`ru%PsBl1tA-n|>geY-@DuM8vP<YNLJQoz6
zD+<pIh3AgK^Mvv|z|QoB@)W_&gm}vbDhA>CqVW7sc>XB700a-=y+DK@Fj)nm@WP-x
zxaA=Jbs%2H$iUDGOwKGIfrjphcO*2!%UKo}shhVug<1L?)0?*b(8`$R!^S#o|DCq2
zzIl1gFF7aCrdLl_w);2gF$*)iZ`kmhbFzA`{*8sK-0K$Dxw3&w0tF{G!@UVB?lLGb
zr)B1pq!#Jrq?V=Tq$KG%l>+&hc}_X``MGXjDWkir;LriOg@Ix5qmOeK1erlhAb>gC
li-F+=IJbe-9J!rjv6ju?5gRPcKonyo!+$q)tCYIm1^`PzQ~>}0

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/stats.toml
new file mode 100644
index 000000000..5dc6ca90e
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Coverage/default_ns/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = false
+nodes = 56
+root_nodes = 12
+avg_fan_out = 0.9285714285714286
+fan_out_99_percentile = 10
+inverse_fan_out_99_percentile = 2
+max_fan_out = 10
+max_depth = 1
+dfs_visit_ratio = 1.1428571428571428
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/annos.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/annos.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a2ab2f0b81bcdef9dbba8db5adcdbc23431674da
GIT binary patch
literal 73
LcmZQzpb`K807(D=

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/impl.cfg
new file mode 100644
index 000000000..e96ea914e
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/impl.cfg
@@ -0,0 +1 @@
+PrePostOrderO16L8V1
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/node_to_order.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/node_to_order.bin
new file mode 100644
index 0000000000000000000000000000000000000000..952be524aa3e9050cf2616ddce55acd4769b4e41
GIT binary patch
literal 1940
zcmZA1)l*hs6vg4+Iq!SUcRr-VBt%qNL`51g2vO+<MKLk3ySux)yF30p8G)O%Zr03m
zwP(-l-NRtYhuRtH4mxp*1V=FzZ>FF_HqKPRRrKSp7o5dByfne(9L8NG*hD+-M8UJ1
z!|4&c$|JmZ!2=w`b_jmv3#Li%4tEKn1Y;SA9}-L>9aAMZhq?I0f*x^rd4iM4z|9d{
z%|=2S1oJ7!PZ!+FA<SyQ88qUi2tMH{{xZR9+{aH4+{b=wr{E<%;x-H3<QCy*!APPA
zY6ZXY4HF1XAr(_CSV|eeT)|e_FeQRMG5E6tT_)j96I{bOf{lU&%qEl}*vlTw8o@<0
zV3rGB=N_R%!Cmaeb_u@b17?iiDNf_I2>#{|PK;of2*P!O?|6?53nnuWyFjprV#4zT
zn`yz63WkU#lrNY_3MN;uf=c{;!9q$1G6h$#6T4P$CJQl>1$#ITy@EHmO^_tGiOu-k
zf|q%PX%)P{Ux*c)N)ArF;2YjzBLtI}fUObiW(m%G!8+=(Wr6`C2nqxZ379Owt?a<-
z6U?I;J4J9AXK+>s=CYK5Q3o^psMoWBV2R)ro@2)e9_Iw6P4FVWa2f<(@De*ha6Iu?
zPcV}#+zP?P)M5(-E#t5=1YzKWvjtbO1#g|;e5&wL1&?qNcctKTx^R*O*V0FLV0-^F
wyt>Cc!y7Mnlw+87!B2e0jT79)cKk-c2Ry`&6dcVM{6&I4_=#I7xPTh$KVDN<)c^nh

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/order_to_node.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/order_to_node.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8a4a431cbde04465ef24982ff9ad8b5b2fe346b1
GIT binary patch
literal 2768
zcmY+F^_LY#6ohN0d*;nU77Nazixb?Jg<TvLcSwN6-AN!JAweRzBsd8~@jwLm^YZS}
z=XBNlGUq#0x9;sbecpSA_fJxJeF%&s9D-yNVIQPp2)n_JBkT-6p0GW#iG*#CPa>>B
zX)<9;6t#pcP;L`2t-Q;Mg_>KVs3LPK#Zqi!QMr!M4tZ;;ZL65+5jr885X`o#j>ub4
z?FWh-qS^X-T~1f{mQ>qC(dlA)!;KAQ>st-qf@*t2G_@Cenbe+&-D8wZmFfo_h;($Q
zwWu1nf};9GG<5)6iPZjzeKl$@(u~Z56cfE>!;rKk^H9Z#!dLVopedMbsx?S95iUl$
zg>Ww1Ho{r(+X?HD?If&2zKd`QO1lXsq1Zz>5#`wdZLt#+$E&#(#Vj&URvZ@_S=3k^
zV;b_AR6A91j2>YovK_%}%b9_EIMq&9Y=~y-J6e}B2Yv?C&Q=_yi(Le_HJGh$Bm8u#
zogdNEh42leE>N5oqim{?`axGB-5hEyY8hO8Kr=6mXzB{MX{0V!T%u8{k=BuUmEs7!
zX6um5A@f?r%FdgDYGa%O`VxMNq#xmzNc$6h3OA7O2>c+z{m6z8?m<43a5qZB2zQ{U
z{Jb|mHpX_8dj_;!ZByK;=3OYN$-Gl>OKfCOn{|wR$a_%jUd2s%gu}=N2eS=%2zh&|
zJ*apfnyqh>F6R^Y?o@kJaicEw3%CKnY<)k2??$!9Bbs^wzALGpD;|qcHq{3Gpl6WQ
zgj$RG8m<dPeHGEv({P<hJ*9Y3qrO4fk<4cm*XuR=9!W1Uf2a7dUeUL}@?f^9{zI~Y
z@EOw8gulbBCHw_`9pQasmGI`rqV6DHPk0-p4TLvPY$W^<<)r~_u|FtYSMx0tOUQgv
z@mg$TQCD@0d&n14?OnwydW46_D*r*vkDcfN^4{;M^6v`#toTzjn|fK7^Bepks{K{*
zk}mcs+?rsvzK`G+Qth7+O??8tfYir|f5a%8>Y{$ozmcvCwHEak+$xHC5z*9FaE+wC
dRD7;c|0117=6@6~=rwzTWEq)XE1r*6^nWzDa6te7

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/stats.toml
new file mode 100644
index 000000000..170f24e0b
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Dominance/syntax/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = true
+nodes = 92
+root_nodes = 4
+avg_fan_out = 0.9565217391304348
+fan_out_99_percentile = 3
+inverse_fan_out_99_percentile = 1
+max_fan_out = 3
+max_depth = 9
+dfs_visit_ratio = 1.0
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/impl.cfg
new file mode 100644
index 000000000..bd39cc239
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/impl.cfg
@@ -0,0 +1 @@
+DiskPathV1_D15
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bfdae90fe44959c37e21f15127fa3b214684d327
GIT binary patch
literal 912
zcmaKrNh||l7=~x6t;nb<YOCGYX=~pJwH2`hX>C>2mQtED>f|6|4Z+PpB5^_!kvI^E
zh?|2(;w*9I<mTdH=JgRLe=^_9^Stj&zC+7|OPKE?qq~%4m8?XMS;(@%A(CK9<Zvp?
zA|xS{6vLIkapX!tp+se9l*1{=Re-{XD$%HdQ<AF&MG$4PEc3hWkIX?uO_eBoBn>ki
zNem^CaM74Nk~m5-;4<Mf<g!2sM7f|uqCE8T;gZM|fKrHt&=~%^u9hMvqDj<i$<@GF
zF~^bAQ=)^b!?Yo3pkxBh4rd_e05uU!ftrb2=ugA7kedM+iQ4|I`y>1CT0e>%R7{0y
z#~eV?Nr?&04A(_&5Y$aH0_q_eMc)F~OKwb9LpbNcWf7hf!p~DGiYor4S#iH}cekkw
z@Hvvz+Qrtq+xYw0leUM`HEp)+{M~wRnRzX9?yjiK7UTXE=k>pGCPi%jvDvV~KW!iF
zclwuI4Hp5OYJ+=_i{BLY<`~uE#~YIt<aD{`9bT>5vEXpqZQ4ekz^=JQx5qQvf~e?B
vHXy!*{Q{c?4WkMYSvbGTGt3G8N9fcT_YT+P`Wt!jck91t`dr%LeXY-5-F{e7

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/paths.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/paths.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0c6f66210a6300f26d75a40df7374e8cc615d3b8
GIT binary patch
literal 13068
zcmeI1OA5kJ3`LuPi1-6RMEu|i+<}Pr0a2{(e>2VtC3Fx@+uY7tLQd{^@LFDrLRa4n
z*r4wTtNCnLeWH6AkO3Kxff@t(woF^qsEB@S%V<K|`eUlwCPMSbfDFihZ@^xDO-Kdy
zcH+R;hrl{y1^uRmfj9neAr-a=Y3n!Tz*~Q~kP6#{wDp^E;JrUwzbfnu<-yJ}IWTpW
zX(MWlG4Pm5JV^(hQi*5jz;i0`2?AqS?6rs`7FY&kp!){w+5HU75?}{x&Bj!<49Gwm
z4PaX)2d4f>V9RfB2+e^%RN_NAa7iWprUU<|#MBvxVPNdnz=c|441A>$AJc(vRN{L&
I@B>Qx2V9#90ssI2

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/stats.toml
new file mode 100644
index 000000000..29d71f537
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/LeftToken/annis/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = false
+nodes = 92
+root_nodes = 60
+avg_fan_out = 0.6521739130434783
+fan_out_99_percentile = 1
+inverse_fan_out_99_percentile = 3
+max_fan_out = 1
+max_depth = 1
+dfs_visit_ratio = 1.3043478260869565
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/impl.cfg
new file mode 100644
index 000000000..bd39cc239
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/impl.cfg
@@ -0,0 +1 @@
+DiskPathV1_D15
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..87efaf3e8beb5ec72f9b2794a27349ed5cb41c68
GIT binary patch
literal 804
zcmZ|M-7AAp90%~{+57C(iftHcc_}Xsk=)RD*tE=Eq@k3OM=e9ixLwgoWPgBCT)9#1
z`~i|HQj+4Hf56Rh!Sj55x7_&E>738+{7&b0I)Ym#{DzscQ54Nqou!h6tds?+p{$UN
zs)6j34RTO+$VoXM7v+T9lne4ujgXhBgM5@1@>4#jf$~F*RD&oA)n_&_=|M?=YJ!@n
z0MtS?LqRG5Nz^zLrbeJqDh!QLqmWFELF1GRO;8dPp(dax6@ezHC^SXI{ysC#WCkVE
zR2)iB(@>HMLMbW<rKuD&L#3e%HKksw#)mo{)RkUCJXT&>>pE8UN`<qd{e@nWz5G&t
zd-e6&cDj4s^VGRj>`32-u7}!-7lBZ5-lZ#D&YBKP+R*`Z*o4KwL_`SbSF^coWlPE_
zJ4$Y4S(5Xrvm0`5b8|h4t^QZTuyzV9yvIIogjzJ9rmFo3!n2`TbN8L<J~8$^8Y{o6
P?fesSUxmZg{N&pY4RuCh

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/paths.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/paths.bin
new file mode 100644
index 0000000000000000000000000000000000000000..5498715275a8d3cb536a8365b1e936c4995a5e19
GIT binary patch
literal 10527
zcmeI0H&O#Z5JlII!H6P3Lbwxf1q4DO0AvVs|D%vRVii?rzqa_BykN~tzwYUsnxgP!
z0|zpY10Rks!3<B}DLjMca0D;l(IcrMjViT4mCH_5Jgur)8`bitRqL@*HGpz39p~s<
z;=TzsH#~6*@8JEuH+}z?2J#gzjNt@c!zsLkgHKITMH*FVgDRJus(4yewbrVRRjFEJ
z`R{siv!)ep;1+&C!v<`?25i6vdN<I#GUNg-;al%h@qjj912$j-HlQ|;pP0c6KEoF{
zhmUaZztJ=!lD;F2eGf>SMNs9kQx#9Es@7VyyR1~(+0y&xpLKR<=66Qe!4LQeSMUK2
m{y;9NA`L2~ohpZ|s%UD}adD+OpDukMPJ=T3+(1<Q6utr9q$j-q

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/stats.toml
new file mode 100644
index 000000000..43895c29a
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Ordering/annis/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = true
+nodes = 44
+root_nodes = 4
+avg_fan_out = 0.9090909090909091
+fan_out_99_percentile = 1
+inverse_fan_out_99_percentile = 1
+max_fan_out = 1
+max_depth = 10
+dfs_visit_ratio = 1.0
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/impl.cfg
new file mode 100644
index 000000000..bd39cc239
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/impl.cfg
@@ -0,0 +1 @@
+DiskPathV1_D15
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..9f564de37af33bda4438b67980e2294b8defba60
GIT binary patch
literal 1213
zcmaLVJ4{ni7{Kv+3#COsitjfH^-%;Ae2XZ3L8~o3pdgA0wHK(i6i90vJ2YB@CML?j
z<RB&{hCyX;a43n5jVT)@Bsvjk>;QvtGa7IIxivKsPjY_A_kHJ{@0`;LI+gJE$QQ*@
zQKiNL)Swm%aRnA(F_z#;EJYob;eM<WMd6=z^^^|aK|F+q@dzHpV|W}JL{U?y5|xv$
zBDWgX;96XV>vP*?KiNQOBW}XYxCOW3HY~>qtVBKT#9g=>_n=|9c2$)2Vl~!qgY4sP
zC$|G_Xvfod1|8_cRy>R6@O*CD?6wyuwP8EDumd~sB6i^=yo}v=1+U^Yj-01FdM&wq
zxoxwh6O<b9B%Z=1G@=R3XhADBV+;0RFS_yia_u~nB<#a}96&GLz?<B_r`*6#Zit*;
z$qkYVkQ-ETDx!cA2}DCmG|c*u++z(ZC6R?AN|Yxrkw<2ilssK>l$57g{;6Kc(=NZA
zHnX7om-NlCsGE@TB9&3m_MrV`?A>j9zh&B<-<_D8cTSmOjxRHQVPR=pdKQfajq_pQ
zt-8eNv)%Wmo?DW=aWNRS=#!7fCYwCg*$<JgpJ#iVsu!*aL&6!8TGLWTm%1j@V7+G!
zd*Y6T_fu^~$zXL)8<z$?ragt`?{|_3$In#USsawAEsJl|k%G**9?G*wXuR9@RH(@B
z^ZJ6)NUcx0CHXw=T2n}ly@MuSAmDGNR6nK}=e|%iGxtO&%l<q5yokK1CbKnb9*yee
WUTf>p+VV{5|D*FqcvK&1`ScrjNQ~kD

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/paths.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/paths.bin
new file mode 100644
index 0000000000000000000000000000000000000000..c0a9c2b9977c39e3c0407aa268d69cab285548cb
GIT binary patch
literal 13915
zcmeI2y%NJ941-DFar^)NLx<WK((BMWArE|6oGca=G6s`=Z=dHH_7Zls>CZrTxPfm0
zXpi_+6!8e$=ALVSn@eO+4b_?Ea*MbSmt0ioaBT7B-VgxXTzfC5hU(0BxkX%vOD?K(
zIQDpRpBn3csmGMXl)u`tR<l;CHFfw~P+UENcefTt8F0=vz0BF2db&PzeKOnR7I7^g
zuCgKlu<Zx~)le<34&ktFm${y**Xwf;vF+yssD|okMFMdlF2to`pzhv@JSLA-Dn9$E
zQ}uc$2_mK=aLzx~Hyr`r{qb^ntP<(pezw_HdA~-qyLKGs0h9sPQSB-UUsP8CzXA3k
Bk^=w$

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/stats.toml
new file mode 100644
index 000000000..ef1a9204c
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/PartOf/annis/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = false
+nodes = 115
+root_nodes = 104
+avg_fan_out = 0.991304347826087
+fan_out_99_percentile = 1
+inverse_fan_out_99_percentile = 26
+max_fan_out = 1
+max_depth = 4
+dfs_visit_ratio = 4.521739130434782
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/impl.cfg
new file mode 100644
index 000000000..bd39cc239
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/impl.cfg
@@ -0,0 +1 @@
+DiskPathV1_D15
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d104934b42b0f26fd65c9312c87c7420135ba93b
GIT binary patch
literal 234
zcmZP&U}As(K`70`$jHvgBn;&Vfq9Bho&uPs59R4GGJ;rOqFC~r{XZoK3&svF2R4u>
zClGTpJXJj`&!EJVmYGwMTBMhgT9%rVlBDNU3gl<zIpyT%=emKVjC_PZn!rXqd%NKP
lgCNXVaLX7N+~G=&{_T33tUsMm4kSVVg#T{nRw;GA4FCc)A-ezo

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/paths.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/paths.bin
new file mode 100644
index 0000000000000000000000000000000000000000..350cf3b3da35ffca44fa7e87f77368b73e8e9fdd
GIT binary patch
literal 5687
zcmeIu!3h8`2m~-4p>%>y(EYdT3_=AT@qMR~2t$AX0Rqzna!voxe*~%mxvLF%0t5&U
YAkZz4r~88a1PBlyK%gp+x7zrWe_5CSc>n+a

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/stats.toml
new file mode 100644
index 000000000..599f469e7
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/Pointing/default_ns/anaphoric/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = true
+nodes = 8
+root_nodes = 4
+avg_fan_out = 0.5
+fan_out_99_percentile = 1
+inverse_fan_out_99_percentile = 1
+max_fan_out = 1
+max_depth = 1
+dfs_visit_ratio = 1.0
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/impl.cfg b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/impl.cfg
new file mode 100644
index 000000000..bd39cc239
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/impl.cfg
@@ -0,0 +1 @@
+DiskPathV1_D15
\ No newline at end of file
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/inverse_edges.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/inverse_edges.bin
new file mode 100644
index 0000000000000000000000000000000000000000..eb64a463c43b6e285fd63dc27b802b90892d0cf1
GIT binary patch
literal 832
zcmZvaNiRcD7>4g@duzBv)ex~T)I3$yJccHEX)0oDYgWDG(h_1=No=raB8GI0mDsTo
z3-J?J60x$8u(3ws^z{)7CppjU``qt6JxwYYG{W~1>8WEZDn=7cDUq=Rrm@1slB+_B
zqg0EOK&b-rm0&ek1J;4{U<25Qo`HILI18MSToGf!@AXnhCSxRxQW9#paOvdokuoUd
zU?LAJ01Lrlummgx%h1cDUMidkE{ohSl9|%<-|ID#Y{Ez@rE(nNg|m_KA=xQSVqywZ
zKtDJGs^Ba*hn`Hm7PtwxPI4VcU6kB^uh&DeA2k=~0f)d5a1<N^$8lIM^&D`0aD(Ks
z;Rg8E@O8*{3;*=Uw=N=NefMc_KUTWjlY-B(Cfs9hD=(7T)%dB<=g!qh{L^|?=-hU?
zm80xt8g4|-F?%_9i@9}f9J<=pblKbTQFzfLy6>-@c@n!1>mz46`IZFLSokcMw0<?P
ztSp%W%Bm9ZdCj&Jp4E9<U|}I>$1CfjZXnXa{WjC9iiJdNB=hz26Zrc(+!iw*XXB0w
W(ef(({0F{=|E24FBb2)$KYRgI>Qq7i

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b02d64e9ea8c6d80351f24c76f290ac1252d084b
GIT binary patch
literal 123
zcmZQzU|`^8=z4MMB7+iBT4qj3YLQ+}YFTPdN|K&aDUhF;=aiG5pX&yeGGbr_s%KyX
g5vea?SHL+3J}un9rEkXp5+DS^e>Zfil)B#r0Hf|2XaE2J

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ed24586ecab403102687114829631ec81693c9e4
GIT binary patch
literal 33
KcmZQzAOHXWApijY

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/paths.bin b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/paths.bin
new file mode 100644
index 0000000000000000000000000000000000000000..292e96c477f57db30e3f77a10b9991dff8550b12
GIT binary patch
literal 13068
zcmeI1F$w}P5JmSFf{369f>@|0@CJfdC>9o4c>kl=Z`kRQhm&vT4CLp%2|;&<!Jq5~
zG|Nsn=F@sy@pl`r0UNM^2?Lr>G}4x}gV3~vg0WZ~v1kX|fDPEdzZuXM+A(%l>p&4>
zKLjx1J*65W#)3Ci-N*oUfl!C5k=Qau*Rs~6f<4*OstI2ig8eb)h!>@WCKZktJ6y{q
zV#|;YRbw;8f-mLVb651p&1X6ruz`#Q^vp;VG3gU^-LnB3uz|n;c7f;}?5bu9dnZv7
zDqNdW{R;P0ugMW(r}|XSX1H>Ps<9bk!PFB0-NUn2ugSd{z=&`4Y=$dyGsaY(Lm~B=
L9jeA=Oa=b{t<4J<

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/stats.toml b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/stats.toml
new file mode 100644
index 000000000..10adbd93f
--- /dev/null
+++ b/graphannis/tests/data/sample-disk-based-3.8/current/gs/RightToken/annis/stats.toml
@@ -0,0 +1,10 @@
+cyclic = false
+rooted_tree = false
+nodes = 84
+root_nodes = 60
+avg_fan_out = 0.7142857142857143
+fan_out_99_percentile = 1
+inverse_fan_out_99_percentile = 8
+max_fan_out = 1
+max_depth = 1
+dfs_visit_ratio = 1.4285714285714286
diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_anno_qname.bin b/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_anno_qname.bin
new file mode 100644
index 0000000000000000000000000000000000000000..993a85532dbc7af56830e37b79dcbf827e54ecd4
GIT binary patch
literal 7167
zcmchbdvH|c8Hdj~B<yB$zj6;D2@sHk+<-t3$ps`K<^mx=B*5lILY8dUY$!s<aUcZc
zCe*uSP@pKIgMwHqS}WpAW5H6;A+1Agt%wyJ(OPRO)UnULc)sk0jqq2QIS>1u^PO|P
z?|py2_dB2sGqfN6xSURRfz#F2(Qck`=z3I)F|oa)_S*Rj^TH}!9}#9utasL_Q?0t5
z7-b|C&0p5uW^Yxe=ji%)n~`bWSgh+w;YRy>hGd!NM(KKTm{HlV)~)U`bbWM);cQVS
zigZ22W~7_9R7j|<&eQd=UN4y|p?ZFmgxTi3B@#|D!#gCLY=)&0PBFs-UDq=FWXLNg
zl7*!9N@q)sc~_Y{WU3jKv%o0B?dqs=ceonVtmo@`s@IxoBve<IN~jXIOxGuPEwNHU
zwZsJyswFOzP%UwhgldV4B~(kSmQc<8P6^fAm+1N^ulG9rS2=(*W#2eWnWj9{vIhFm
zL*(^((E92j&{q%PzIup|{HvE(Bvi?=N~n?*B%w-{O+uBdU<p;SLL^kl3YAbLE6jhP
z(|rp)m1Ir1nPjyq$tYD!^HoKz6iY2MLqfICObOLOvm{gt&6ZFtbdrQ>p_3(43!NgN
zT4;{05BE|<n`#H@-K+fzCC0ZM#Q7<I^!4S}Z|K5Ij|;QJg(_FEz8Z=5)yPOGol4yx
z3DxumOQ=#eL_(Fip%SXp4U<r%E=od`y5SP4)Q#}3kwjmO+#;b`$^_rCr%E-{+>=Pp
z<R2B3nI2P6YMsgQB(=`b5~_8Mkx;F3tb}TvDH5u6j+0QWbG$ST^<<ZSJxur2!wf%q
zn0UP&2J~$QxB8ZPwr{C#mkOwK-F9<o$`&=LOwE*ssP*JYs4|r=p~_T&gep^M5~@rU
zN~kh5%fBL)`zoT@R}m|H711KeS4*@@sODZPp_+T0glg{f5~{g3+?<+7kEtfL&PI8X
zTIUK0)jFGW-4bEwb$vfE;&g-W3CJe|WQ&2c6i63=XvZ3D)Z5+mc4vpHuEE?_*%gNB
zd@6XzUl3tfLycBvy&A*L6r&q#7h0_85Z(Ja8xYsoAs`zAWIupx1CZ_yY5I^}4{7m`
z{RXnlK(;H0=^?!f(&QlRr=NGy^|(&QfixOOe}S|WNH>9W4M?+qbS_B4f(gP=!bD+`
z5F>LH|Mv4vtju*B3^6XmuMnF;+6|=3K$;6gvJme<_Akih0yBkdCH!UwfovT8ypuhI
z>uep69RsrUK{heS-UVp@5dT8l53wi2ZV;Ejd?B9VH{O9*rk{6WeXiqpI7e6{EEdic
zmI&_<mI}*+<-!W#JmGv{rEo#N@2ryR3x$h>i-py~JB3SxHNvIBWy0k`{K|Z5g}9U7
zm=b!t=wG*3jq5lJVk}65fb<1OJFvtUMtwu0y~E*N*4l1b20LIAL=O=zM6wXcLWBts
z7o_x%@<0lssedQ<c`Au!6e-50tFU*I)8Slkm-$Z2$5ku_twJouZ<`Q{@f(XlEC#U{
z#9|POLCgg)6vQ@ObGiO)7>ZFC3SuaTp&*8W7z$!2h@l{cg2){rLx=z&0)z+<B0z`$
zfAq~rlTk<$B14D_Ap(R*4<bE?^dQoMNDm@Ci1Z-RgNVw?zm)E3b=%jPnZ)q4!zAW0
z$ZTePMkLFK(ZVspvBDJLIN^BVEkZ2M6R|wR<`A1hY!0zG#OBar^Zwq7(HVi!Ax4K7
z9b$Bd(IHlXSP5b!h+!axffxp27>HpYhJhHS-?w5TMqnd|ji6nKjrfg?AU1;72x1wC
zWgwP;SO#Jl=&{UTV?dp=)$OvkyHgvTb-n)(#y0U{5^bcJ_<lxW7)D|UYnb77wl&w8
zwZn6<;xLGxAbzm~8v(bOBhY%Js1l+ah!U*9Mv-zaPVx*+ghk9_%>Ye17@%q0ftvQs
zKuwFaYT6}^njlR(%Q4HQX~#HHgEeh0M@opMy~vRks%bBAObFAoeH;_RH7(Bb;I-es
zwnmFKR~XRyVyqV>_JGLQ8e!Pm+8XRGH8Y$$NIVAd0Yul#y;g0+_?eq=9>g6GwOeD1
zI_Ii3M{}Ls-B9nDg69v7VVIZ~Vm*jqcxS^}J7d)o=_T4=HQ7EFX#-3G6GM-cf{kYP
zwWn(u&4$<0RA7P-`}5sEGf@u@l0Hfsp_|eKAl7A!yUE^aK8M!DdD;u4IY2CJ<4r0p
z7N)5BcYwR8_X=%{@w6I9YY4I#PV=#0G9@|{)1#X}x&ulHyn#}}e<+Ii9BP8D>3fix
zz`ix{=t-43uy3ud75!Sp=#0$cTHO!;zi+mBks(I2Q>CoGLQ<t0RLpCly_Nhw)sPzf
zc)h$P=T>#RzJdeO%-Zzl^}<SX1kxePycQx8h)g`)pwt|J{5XsJAo7FAj}gVi<_L_`
zJF|hk=h+t-jkOqEs-B0RI7=BI#(`7|dh8Qol+P<SAB%sY#Xk`LK#zYyjCuL<%pur^
z)7S@MALy}9uu)lNjzHyJbL&0dCu`hOm2ckTw};qlMzuKrLGvnv50NiK{t)>>OaPHD
z#0nOhVKHyy$y%h@OVnAa6d^qej+EaxfZudyhy$RfN85}9^Fhpj#!c}c35BErQVK}P
zAo+vL5_-}QrCzKagRr^C5+E9Y)CH1JNc|v{hb$11HApca#eggsq8vyyA$5RM-QNol
z5Esc1r0GEN1Q8vicn~?jQX%={H;oA*5KEY0%_}Tc5*aOxZH$nw2n@tm1YF45RYUZ-
zW072E*Jf0`Xr)v=Em?FpPM9g2B3vNkdjXZQ>%$e?8EVWcD^os7lqaSO`3B+X%cWxT
z&5EUXC&a2w-y%!n%Z3F`74mh$?;2sFu$jgkV&vOh=Cj7i3nmE{3mc@jSC|8mWR`qm
zF;~8Ypk3BlrPey0>8PG;?L9w1MyCpMgwuq%!X-k!h<M^MVUuvBu!RDJ8bMiUS>`+(
zQZieg|G*FS`aeCVt`n*oR!gY<3&F)K`E!V)oL^sXJc^1Z@kR0njuyU3zU6T7WpaVT
z&ezG;92>D#nCHeFEg5@PJhJCpPxkPS_MQ23$g>%r>@f--{~%y?+_;7Nj!ztZ>Q5U-
z?oQjJA6si|JeU9Rxj$b_35@*9_9BbzTg_pAwqotyzO0GwGER5KZFoNK!`VTvCq$3P
zsUM|Rt{w8#{eMr$scKzR`1?r#g&$2m`prW3<-6YsT2!Kq=&XMAbmwk+!OOo2ICLg`
z^^ULpy5M9<;2-NlW<1<6!Ts-lJycQMlXE8b(nnEsb<GXWHzsr+51M=B<KyncmWxA&
z)O6|b?_@j9hn;+VL1*FFNkcNn{`|$`hwC5yZ^*1IuA>{Dn^_(7z+;Q|z14Hypx*}c
zbY<*Z{N=dL&cQtg+$Gze*!@D;)E)H$=6pYS+I!LK2bXTXc-8Up{SnDeZSB@F#^}cG
zpPr3vsi<B$_}7nhf1Tg_hGTe4<~<dguH-#=rFrJI1M82SxqHcJ%jq||vs-qpJo%qh
z@zWoQ%h2-9x@T@myfxPNFly5~LF*r{Tz;_ppKt!AGO{FpU1)Q5d)N7}&$mTCH9dLd
zQ-L2RJb3JlJ+s$*H@c`K_sN|#wiAc<=B{47EdI#$89N4zUsc}oOyY=w(eIAyiM(}g
z<E4=2TVGB7bit;Q!|{(^vSdxYaL2&IV=lfkZonnm;sYDYyAE!!y!%$w!On=gKMOwM
zTGz5UVgI_^%_DcsD4I3w(1p~`6Se2E0>ZljPc;^uxcuOj{c#gBv|l(L9a=oCV0p&M
z;`9w$^f%ihQ}%rKlaaMidsm#hGyIcZzFan_Am2LdH9d7p#EIPGt$CIa%e$B6IunxT
zrVdOVxo`gm@sU^cX(tcdlRjeqm1pmLDRoQz&z^qa>E`iUhkW)#>#++lVW;jpvH8J*
zCfj>oo^E=kaQzNj&gB6;2lL;X^KfEAQ|;4^H@Agsxi9<jgl*Fo{iC~W@JTIdTEU)A
z$KSp^F)!uF&a_vm3NtI7bFK@>kA5XR_TKKJ!=~?doO(TOqGf5)nHL@_K0k9s;?B;p
zwpi<~f`pWnVMj`=p$m|<rX9rup5twd5Uf4&c~^jz7|_`4a5uQp91Uw49QCzn`5nZ~
zt@#e8bJc9QlzBD%eH@}`+^K1E4}I~r7L9}CkJlIYKy6!sB|O06*j^KAyAm!&ct*;-
f$A1@G`2OUK2d1e1kGu91e_VL=o~_9px1awXFH%>-

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_container.bin b/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/by_container.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2d8ced0af904f2d84f9fd81d6e63880de715aa02
GIT binary patch
literal 11833
zcmb_i32+nF8Q!PU@+lLrsWD*57~4P$J|M;%!NRfefh`{~cZ@8_w=JO~F-+Pqw+*<^
zi_n&&LkLYs8A54*wi!B(fk39t0bw#N9i|+ONr6xrI>RKHWcsh<_jXrW@4kh2#+GF1
zec${4egF3#E5?}0g!~|r$K~=Zc6r=>?@X`1G59Qdro&~=N%ht_yM5VlC>o(ihs|g6
zy8IryQ*TH`7)5+}4!b;e^yMn)<pK8W(U+^Gmj~GMMqQq!A<I)GY%i^C@VaeXV7(3@
zO_$5z1V5<256&i=zr)wi<&8r|HHx*nx_lm+*Y{+T%kKB;4Q4HYPmzH4*16ilush*0
zw75E4&FkZkMuSwPUdD)bi($1$)d*YB>X5qLvN8^-^$4Zt4QV<;OJP-xlvR2)()wKP
zRy!D@C#Z79ghBF+(P@yz*U}0u(`oTDjSi`b>WBg62x_WaZug8adSn@14ws0}Y|tRR
zv)k6`?f|D5w7eHAe`O`dZ=mMS<M@MTf=|er3a^2UH8uE*y;wDt8jdTHnrpTcSBwTl
z*<D~;tKH^vIv8Wj2rkCg3f_l91vZzJar=m(-p3rNeFQJBtt)7=uZ4ZaQkC-Di?0)d
z<WAC4QdDeMwE6%Z;(&c*U2CneGDa*4m?sz(X6zAp+I%Tm_88gRZl}!yXE;KObio(l
zacGXEcIjLxLe@^7D`;on5!_i<MXp?G9HEgGa8Q|9j1h|iUL{a8{^@0DS$SdI8+ZiT
zdP^Kq=@A3EZ=`3H6gRa7p{KzZZSVmUHj8k?({a>-IP-)cRGb;3c?4$+!4o}rBv-lG
zALZb2SL6x8BYH(JkIcS7$;T2PQ6(P%dgS~xj=;<LgLerzoyE&TbT*i-xPudxpArZy
zg;in6gr9Z!`6-d0${7;|$v^h8`H4GD<Ea_LPA8v{J}>=%n(*=SQW7;6FE4Rt11}Fb
z8zE^b4#mjjr3uvgc<j8COie|bAQ&UAMKHEhzL-eO|5*8A5}}b)C$=h`C6|}N69i+#
zB|YW|=8MVntUR#^FRCQ+g*k<Ah)MuyCLUMcO@v3(gP13v?_9>ks{|RBQ!~9G9p@<q
zrHS<mIB)noe!I`_aSHqtp+Y+E`YxZX8?IRd($rK{)d3{w^5`H1(#VI@L?BgdH2_W_
zwo~m<pnRc#l~w>uQwGRWHZHm_Mu-mzi^^Drsmd65dIklaIAppqNP#qkLe^Sr9@u4u
zGDu#9!Y(`xefTAaynjORdj%TBhtMS6fm*Ks1gHS|nR@60nxMvC0Yz~uRKjPWI1SXN
zLz}*~bB@HD;90GXK&Te>m#J)j0*U0C%$}scg`cJjP@s9Cfc0>lGo};3sZ`-3P&Yo}
zh{+i<6aw-Bl5;5r;klG`L14#EDkP8$PZrUfMG!?LFd!l9K;&^Ha3<jql|alRQ{pm4
z97wYWo~W$gDU=fF%T@*{>Dz$x<tPJ`m%dzOfIR6F7beb~dCHJ!ByYAdNE*qTqYRQK
zc|!ZKL_MVpPe!6RZ{{n*l~<~{1W;6|!cM!aQWYo!<SCWVl|-86DMXYEpL&?e;2b@l
z0FU1N<)uDyPrHD?%L$;s2kbnT-;~KVu?q<gg&6|pgpQ8+P3+SIRc`k=Tw#e92?|TT
zF|s`^B#uSY%mQ5t^&5RNTS!eP&^;j`=L8H#Gjp|2QwxMqNDaNKaOO(U%r2%T7RYBP
zarlsxyE%X<LT0W-)XV}^3}vPtem+A@D3HrgLdoGLLvCUf6E?|m<Pm#XGLS5xr;$__
zi6%B}3E>bGmXYQ%TpzjF3q}A*<J!wnZc&vIHc`RBtt{OwDv6n28R4O@idrtq398)g
z!yq|*<TM1sQQV@cpk|g(FX+^1<Z3B3momAE)uL!}&#2Ya#fb=J1QFri7ptV+h(h^-
zgA<}xRn%19hhA|rT`M*3qstcDJ*g%%qI$&-J3;LyCXd_<js>lSLBM(#1YCeYKmrT_
z>@Wm40rzhI9KkSyFbMb~3<A!<AmBR~1n6M`fZ#?@U=Xl&*}2u+3BVf^-t?)V-}a;G
zUwSR!7`+pOZv-gm1nZPR3bcZo3vkZCcz7oEtTKE-0mb1%5R8mQ%=H9NPS*rjp{Iko
z7)xJ92&tSd&@{dU?BtQ1x^cbJa%IQ@1rtMtNz@8ukOD0Vg&Y~atyBgn+lj7Hh9}T4
z{uYBLg7fBTWw-*33dIdfpw<w;(K}Iov<cLRkBDbx&nW~H=mZ~-bfOIk5l0E%V0f{e
zD4sd3CBUMG9?!6pD2+`Sq~uluGp9ynfC4!P^=4q|YF7rRAkrPmkZB^_sSJ`P(wme)
z%0_y#GCb-?Z&8LTAL*?GP&CrRPP=TRw<!dakMwqhh?3#MMmp3ufu9Zn9Le1ofWa3+
zI(y*zj_~ar15)Q^<?0QIo#eygxDWHb5XoPsHzaqFk0*(Koc#hMn9k?YMIsIRUs&y;
zSPf>1tj>Y0y2)zMtqK+b**fxZ(X9rK04RtESuSJ5cd$JaAfOr|Ac3LuM5s?op+0>H
z>eGM0jjCLzPd7q+`XSV(Nl=?YiOMiV?T(|3hhFrYRo!zucWr;}n0wQT-m4oP9P_`$
zZ+BL_TKq}R2W3C++@LFYcwlJi+JwnfNuANnga27{yS=`sS)2N~?ciq1MQeiH+N-*+
z@<kLM)y`_%`{12~41Gey@JaRFj`c6Bv7W8|>elZot?s&-%FdO|gUI?`gZ-LmPtNnP
zi;O!OE9Okm+?u(k_@x6)nt}TZh8DhZz9(r<T*9@>IVX#pC){nfc1-$Fou{FE>-m}s
zYs&O7T|M>FDtBJWuh2D^)#hUxa{8<ve&ZfAr#>t<eG<{K^kS@QN@e?miN_Pt&ugk;
z4#pf#)jV((&&hZ*YWa;p`<BavcZ!EC(;XG_3ftV3+tLqq?b{Ij+2xO$wbxZCWAs}}
zl5gnxGA9&N^nKQoy&`XH&(_y6n_q5tfA>w}#>B&4SbtGrm}fYfz4c<*o~u7OZE8%s
zdV1ecLt(}e)6)1e3l49++uQf%$zRXk_-(|W()+*c)tC-s-55a2Z@d_Lb6>CTNPK_k
zr?qE)`9$?!`&%X@A6=eoPMQ0&bNS`*Jt*n7gWv7j6`%Iy-j|tz>LcB|(qi06(|gpP
zR}W=XB);+1=J&VN_pXoIdGFAn7e2juuI3eWX(g)OY@gYCY2*F<Q5&}B_nrCp$9X^e
z%~x+^ABsu*)$Pf1(==;8+Lqb31--uK@6joB1B2sxkL=%)*l$>(8rnTO7G2obsoHz@
z)}{it`Y*pbHD=RCcfanRc>830jCI$-*p$r&cO%23wjJ-L{WMe8J;U<uJ6jiJC*QMl
zs#6Wl=st_K_8Cw4k?1l<k$>QNej6HU8)#{FBrn$NyHk5%#?bY+Y2A&MEYqF0i(bE!
zmN~ZYkY8K8{rCy<gJap{UmW{yPPfT4U0+(Lnj6u_q?^#i>MJ)@XkFA%XST+Xc=6`n
zQ1||IrEi&r8k&cXoo^eu<aGbL#P1$A<F!BZSZ<cgA81{@)O~8a`C3Ox|AU71e!UaU
zIKzyElh6KU!0BgyKY-sT=E~^@35;3O)Y{>5da^p4&pSIDjae2yyl(BXbhuod#euhS
zUVU~L+Bs<0*gq$}J>_NMA%p@y86%JIR&R<vQiGc}Ez_T}h5ZMP*+^#ZqV0jNZ3R$<
bYffzHUzL4kRpAJL#KW?0dUj6nm;B@Z=6CZ<

literal 0
HcmV?d00001

diff --git a/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/custom.bin b/graphannis/tests/data/sample-disk-based-3.8/current/nodes_diskmap_v1/custom.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a5cb3e4c9ae10ba26c822c00faa768e5fa9de3a2
GIT binary patch
literal 11732
zcmeI2O>^5e5J1VMY11~-$sxJql#Hi6MTtczA3k;B<YFIe*V6dViwB|LN)w6DkWgd)
z`%(gyUQy!i!f`U4l)i8YJ}yBBep-=;^T_@E;<rEh?Ll_Y8Q6T0T4)*R3$>Wq^Q^e%
znY(_QnylD@?fS!a{B~qgI74OGr1-v;W~#1qzDA7egr3b5oK~o*edd9m{5CTqeV(bA
zUQ1JKqwLjuI|bdHo2rNSQV)ExTu)bb(fLQ`#l&Q5SByObxvcrPb{zWs&~xSS&ZqIj
z>SC@^-BA~|HWF~cO+Kby{YA=;$>yrCoePsLiqCI_JF;dzsXkfZ4l`xoGk*L2@Z?Lx
z-}xaqI=P()`gnSLJe0&RKL(!}z_kYZDwo7?w}U@r(OES<Unhe9mnK!fJ(kD2C{9q^
zLvbIbXtP>2?e=Z8%uGHXo2mKx3ht|IX-C89@BcUxe(>fmt}?4G+r21iN8-OU3#xCd
zrcxIwhChe^Zb;4NTIKRRC<;F4+lioBP0egRouta@k*w){RkHdM%L4mH-=;5B*{Fkd
z;dUmLj-YlZ&8S0SJ60L+viSPvj~zG1&|7QhNcQ^&DuCP=`>w1D7R4h!rq?Ymip2+a
ztoxWc_+nXt?eN6G;Jo}zlLes3Q$A}kD0B8C;80)M?h`@Aq50SqB`mpJDJZ>iBJ5#L
z$n}NXOCk44$h{VFZ-m_5Z;|`kIzinWELKk`7P;>J9s=(M(P6ohkZuRTy^zfP5cp*X
z{3--~9Rj}zf%jNNC9I;79kHS+lvh-;BUDrg$`rK(C5rMKuP6`ligF;YD97=NavZNH
z$MK4C9Iq(H@rrUBuPDdyiYhs|qA<)W3e&uzFwI*B(Lpe;D3qC36oPq0A(&Sbf_X(D
zm{$~n_oxypq5PFBzutP3zdRxF*E&l2%cH!%0A&4DfsDT}NBRqcq`weI`U`QSzYs_I
z3vr~s5J&n8aiqTxNBRqKq`!a@`D+6v{k1`p{@S2PCDrL5nDkdYGwH87O!})1lm4p1
zq`&Gg>90CWmRKve&nl`fDzSVrZ!=y|eL;!kbMhrt1xgfEUU^xH@rv>wuP6udigFyU
zD97=NavZNH$MK4C9Iq(H@rrUBuc(rfD+<HBqA<-X3e&uG5FG^bib9!rMIo416oPq0
zA(&Sbf_X(DcyI0LSL1tqcnq4Zo_u|YJ$Sx0QdSjak*E4*E$}39xtgJkqo^GR_h73X
zhjtvkg*kA?Kc{V2VHNws&xrtDhrA!kqJtyg=;-hexNXr<m4%Do7Eex3#Np}wDX@(p
z^lo@8i_Z3e+aAFM&-VYi)$oD;49Qh;o{8Yf<tn(lzHPO%4i>$^JfL2A!fhTDbOf&r
z-K)_eXcsbbuSaF?IAs)cuzl<4W(Ix}`p{37a^A>RPr`(|=@us9lQxK=8x8_Y`*^Kt
z-&nJ|et4s(k3kfhEBj(7x~V11ag%ed6W14DRHxLwwqo%#ta3N3evL)etol)3Bnh+n
EUn)f5g8%>k

literal 0
HcmV?d00001


From 33f43116b6cf8a21ba6426f4f68859c9168298b6 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:29:34 +0200
Subject: [PATCH 07/16] Get popullatiion estimate from the component statistics

---
 .../src/annis/db/aql/operators/edge_op.rs     | 39 ++++++++++++++-----
 1 file changed, 30 insertions(+), 9 deletions(-)

diff --git a/graphannis/src/annis/db/aql/operators/edge_op.rs b/graphannis/src/annis/db/aql/operators/edge_op.rs
index dd9473cbb..175245a7c 100644
--- a/graphannis/src/annis/db/aql/operators/edge_op.rs
+++ b/graphannis/src/annis/db/aql/operators/edge_op.rs
@@ -41,26 +41,47 @@ impl BaseEdgeOp {
             let gs_for_component = db.get_graphstorage(c).ok_or_else(|| {
                 GraphAnnisError::ImpossibleSearch(format!("Component {} does not exist", &c))
             })?;
+
             gs.push(gs_for_component);
         }
+        let any_node_count = db.get_node_annos().guess_max_count(
+            Some(&NODE_TYPE_KEY.ns),
+            &NODE_TYPE_KEY.name,
+            "",
+            &char::MAX.to_string(),
+        )?;
         let all_part_of_components = spec
             .components
             .iter()
             .all(|c| c.get_type() == AnnotationComponentType::PartOf);
-        let node_type = if all_part_of_components {
-            "corpus"
+        // Use the single graph storage to get an estimate of population of nodes that can be found.
+        let max_nodes_estimate = if gs.len() == 1 {
+            gs[0]
+                .get_statistics()
+                .map(|s| s.nodes.saturating_sub(s.root_nodes))
+                .unwrap_or(any_node_count)
         } else {
-            "node"
+            // Use all nodes regardless of the component as population estimate
+            if all_part_of_components {
+                db.get_node_annos().guess_max_count(
+                    Some(&NODE_TYPE_KEY.ns),
+                    &NODE_TYPE_KEY.name,
+                    "corpus",
+                    "text",
+                )?
+            } else {
+                db.get_node_annos().guess_max_count(
+                    Some(&NODE_TYPE_KEY.ns),
+                    &NODE_TYPE_KEY.name,
+                    "node",
+                    "node",
+                )?
+            }
         };
         Ok(BaseEdgeOp {
             gs,
             spec,
-            max_nodes_estimate: db.get_node_annos().guess_max_count(
-                Some(&NODE_TYPE_KEY.ns),
-                &NODE_TYPE_KEY.name,
-                node_type,
-                node_type,
-            )?,
+            max_nodes_estimate,
             inverse: false,
         })
     }

From a79de1cccf4db21470ed820b22babcf3a69efbe6 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:41:27 +0200
Subject: [PATCH 08/16] Add changelog entry

---
 CHANGELOG.md | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7497f2ac4..57f086dbc 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,9 +10,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - New optional `file` option for the `[logging]` section in the webservice
 configuration. Can be used to additionally output all log messages to the given
 file.
+- Add number of root nodes to graph storage statistics. This changes the way
+most of the graph storages store their statistics. You can use old imported data
+files, but to make use of the new information you queries, you have to
+**reimport** your corpora.
 - `Graph:ensure_loaded_parallel` returns the actually loaded components that did
 exist.
-- Add number of root nodes to graph storage statistics.
 
 ### Fixed
 

From 29cdb024c81dab048b0b7215d5faf2dda102bce3 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:52:13 +0200
Subject: [PATCH 09/16] Only run verify workflow on pull requests

---
 .github/workflows/verify.yml | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/.github/workflows/verify.yml b/.github/workflows/verify.yml
index f3410a9ac..195ef6073 100644
--- a/.github/workflows/verify.yml
+++ b/.github/workflows/verify.yml
@@ -1,9 +1,6 @@
 name: Verify
 
 on:
-  push:
-    branches:
-      - main
   pull_request:
 
 jobs:
@@ -66,4 +63,3 @@ jobs:
             ```
             ${{env.COVERAGE_INFO}}
             ```
-      
\ No newline at end of file

From 41db5069f78b6cea8fa130229a817e3ade5afb79 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 17:53:48 +0200
Subject: [PATCH 10/16] No manual default value for serde needed

When an old statistics struct is loaded, it is derived from a different struct and then manually copied over
---
 core/src/graph/storage/mod.rs | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/core/src/graph/storage/mod.rs b/core/src/graph/storage/mod.rs
index 3bb627af8..52f0498fc 100644
--- a/core/src/graph/storage/mod.rs
+++ b/core/src/graph/storage/mod.rs
@@ -32,7 +32,6 @@ pub struct GraphStatistic {
     pub nodes: usize,
 
     /// Number of root nodes in this graph storage.
-    #[serde(default = "default_number_root_nodes")]
     pub root_nodes: usize,
 
     /// Average fan out.
@@ -52,10 +51,6 @@ pub struct GraphStatistic {
     pub dfs_visit_ratio: f64,
 }
 
-fn default_number_root_nodes() -> usize {
-    1
-}
-
 impl std::fmt::Display for GraphStatistic {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(

From e380fdd485d876d30fc71037b6fbd964f852a6fd Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Mon, 12 May 2025 18:04:11 +0200
Subject: [PATCH 11/16] Mark test data folder as generated

---
 .gitattributes | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/.gitattributes b/.gitattributes
index 3c507eb27..7500e2c01 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -1 +1,3 @@
-*.graphml text eol=lf
\ No newline at end of file
+*.graphml text eol=lf
+third-party-licenses.html linguist-generated
+/graphannis/tests/data/ linguist-generated=true

From 32be65405281e4557be84c3865dd2e635fb9d82d Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Tue, 13 May 2025 11:02:22 +0200
Subject: [PATCH 12/16] Documen fields of the cost estimate

---
 graphannis/src/annis/db/exec/mod.rs | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/graphannis/src/annis/db/exec/mod.rs b/graphannis/src/annis/db/exec/mod.rs
index eb02fb017..8965e29ed 100644
--- a/graphannis/src/annis/db/exec/mod.rs
+++ b/graphannis/src/annis/db/exec/mod.rs
@@ -12,8 +12,12 @@ use std::sync::Arc;
 
 #[derive(Debug, Clone)]
 pub struct CostEstimate {
+    /// The estimated number of tuples produces by this execution step.
     pub output: usize,
+    /// Sum of all processed tuples including the ones of the sub-steps.
     pub intermediate_sum: usize,
+    /// The estimated number of tuples that are processed in a join in this
+    /// execution step.
     pub processed_in_step: usize,
 }
 

From 1a6a11f2b9fcbaeca2f2eb2fb7ff3a78b42658d0 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Tue, 13 May 2025 13:09:48 +0200
Subject: [PATCH 13/16] assume all annotation can be matched in regex search
 without a prefix

---
 CHANGELOG.md                               |  2 ++
 core/src/annostorage/inmemory.rs           |  9 ++++++---
 core/src/annostorage/ondisk.rs             |  9 ++++++---
 graphannis/src/annis/db/aql/conjunction.rs |  1 +
 graphannis/src/annis/db/exec/mod.rs        |  9 ++++-----
 graphannis/src/annis/db/exec/nodesearch.rs | 10 +++++-----
 6 files changed, 24 insertions(+), 16 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index e904a8319..c722a843d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -27,6 +27,8 @@ exist.
 
 - Less frequent corpus cache status updates in log. Before, every corpus access
 could trigger an entry into the log which is not desired under heavy load.
+- Improve query execution planning by assuming all annotations can be matched in
+regular expressions without a prefix.
 
 ## [3.7.1] - 2025-04-14
 
diff --git a/core/src/annostorage/inmemory.rs b/core/src/annostorage/inmemory.rs
index 6d881061f..f75599652 100644
--- a/core/src/annostorage/inmemory.rs
+++ b/core/src/annostorage/inmemory.rs
@@ -752,6 +752,10 @@ where
     fn guess_max_count_regex(&self, ns: Option<&str>, name: &str, pattern: &str) -> Result<usize> {
         let full_match_pattern = util::regex_full_match(pattern);
 
+        // Get the total number of annotations with the namespace/name. We
+        // can't get larger than this number
+        let total = self.number_of_annotations_by_name(ns, name)?;
+
         // Try to parse the regular expression
         let parsed = regex_syntax::Parser::new().parse(&full_match_pattern);
         if let Ok(parsed) = parsed {
@@ -770,11 +774,10 @@ where
                         guessed_count += self.guess_max_count(ns, name, lower_val, &upper_val)?;
                     }
                 }
+            } else {
+                guessed_count = total;
             }
 
-            // Get the total number of annotations with the namespace/name. We
-            // can't get larger than this number
-            let total = self.number_of_annotations_by_name(ns, name)?;
             Ok(guessed_count.min(total))
         } else {
             Ok(0)
diff --git a/core/src/annostorage/ondisk.rs b/core/src/annostorage/ondisk.rs
index f947dce6d..140c00b09 100644
--- a/core/src/annostorage/ondisk.rs
+++ b/core/src/annostorage/ondisk.rs
@@ -895,6 +895,10 @@ where
     fn guess_max_count_regex(&self, ns: Option<&str>, name: &str, pattern: &str) -> Result<usize> {
         let full_match_pattern = util::regex_full_match(pattern);
 
+        // Get the total number of annotations with the namespace/name. We
+        // can't get larger than this number
+        let total = self.number_of_annotations_by_name(ns, name)?;
+
         // Try to parse the regular expression
         let parsed = Parser::new().parse(&full_match_pattern);
         if let Ok(parsed) = parsed {
@@ -913,11 +917,10 @@ where
                         guessed_count += self.guess_max_count(ns, name, lower_val, &upper_val)?;
                     }
                 }
+            } else {
+                guessed_count = total;
             }
 
-            // Get the total number of annotations with the namespace/name. We
-            // can't get larger than this number
-            let total = self.number_of_annotations_by_name(ns, name)?;
             Ok(guessed_count.min(total))
         } else {
             Ok(0)
diff --git a/graphannis/src/annis/db/aql/conjunction.rs b/graphannis/src/annis/db/aql/conjunction.rs
index d7adc63e7..5551f63c7 100644
--- a/graphannis/src/annis/db/aql/conjunction.rs
+++ b/graphannis/src/annis/db/aql/conjunction.rs
@@ -119,6 +119,7 @@ fn get_cost_estimates<'a>(
     }
 }
 
+/// Returns true if it is estimated to switch the operands in a join.
 fn should_switch_operand_order(
     op_spec: &BinaryOperatorSpecEntry,
     node2cost: &BTreeMap<usize, CostEstimate>,
diff --git a/graphannis/src/annis/db/exec/mod.rs b/graphannis/src/annis/db/exec/mod.rs
index 8965e29ed..fe27ea482 100644
--- a/graphannis/src/annis/db/exec/mod.rs
+++ b/graphannis/src/annis/db/exec/mod.rs
@@ -16,8 +16,7 @@ pub struct CostEstimate {
     pub output: usize,
     /// Sum of all processed tuples including the ones of the sub-steps.
     pub intermediate_sum: usize,
-    /// The estimated number of tuples that are processed in a join in this
-    /// execution step.
+    /// Simplistic estimated number of tuples that are processed in a join.
     pub processed_in_step: usize,
 }
 
@@ -58,13 +57,13 @@ impl ExecutionNodeDesc {
     pub fn empty_with_fragment(
         node_nr: usize,
         query_fragment: String,
-        est_size: Option<usize>,
+        estimated_output: usize,
     ) -> ExecutionNodeDesc {
         let mut node_pos = BTreeMap::new();
         node_pos.insert(node_nr, 0);
 
-        let cost = est_size.map(|output| CostEstimate {
-            output,
+        let cost = Some(CostEstimate {
+            output: estimated_output,
             intermediate_sum: 0,
             processed_in_step: 0,
         });
diff --git a/graphannis/src/annis/db/exec/nodesearch.rs b/graphannis/src/annis/db/exec/nodesearch.rs
index 1048c0538..206d3d7dc 100644
--- a/graphannis/src/annis/db/exec/nodesearch.rs
+++ b/graphannis/src/annis/db/exec/nodesearch.rs
@@ -569,7 +569,7 @@ impl<'a> NodeSearch<'a> {
                     desc: Some(ExecutionNodeDesc::empty_with_fragment(
                         common_args.node_nr,
                         common_args.query_fragment,
-                        Some(est_output),
+                        est_output,
                     )),
                     node_search_desc: Arc::new(NodeSearchDesc {
                         qname: (
@@ -676,7 +676,7 @@ impl<'a> NodeSearch<'a> {
             desc: Some(ExecutionNodeDesc::empty_with_fragment(
                 common_args.node_nr,
                 common_args.query_fragment.clone(),
-                Some(est_output),
+                est_output,
             )),
             node_search_desc: Arc::new(NodeSearchDesc {
                 qname: (qname.0, Some(qname.1)),
@@ -772,7 +772,7 @@ impl<'a> NodeSearch<'a> {
             desc: Some(ExecutionNodeDesc::empty_with_fragment(
                 common_args.node_nr,
                 common_args.query_fragment,
-                Some(est_output),
+                est_output,
             )),
             node_search_desc: Arc::new(NodeSearchDesc {
                 qname: (qname.0, Some(qname.1)),
@@ -920,7 +920,7 @@ impl<'a> NodeSearch<'a> {
             desc: Some(ExecutionNodeDesc::empty_with_fragment(
                 common_args.node_nr,
                 common_args.query_fragment.clone(),
-                Some(est_output),
+                est_output,
             )),
             node_search_desc: Arc::new(NodeSearchDesc {
                 qname: (
@@ -978,7 +978,7 @@ impl<'a> NodeSearch<'a> {
             desc: Some(ExecutionNodeDesc::empty_with_fragment(
                 common_args.node_nr,
                 common_args.query_fragment.clone(),
-                Some(est_output),
+                est_output,
             )),
             node_search_desc: Arc::new(NodeSearchDesc {
                 qname: (

From 540a8176635b14b21a4db79b890e80bcb6fcfa2a Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Tue, 13 May 2025 14:33:30 +0200
Subject: [PATCH 14/16] Only assume non-root nodes as population for PartOf
 searches, not any other base edge operator searches

---
 cli/src/bin/annis.rs                          |  4 +--
 .../src/annis/db/aql/operators/edge_op.rs     | 31 +++++++++----------
 2 files changed, 17 insertions(+), 18 deletions(-)

diff --git a/cli/src/bin/annis.rs b/cli/src/bin/annis.rs
index a055918e2..9205f43e2 100644
--- a/cli/src/bin/annis.rs
+++ b/cli/src/bin/annis.rs
@@ -321,7 +321,7 @@ impl AnnisRunner {
                 format = ExportFormat::GraphMLZip;
             } else if file_ext.to_string_lossy() == ".graphml" && self.current_corpus.len() != 1 {
                 bail!(
-                    r##"You need to select a *single* corpus first with the \"corpus\" command when exporting to a GraphML file. 
+                    r##"You need to select a *single* corpus first with the \"corpus\" command when exporting to a GraphML file.
                 To export multiple corpora, select a directory as output or a ZIP file (ending with .zip)"##
                 );
             }
@@ -442,7 +442,7 @@ impl AnnisRunner {
                 "unsorted" => ResultOrder::NotSorted,
                 _ => {
                     return Err(anyhow!(
-                        "Non-existing order with name {}. 
+                        "Non-existing order with name {}.
                         Must be one of \"normal\", \"inverted\", \"random\", \"unsorted\"",
                         args
                     ));
diff --git a/graphannis/src/annis/db/aql/operators/edge_op.rs b/graphannis/src/annis/db/aql/operators/edge_op.rs
index 175245a7c..c5722d6d5 100644
--- a/graphannis/src/annis/db/aql/operators/edge_op.rs
+++ b/graphannis/src/annis/db/aql/operators/edge_op.rs
@@ -54,29 +54,28 @@ impl BaseEdgeOp {
             .components
             .iter()
             .all(|c| c.get_type() == AnnotationComponentType::PartOf);
-        // Use the single graph storage to get an estimate of population of nodes that can be found.
-        let max_nodes_estimate = if gs.len() == 1 {
-            gs[0]
-                .get_statistics()
-                .map(|s| s.nodes.saturating_sub(s.root_nodes))
-                .unwrap_or(any_node_count)
-        } else {
-            // Use all nodes regardless of the component as population estimate
-            if all_part_of_components {
+        let max_nodes_estimate = if all_part_of_components {
+            // PartOf components have a very skewed distribution of root nodes
+            // vs. the actual possible targets, thus do not use all nodes as
+            // population but only the non-roots.
+            if gs.len() == 1 {
+                gs[0]
+                    .get_statistics()
+                    .map(|s| s.nodes.saturating_sub(s.root_nodes))
+                    .unwrap_or(any_node_count)
+            } else {
+                // If multiple PartOf graph storages are combined, we can guess
+                // the non-root nodes by estimating the number of nodes in the
+                // corpus grah.
                 db.get_node_annos().guess_max_count(
                     Some(&NODE_TYPE_KEY.ns),
                     &NODE_TYPE_KEY.name,
                     "corpus",
                     "text",
                 )?
-            } else {
-                db.get_node_annos().guess_max_count(
-                    Some(&NODE_TYPE_KEY.ns),
-                    &NODE_TYPE_KEY.name,
-                    "node",
-                    "node",
-                )?
             }
+        } else {
+            any_node_count
         };
         Ok(BaseEdgeOp {
             gs,

From 2f353b5f8fc41a617adcb81a59f14a51fdb3ddec Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Tue, 13 May 2025 15:17:15 +0200
Subject: [PATCH 15/16] Only estimate 5% of regex without prefix actually do
 match

---
 core/src/annostorage/inmemory.rs              |  6 ++-
 core/src/annostorage/ondisk.rs                |  5 ++-
 .../src/annis/db/aql/operators/edge_op.rs     | 40 ++++++++-----------
 3 files changed, 24 insertions(+), 27 deletions(-)

diff --git a/core/src/annostorage/inmemory.rs b/core/src/annostorage/inmemory.rs
index f75599652..7b8446a38 100644
--- a/core/src/annostorage/inmemory.rs
+++ b/core/src/annostorage/inmemory.rs
@@ -720,7 +720,6 @@ where
                 if let Some(anno_key) = self.anno_keys.get_symbol(&anno_key) {
                     if let Some(histo) = self.histogram_bounds.get(&anno_key) {
                         // find the range in which the value is contained
-
                         // we need to make sure the histogram is not empty -> should have at least two bounds
                         if histo.len() >= 2 {
                             sum_histogram_buckets += histo.len() - 1;
@@ -775,7 +774,10 @@ where
                     }
                 }
             } else {
-                guessed_count = total;
+                // For regular expressions without a prefix the worst case would be `.*[X].*` where `[X]` are the most common characters.
+                // Assume that a generic percentage (here 5%) of all nodes match the regex.
+                // TODO: find better ways of estimating this constant
+                guessed_count = (0.05 * (total as f64)) as usize;
             }
 
             Ok(guessed_count.min(total))
diff --git a/core/src/annostorage/ondisk.rs b/core/src/annostorage/ondisk.rs
index 140c00b09..069588830 100644
--- a/core/src/annostorage/ondisk.rs
+++ b/core/src/annostorage/ondisk.rs
@@ -918,7 +918,10 @@ where
                     }
                 }
             } else {
-                guessed_count = total;
+                // For regular expressions without a prefix the worst case would be `.*[X].*` where `[X]` are the most common characters.
+                // Assume that a generic percentage (here 5%) of all nodes match the regex.
+                // TODO: find better ways of estimating this constant
+                guessed_count = (0.05 * (total as f64)) as usize;
             }
 
             Ok(guessed_count.min(total))
diff --git a/graphannis/src/annis/db/aql/operators/edge_op.rs b/graphannis/src/annis/db/aql/operators/edge_op.rs
index c5722d6d5..6c5ab066b 100644
--- a/graphannis/src/annis/db/aql/operators/edge_op.rs
+++ b/graphannis/src/annis/db/aql/operators/edge_op.rs
@@ -44,12 +44,7 @@ impl BaseEdgeOp {
 
             gs.push(gs_for_component);
         }
-        let any_node_count = db.get_node_annos().guess_max_count(
-            Some(&NODE_TYPE_KEY.ns),
-            &NODE_TYPE_KEY.name,
-            "",
-            &char::MAX.to_string(),
-        )?;
+
         let all_part_of_components = spec
             .components
             .iter()
@@ -57,25 +52,22 @@ impl BaseEdgeOp {
         let max_nodes_estimate = if all_part_of_components {
             // PartOf components have a very skewed distribution of root nodes
             // vs. the actual possible targets, thus do not use all nodes as
-            // population but only the non-roots.
-            if gs.len() == 1 {
-                gs[0]
-                    .get_statistics()
-                    .map(|s| s.nodes.saturating_sub(s.root_nodes))
-                    .unwrap_or(any_node_count)
-            } else {
-                // If multiple PartOf graph storages are combined, we can guess
-                // the non-root nodes by estimating the number of nodes in the
-                // corpus grah.
-                db.get_node_annos().guess_max_count(
-                    Some(&NODE_TYPE_KEY.ns),
-                    &NODE_TYPE_KEY.name,
-                    "corpus",
-                    "text",
-                )?
-            }
+            // population but only the non-roots. We can guess the non-root
+            // nodes by estimating the number of nodes in the corpus grah.
+            let result = db.get_node_annos().guess_max_count(
+                Some(&NODE_TYPE_KEY.ns),
+                &NODE_TYPE_KEY.name,
+                "corpus",
+                "datasource",
+            )?;
+            result
         } else {
-            any_node_count
+            db.get_node_annos().guess_max_count(
+                Some(&NODE_TYPE_KEY.ns),
+                &NODE_TYPE_KEY.name,
+                "node",
+                "node",
+            )?
         };
         Ok(BaseEdgeOp {
             gs,

From 7966e8ddbecbbc610a55f8d1fc5e70d43869a4b1 Mon Sep 17 00:00:00 2001
From: Thomas Krause <thomas.krause@hu-berlin.de>
Date: Tue, 13 May 2025 15:41:21 +0200
Subject: [PATCH 16/16] Use more accurate root_node statistic

---
 .../src/annis/db/aql/operators/edge_op.rs     | 24 +++++++++++--------
 1 file changed, 14 insertions(+), 10 deletions(-)

diff --git a/graphannis/src/annis/db/aql/operators/edge_op.rs b/graphannis/src/annis/db/aql/operators/edge_op.rs
index 6c5ab066b..c22f204ac 100644
--- a/graphannis/src/annis/db/aql/operators/edge_op.rs
+++ b/graphannis/src/annis/db/aql/operators/edge_op.rs
@@ -49,18 +49,22 @@ impl BaseEdgeOp {
             .components
             .iter()
             .all(|c| c.get_type() == AnnotationComponentType::PartOf);
-        let max_nodes_estimate = if all_part_of_components {
+
+        let max_nodes_estimate = if all_part_of_components && gs.len() == 1 {
             // PartOf components have a very skewed distribution of root nodes
             // vs. the actual possible targets, thus do not use all nodes as
-            // population but only the non-roots. We can guess the non-root
-            // nodes by estimating the number of nodes in the corpus grah.
-            let result = db.get_node_annos().guess_max_count(
-                Some(&NODE_TYPE_KEY.ns),
-                &NODE_TYPE_KEY.name,
-                "corpus",
-                "datasource",
-            )?;
-            result
+            // population but only the non-roots.
+            if let Some(stats) = gs[0].get_statistics() {
+                stats.nodes - stats.root_nodes
+            } else {
+                // Fallback to guessing by using the node type
+                db.get_node_annos().guess_max_count(
+                    Some(&NODE_TYPE_KEY.ns),
+                    &NODE_TYPE_KEY.name,
+                    "corpus",
+                    "datasource",
+                )?
+            }
         } else {
             db.get_node_annos().guess_max_count(
                 Some(&NODE_TYPE_KEY.ns),