harvest, document reporting, sorted topic register

2 files changed, 152 insertions, 16 deletions

diff --git a/org/doc_reform.org b/org/doc_reform.org
index b13d086..b4bce52 100644
--- a/org/doc_reform.org
+++ b/org/doc_reform.org
@@ -124,17 +124,50 @@ void main(string[] args) {
     string[] _document_topic_register;
     string[] _topic_register;
     string[] _sub_topic_register;
+    string[][string][string][string][string] subject_trees;
     foreach(k, doc_harvest; harvests) {
       _topic_register = [];
       foreach(topic; doc_harvest.topic_register_arr.sort) {
         _sub_topic_register = [];
         string _spaces;
-        foreach (i, _top; topic.split(mkup.sep)) {
-          _sub_topic_register ~= format(
-            "  %s- %s",
-            "  ".repeat(i).join,
-            _top,
-          );
+        string[] subject_tree = topic.split(mkup.sep);
+        switch (subject_tree.length) {
+        case 1:
+          if (subject_tree[0] in subject_trees) {
+            subject_trees[subject_tree[0]]["_a"]["_a"]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]]["_a"]["_a"]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 2:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]]["_a"]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]]["_a"]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 3:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]
+          && subject_tree[2] in subject_trees[subject_tree[0]][subject_tree[1]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 4:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]
+          && subject_tree[2] in subject_trees[subject_tree[0]][subject_tree[1]]
+          && subject_tree[3] in subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]][subject_tree[3]] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]][subject_tree[3]] = [doc_harvest.uid];
+          }
+          break;
+        default:
+          break;
         }
         _topic_register ~= _sub_topic_register.join("\n");
       }
@@ -150,8 +183,43 @@ void main(string[] args) {
         (doc_harvest.date_published.length > 0) ? " (" ~ doc_harvest.date_published ~ ")" : "",
         _topic_register.sort!("toUpper(a) < toUpper(b)", SwapStrategy.stable).release.join("\n"),
       );
-      foreach(_dtr; _document_topic_register.sort) {
-        writeln(_dtr);
+    }
+    foreach(k0; subject_trees.keys.sort) {
+      if (k0 != "_a") {
+        writeln("", k0);
+        if ("_a" in subject_trees[k0]) {
+          foreach (uid; subject_trees[k0]["_a"]["_a"]["_a"].sort) {
+            writeln("- ", uid);
+          }
+        }
+      }
+      foreach(k1; subject_trees[k0].keys.sort) {
+        if (k1 != "_a") {
+          writeln("  ", k1);
+          if ("_a" in subject_trees[k0][k1]) {
+            foreach (uid; subject_trees[k0][k1]["_a"]["_a"].sort) {
+              writeln("  - ", uid);
+            }
+          }
+        }
+        foreach(k2; subject_trees[k0][k1].keys.sort) {
+          if (k2 != "_a") {
+            writeln("    ", k2);
+            if ("_a" in subject_trees[k0][k1][k2]) {
+              foreach (uid; subject_trees[k0][k1][k2]["_a"].sort) {
+                writeln("    - ", uid);
+              }
+            }
+          }
+          foreach(k3; subject_trees[k0][k1][k2].keys.sort) {
+            if (k3 != "_a") {
+              writeln("      ", k3);
+              foreach (uid; subject_trees[k0][k1][k2][k3]) {
+                writeln("      - ", uid);
+              }
+            }
+          }
+        }
       }
     }
   }
diff --git a/src/doc_reform/doc_reform.d b/src/doc_reform/doc_reform.d
index 663fb4d..9a17f2e 100755
--- a/src/doc_reform/doc_reform.d
+++ b/src/doc_reform/doc_reform.d
@@ -829,17 +829,50 @@ void main(string[] args) {
     string[] _document_topic_register;
     string[] _topic_register;
     string[] _sub_topic_register;
+    string[][string][string][string][string] subject_trees;
     foreach(k, doc_harvest; harvests) {
       _topic_register = [];
       foreach(topic; doc_harvest.topic_register_arr.sort) {
         _sub_topic_register = [];
         string _spaces;
-        foreach (i, _top; topic.split(mkup.sep)) {
-          _sub_topic_register ~= format(
-            "  %s- %s",
-            "  ".repeat(i).join,
-            _top,
-          );
+        string[] subject_tree = topic.split(mkup.sep);
+        switch (subject_tree.length) {
+        case 1:
+          if (subject_tree[0] in subject_trees) {
+            subject_trees[subject_tree[0]]["_a"]["_a"]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]]["_a"]["_a"]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 2:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]]["_a"]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]]["_a"]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 3:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]
+          && subject_tree[2] in subject_trees[subject_tree[0]][subject_tree[1]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]["_a"] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]["_a"] = [doc_harvest.uid];
+          }
+          break;
+        case 4:
+          if (subject_tree[0] in subject_trees
+          && subject_tree[1] in subject_trees[subject_tree[0]]
+          && subject_tree[2] in subject_trees[subject_tree[0]][subject_tree[1]]
+          && subject_tree[3] in subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]]) {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]][subject_tree[3]] ~= doc_harvest.uid;
+          } else {
+            subject_trees[subject_tree[0]][subject_tree[1]][subject_tree[2]][subject_tree[3]] = [doc_harvest.uid];
+          }
+          break;
+        default:
+          break;
         }
         _topic_register ~= _sub_topic_register.join("\n");
       }
@@ -855,8 +888,43 @@ void main(string[] args) {
         (doc_harvest.date_published.length > 0) ? " (" ~ doc_harvest.date_published ~ ")" : "",
         _topic_register.sort!("toUpper(a) < toUpper(b)", SwapStrategy.stable).release.join("\n"),
       );
-      foreach(_dtr; _document_topic_register.sort) {
-        writeln(_dtr);
+    }
+    foreach(k0; subject_trees.keys.sort) {
+      if (k0 != "_a") {
+        writeln("", k0);
+        if ("_a" in subject_trees[k0]) {
+          foreach (uid; subject_trees[k0]["_a"]["_a"]["_a"].sort) {
+            writeln("- ", uid);
+          }
+        }
+      }
+      foreach(k1; subject_trees[k0].keys.sort) {
+        if (k1 != "_a") {
+          writeln("  ", k1);
+          if ("_a" in subject_trees[k0][k1]) {
+            foreach (uid; subject_trees[k0][k1]["_a"]["_a"].sort) {
+              writeln("  - ", uid);
+            }
+          }
+        }
+        foreach(k2; subject_trees[k0][k1].keys.sort) {
+          if (k2 != "_a") {
+            writeln("    ", k2);
+            if ("_a" in subject_trees[k0][k1][k2]) {
+              foreach (uid; subject_trees[k0][k1][k2]["_a"].sort) {
+                writeln("    - ", uid);
+              }
+            }
+          }
+          foreach(k3; subject_trees[k0][k1][k2].keys.sort) {
+            if (k3 != "_a") {
+              writeln("      ", k3);
+              foreach (uid; subject_trees[k0][k1][k2][k3]) {
+                writeln("      - ", uid);
+              }
+            }
+          }
+        }
       }
     }
   }