From e604b131604179aaa7ddc9398c5a4def35a3b944 Mon Sep 17 00:00:00 2001
From: Yu Cong <sxlxcsxlxc@gmail.com>
Date: Mon, 13 Oct 2025 00:23:29 +0800
Subject: [PATCH] yay

---
 filter.lua  | 82 +++++++++++++++++++++++++----------------------------
 input.md    |  4 +--
 output.html | 29 ++++++++++++++++---
 3 files changed, 65 insertions(+), 50 deletions(-)

diff --git a/filter.lua b/filter.lua
index 4abbaa7..ea255a0 100644
--- a/filter.lua
+++ b/filter.lua
@@ -4,8 +4,8 @@ local function show(s)
 end
 
 ------------------------------------------------
-local label_map = {} -- label has to be unique!
-local adj = {}       -- graph
+local adj = {} -- graph
+local include_nodes = {}
 local function add_edge(u, v)
     if not adj[u] then
         adj[u] = {}
@@ -14,7 +14,7 @@ local function add_edge(u, v)
 end
 
 ---------------------------------------------------------
--- Topological sort (DFS-based, with cycle detection)
+-- Topological sort (DFS-based, with cycle detection) -- chatgpt
 ---------------------------------------------------------
 local visited = {}  -- "perm" mark: node completely processed
 local in_stack = {} -- "temp" mark: node currently in recursion stack
@@ -49,28 +49,21 @@ local function _dfs(u, path)
 end
 
 local function topo_sort()
+    local rank = {}
     for u, _ in pairs(adj) do
         if not visited[u] then
             _dfs(u, {})
-            if cycle then return nil, cycle end
+            if cycle then return rank, cycle end
         end
     end
 
     -- build rank map
-    local rank = {}
     for i, u in ipairs(order) do
         rank[u] = i
     end
     return rank
 end
 
-local function collect_labels(blk)
-    if blk.identifier and blk.identifier ~= "" then
-        label_map[blk.identifier] = blk:clone()
-    end
-    return nil
-end
-
 local function words(s)
     local res = {}
     for part in s:gmatch("[^,]+") do               -- split by commas
@@ -83,12 +76,12 @@ local function words(s)
 end
 
 local function dfs(blk, stack) -- depth first search on a top level blk
-    -- check for identifier and build label_map
     -- look for 2 types of AST node: divs with include attr and divs with labels
     local labelled = false
     local include = false
     if blk.attributes and blk.attributes["include"] then -- this must be a leaf node
         include = true
+        table.insert(include_nodes, blk.identifier)
         -- labels in include may appears later in the dfs than this include-node
         -- but we assume every label will be there and build the graph now
         -- This is a directed bipartite grpah.
@@ -107,7 +100,6 @@ local function dfs(blk, stack) -- depth first search on a top level blk
     elseif blk.identifier and blk.identifier ~= "" then
         -- collect labelled nodes & maintain the stack
         labelled = true
-        label_map[blk.identifier] = blk:clone()
         table.insert(stack, blk.identifier)
     end
 
@@ -125,22 +117,31 @@ local function dfs(blk, stack) -- depth first search on a top level blk
     end
 end
 
-local function replace(e)
-    local include = e.attributes["include"]
-    if include then
-        local blocks = {}
-        -- Replace the Div contents
-        for _, l in ipairs(words(include)) do
-            if label_map[l] then
-                table.insert(blocks, label_map[l]:clone())
-                show("insert [" .. label_map[l].identifier .. ']\n')
-            else
-                io.stderr:write("Cannot find AST node with label " .. l .. "\n")
+-- read doc and return the concatenation of target nodes for an include-node
+-- Things become messy here since the include list is ordered.
+-- one has to again transverse the AST to collect labelled nodes and then
+-- do the concatenation...
+local function collect_node(doc, l)
+    local blocks = {}
+    local targets = words(l.attributes["include"])
+    local nodes = {}
+    doc:walk {
+        Div = function(div)
+            if div.identifier and div.identifier ~= "" then
+                nodes[div.identifier] = div:clone()
             end
+            return nil
+        end
+    }
+    for _, t in ipairs(targets) do
+        if nodes[t] then
+            table.insert(blocks, nodes[t]:clone())
+            -- show("insert [" .. nodes[t].identifier .. ']\n')
+        else
+            io.stderr:write("Cannot find AST node with label " .. t .. "\n")
         end
-        return pandoc.Div(blocks, e.attr)
     end
-    return e -- no change
+    return pandoc.Div(blocks, l.attr)
 end
 
 return {
@@ -151,30 +152,23 @@ return {
             dfs(blk, {})
         end
 
-        -- sanity check
-        show("edges:")
-        for u, vs in pairs(adj) do
-            for _, v in ipairs(vs) do
-                show(u .. "->" .. v)
-            end
-        end
-
         -- topological sort
         local rank, cycle = topo_sort()
         if cycle then
             error("Cycle detected:" .. table.concat(cycle, " -> ") .. "\n")
         end
 
-        -- sanity check
-        show("ranks:")
-        for k, v in pairs(rank) do
-            show(k .. "--" .. v)
-        end
-
         -- replace
-        -- for i, blk in ipairs(doc.blocks) do
-        --     doc.blocks[i] = replace(blk)
-        -- end
+        table.sort(include_nodes, function(x, y)
+            return rank[x] > rank[y]
+        end)
+        for _, v in ipairs(include_nodes) do
+            doc = doc:walk { Div = function(div)
+                if div.identifier and div.identifier == v then
+                    return collect_node(doc, div)
+                end
+            end }
+        end
         return doc
     end
 }
diff --git a/input.md b/input.md
index def3579..c329951 100644
--- a/input.md
+++ b/input.md
@@ -1,6 +1,6 @@
 
 ::: {#structure}
-inc (need thm2 and a fake label)
+inc (need thm2, thm1 and a fake label)
 
 thm1 - inthm1
 
@@ -8,7 +8,7 @@ thm2 - inthm2 (need thm1)
 
 ::::::::::::::::
 
-:::{include="thm2 , fakelabel" #inc}
+:::{include="thm2, thm1 , fakelabel" #inc}
 This line will be ignored
 :::::::::::::::::::::::::::::::::::::::::::::::::::::::
 
diff --git a/output.html b/output.html
index 308a0a0..102d986 100644
--- a/output.html
+++ b/output.html
@@ -1,10 +1,26 @@
 <div id="structure">
-<p>inc (need thm2 and a fake label)</p>
+<p>inc (need thm2, thm1 and a fake label)</p>
 <p>thm1 - inthm1</p>
 <p>thm2 - inthm2 (need thm1)</p>
 </div>
-<div id="inc" data-include="thm2 , fakelabel">
-<p>This line will be ignored</p>
+<div id="inc" data-include="thm2, thm1 , fakelabel">
+<div id="thm2" class="Theorem">
+<p>test thm2</p>
+<div id="inthm2" data-include="thm1">
+<div id="thm1" class="Theorem">
+<p>test thm1</p>
+<div id="inthm1" class="Definition">
+<p>something</p>
+</div>
+</div>
+</div>
+</div>
+<div id="thm1" class="Theorem">
+<p>test thm1</p>
+<div id="inthm1" class="Definition">
+<p>something</p>
+</div>
+</div>
 </div>
 <div id="thm1" class="Theorem">
 <p>test thm1</p>
@@ -15,7 +31,12 @@
 <div id="thm2" class="Theorem">
 <p>test thm2</p>
 <div id="inthm2" data-include="thm1">
-<p>i need theorem 1</p>
+<div id="thm1" class="Theorem">
+<p>test thm1</p>
+<div id="inthm1" class="Definition">
+<p>something</p>
+</div>
+</div>
 </div>
 </div>
 <div id="thm3" class="Theorem">