From 7de936fc5ebed67ff527bc53e2d72eaf3291bab5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?n=CE=B9c=D0=BDola=D1=95=20w=CE=B9lde?= <ncwilde43@gmail.com>
Date: Sun, 14 Jun 2026 12:18:21 -0700
Subject: [PATCH] feat(scripts): Add find_missing_sources.py and
 check-missing-sources Taskfile command

---
 Taskfile.yaml                   |  4 +++
 scripts/README.md               | 12 +++++++-
 scripts/find_missing_sources.py | 51 +++++++++++++++++++++++++++++++++
 3 files changed, 66 insertions(+), 1 deletion(-)
 create mode 100755 scripts/find_missing_sources.py

diff --git a/Taskfile.yaml b/Taskfile.yaml
index 5b313363..90e83c99 100644
--- a/Taskfile.yaml
+++ b/Taskfile.yaml
@@ -13,6 +13,10 @@ tasks:
     desc: Check recipes for missing iamges.
     cmds:
       - 'uv run scripts/check_missing_images.py'
+  check-missing-sources:
+    desc: Find recipes missing a source section.
+    cmds:
+      - 'uv run scripts/find_missing_sources.py'
   build:
     desc: Build the static site
     cmds:
diff --git a/scripts/README.md b/scripts/README.md
index 4b4842a9..4a884259 100644
--- a/scripts/README.md
+++ b/scripts/README.md
@@ -20,6 +20,7 @@ Where possible, run these via `Taskfile.yaml` using the `task` runner.
 | `scripts/scrape_to_cook.py` | - | Scrapes a recipe webpage URL and compiles it into a CookLang `.cook` file while downloading the image. |
 | `scripts/import_recipe_workflow.py` | - | Orchestrates the entire recipe import workflow (scrape, move, emojis, units, spellcheck, whitelist). |
 | `scripts/whitelist_typos.py` | - | Whitelists words in the typos dictionary, sorts the file, and rebuilds typos config. |
+| `scripts/find_missing_sources.py` | `task check-missing-sources` | Scans markdown recipes to find files that do not have a Source/Sources section. |
 
 ---
 
@@ -58,7 +59,10 @@ Where possible, run these via `Taskfile.yaml` using the `task` runner.
 #### [import_recipe_workflow.py](import_recipe_workflow.py)
 
 * **Usage**: `uv run scripts/import_recipe_workflow.py <URL_or_issue> [category]`
-* **Description**: Orchestrates the entire single recipe import pipeline: scrapes a recipe URL or extracts it from a GitHub issue, moves it to the appropriate category, runs emoji verification and auto-fixing, converts volumetric measurements to weights, and runs the spellchecker (with auto-whitelisting for proper nouns).
+* **Description**: Orchestrates the entire single recipe import pipeline: scrapes a recipe URL or
+  extracts it from a GitHub issue, moves it to the appropriate category, runs emoji verification
+  and auto-fixing, converts volumetric measurements to weights, and runs the spellchecker (with
+  auto-whitelisting for proper nouns).
 
 ---
 
@@ -147,6 +151,12 @@ Where possible, run these via `Taskfile.yaml` using the `task` runner.
 * **Description**: Queries GitHub Discussions via `gh api` to locate recipes with active comments,
   automatically setting `comments: true` in their front-matter.
 
+#### [find_missing_sources.py](find_missing_sources.py)
+
+* **Usage**: `uv run scripts/find_missing_sources.py` or `task check-missing-sources`
+* **Description**: Scans all markdown recipe files under `docs/` (excluding configuration, index, and reference files)
+  to identify those missing a Source/Sources heading.
+
 ---
 
 ### 4. Helper Libraries
diff --git a/scripts/find_missing_sources.py b/scripts/find_missing_sources.py
new file mode 100755
index 00000000..8756bb52
--- /dev/null
+++ b/scripts/find_missing_sources.py
@@ -0,0 +1,51 @@
+#!/usr/bin/env python3
+import os
+import re
+
+def main():
+    docs_dir = "docs"
+    exclude_files = {"index.md", "tags.md"}
+    exclude_dirs = {"reference", "assets"}
+    
+    # Regex to match Source or Sources with optional emoji prefix
+    # Matches: ## :link: Source, ## Source, ## :link: Sources, ## Sources
+    source_pattern = re.compile(r'^##\s+(?::[a-zA-Z0-9_-]+:\s+)?Sources?(?:\s+|$)', re.IGNORECASE)
+    
+    missing_files = []
+    total_checked = 0
+    
+    for root, dirs, files in os.walk(docs_dir):
+        # Exclude specific directories
+        dirs[:] = [d for d in dirs if d not in exclude_dirs]
+        
+        for file in files:
+            if not file.endswith(".md") or file in exclude_files:
+                continue
+                
+            file_path = os.path.join(root, file)
+            total_checked += 1
+            
+            has_source = False
+            try:
+                with open(file_path, "r", encoding="utf-8") as f:
+                    for line in f:
+                        if source_pattern.match(line.strip()):
+                            has_source = True
+                            break
+            except Exception as e:
+                print(f"Error reading {file_path}: {e}")
+                continue
+                
+            if not has_source:
+                missing_files.append(file_path)
+                
+    print(f"Checked {total_checked} markdown files.")
+    if missing_files:
+        print(f"Found {len(missing_files)} file(s) missing a Source section:")
+        for path in sorted(missing_files):
+            print(f"- [ ] {path}")
+    else:
+        print("All recipes contain a Source section!")
+
+if __name__ == "__main__":
+    main()